BIOT 1

In silico design of integrated downstream processes for non-mAb biologics produced in complex expression systems

Nicholas Vecchiarello2, [email protected], Chaz Goodwine3, Steven M. Cramer1. (1) Ricketts Bldg, Rensselaer Polytechnic Inst, Troy, New York, United States (2) Rensselaer Polytechnic Institute, Troy, New York, United States (3) Chemical and Biological Engineering, Rensselaer Polytechnic Institute, Troy, New York, United States

For biomolecules not amenable to affinity capture, determining optimal downstream processes can be a major challenge due to the broad range of experimental design space available with regard to resins, buffer pH, and salt. Previous work in our group has addressed this issue for products produced in Pichia pastoris by chromatographically characterizing process-related impurities found in spent cultivation fluids on a set of multimodal, HCIC, and ion exchange resins for a wide range of conditions using UP-RPLC. The collected data was used in concert with product retention data and a custom in silico tool to generate and rank integrated downstream processes for three non-monoclonal antibody products expressed in P. pastoris. In this work, we have chromatographically characterized the host-related impurity behavior for a CHO cell line containing a more significant HCP burden. In an effort to expedite process development, a method for assessing global orthogonal selectivity for different process-related impurities between resin pairs was employed in order to prune the number of resins required for screening. This strategy was then applied to the purification of a product with a significant variant challenge. To further reduce the number of resins to screen for product retention, sequence understanding and biophysical properties were utilized to focus on regions that would likely remove host- related impurities and offer variant selectivity. Product and variant retention data from these focused screens were used as initial inputs for the in silico tool and a fully inclusive list of integrated downstream processes was generated and ranked based on overall purification sequence orthogonality. Top process outputs were then subjected to experimental process development and refinement. This approach represents a strategy for rapidly designing integrated downstream processes in an efficient manner for non-mAb biologics produced in complex host expression systems.

BIOT 2

Streamlining early DSP development through evolving integrated mechanistic models

Alexander T. Hanke, [email protected], Rushd Khalaf, Lars W. Pampel. BTDM, Novartis Pharma AG, Basel, Switzerland

Efficient development of chromatographic purification processes requires knowledge on how the choice of resin, buffers and operational parameters influence the behavior of both the product and its impurities. High-throughput systems capable of performing batch adsorption and parallel micro-column experiments have become staples of most process development labs due to their ability to rapidly explore this large potential process space and identify promising conditions.

Even with such technologies in place it remains a challenging task for developers to find the right combination of operations to robustly and efficiently meet quality expectations. Due to the non-linear nature of preparative chromatography and changes is the upstream process during the project lifecycle, large amounts of the data generated during early process screenings are rendered obsolete once the process intermediate they were generated with is no longer representative of what the step will be challenged with in the final process.

We present how predictive adsorption models can reduce this waste by narrowing screening spaces to specific windows of interest tailored to individual molecules. First experimental results generated within these windows serve to correct first prediction errors and can be fed directly into mechanistic models integrating all potential operation sequences. As these models are refined during development they allow to assess the impact of and intermediate composition changes and non-linearity effects, effectively reducing the experimental and analytical burden of conventional sequential process development.

BIOT 3

Using a Bayesian framework to account for scale-down model offsets during process characterization

Matthew Stork1, [email protected], Aili Cheng2, Brad Evans2, Peter Slade1, Erwin Yu1. (1) Bioprocess Research and Development, Pfizer, Inc, Andover, Massachusetts, United States (2) Non Clinical Statistics, Pfizer, Andover, Massachusetts, United States

One of the primary challenges during process characterization is establishing the suitability of the small scales models used to generate the data. The traditional approach to qualification of scale-down models has been to verify that lab-scale results match large-scale results to within an arbitrary precision. An alternate approach is to directly incorporate scaling offsets into predictive models. However, obtaining a precise estimate of scaling offsets can be difficult when large-scale datasets are relatively small. As such, it is essential to use a statistical method that can account for this uncertainty. This talk will explore the use of Bayesian modeling to account for scaling offsets in predictive models. Bayesian methods are well-suited to this task because they inherently account for uncertainty, while offering a flexible platform for the implementation of hierarchal models. This talk will explore the use of Bayesian statistics for both upstream and downstream applications. In a downstream application, Bayesian statistics were used to account for scale-offsets in a high-throughput chromatography system using miniature chromatography columns. In the past, concerns about the scalability of high-throughput methods have been a barrier to using high-throughput data in process characterization models. Using Bayesian techniques, chromatography data from high-throughput and traditional lab-scale DOE studies were combined with manufacturing-scale data to generate models accounting for scaling-offsets. In an upstream application, a hierarchal Bayesian model was used to generate a predictive model to account for scaling offsets in bioreactor performance parameters. Bioreactor scale-up is not a simple linear process, and a hierarchal Bayesian approach enables an assessment of how the effects of input process parameters may differ across scales. Overall, the Bayesian approach is a flexible method that facilitates more informed predictions of large-scale performance by enabling the incorporation of scaling offsets and prior knowledge into predictive models.

BIOT 4

Mechanistic modeling analysis of chromatography scale-down models

Steven Benner, [email protected], John Welsh, Michael Rauscher, Jennifer Pollard. Merck, Kenilworth, New Jersey, United States

Chromatography has been a cornerstone of downstream process development (PD) for years, and there is an ever increasing demand for improved speed and efficiency. Scale-down models are used in process development to optimize operating conditions and study process robustness while expending as little time and material as possible. The advent of automated liquid handling systems and miniature columns has taken the efficiency of process development to another level by allowing up to eight column runs in parallel with column volumes under 1 ml. However results between these miniature columns and typical lab scale columns can deviate, resulting in the need for a better understanding of the differences between columns and systems. Mechanistic models can be used to understand the physics of the process (fluid flow, mass transfer, etc.) as a function of scale. We have used mechanistic modeling to study the factors leading to differences in pool sizes observed between scales, and to make predictions on lab scale pool sizes from miniature column data. Results indicate that changes in mass transfer parameters, specifically axial dispersion and film mass transfer rates, between scales leads to the observed differences in pool size. Additionally, we have studied the effect of system differences between automated liquid handling systems and conventional preparative chromatography systems on process performance. This work provides new insight on the ability of high-throughput process development to be used for scale down modeling and process characterization.

BIOT 5

In-silico model formulation, calibration and application for commercial CEX chromatography

Christian Kunert3, [email protected], Fabrice Schlegel1, Karin Westerberg3, Oliver Kaltenbrunner2, Pablo Rolandi4, Xiaoxiang Zhu5. (1) Process Development, Amgen, Cambridge, Massachusetts, United States (2) Amgen Inc, Thousand Oaks, California, United States (3) Amgen, Cambridge, Massachusetts, United States (4) PD, Amgen, Cambridge, Massachusetts, United States

Cation exchange chromatography (CEX) is an important step in the purification of some biologics such as monoclonal antibodies (mAb). Commercial Process Development and Process Characterization require a large set of experiments to establish desirable operating conditions. Lately, the biopharma industry has shifted toward developing mechanistic numerical models that can be calibrated with a limited set of experimental runs and are able to accurately predict process conditions. These mechanistic models enable the characterization of process robustness and help establishing a design space that ensures product quality by the means of high performance computing.

In this talk, we present the development of a highly predictive model for a mAb CEX process. The model uses a novel pH dependent multi-state steric mass action isotherm, calibrated with a limited set of targeted bench scale experiments. The model is validated against a larger set of bench scale experiments. Finally, we apply the model to the commercial scale, the results of which can be used for regulatory filing.

BIOT 6

A scale-down methodology using mini-columns to predict chromatography performance in development and production scale columns

Steven T. Evans1, [email protected], Maxwell Nagarajan2, Tedrick Lew3, Claude Lupis2, Gisela Ferreira1, Guillermo Miro-Quesada1, David Robbins1. (1) MedImmune, Clarksburg, Maryland, United States (2) Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts, United States

An evaluation of the suitability of high throughput mini-columns as a sufficiently predictive scale-down model for process characterization for regulatory submissions was performed. While the mini-columns have been widely used for screening studies in early stage development, the operating conditions with these columns can be quite different from the process scale conditions. This may include differences in bed height, flow distribution, extra column effects, wall effects, porosity, and packing differences. This can present challenges to their use for process characterization which requires a predictive scale down model for an intensive analysis of an existing process to determine the multidimensional combination and interaction of process parameters on product quality and process performance. When an experimental scale-down model is used to justify control strategy and parameter criticality assessment and ranges, regulatory agencies expect the applicant to demonstrate the relevance of the characterization studies to the proposed production scale process. This work applied an engineering understanding, through theoretical analysis and experimental evaluations, to assess the suitability of these systems for this purpose. A cation exchange chromatography purification of a monoclonal antibody was used as a case study. Studies were performed using the high throughput mini-columns as well as conventional bench scale systems that are more typically used for process characterization. Steric Mass Action model parameters were obtained for both systems. Statistical analysis of the model simulations using these parameters and experimental results identified that for both systems, the same process parameters were identified as having a quantifiable impact on the process outputs. The work improved our capability to make appropriate predictions from chromatography purifications performed in a small column to the performance of a large column (for yield, column volume, and aggregate clearance).

BIOT 7

Model-based quality by design: Challenges, pitfalls and best practices

Thiemo Huuk1, [email protected], Maria Casals-Peralvarez1, Tobias Hahn1, Teresa C. Beck1, Juergen Hubbuch2. (1) GoSilico GmbH, Karlsruhe, XX, Germany (2) KIT, Karlsruhe, Germany

The quality-by-design (QbD) initiative of the regulatory authorities grants biopharmaceutical manufacturers a higher flexibility once a design space has been defined based on an extensive process understanding. In addition, this process understanding opens up the path to a much more rational way of process development and production facility operation. Two possibilities to create process understanding are high-resolution screening experiments based on Design-of-Experiments (DoE) and mechanistic modeling. The first requires a high sample amount, relies on the user’s expertise in preselecting relevant parameters, and is finally restricted to the calibrated experimental range. Mechanistic modeling relies on a much smaller set of experimental data and on sophisticated mathematical methods, allowing to extrapolate to processes outside the calibrated range, and to mathematically describe the relation of critical process parameters (CPP) and critical quality attributes (CQA)

This contribution discusses the challenges, pitfalls, and best practices in using mechanistic modeling to fulfil the latest QbD requirements. Conventional experimental and DoE-based QbD studies start with the preselection of the most relevant process parameters affecting process performance and product quality. In contrast, the model-based QbD presented in this work, does not need a preselection of parameters, eliminating a time consuming and uncertain decision process from the workflow. Next, a mechanistic chromatography model is calibrated based on 3-4 experiments, and the quality of the model parameters is assessed by means of parameter confidence intervals and parameter correlations. Subsequently, the model parameters uncertainties and the typically observed variabilities of the process parameters are used to simulate a virtual sensitivity analysis. The probability of occurrence and the impact of an out of specification run resulting from this analysis are used to compute a CQA impact score and to cluster the process parameters into non-relevant, key, and critical parameters. Finally, a fine Latin Hypercube sampling is carried out for the key and critical parameters with respect to the key performance indicators and quality attributes. This high dimensional data set is used to define a process design space.

This talk will present the technological workflow of model-based QbD by means of an industrial study of purifying an antibody from a mixture containing multiple product related impurities.

BIOT 8

Artificial intelligence in protein chromatography: Root cause investigation of process deviations

Gang Wang1, [email protected], Till Briskot1, Tobias Hahn2, Pascal Baumann1, Juergen Hubbuch1. (1) Biomolecular Separation Engineering, Karlsruhe Institute of Technology, Karlsruhe, Germany (2) GoSilico GmbH, Karlsruhe, Germany

In protein chromatography, process variations, such as aging of column or process errors, can result in deviations of the product and impurity levels. Consequently, the process performance described by purity, yield, or production rate may decrease. Based on visual inspection of the UV signal, it is hard to identify the source of the error and almost unfeasible to determine the quantity of deviation. The problem becomes even more pronounced, if multiple root causes of the deviation are interconnected and lead to an observable deviation.

In the presented work, a novel method based on the combination of mechanistic chromatography models and the artificial neural networks is suggested to solve this problem. In a case study using a model protein mixture, the determination of deviations in column capacity and elution gradient length was shown. Maximal errors of 1.5% and 4.90% for the prediction of deviation in column capacity and elution gradient length respectively demonstrated the capability of this method for root cause investigation.

BIOT 9

Designing biology for health and the environment

Pamela Silver, [email protected]. Harvard University, Boston, Massachusetts, United States

The engineering of Biology presents vast opportunities for new therapeutic design, prevention of disease, new chemistry and solutions to environmental problems. We use what we know from Nature to engineer systems with predictable behaviors. We also seek to discover new natural strategies to then re-engineer. Here, I will present concepts and experiments that address how we and our collaborators approach these problems in a systematic way. In one instance, we engineer cell compartments to create nanofactories for chemical production, new materials and environmental sensors. Secondly, we have co-developed the ‘bionic leaf’ to channel sunlight via electrocatalysts to commodity producing bacteria. In doing so, we have discovered a wide spread strategy by which all prokaryotes sequester chemical reactions to protect from toxic intermediates. These results have far-reaching implications for cell-based manufacturing and sustainability.

BIOT 10

Process analytical technology for flow synthesis of antisense oligonucleotides: Real-time starting material assurance and ASO sequence confirmation

Xiaoye Su, [email protected], Jim Yang, Austen Ng, Elliott Schmitt, Jessica Stolee. Biogen, Cambridge, Massachusetts, United States

Background Antisense oligonucleotide (ASO) synthesis is conducted on a solid-phase support by flowing phosphoramidites (ASO building blocks) one unit at a time to build the chain of nucleotides which make up the ASO. ASO sequences in Biogen’s pipeline contain up to eight different phosphoramidites. The crude intermediate (after synthesis) is sampled and tested off-line by high resolution mass spectrometry to confirm that the correct sequence was made.

Goal/Opportunity: Develop a robust in-line Process Analytical Technology (PAT) method to identify incoming phosphoramidites during oligonucleotide synthesis in order to obtain real-time ASO sequencing. This would eliminate the need for a laborious sequencing release test and can be used to identify synthesis errors in real time thus reducing downstream efforts and the associated costs.

Methods: A principal component analysis (PCA) model based on different types of phosphoramidites from different lots and vendors was established via off-line mid- infrared measurements. A high-pressure attenuated total reflection (ATR) flow cell was connected to the ASO synthesizer to collect real-time spectra as a proof-of-concept study. However, this flow cell created extra volume that may impact the process. Therefore, a customized transmission flow cell combined with fiber optics has also been evaluated.

Results: The off-line PCA model was optimized via a series of data processing and key parameter optimizations. All eight phosphoramidites were uniquely identified by this model. Both high-pressure ATR and transmission flow cells were shown to be amenable for in-line implementation with the synthesizer. The phosphoramidites spectra were collected in real-time and unknown amidites were successfully identified.

Conclusions: Real-time phosphoramidites identification and ASO sequence confirmation was achieved via an in-line mid-IR PAT method. These results provide evidence that PAT can be used in ASO manufacturing. Furthermore, implementation of this method removes the need for the more complex sequencing release test which provides data after the downstream process has already been completed.

BIOT 11

Paving the road towards real-time release testing

Gunnar Malmquist, [email protected]. R&D, GE Healthcare, Uppsala, Sweden

The first edition of the Biomanufacturing Technology Roadmap developed by the BioPhorum Operations Group (BPOG) was released in July 2017 and put a ten-year perspective on the biopharmaceutical industry. One area where significant changes can be predicted is process control and release testing. The industry wants to move testing from the QC laboratory (off-line analysis) to the manufacturing floor (in-line, on-line or at-line monitoring) as off-line analysis in the QC laboratory increases the total cost to supply and time to release the product. Another benefit of “timely measurements” is the ability to implement new feedback or feed forward control strategies. However, significant investment is required to develop new in-line methods or convert off-line methods to rapid at-line methods and guidance is needed to determine where to focus the development effort.

The BPOG Technology Roadmap In-line-monitoring/Real-Time Release project is a cross company forum for collaboration that gathers end-users and supply partners. The project team has been working through the monoclonal antibody manufacturing process to evaluate the current set of release tests and in process controls by focusing on the critical control points for each attribute in the process. Current testing regimes are compared to a desired future state to establish the Critical Quality Attributes and in- process controls that should be prioritized for development of new methods, while recognizing that real-time release does not always predicate real time testing (in-time release is probably a more appropriate term).

The next step will entail development of User Requirement Specifications (URS) for the prioritized attributes that will enable innovators to focus on areas relevant to the industry. The project team seeks industry wide feedback and discussion regarding the prioritized list to ensure broad applicability across the biopharmaceutical industry. The successful outcome of the project is a consolidated URS list that will initiate external innovation and development of tools that expedites the journey towards real-time release testing.

BIOT 12

Integrating metabolomics and process analytical technology (PAT) for rapid monitoring of protein glycosylation in a CHO bioprocess Elaheh Kamaloo1, [email protected], Eric Hodgman1, Amanda Lewis1, Jay West2, Deborah Mohammed2, Barry Drew2, Nelly Aranibar3, Bethanne M. Warrack3, Michael D. Reily3, Eric Garr1. (1) Manufacturing Sciences & Technology, Bristol-Myers Squibb, Devens, Massachusetts, United States (2) Process Development, Bristol-Myers Squibb, Devens, Massachusetts, United States (3) Research & Discovery, Bristol-Myers Squibb, Princeton, New Jersey, United States

Here we present a strategy to leverage metabolomics analysis as a means to identify critical quality attribute biomarkers for a CHO Fc-fusion protein production process. This strategy has a direct impact on monitoring and control of a critical quality attribute which is linked with batch harvest time. We used untargeted metabolomics to characterize more than 150 metabolites in the cell culture supernatant across multiple time points. Among all the metabolites, two showed strong correlations with the amount of a specific glycosylation attribute and had the structural characteristics required for separation and quantification by ultra-high pressure liquid chromatography (UHPLC). After identifying these two biomarkers, we developed a fast assay to quantify them via UHPLC combined with mass spectrometric detection. While the current process assay for analysis of the CQA takes more than 12 hours, our developed rapid assay takes only fifteen minutes. In addition, it does not require any pre-purification and it has the capability of being implemented as an online assay in manufacturing with commercially available autosamplers. The assay was utilized to characterize numerous cell culture samples from lab scale, pilot scale, and manufacturing scale bioreactors. Using this sample set, we developed a computer model to combine the concentrations of the two metabolites to predict the amount of glycosylation attribute of the product. Implementation of this method will allow for online monitoring of a glycosylation CQA known to approach the specification limit as culture duration progresses, therefore maximizing culture duration and product yield.

BIOT 13

Late-phase process improvement: Methods to increase titer while maintaining product quality

Michael Bennett, [email protected], Matthew Zustiak. Pharma Services Group, Thermo Fisher Scientific, St. Louis, Missouri, United States

During the early phase development, a phase-appropriate approach often focuses on development speed rather than on process performance. As a program heads toward phase 3/registration, in addition to the process performance qualification exercise, companies may seek opportunities to improve the process economics and increase supply without changing production scale or the host cell line. In this case study, a process improvement project was undertaken starting with media and feed screening with a goal of increasing the titer by up to two fold while maintaining comparable product quality. The ambr®15 microbioreactor was used as a scale down model to execute the initial media and feed screens. The top performing media and feed combination that maintained the desired product profile was then taken into a second round of ambr screens for fed-batch process development. A methodology for modulating the product profile was identified during the second round of ambr screens, in which optimization of the ratio of multiple commercially available basal media resulted in the ability to either increase or decrease the overall percentage of a key product quality attribute in the molecule’s product profile while maintaining high titer. Further experiments were conducted at the 10L scale to demonstrate successful scale up, increase robustness and determine the process parameter ranges for the improved media process. This improved process was then confirmed at the 10L scale and executed at the 2000L.

BIOT 14

Differential expression and activity of metabolic enzymes in cell culture fluid correlates with increased disulfide reduction of mAb products

Anthony J. Cura2, [email protected], Srinivas Chollangi2, Kathleen McWade2, Joon Chong Yee2, Michael Peck2, Tyler Hageman1, Yunping Huang1, Xuankuo Xu2, Sanchayita Ghose2. (1) Bristol Myers Squibb, Pennington, New Jersey, United States (2) Process Development, Bristol-Myers Squbb, Devens, Massachusetts, United States

With the increased demand for immuno-oncology pharmaceuticals to target a wide range of diseases, modern development efforts are heavily focused on maximizing monoclonal antibody (mAb) product yield while controlling process-related impurities. One of the unanticipated side-effects of higher host-cell density, titer, and productivity is the increased presence of host-cell proteins (HCP) in harvested cell culture fluid (HCCF). Not only does increased HCP further challenge downstream impurity clearance efforts, but can also impact mAb stability even before purification begins. One such consequence of increased HCP is the thioredoxin reductase (TrxR)-induced disulfide reduction of intact mAbs and subsequent formation of low molecular weight species (LMW). LMW formation is a direct risk to product stability, potency, and patient safety; and though the enzymatic mechanism of reduction is well established, the cellular mechanisms leading to increased TrxR activity in HCCF remain unclear. Further, batch to batch variability in LMW formation makes studying such mechanisms a challenge. In this study, we observed differences in culture performance profiles across lab- and manufacturing scale batches of a monoclonal antibody where HCCF either did or did not demonstrate disulfide reduction. Using comparative proteomics, we identified variations in HCP content between a highly reducing and nonreducing batch of HCCF, particularly in proteins involved in redox pathways. Subsequent protein expression and enzyme activity analysis identified higher levels of GAPDH expression as well as increased TrxR and glucose-6-phosphate dehydrogenase (G6PD) activity in LMW containing batches, with no significant difference in TrxR or thioredoxin expression. These results suggest that metabolic alterations during cell culture can lead to changes in protein expression and enzyme activity that yield increased disulfide reduction capacity in HCCF. Further, the observed changes in glycolytic enzyme expression and activity may be useful as markers to further elucidate the mechanism of cellular metabolism and its effect on LMW formation in HCCF. BIOT 15

Steering N-glycosylation of recombinant proteins using systems engineering

Meghan G. McCann1, [email protected], Tung S. Le1, Christopher Stach2, Xinning Chen3, Nikunj Somia4, Liang Zhao3, Michael Smanski2, Wei-Shou Hu1. (1) Chemical Engineering, University of Minnesota, Twin Cities, Saint Paul, Minnesota, United States (2) Biochemistry, Molecular Biology & Biophysics, University of Minnesota, Twin Cities, Saint Paul, Minnesota, United States (3) State Key Laboratory of Bioreactor Engineering, East China University of Science & Technology, Shanghai, China (4) Genetics, Cell Biology, and Development, University of Minnesota, Twin Cities, Minneapolis, Minnesota, United States

Glycan profiles of therapeutic proteins affect their stability, half-life, and biological activities, making them an important quality attribute which must be matched in the development of a biosimilar. However, glycosylation is a complicated system with a distributed reaction network spanning the Golgi apparatus. Glycan structures are influenced by glycosylation enzymes, the supply rate of nucleotide sugars, and competition between both substrates and enzymes within the system. We have taken a systems engineering approach to identify, through model assisted design, the genetic manipulations that may steer the glycan flux to obtain desirable glycoforms.

As the glycosylation pathway lacks key regulated steps as easily identifiable targets for genetic alteration to redirect the flux, multiple genetic alterations are likely needed to obtain a target glycan distribution. Although model prediction can reduce genetic targets, a scheme of rapid construction of gene combinations to facilitate genetic engineering of the cell is necessary. A golden gate assembly workflow was established to produce multi-gene constructs for engineering the glycan biosynthesis pathway. Libraries containing promoters, terminators, and glycosylation related coding sequences of interest, all refactored to be devoid of type IIS restriction sites, were synthesized. From these components, single gene and multi-gene cassette libraries were rapidly built.

To apply this approach, the N-glycosylation pattern of a recombinant IgG produced in CHO cells was modified. A stoichiometric network visualization tool (GlycoVis) helped identify reaction pathways most likely utilized by the cell. Cassettes from the library were selected to redirect flux within the IgG producing cell. Multiple cassettes successfully directed glycosylation to produce antibody with desirable glycoforms. These results refined our model parameters and sharpen its predictive capabilities. This combination of systems analysis and synthetic glycoengineering enhances our capability to obtained targeted N-Glycan patterns and control the quality of therapeutic proteins for innovator and biosimilar production alike.

BIOT 16 An E2E approach for controlling aggregation through a monoclonal antibody (mAb) process

Sonal Saluja, [email protected], Greg Evangelist, Chongfeng Xu, Ravali Raju, Chris Kwiatowski. Biogen, Cambridge, Massachusetts, United States

Aggregation in biotherapeutics is a key concern due to the potential to cause immunogenic reactions in patients. From a drug substance formulation perspective, it is one of the major degradation pathways that results in limiting shelf-life. Controlling aggregation throughout the entire production process can result in lower aggregate levels going into drug substance. In addition, control of aggregation in purification can increase yields by reducing the burden on polishing columns. Work will be presented demonstrating the use of novel excipients for the control of aggregation in mAb production processes, including cell culture, downstream and formulation development. A number of mAbs were spiked with excipients at different steps during the development process and into drug substance. Product quality was assessed by measuring purity and changes in aggregation rates. Results exhibited lower aggregation in both the downstream process and in drug substance compared to non-spiked controls. In addition, increased process yields were observed for the polishing columns in the purification process while maintaining acceptable aggregation levels. The work presented here can be used as a platform approach to improve product quality and enable increased controls to hit required targets by implementing stage appropriate levers from cell culture to formulation development.

BIOT 17

Monitoring and control of capture steps by spectroscopy combined with PLS modeling

Laura Rolinger, [email protected], Matthias Rudt, Nina Brestrich, Juergen Hubbuch. Institute of Engineering in Life Sciences, Karlsruhe Institute of Technology, Karlsruhe, Germany

In the biopharmaceutical industry, process control is required to achieve Quality by Design (QbD). Therefore, authorities support the implementation of process analytical technologies (PAT) for process monitoring and control. For Protein A chromatography in particular, monitoring the product breakthrough in the column effluent would enable a better capacity usage of the expensive Protein A resin and eliminate the need for product titer analytics prior to the load step to determine the load volume. However, due to the variability from cell culture, capture steps are especially affected by changing antibody titer, host cell protein (HCP) concentration and composition which makes monitoring of those steps challenging. In this study, partial least squares regression (PLS) modeling on UV/Vis absorption spectra in a wavelength range from 200-410 nm was applied to monitor and control the antibody concentration in the effluent of a Protein A capture step during the load phase. In a first set of experiments, a PLS model was calibrated based on several breakthrough curves of a Protein A column with variable residence times, mAb titers and HCP composition in the HCCF. The model was used to terminate one run at 10% mAb breakthrough and one run at 50% mAb breakthrough in real-time as a model validation. To evaluate the validity of the proposed approach with a different shape of the breakthrough curve, a Protein A membrane instead of a column was used as capture step. Again, the mAb titer in the feedstock was varied. The predicted mAb concentration in the breakthrough was subsequently used to control the chromatography system and terminate the load step in real-time. Usually no full UV spectra can be recorded in manufacturing, because only a few single wavelength UV detectors are implemented. Therefore other methods with fewer input parameters were compared to the proposed method with PLS modeling on UV/Vis spectra. These methods include wavelength selection and reduction, single wavelength derivation and combination with other orthogonal sensors. In summary, UV/Vis spectroscopy combined with PLS modeling is a promising tool for the real-time control of purification processes. Even for Protein A capture steps that still heavily relies off-line titer assays and volume based loading, UV/Vis spectroscopy combined with PLS modeling can pave the way to a more robust process.

BIOT 18

Rapid mapping of glycoprotein structure-activity relationships by shotgun scanning glycomutagenesis

Xiaolu Zheng, [email protected], Matthew P. DeLisa. Robert Frederick Smith School of Chemical and Biomolecular Engineering , Cornell University, Ithaca, New York, United States

It remains a significant challenge to predict a priori what sites in a protein can/will become glycosylated and what effect glycan installation at a particular site will have on protein structure and function. Shotgun scanning mutagenesis is an extremely rapid method for analyzing the effect of many side chains simultaneously without the need for protein purification or biophysical analysis of each individual mutant. At present, while the presence/absence of glycans can be assessed by a one-site-at-a-time approach, there is no analogous shotgun scanning technique for rapid, high-throughput determination of many glycan sites in a single experiment. To address this technology gap, we have developed an experimental procedure termed shotgun scanning glycomutagenesis (SSGM) for comprehensively identifying the sites/structural motifs within a target protein that can/cannot tolerate glycan installation. SSGM uses combinatorial protein libraries to introduce a glycosylation acceptor site at every possible position along the protein backbone. The library of sequon variants is interrogated for the presence/absence of glycosylation using a high-throughput screen, such as our previously described glycoSNAP (glycosylation of secreted N-linked acceptor proteins) technology. As proof-of-concept, we used SSGM to investigate the permissiveness to glycosylation of the Im7 immunity protein, a four-helix bundle protein that has been extensively studied including with respect to limited site-specific glycosylation. Following library construction and screening, a surprisingly large number of sites along Im7 were found to be glycosylated, with efficiencies ranging from 30- 100%. Because our screen distinguishes positive and negative clones, we are able to tabulate an exhaustive map of the sites in the target protein that can tolerate glycan installation as well as those that cannot. We anticipate that SSGM will become a useful new tool for deep exploration of glycoprotein structure-activity relationships (glycoSARS) as well as for developing glycoengineered protein variants with new or improved functions.

BIOT 19

High-throughput single molecule FRET for protein structure-dynamics-function screening and molecular evolution

Kambiz Hamadani, [email protected]. Chemistry and Biochemistry, California State University San Marcos, San Marcos, California, United States

Single-molecule methods access biomolecular distributions, transient states, and asynchronous dynamics inaccessible to standard ensemble techniques. Although extremely powerful, the ability to screen large biomolecular libraries using fluorescence- based single-molecule detection platforms and evolve such properties in-vitro remains a challenge due to the lack of suitable approaches for the generation and screening of large libraries of dye-labeled proteins. By combining purified and reconstituted in-vitro translation, quantitative unnatural amino acid incorporation, and bio-orthogonal click chemistries (to attach dyes to in-vitro generated target libraries) we overcome these bottlenecks. Here we apply this approach to generate libraries of dual-labeled proteins and ribosome-nascent-chains (RNCs) suitable for single-molecule FRET-based structural phenotyping. Importantly, dual-labeled RNC libraries enable single molecule co-localization of genotypes with phenotypes, and thus multiplexed single molecule screening of protein libraries. Such an approach to high-throughput single molecule screening may be useful for the in-vitro directed evolution of proteins with designer single molecule phenotypes.

BIOT 20

A novel widespread toxin-antitoxin module

Frank Piscotta, A James Link, [email protected]. Princeton University, Princeton, New Jersey, United States

In our efforts in lasso peptide genome mining, we have found two lasso peptide gene clusters that contain genes encoding a toxin-antitoxin (TA) module. In one case, the TA module was the well-characterized HipAB pair, but the other TA module was characterized only from genetic and bioinformatic perspectives. We have carried out bioinformatic studies that suggest that this TA module is exceptionally widespread, appearing in over 2000 different bacterial strains. We have heterologously expressed this novel TA module in E. coli and confirmed that the toxin causes a growth arrest phenotype to which the antitoxin can serve as an antidote. Mutagenesis of the toxin at conserved residues led to identification of residues key for the toxic phenotype. Since the mechanism of the toxin remained unclear, we solved the crystal structure of the toxin in complex with a fragment of the antitoxin. This talk will discuss this structure as well as the cellular targets and function of the toxin.

BIOT 21

Cholesterol dependent ligand binding and downstream signaling of the adenosine A2a receptor

Claire McGraw, [email protected], Anne S. Robinson. Chemical and Biomolecular Engineering, Tulane University, New Orleans, Louisiana, United States

G-protein coupled receptors (GPCRs) represent the largest family of receptor proteins in the living world, having approximately 800 human genes predicted; however the high- resolution crystal structures of only 41 unique GPCRs have been reported to date. The first human GPCR to be crystallized was the β2-adrenergic receptor (β2AR) in 2007. Shortly thereafter an alternate crystal form of the β2AR revealed a specific cholesterol binding site between helices I, II, III and IV. From this work a cholesterol consensus motif (CCM) was established, which defined specific interactions between cholesterol and the receptor. Utilization of this CCM predicted that as many as 25% of all class A GPCRs could have a specific interaction with cholesterol at this site, including the Adenosine A2a receptor. However, the first crystal structure of the human Adenosine A2a receptor (bound to an antagonist) revealed a lipid, not cholesterol bound at the CCM. Molecular dynamics simulations have given insight into other potential binding sites for cholesterol on the Adenosine A2a receptor’s helices 5 and 6, based on the crystal structure bound to an antagonist. Previous work from our lab has shown that point mutations to certain amino acids within the CCM of A2aR lead to a disruption in the downstream signaling of the receptor following agonist stimulation, measured via cAMP assay. In a similar experiment, point mutations to amino acids within the helix 5/6 binding site did not lead to a significant change in the downstream signaling behavior of A2aR following agonist addition. All atom simulations have suggested that A2aR binds cholesterol in a ligand dependent manner; when agonist is bound, the active receptor binds cholesterol at the CCM, while the inactive or antagonist bound receptor binds cholesterol at helices 5 and 6. This talk will focus on continuing work from our lab on the effect of point mutations to both the CCM and the helix 5/6 binding sites to ligand binding of fluorescent and radiolabelled agonist and antagonists in mammalian cells to test the model of cholesterol-mediated receptor activation. We also describe efforts to determine G protein coupling of purified Gαs to purified A2aR and variants by surface plasma resonance.

BIOT 22 Comparing apples and oranges: Identifying initiator peptides of non-amyloid prions

Christopher Jakobson1, [email protected], James Byers2, Thomas Lozanoski3, Zachary Harvey1, Ray Futia4, Sharifa Sahai1, Michael Swift1, Daniel F. Jarosz1,2. (1) Chemical and Systems Biology, Stanford University School of Medicine, Stanford, California, United States (2) Developmental Biology, Stanford University School of Medicine, Stanford, California, United States (3) Bioengineering, Stanford University, Stanford, California, United States (4) Biology, Stanford University, Stanford, California, United States

Prions are proteins that can adopt a stable alternate conformation and recruit proteins from the native fold into this alternate state. Prion-like infectious conformations are classically associated with neurodegenerative disease, but are now understood to be ubiquitous in the normal function of S. cerevisiae and other organisms. Classical prions (oranges) form amyloid fibrils, aggregating the protein in question and sequestering its activity. The 46 new yeast prions (apples) discovered in our laboratory are fundamentally different: they form non-amyloid oligomers and often mediate a gain, rather than loss, of molecular function. Non-amyloid prions confound existing computational analyses of the sequence determinants of prion-like behavior. We therefore use unbiased library generation and selection for prion phenotypes to discover the peptides that can initiate the prion fold. We find that non-amyloid prions can be initiated by the overexpression of multiple peptides that are discontinuous in sequence space, and that de novo prion initiation leads to distinct alleles of the “protein gene” encoded by the prion-like state. Our findings suggest new avenues for controlling protein self-assembly, and offer the possibility of conferring non-amyloid prion-like behavior on heterologous proteins, as has been demonstrated for classical amyloid prion domains.

BIOT 23

Deciphering structural and chemical determinants of pathological tau conformation using yeast surface display

Shiyao Wang, Yongku Cho, [email protected]. Chemical and Biomolecular Engineering, University of Connecticut, Storrs, Connecticut, United States

Microtubule-associated protein tau is expressed at high levels in neuronal cell types and is essential for stabilizing axonal microtubule assembly. In patients with neurodegenerative disorders such as Alzheimer’s disease, frontotemporal dementia, and Parkinson’s disease, tau aggregates into paired helical filaments (PHFs), which constitutes neurofibrillary tangles. Biochemical and structural studies have provided evidence that PHF tau assumes a characteristic paper-clip conformation. However, determining factors that lead to such pathological conformation of tau remain to be elucidated. In order to decipher the structural and chemical determinants of the pathological conformation of tau, we expressed full length human tau on yeast cell surface. Physiologically important isoforms of the human tau, including the 441-amino acid full-length isoform (2N4R) can be displayed at high levels on yeast cells as assessed using anti-tau antibodies. In the full-length tau displayed on yeast surface, majority of phosphorylation sites found in PHF tau are not phosphorylated, potentially due to endogenous yeast phosphatases. Interestingly, however, tau displayed on yeast cells show clear interaction with two independent anti-tau antibodies specific to the pathological conformation of tau. Moreover, the tau displayed on yeast do not interact with tubulins which constitute microtubules. These results indicate that tau displayed on yeast possesses the paper-clip conformation found in PHF tau. Based on previous finding that increased ionic strength and negatively charged species induce PHF formation, we hypothesize that the negative surface charge in yeast cell wall is responsible for PHF tau conformation. We will present our recent progress on testing this hypothesis by modulating yeast surface charge. The impact of various tau mutations that mimic phosphorylation and acetylation of tau on PHF tau conformation will also be presented. Taken together, yeast surface display provides a novel platform to efficiently identify factors that induce tau conformational change, and may lead to new discoveries on how to detect and prevent the formation of such structural variants.

BIOT 24

Interactions of arginine.HCl and NaCl with protein surfaces: Effect on self- association, aggregation, and viscosity of mAb formulations

Chaitanya Sudrik1, [email protected], Theresa Cloutier1, Phuong Pham3, Hardeep Samra2, Bernhardt L. Trout1. (1) Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts, United States (2) MedImmune, Gaithersburg, Maryland, United States (3) Massachusetts Institute of Technology, Cambridge, Massachusetts, United States

Aggregation and viscosity are significant challenges in the development of high- concentration antibody formulations. Excipients like sugars, inorganic salts, and amino acids are used to limit these instabilities to an acceptable range. These excipients affect the dynamics of proteins in solution through their weak interactions with the protein surface. Timasheff and co-workers have developed the preferential interaction framework to describe such weak interactions. According to this theory, excipients that are preferentially excluded from the surface of protein molecules disfavor an increase in the protein-water interface and are, thus, conformationally stabilizing and colloidally de- stabilizing. Excipients that are preferentially bound, on the other hand, promote an increase in the protein-water interfacial area and are, consequently, conformationally de-stabilizing and colloidally stabilizing. Electrolytes are often used to reduce aggregation and viscosity of antibody formulations. To gain further insight into the interactions of electrolytes with protein surfaces, we have characterized the preferential interactions of L-arginine.HCl and NaCl with three IgG1 antibody molecules. We find that NaCl is excluded from the surface of all three antibody molecules. The degree of exclusion increases linearly with increasing NaCl concentration in the range of 0-0.4 molar. For L-arginine.HCl, the degree of exclusion from the protein surface increases non-linearly in the concentration range of 0-0.4 molar. At low concentrations (<0.1molar), L-arginine.HCl is weakly excluded from the surface of the three antibodies, whereas, it is strongly excluded at higher L- arginine.HCl concentrations. In the context of these interactions, we have also characterized the effect of excipient addition on the reversible self-association, aggregation, and viscosity behavior of high-concentration mAb formulations. The data presented in this study provides further insight into the conformational and colloidal stabilization of antibody molecules by NaCl and L-arginine.HCl through their preferential interactions with the protein surface.

BIOT 25

Vibrational Stark effect spectroscopy as a tool for enhancing understanding of electrostatics in complex protein systems

Elisa T. Novelli1, [email protected], Lauren J. Webb2. (1) Chemistry, The University of Texas at Austin, Austin, Texas, United States (2) University of Texas at Austin, Austin, Texas, United States

Vibrational Stark effect (VSE) spectroscopy is a technique that allows us to probe the local electrostatic field at specific locations in and around proteins. In these experiments, we use nitrile vibrational probes that are site-specifically placed in proteins by post-translational modification of cysteine residues. Shifts in vibrational energy of the nitrile probe are related linearly to changes in local electrostatic fields caused by changes in structure, binding, catalysis, and inhibition. This work focuses on the application of VSE spectroscopy as a measurement tool in three well-studied protein systems, p21Ras, Arf1, and Staphylococcal Nuclease. When analyzed along with a suite of other analytical characterization techniques and activity measurements, VSE also allows us to measure the role of electrostatics in catalysis (in the case of p21Ras), folding within the interior of proteins (in the case of Staphylococcal Nuclease), and protein-protein interactions (in the case of the Arf1 system) and in order to better understand these and related classes of proteins and their role in complex biological processes. In the case of p21Ras, we probed the electrostatic mechanism of intrinsic guanosine triphosphate (GTP) hydrolysis in the wild-type and oncogenic mutations of the protein. We showed that different mutations to position 61 of p21Ras resulted in different electrostatic fields at the GTP binding site depending on the identity of the residue. Additionally, we analyzed the relationship between the change in the electrostatic field and the change in intrinsic hydrolysis rate, which is related to cellular apoptosis. In Staphylococcal Nuclease, we related our measurements of the electrostatic field in the interior of the protein to previous measurements of pKa shifts of ionizable residues buried within the hydrophobic core that are crucial to structure and function. Arf1 experiments are ongoing, but the goal is to measure the changes in electrostatics when Arf1 docks to its partner protein and also when a small molecule natural product, Brefeldin A (BFA), inhibits the complex. Altogether, these are well-studied model systems that represent diverse classes of proteins in which structure and function are driven by electrostatic forces that can be understood and quantified with VSE spectroscopy to enhance understanding of electrostatic forces in biological processes.

BIOT 26

Investigating the global landscape of chromatin accessibility to improve production Chinese hamster ovary (CHO) cell line stability

Zion Lee, [email protected], Wei-Shou Hu. Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, Minnesota, United States

A major obstacle in the engineering of CHO cells for therapeutic protein production is the unpredictable nature of long-term cell line instability. Often, this instability results from loss or silencing of the product gene in the genomic integration site. We can now rationally select the locus of integration due to the advent of many site-specific technologies, but little is known about what makes a genomic region amenable for stable transgene expression. We employed Assay for Transposase-Accessible Chromatin sequencing (ATAC-seq) to survey the epigenetic landscape of several CHO cell lines and elucidate the characteristics of favorable integration sites. Our hypothesis is that long stretches of highly accessible chromatin will minimize the risk of transgene epigenetic silencing while supporting robust expression.

ATAC-seq provides a measure of the regulatory activity in the genome. Open chromatin is often the result of RNA polymerase or transcription factor binding events, which displace the local nucleosomes and expose the nearby DNA. In ATAC-seq, a hyperactive Tn5 transposase inserts sequencing adapters preferentially into exposed DNA, and the resulting fragments are subjected to Next-Gen sequencing. Upon mapping ATAC-seq reads to the Chinese hamster genome, we found clusters of peaks in many gene-dense regions, which were corroborated by RNA-seq data, and putative super-enhancers, which are known to mediate high transcription of cell-type specific genes. We also interrogated which areas of the genome are particularly devoid of ATAC-seq signal to identify potential heterochromatic regions. We synthesized these analyses into global comparisons between many cell lines to gain insight into the relationship between phenotype and chromatin accessibility. Increasing our understanding of how CHO cell genomes are organized will allow production of more stable cell lines and streamline cell line development processes.

BIOT 27

Advancing mammalian metabolic engineering through kinetic model optimization

Conor M. O'Brien, [email protected], Andrew Allman, Prodromos Daoutidis, Wei- Shou Hu. Chemical Engineering and Material Science, University of Minnesota, Saint Paul, Minnesota, United States Fast growing cells, including stem cells, cancer cells, and cell lines, have a high glycolysis rate and convert most of the glucose they consume to lactate. In cell culture based biomanufacturing, excessive lactate production has been associated with reduced productivity. Prior efforts to genetically engineer metabolism to eliminate or drastically reduce the conversion of glucose to lactate while maintaining a fast growth rate have not been successful. Given the complex roles of energy metabolism in anabolism and redox homeostasis, successful suppression of lactate production in proliferating cells will likely require simultaneous alteration of multiple metabolic genes. Considering the complexity and number of reactions, we have taken a systems approach to search the parameter space intelligently for key changes to alter cellular metabolic behavior.

Using a nonlinear kinetic model of cell metabolism, we implemented an optimization algorithm to identify combinations of changes to genes of glucose metabolism that will maintain rapid growth with reduced lactate production. The model encompasses glycolysis, the pentose phosphate pathway, and the citric acid cycle, and includes the known allosteric regulations. The formulated multi-objective optimization problem was solved using a local nonlinear optimizer in GAMS. Constraints were chosen to maintain cellular requirements for growth. A penalty term was added to the objective function to reduce the number of genetic alterations to ease experimental design. The resulting solutions contain small sets of changes to enzyme expression that are being evaluated experimentally.

We have demonstrated a method using kinetic model optimization to rationally guide metabolic engineering. The optimization of kinetic models can identify sets of parameter changes that have desirable outcomes and thus greatly reduce the effort required to engineer cell metabolism, providing multiple metabolic engineering strategies and deep insight into the reaction networks and their behavior.

BIOT 28

Leveraging –omics techniques to enhance productivity of biologics in microbes

Joseph R. Brady1,2, [email protected], Charles A. Whittaker2, Melody Tan2, Noelle Colant1,2, Neil Dalvie1,2, Kerry Love2, John C. Love1,2. (1) Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts, United States (2) Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology, Cambridge, Massachusetts, United States

Market sizes for novel breakthrough therapies and growing demand for existing treatments in emerging markets promise to challenge the current capacity for production of biologics. These trends dictate the need for a concomitant paradigm shift in biomanufacturing toward greater productivity for lower cost. Strain engineering is a promising means to realize the greatest returns by increasing the product titer going into downstream processes. Current cellular hosts are approaching saturation of optimal productivity due to lack of deep biological understanding or limitations of the host’s intrinsic secretion capacity. We demonstrate an approach informed by functional genomics to engineer the methylotrophic yeast, Pichia pastoris. Omics-based techniques such as RNA-Seq and ATAC-Seq afford greater upfront understanding for subsequent optimized strain engineering on a product-by-product basis. In turn, design of the transgene cassette grants tight control and fast in silico optimization of wide- ranging factors such as gene dosage balance, translation efficiency, and balanced cell kinetics leading to high-quality protein production. Collectively, these tools enable faster design of a new heterologous protein-producing strain and more robust optimization for engineering higher productivity cell lines.

BIOT 29

Mapping the energy profile of cell culture through advanced lipidomics and metabolomics

Amr S. Ali1, [email protected], Alan Gilbert2, Li Zang2, Ravali Raju1. (1) Cell Culture Development, Biogen, Cambridge, Massachusetts, United States (2) Biogen Idec, Cambridge, Massachusetts, United States

As the demand and success of biological therapeutic proteins soar, there is an increasing need for an optimized, robust and highly efficient bioprocess. Specifically, maximizing protein production by optimizing the cellular nutritional and metabolic needs. A comprehensive lipidomics analysis was performed on Chinese Hamster Ovary (CHO) cells producing an IgG1 antibody in 5 L bench-scale bioreactors. The study demonstrates the dynamic nature of the CHO lipidome as well as the impact lipids have on cellular growth and specific productivity. A robust liquid chromatography mass spectrometry (LC-MS) method using positive and negative mode electrospray ionization (ESI) was developed. 377 lipids were identified and quantitated using two software programs utilizing two different approaches, MS-Dial and GREAZY. The analysis revealed large changes in lipid features such as accumulation of triacylglycerol (TG) (>10 fold increase) and lysophospholipid species (8 fold increase) while depletion of diacylglycerol (DG) species (7 fold decrease). Combining the lipidomics information with results from 30 polar metabolites and RNA sequencing data from 15 genes in a pathway analysis approach demonstrated the difference in lipid metabolism between the various stages of cellular growth and highlighted potential bottlenecks. The study demonstrates the importance of lipidomics in the expanding role of omics methodologies thus gaining insight into cellular behavior during protein production, potentially leading to further opportunities for bioprocess optimization.

BIOT 30

Interrogation of critical metabolic pathways important for compartment-specific redox homeostasis in cancer cells

Mehmet Badur1, [email protected], Dongxin Zhao1, Jens Luebeck2, Jose Magana1, Amanda Birmingham4, Roman Sasik4, Trey Ideker3, Christian Metallo1, Prashant Mali1. (1) Department of Bioengineering, UCSD, La Jolla, California, United States (2) Bioinformatics and Systems Biology, UCSD, La Joola, California, United States (3) Department of Medicine, UCSD, La Jolla, California, United States (4) Center for Computational Biology and Bioinformatics, UCSD, La Jolla, California, United States

The metabolic pathways fueling tumor growth have been well characterized, but the specific impact of transforming events on network topology and enzyme essentiality remains poorly understood. Importantly, cancer cells must reprogram redox pathways to support increased biosynthesis and withstand environmental stresses. To this end, we performed combinatorial CRISPR-Cas9 screens on a set of 51 carbohydrate metabolism genes that represent glycolysis and the pentose phosphate pathway (PPP). This high-throughput methodology enabled systems-level interrogation of metabolic enzyme dispensability, interactions, and compensation within carbohydrate metabolism in two cell lines. The metabolic impact of specific combinatorial knockouts were validated using 13C and 2H isotope tracing. Quantitative analysis of fitness single knockout fitness in A549 and HeLa cells revealed a key role for the KEAP1-NRF2 signaling pathway in influencing the cellular response to knockout of oxidative pentose phosphate pathway (PPP) enzymes. Specifically, by augmenting NADPH regeneration and glutathione synthesis pathways, loss or mutation of KEAP1 facilitates growth by partially mitigating the deleterious effects of PPP gene knockout. Additionally, 2H tracing was used to understand how the PPP supports compartment-specific reprogramming of the redox network in response to routine hypoxic insults. These results together highlight the critical importance of the PPP in supporting redox homeostasis and demonstrate the need for further interrogation of redox metabolism at compartment- and network-level.

BIOT 31

From genotype to fluxome to a comprehensive kinetic model of Escherichia coli

Maciek R. Antoniewicz, [email protected]. University of Delaware, Newark, Delaware, United States

Metabolic and physiological responses to genetic perturbations are of fundamental interest to systems biology and metabolic engineering. They reveal underlying information about network structure including kinetics, regulation, and the presence of otherwise “hidden” reactions. Quantitative prediction of these responses is essential for rational strain design, but has been a persistent challenge in the field. There has been much work done to develop in silico predictive models of metabolism, most notably the constraint-based models. Progress in this area has been limited, however, by a lack of high-quality, reproducible experimental metabolic flux (fluxomic) data of genetically perturbed strains.

In this study, state-of-the-art 13C metabolic flux analysis (13C-MFA) was applied to approximately 50 Escherichia coli knockout strains spanning the major pathways of central carbon metabolism. A full physiological characterization of each strain during aerobic exponential growth was performed, including uptake and secretion rates as well as biomass composition. Flux estimates were obtained using an optimized parallel experimental design, coupled with mass spectrometry measurement of metabolite isotopic labeling and 13C-MFA. The flux and physiological responses of these strains collectively provide valuable new insights into the robustness of central carbon metabolic pathways, as well as areas of likely kinetic limitation. Severely growth- impaired knockout strains identify the most important enzymes in central metabolism and the adaptability of E. coli to extreme perturbations. Particularly surprising responses, including the identification of novel enzymatic activities, will be emphasized. Next, we used this comprehensive flux data set, combined with complementary transcriptomic and metabolomic data, to parametize a genome-scale kinetic model of E. coli metabolism. In addition to having significant value in advancing fundamental biological knowledge, this model provides a valuable new toolkit for industrial biotechnology.

BIOT 32

Genomic scale metabolic network reconstruction of Nostoc sp. PCC 7120 for the analysis of cyanophycin production

Daniel Norena-Caro, [email protected], Michael G. Benton. Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana, United States

Cyanobacteria are emerging as solar cell factories for chemical and pharmaceutical production. Nostoc sp. PCC 7120 (also known as Anabaena sp. UTEX 2576) is a diazotrophic filamentous cyanobacteria capable of producing metabolites from readily available nitrogen (N2), carbon dioxide (CO2), and sunlight. Diazotrophic cyanobacteria are noteworthy from a biochemical production standpoint because they can access a vast pool of nitrogen, which is essential to produce amino-acids, nucleic acids, and bioactive compounds. Nostoc sp. PCC 7120 is part of the order Nostocales, a group of cyanobacteria capable of increased accumulation of cyanophycin polypeptides. The poly-aspartic acid backbone of cyanophycin can be used as a biodegradable alternative to polyacrylate salts, which are widely utilized in personal hygiene products and wastewater treatment processes. We are presenting the genome-scale metabolic reconstruction of Nostoc sp. PCC 7120 comprising 1185 reactions, 1362 metabolites, and 850 genes. The complete genome sequence of Nostoc sp. PCC 7120 was used to formulate the model by means of manual and automatic reconstruction methodologies. The growth kinetics and biomass equation were obtained using our own experimental data, as well as the chemical composition information from closely related cyanobacteria. Flux balance analysis (FBA) and flux variability analysis (FVA) were used to determine the metabolic distribution under autotrophic growth, and minimization of metabolic adjustment (MOMA) was employed to obtain a suboptimal flux distribution of a theoretical mutant expressing increased accumulation of cyanophycin. These metabolic distributions were analyzed for heterocysts (i.e., nitrogen fixating cells) and vegetative cells, and the model predictions are close to the observed growth rates and cellular compositions. These results suggest that the genome scale metabolic model of Nostoc sp. PCC 7120 can be used to advance the photosynthetic production of specialty chemicals in filamentous cyanobacteria.

BIOT 33

Quantitative multi-dimensional single-cell analysis of autophagy during myogenesis

Heather Brown2, [email protected], Edgar A. Arriaga1. (1) Univ of Minnesota, Minneapolis, Minnesota, United States (2) Chemistry, University of Minnesota, Minneapolis, Minnesota, United States

Myogenesis is a crucial biological process for the development and maintenance of muscle mass, and the loss of regenerative potential leads to debilitating musculoskeletal disorders. Differentiation of progenitor cells (myoblasts) into mature myotubes is a dynamic process during which cell identity and function shift over time. Traditional, bulk biochemical analyses lack the resolution to tease apart the highly heterogenous myoblast response to differentiation conditions, where subpopulations of cells undergo progressive phenotypic and functional changes. Autophagy, a system of intracellular recycling, has been implicated as essential for myoblast differentiation. Via autophagy, long-lived proteins or damaged organelles are degraded to their amino-acid constituents, providing building blocks for synthesis of new or different subcellular components. A high-resolution study of progressively differentiating myoblasts will illuminate which phenotypically unique subpopulations increase autophagic flux. We use mass cytometry, a single-cell inductively-coupled plasma mass spectrometry technique, to quantify changes in phenotypic and autophagy markers during myoblast differentiation in vitro. The technique can measure upwards of 50 parameters per particle through the detection of isotopically pure lanthanide ions chelated by antibody- based probes specific towards protein targets of interest.

In mass cytometry analysis two major issues which can bias detected signal are, i) antibody binding specificity, and ii) variation in cell size or cell permeabilization. To address these issues, we make use of two control antibodies: an isotype to assess non- specific binding, and a “loading control” to normalize specific antibody signal. Individual cell data is mathematically transformed post-acquisition to yield corrected and normalized signals for biological analysis. We demonstrate the use of this method to reveal accurate levels of autophagy-related proteins in genetically or pharmacologically manipulated myoblast models of autophagy dysregulation. Unaltered data show linear titration curves while transformed data can be accurately fit to Langmuir isotherms. With these data, we have accurately quantified the fold-change of autophagy-related proteins in progressively differentiating myoblasts in during normal myoblast differentiation. Analyses suggests an increase in mitochondrial-specific degradation via autophagy, which may indicate a response to changing metabolic needs.

BIOT 34 Use of polyamines for enhanced retention of plasmid DNA isoforms during ultrafiltration

Ying Li, Ivan Manzano, [email protected], Andrew L. Zydney. Dept of Chemical Engineering, Penn State University, University Park, Pennsylvania, United States

Spermidine and spermine are two naturally occurring polyvalent cations that play a critical role in compacting and controlling the structure of supercoiled plasmid DNA in living organisms. These polyamines interact with DNA in two ways: the positively charged amine groups can interact electrostatically with the negatively charged phosphates along the DNA backbone and the polyamines can bind in the major and minor grooves of DNA involving hydrogen bonds with bases from the opposing strands of the DNA double helix. The objectives of this work were to investigate the effects of the polyamines spermidine (with three amine groups) and spermine (with four amine groups) on the ultrafiltration behavior of plasmid DNA isoforms and to explore the opportunity of using these poly-cations to enhance the purification of the desired supercoiled plasmid. Experiments were performed using a 9.8 kbp plasmid in the supercoiled, open circular, and linear isoforms. DNA solutions were prepared in 10 mM Tris + 1 mM EDTA buffer with different NaCl concentrations. Polyamines were added at a concentration of 2 – 30 µM and allowed to equilibrate. Ultrafiltration was done in a stirred cell using Biomax® polyethersulfone membranes with nominal molecular weight cutoffs of 100 and 300 kDa over a range of filtrate flux from 30 to 100 µm/s. The addition of spermine caused a dramatic reduction in the transmission of the supercoiled plasmid during ultrafiltration. For example, the DNA sieving coefficient at a filtrate flux of 50 µm/s decreased from more than 70% to less than 20% when the spermine concentration was increased from 2 to 20 µM. Higher polyamine concentrations were needed to obtain the same effect at higher NaCl concentrations and at higher plasmid concentrations, with the data largely collapsing to a single curve when plotted as a function of the ratio of the spermine to DNA concentration. The polyamines also had a significant effect on the transmission of the linear and open circular isoforms, although the open-circular isoform was highly retained even up to filtrate flux as high as 80 µm/s when using low NaCl concentrations. These results demonstrate the potential of using natural-occurring polyamines to control the retention of plasmid DNA during ultrafiltration.

BIOT 35

Scalable nanofiber based lentiviral vector purification

Jelena Ruscic1, Christopher Perry1,2, Yasu Takeuchi2,3, Tarit Mukhopadhyay1, Daniel G. Bracewell1, [email protected]. (1) Department of Biochemical Engineering, University College London, London, United Kingdom (2) Division of Infection and Immunology, University College London, London, United Kingdom (3) Advanced Therapy Division, National Institute for Biological Standards and Control, London, United Kingdom Lentiviral vectors (LVs) are indispensable part of gene therapy clinical trials due to their ability to transduce both dividing and non-dividing cells. Current cultivation methods produce titers of 105 to 107 TU/mL thus it is necessary to concentrate LV as well as to remove process and product related impurities.

In this work we used a continuous producer cell line WinPac-RD for LV production. Recovery through the purification process was monitored by several orthogonal methods: an infectivity assay utilized GFP expression determined by flow cytometry, LV RNA genome was quantified via RT-qPCR using primers specific for the GFP gene, LV particles were detected with p24 ELISA and SYBR Green I-based product-enhanced reverse transcriptase (SG-PERT) assays. Using tangential flow filtration (TFF) we were able to remove significant amount of cell culture proteins, but LV recovery was only 20%. This led us to pursue a TFF-free nanofiber bind/elute step based on ion-exchange chromatography. The cellulose nanofiber based purification platform provides an accessible high surface area adsorbent well suited to viral vector sorption. Using this method, we were able to concentrate LV 100 fold while also achieving a 2 log removal of host cell protein and maintaining up to 90% yield of a functional vector.

BIOT 36 mRNA vaccines and therapeutics: On the progress from promise to reality

Hari Pujar, [email protected]. Moderna Therapeutics, Cambridge, Massachusetts, United States

Messenger RNA offers unparalleled breadth and depth for the discovery and development of novel drugs and vaccines. The potential to direct specific tissue translation of both wild type and engineered intracellular, membrane-bound, and secreted proteins (and combinations thereof), coupled with rapid transition from preclinical to clinical development, has enabled Moderna and its partners to progress several development candidates into the clinic. This potential is being realized by the parallel development of different modalities (e.g. prophylactic and therapeutic vaccines, paracrine and systemic drugs). This talk will review the scientific and engineering accomplishments enabling pharmaceutical development.

BIOT 37

Virus purification in aqueous two-phase system at varying tie line lengths

Caryn Heldt, [email protected], Pratik Joshi, Matthew Weiss. Michigan Technological University, Houghton, Michigan, United States

Virus purification has been a challenging task due to the complexity of the broth, particle size, and the sensitivity of viral particles to processing conditions. An alternative unit operation that is not typically employed industrially is aqueous two-phase systems (ATPS). ATPS are biocompatible, environmentally friendly and easy to scale-up. However, there remain challenges to understanding the dominate parameters in ATPS before industrial implementation will be realized. Our goal is to understand the dominate forces in ATPS and to reduce the large experimental space to a smaller set of parameters that would increase industrial adaption of the technology. From our previous work, we have determined that viral hydrophobicity is likely the dominate separation force in ATPS. In order to test this theory, we are studying the separation of the non- enveloped virus, porcine parvovirus (PPV), using a polyethylene glycol (PEG) 12kDa- citrate system at pH 7. High PEG molecular weight provides a hydrophobic environment for PPV and the highly hydrated citrate provides the repulsive force to propel the PPV into the PEG-rich phase. To systematically study the hydrophobic changes, we are studying the effect of tie line lengths (TLL) and the ratio of phases along the tie line. Longer TLL, which has a higher concentration of PEG and citrate, allows for a greater recovery of PPV in the PEG-rich phase, demonstrating that both the repulsion of the citrate and the attractive force of the PEG-phase are needed for high virus recovery. Movement along the tie line changes recoveries by changing the volume ratio of the two aqueous phases. The PEG-rich phases of the longer tie lines studied were free of contaminant proteins, as observed by SDS-PAGE. A recovery of 75% of infectious PPV was found. Our work continues to support the idea that virus hydrophobicity is the dominate force in the ATPS separation of viral particles.

BIOT 38

Development of a scalable chromatography process to separate empty and full adeno-associated vector (AAV) particles

Ryan Dickerson, [email protected], Meisam Bakhshayeshi. Technical Development - Gene Therapy, Biogen, Somerville, Massachusetts, United States

Recombinant adeno-associated viral (rAAV) vectors have emerged as one of the most popular delivery vehicles for gene therapy applications. During the cell culture process to manufacture, AAV vectors, undesired empty particles are formed. These particles are capsid shells that are similar to the desired product, but lack the therapeutic gene inside them. The presence of empty particles at high levels could reduce efficient gene transduction in the targeted cells and are believed to cause immunological responses. To ensure high therapeutic potency of AAV vector products and enhance their safety profile, it is critical that the manufacturing process can consistently produce viral vector products that contain low levels of empty particles.

Gradient ultracentrifugation using cesium chloride or iodixanol has proven to be effective in removing empty particles, but the process scale-up for large-scale AAV manufacturing remains a significant challenge. In this work, we present a scalable anion exchange chromatography method to separate full and empty particles. The presence of DNA inside the full AAV particles reduces the isoelectric point of the capsid by approximately 0.4 units, and that minor surface charge difference was exploited to achieve the desired separation. Very shallow linear gradient elutions using chlorine salts was applied to an anion exchange monolith column to achieve the high degree of resolution required for separation of these similarly charged particles. There are several challenges in implementing a shallow gradient on a monolith column at large-scale manufacturing, including consistent generation of gradients using disposable flow-path chromatography skids, effective fractionation and pooling, and challenging column scale-up. To overcome these limitations, a novel column process with isocratic wash and elution was developed which enabled separation of empty particles with a higher degree of resolution while maintaining high product recovery.

BIOT 39

In silico prediction of physicochemical liabilities in Adeno-associated virus capsids

Kristine Manibog, [email protected], Sandeep Kumar. BTX Pharmaceutical Sciences, Pfizer Inc, Chesterfield, Missouri, United States

Successful clinical trials using Adeno-associated viruses (AAVs) have prompted biotechnology and pharmaceutical companies to invest in gene therapeutics. However, industrialization of gene therapy will involve large scale manufacturing of gene therapy products; and their stabilization against physicochemical degradations that can potentially occur during manufacturing, storage and shipping. To enable this, appropriate chemistry, manufacturing and control (CMC) processes and work flows need to be devised. Studies from CMC processes developed for commercial development of biologic drug products can offer useful guidance in this regard. Specifically, analogous experimental and computational methods can be applied to study physicochemical degradations of AAV capsids. In this work, we strive to gather theoretical insights into aggregation and potential chemical degradation susceptibilities present in amino acid sequences of 15 AAV serotypes. AAV capsid sequences contain several (14-17) potential aggregation prone regions (APRs), each containing 5-12 contiguous amino acid residues. Interestingly, almost all (99.9%) of the predicted APRs across 15 AAV serotypes are located in their VP3 regions and several of them (~10 APRs) are conserved among 12-15 AAV serotypes. Two of the APRs conserved in 12 out of 15 AAV serotypes lay within the EF and GH loops. The EF and GH loops have been implicated in A20 mAb binding, and in the competitive binding inhibition of heparan sulfate proteoglycan to AAV2, respectively. In addition, a few sequence motifs (DG, DS, NG, and NS), vulnerable to deamidation and isomerization, are solvent exposed and located near the 5-fold DNA pore. These computational modeling studies have improved our understanding of potential degradation routes of AAV capsids structure. Insights gained for these studies will help guide formulation development of AAVs and also enable rational design of AAV capsids for improved develop-ability.

BIOT 40

Single-use purification of cell culture-derived virus particles by steric exclusion chromatography Pavel Marichal-Gallardo1, Keven Lothert2, Tanja Grein3, Udo Reichl1,4, Peter M. Czermak2,5, Michael W. Wolff2,1, [email protected]. (1) Bioprocess Engineering, Max Planck Institute for Dynamics of Complex Technical Systems, Magdeburg, Saxony-Anhalt, Germany (2) Institute of Bioprocess Engineering and Pharmaceutical Technology, University of Applied Sciences Mittelhessen, Giessen, Hesse, Germany (4) Chair of Bioprocess Engineering, Otto-von-Guericke University Magdeburg, Magdeburg, Saxony-Anhalt, Germany (5) Project group Bioresources, Fraunhofer Institute for Molecular Biology and Applied Ecology , Giessen, Hesse, Germany

The variety of purification techniques for industrial-scale downstream processing of large biomolecules such as viruses, virus-like particles, and gene therapy vectors is more or less restricted to centrifugation, chromatography, and filtration-based operations. For the development of platform purification technologies, adsorption chromatography is particularly challenging as specific common ligands do not exist or are at least rare. Selectivity of steric exclusion chromatography (SXC) is strongly linked to target size, as for the classical size exclusion chromatography. However, in contrast to the latter, it is also suitable to capture and therefore concentrate biomolecules. SXC is carried out by mixing the crude sample with polyethylene glycol (PEG) and loading it onto a hydrophilic matrix. Based on the applied PEG molecular weight and concentration, the target species are captured without a direct chemical interaction by the mutual steric exclusion of PEG between the target particle and the stationary phase. Smaller impurities, i.e. proteins and DNA, are washed out. Finally, the product is eluted by reducing the polymer (PEG) concentration and can be recovered in a variety of buffers. Unmodified regenerated cellulose filters can be used as stationary phase in SXC. Here, we demonstrate the capture of cell culture-derived influenza A and oncolytic measles virus particles produced in mammalian cells under controlled conditions in serum-free medium. Recovery of influenza virus particles was approximately 100% and the purified virus particles were monodisperse at about 84 nm. Protein and DNA depletion were 92.4% and 99.7%, respectively. The recovery of measles virus particles was >95% and the contaminant depletion is expected to be comparable to the influenza virus purification process. Our experimental setups show that SXC has the potential as a promising and affordable capture platform technology to concentrate, purify, and exchange buffers in a single step.

BIOT 41

Getting more for less: Leveraging epigenetics to increase process yield

Chris Kwiatkowski, [email protected]. Cell Culture Development, Biogen, Cambridge, Massachusetts, United States

Inherent in the approach to phase-appropriate development is the need for multiple process development cycles. A major advantage in this approach is that resources can be minimized in the development of early phase programs, which can be then put toward later phase programs that have a higher probability for commercial success. In this paradigm, it is the goal for early development to quickly transition from research to the clinic with clinical timelines being the main development constraint. Subsequently, the goal of late stage development is the generation of a commercially viable process with matching early-phase product quality being the main development constraint. The length of clinical timelines can have additional impact on how development cycles are assessed and initiated where short timelines pressure early phase processes to deliver commercial ready processes and long timelines push late phase processes to higher titers. During late-stage development, media components are often screened and balanced that can have a positive impact to process yield and/or robustness. Here, we show the results of a novel epigenetic inhibitor that improved process yield by approximately two-fold while also having a positive impact to process robustness. An omics approach was then applied to determine the mechanism for this improvement. Rationale media design was then applied to determine the optimal concentration and delivery method for optimizing the fed-batch process. Finally, we explore the molecules impact on product quality and how a robust end-to-end protein platform can mitigate potential product quality comparability risks between development cycles.

BIOT 42

Media optimization towards a high-titer platform process

Brian Y. Wong, [email protected], Robert Shawley, Martin Gawlitzek, Laurel Zhang. Late Stage Cell Culture PTD, Genentech, South San Francisco, California, United States

Roche/Genentech’s current Global Cell Culture Platform using a CHO host cell line for producing monoclonal antibodies (mAbs) in chemically-defined media has been successfully used in process development and clinical manufacturing. Further cell culture process development, in particular media optimization, is required to optimize lactate metabolism for some lactogenic cell lines and to improve productivity for some high titer demanding projects. We mainly approached these issues through sequential DOEs to identify critical media components and optimize feed strategies to achieve titers of > 10 g/L in a traditional fed-batch process. The optimized proof-of-concept process developed work may ultimately lead to our next generation platform process.

BIOT 43

Elucidating amino acid metabolism in CHO cells

Maciek R. Antoniewicz1, [email protected], Jaqueline Gonzalez1, Brian McConnell1, Harnish Naik2, Venkata Dhara2, Michael J. Betenbaugh2. (1) University of Delaware, Newark, Delaware, United States (2) Johns Hopkins University, Baltimore, Maryland, United States CHO cells require complex media for cell growth and protein production. The major components of industrial media are amino acids, however, relatively little is known about the metabolism of amino acids in CHO cell cultures. Here, we applied advanced 13C-flux analysis tools to elucidate the metabolic flow of the amino acids in a fed-batch CHO culture that overproduced IgG. Carbon flows were tracked throughout the growth phase and changes in metabolism were quantified when cells transitioned from growth phase to stationary phase. In addition, we quantified how changes in amino acids profiles in the medium translated to changes in cell growth, protein production and product quality attributes. To trace each amino acid individually, custom media formulations were used, where each medium formulation was depleted of a specific amino acid. A labeled 13C variant of the depleted amino acid was then added to the medium at the desired concentration. CHO cells were then grown in fed-batch culture. As the cells metabolized the labeled amino acids, this resulted in a redistribution of 13C- atoms which we quantified using GC-MS for both extracellular metabolites (including lactate, amino acids and the IgG product) and intracellular metabolites (including free intracellular metabolites, cell proteins, lipids and carbohydrates). We then estimated metabolic fluxes using state-of-the-art 13C-metabolic flux analysis. This allowed us to calculate the fraction of each amino acid that was used for cell growth, protein production, lactate formation and energy generation. We also investigated the effects of labeling in both the batch and fed-batch stationary phase. Finally, we investigated the effects of varying amino acid concentrations. Each 13C-labeled amino acid was added to the medium at a lower or higher concentration compared to the base medium. 13C- metabolic flux analysis was again performed and changes in fluxes were compared in order to determine the precise impacts of amino acid concentration changes on the flux profiles. Taking all of this data together, we are now building a predictive kinetic model that relates how the metabolism of CHO cells can be predicted from amino acid profiles. In future work, model predictions will be experimentally validated as a means of optimizing the amino acid composition of industrial culture media.

BIOT 44

What cells want vs. what cells need: Simplification of chemically defined media for optimal performance

Matej Krajcovic, [email protected], Yelena Ilin, Shelby Hutchins, Alissa Borshchenko, Woo S. Ahn, Lia Tescione, Duane Inlow, Canghai Lu. Process Science, Global MSAT, Sanofi, Framingham, Massachusetts, United States

Optimization of cell culture media is a key part of process development for production of therapeutic proteins in mammalian systems. While historical cell culture media contained serum or hydrolysates, modern processes utilize chemically defined media. During development of chemically defined media, components are often added to boost performance, but there is no concerted effort in simplifying the medium formulations. The benefits of media simplification include increased stability and reduced nutrient and waste accumulation. Here, we present a systematic approach for examining the limiting concentrations of key media components, based on biomass yield and nutrient consumption rates for fed-batch cultures. We present the results that lead to optimal cell culture performance and product quality, and discuss the rationale for selecting either minimal or excess concentrations of key nutrients.

BIOT 45

Media formulation optimization based on multi-scale modeling of heterogeneity in mammalian cell culture process

Shaun Galbraith, [email protected], Huolong Liu, Seongkyu Yoon. University of Massachusetts Lowell, Lowell, Massachusetts, United States

A multi-scale cell culture model is developed to identify critical media components that impact cell growth cycle and thus predicts productivity. Laboratory based media optimization requires an extensive DOE whereas models can perform the same optimization with less experimentation saving resources. Conventionally, cell culture models are categorized into intracellular or extracellular. Intracellular models are average cell models capturing metabolic pathway reactions, this means they must assume all cells in the culture are following the same metabolic pathway. Extracellular models use population balance models (PBM) to account for cell cycle propagation, however, they cannot explicitly consider intracellular metabolism. Here, a multi-scale modeling approach is adopted to unify the understanding of intracellular metabolism and the probabilistic nature of cell heterogeneity due to the cell cycle. The culture dynamics are described using an unstructured model encompassing cell growth, cell death, nutrient consumption, metabolite, and protein production and their dependency on media composition. A one-dimensional, volume-based PBM is formulated for three identifiable cell cycle phases G1/G0, S, and G2/M simultaneously in the culture. The cell metabolism for each cell cycle phase is modeled differently to account for the cell heterogeneity and cycle specific intracellular activities. By the cell-cycle phase specific metabolism, the cell growth and cell density in the high productivity cycle phase could be controlled by media composition. Future work will involve the collection of data from AMBIC (Advanced Mammalian Biomanufacturing Innovation Center) cell lines in order to parameterize and validate the developed model. This work fits well within the “Computational strategies to enhance bioprocess performance: From systems biology to process modeling” session, specifically looking at mathematical modeling of bioprocess upstream.

BIOT 46

Understanding glycosylation variability in CHO cell cultures

Patrick Ahyow, [email protected], Anh Nguyen Dang, Kathryn Mains, Melissa Mun. Late Stage Cell Culture, Genentech, South San Francisco, California, United States

Glycosylation control of recombinant monoclonal antibodies is important to maintain consistent effector function and therapeutic efficacy. Multiple cell culture factors can impact glycosylation, including many trace metals, which are co-factors for enzymes in the glycosylation pathway. Manganese is a co-factor of β-1,4-galactosyltransferase, which catalyzes the addition of terminal galactose to N-acetylglucosamine (Witsell et al. 1990), and can impact galactosylation levels in combination with other components involved in these processing steps (Gramer et al. 2011). In this case study, variation in glycosylation was observed across bioreactor production cultures for a single cell line. Variation in manganese levels was identified as a contributing factor and the effect of manganese on glycosylation was further confirmed in small-scale studies.

To understand the sources of manganese variation in the process, lot-to-lot variability was examined for several raw materials used in the cell culture media, and media preparation procedures were evaluated to quantify manganese loss. Manganese levels in several media components, including hydrolysate, were found to vary across lots. In addition, variable manganese loss was observed across media preparation steps. Together, these factors largely account for the observed variation in manganese levels in the production culture.

In order to mitigate manganese variability, modifications to media preparation procedures were evaluated in small-scale and pilot scale studies. Small-scale bioreactor studies were subsequently performed with media prepared with the optimized media preparation procedure to ensure no impact to cell culture performance or product quality.

These learnings contribute to our overall understanding of cell culture factors that impact glycosylation and methods to potentially minimize manganese variability.

BIOT 47

Understanding the effect of preparation methods on bioavailability of components in a cell culture feed

Mark Berge, [email protected], Erica Hackner, Jeong Lee, Jason Reier, Mao- Shih Liang. MedImmune, Gaithersburg, Maryland, United States

Medium and feed are key factors in the success of modern high titer cell culture processes. Ensuring that all vital nutrients are available when the cells need them is a critical part of developing a commercially viable large scale production process. A key aspect of developing an upstream process is understanding the effects of scaling the unit operations and especially how the medium and feed is prepared. When the preparation methods at differing scales alter the availability of nutrients, the cell metabolism can be affected which alters cell growth, product titers, and even product quality. Recently, it has been discovered that the large-scale preparation method for a process feed has shown to have deleterious effects on its corresponding large-scale process. The most surprising discovery was that although one of the affected feed components was measurable at the expected concertation, it was not bioavailable to the cells. Details of the study and the resulting corrective actions will be discussed. BIOT 48

Dynamic and chemically heterogeneous polymer brushes stabilize protein conformation

David F. Marruecos, Hye H. Kim, Daniel K. Schwartz, Joel Kaar, [email protected]. Univ of Colorado, Boulder, Colorado, United States

Polymer brushes consisting of poly(ethylene glycol) (PEG) have been extensively studied as coatings to minimize protein adsorption and ultimately the unfolding of proteins on biomaterials. However, while PEG brushes reduce protein adsorption, such surfaces may actually enhance protein unfolding as well as stabilize the presence of unfolded protein molecules on the surface. In this work, we investigated the extent to which protein conformation and dynamics may be controlled on polymer brushes that are chemically heterogeneous. To determine this extent, the conformation and dynamic behavior of fibronectin (FN) was characterized on random copolymer brushes consisting of PEG and poly(sulfobetaine) (PSB) using novel single-molecule methods. Through SM studies, we showed that the extent of fibronectin unfolding and the retention of fibronectin on PEG brushes was reduce by incorporating PSB. Perhaps most interestingly, the relationship between the fraction of PSB in the brush and fibronectin unfolding and accumulation in the denatured state was non-monotonic. This suggests there is an optimum PSB-to-PEG ratio that diminished interactions between unfolded fibronectin and the random block copolymer brush. Such an optimum may result from matching the heterogeneity of the brush to that of surface of the protein, which contains residues with a broad range of chemical identities. We hypothesize that, by matching the heterogeneity of the brush to that of the protein, the protein may act as a nanoscale template for the self-assemble of surrounding chemical groups within the brush layer, resulting in favorable nano-environments that stabilize the protein. The stabilization of proteins via protein-templated nano-scale self-assembly of heterogeneous surfaces represents a new paradigm for the design of biocompatabile materials.

BIOT 49

Mechanistic origin of the combined effect of surfaces and mechanical agitation on protein aggregation

Paolo Arosio, [email protected]. Chemistry and Applied Biosciences, ETH Zurich, Zurich, Switzerland

Interactions between proteins and surfaces in combination with hydrodynamic flow and mechanical agitation can often trigger the conversion of soluble peptides and proteins into aggregates. Despite the empirical effects of surfaces and mechanical forces on protein aggregation are well documented, the molecular details of the mechanisms underlying this behaviour are still largely elusive. The analysis is complicated by the fact that the formation of the aggregates is the consequence of a complex reaction network, in which several microscopic reactions of nucleation and growth can occur both at the interfaces and in bulk. In this work, we design an experimental assay based on nanoparticles to investigate independently the effect of surfaces and mechanical forces on the formation of amyloid fibrils from human insulin, and we apply a chemical kinetic platform to analyse the molecular mechanisms at the origin of this effect. By considering a variety of polymeric nanoparticles with different surface properties we explore a broad range of repulsive and attractive interactions between insulin and surfaces. Our analysis shows that hydrophobic interfaces induce the formation of amyloid fibrils by specifically promoting the primary heterogeneous nucleation rate. In contrast, mechanical forces accelerate the formation of amyloid fibrils by favouring mass transport and further amplify the number of fibrils by promoting fragmentation events. Thus, surfaces and agitation have a combined effect on the kinetics of protein aggregation observed at the macroscopic level but, individually, they each affect distinct microscopic reaction steps: the presence of interfaces generates primary nucleation events of fibril formation, which is then amplified by mechanical forces. These results suggest that the inhibition of surface-induced heterogeneous nucleation should be considered a primary target to suppress aggregation and explain why in many systems the simultaneous presence of surfaces and hydrodynamic flow enhances protein aggregation.

BIOT 50

Neutron scattering to characterise protein interactions with solid-liquid interfaces in bioprocessing

Maria Papachristodoulou3,4, [email protected], Daniel G. Bracewell2, Christopher J. Roberts4, Luke Clifton5, Paul Butler1, James Doutch5. (1) NIST, Gaithersburg, Maryland, United States (2) UCL Dept Biochemical Engr, London, United Kingdom (3) Biochemical engineering, University College London, London, United Kingdom (4) University of Delaware, WASHINGTON DC, Maryland, United States (5) Rutherford Appleton Laboratory, Oxford, Didcot, Oxford, United Kingdom

Affinity chromatography is a crucial method of protein purification, used during the manufacture of many therapeutic antibodies. The protein A ligand is the most common choice for selective purification of IgG molecules. Aggregated product is often found in protein A elution pools; this is generally attributed to the effect of low pH (elution) on protein structure. However, there is evidence that other parameters of the chromatography process influence aggregation phenomena. Physical and chemical transitions, like concentration of protein on the adsorbent surface and change in buffer composition, may challenge the structural integrity of proteins, increasing their propensity for aggregation. To advance understanding of how adsorption processes might impact antibody stability, neutron reflectivity was used to characterize the structure of adsorbed IgG on model surfaces. In the first model system IgG was adsorbed directly to silica and demonstrated a flat-on-surface orientation. In the second model system, protein A was attached to a silica surface to produce a configuration representative of a silica chromatography resin. Adsorbed IgG structures extended up to 230 Å away from the surface and are suggestive of two IgG molecules bound to protein A with a skewed orientation. To complement our simplified planar NR model systems, SANS studies on model packed columns where the concentrations, interactions, and flow fields experienced are much more comparable to typical chromatographic separations, were carried out. To that end, we have worked closely with ISIS and NIST staff to construct a SANS compatible flow cell to closely mimic an affinity column and provide the ability to both change buffer conditions in situ on the same sample and study the effect of flowing buffers through. The flow through cell was packed with Prosep-Ultra plus resin (Protein A coated silica beads) and loaded with increasing levels of IgG representative of bioprocessing. This experiment suggests that the resin has a complex structure composed of a range of pore sizes and within this structure the IgG does not adsorb uniformly across these pore surfaces – a process largely dependent on its concentration. Upon increasing concentration, IgG molecules migrate from the external pores (i.e. closest to the bead surface) onto the internal, smaller ones. This hypothesis can be supported by the fact that an additional peak is visible during the SANS experiment at higher IgG concentrations.

BIOT 51

Adsorption of non-ionic surfactants and proteins on hydrophobic surfaces studied by neutron reflectometry

Zhenhuan Zhang3,5, [email protected], Sara V. Orski2, Ann M. Woys6, Isidro Zarraga7, Norman J. Wagner4, Yun Liu1. (1) NIST, Gaithersburg, Maryland, United States (2) Materials Science & Engineering, NIST, Montgomery Village, Maryland, United States (3) Chemical Engineering , University of Delaware, Frederick , Maryland, United States (4) Dept of Chemical Engineering, University of Delaware, Newark, Delaware, United States (5) Center for Neutron Research, National Institute of Standards and Technology, Gaithersburg, Maryland, United States (6) Genentech, San Francisco, California, United States

Monoclonal antibodies (mAbs) have become increasingly popular for commercial pharmaceutical products development because of their excellent target selectivity and fewer side effects. As one of many possible routes, surface adsorption can introduce physical degradations such as unfolding and denature. Those physical degradations near surfaces can further trigger the instability of bulk mAbs. Therefore, it’s very important to have a clear understanding of interactions between mAbs formulation molecules and different type of liquid-solid surfaces. Among these surfaces, hydrophobic surfaces can be easily found in many medical devices such as polymeric drug containers. In this work, neutron reflectometry is used to investigate the structure information of the adsorption process of two non-ionic surfactants (polysorbate 20 and Poloxamer 188) and proteins (monoclonal antibody X and lysozyme) onto both polystyrene-liquid surface and Surfasil coated oil/water interface. Adsorbed layer thickness and molecular orientation of non-ionic surfactants are obtained with the neutron reflectometry through deuterium labeling. We found totally different adsorption behavior of two types of proteins onto the same hydrophobic polystyrene surface. The adsorption of one protein is strongly adsorbed on the surface while there is no adsorption observed for another protein. The possible physical mechanisms of the difference of the adsorption of these two proteins will be discussed. These experimental results can provide significant insight to the adsorption behavior of proteins on other liquid-solid interfaces.

BIOT 52

Peptide-functionalized surfaces for rationally designed biotechnology

Whitney Fies1, [email protected], Lauren J. Webb2. (1) Chemistry, The University of Texas at Austin, AUSTIN, Texas, United States (2) University of Texas at Austin, Austin, Texas, United States

Proteins possess an extraordinary level of functional specificity and efficiency, making it advantageous to immobilize them onto inorganic substrates and exploit their properties for applications in sensing or catalysis. Successful immobilization of biomolecules requires the preservation of conformation when bound at surfaces that deviate drastically from their native cellular environment. We are developing a biomimetic method to immobilize proteins to inorganic surfaces that specifically aims to preserve native protein conformation. Our method uses short, alpha-helical peptides covalently bound to a self-assembled monolayer on gold to mimic a protein’s natural cellular environment. The goal is to exploit specific biological peptide-protein electrostatic interactions, the mechanism proteins naturally use for in vivo stabilization, to mediate controlled protein adhesion for many types of proteins with varying functions. To accomplish this, we have successfully bound two peptides with two solution-facing surface chemistries: positively charged and hydrophobic. These surfaces have been extensively characterized to confirm that we have alpha-helical peptides covalently bound parallel to our gold surfaces with the desired charge or hydrophobicity. Additionally, we have collaborated with scientists at Oak Ridge National Laboratory and measured these surfaces with neutron reflectometry (NR) to develop a more comprehensive understanding the role of water in peptide structure stabilization. NR possesses the powerful and unique ability to elucidate bound water within peptides, proteins and other functionalized layers on inorganic surfaces. Our data thus far shows unprecedented water penetration within the self-assembled monolayer underneath the peptide, which is traditionally expected to be a complete physical barrier to the gold surface. This information, and future reflectometry analysis, will fill a vital gap in understanding that stands in the way of developing a truly comprehensive knowledge of how biomolecules behave at bio/abio interfaces. Additionally, ongoing steps to immobilize full proteins on these surfaces are paving the way for intelligent biosensor and two-phase biocatalyst design.

BIOT 53

Computational approaches to understand a living/inorganic interface: Engineering bacterial fimbrial tip protein FimH to bind gold surfaces Meagan C. Small1, [email protected], Jessica Terrell2, Dimitra N. Stratis- Cullum2, Margaret Hurley1. (1) Sensors and Electronic Devices Directorate, US Army Research Laboratory, Aberdeen Proving Ground, Maryland, United States (2) Sensors and Electronic Devices Directorate, US Army Research Laboratory, Adelphi Laboratory Center, Maryland, United States

Bio-inorganic hybrid (or living) materials have many potential applications in medicine, electronics, and nanotechnology. Thus, biomolecules with predictable and controllable binding to inorganic surfaces are of great interest. Bioengineered proteins can be introduced into bacterial cells to confer nonnative functionality and thereby alter whole- cell behavior. With this is mind, our work has been focused on engineering E. coli fimbrial tip protein FimH so that it binds to inorganic interfaces, namely Au(111) surfaces. Imperative to this has been the understanding of protein dynamics using computational approaches to predict the impact of a modification on native functionality, and further, verify its conformational accessibility. In this work, we present our computational studies of a number of bioengineered FimH proteins that bind to Au(111) surfaces to understand the molecular details underlying biotic-abiotic interface interactions. First, we demonstrate that integration of a series of gold binding peptides (GBPs) into the FimH lectin domain at I52 disrupts bacterial cell binding to mannose, the native ligand, while imparting nonnative gold-binding capability. Second, we show from computational models and simulations of select mannose-bound engineered FimH variants that the inserted GBPs perturb the secondary structure around the mannose binding site, thereby disrupting pivotal hydrogen bonds between mannose and the lectin domain leading to mannose disassociation. Third, we provide insights from simulations of the full series of engineered FimH variants into structural features that enhance binding to Au(111). To conclude, we will discuss these findings in the context of our work in understanding and controlling living/inorganic interfaces to enable reconfigurable, switchable materials.

BIOT 54

Short wave radio frequency resonators for transducing protein and cell surface interactions in closed systems

Sadaf Charkhabi, Andee Beierle, Nigel Reuel, [email protected]. Chemical and Biological Engineering, Iowa State University, Ames, Iowa, United States

This presentation focuses on the development of chipless (lacks traditional integrated circuit), passive (no power source), short wave radio frequency biosensors that can be embedded both in or on materials for wireless monitoring of biologic interactions. The flexible, spiral antennas (Fig_a) can be screen printed with conductive silver ink or patterned on copper-coated polyimide substrates and etched with a solution containing hydrogen peroxide and hydrochloric acid. The antenna’s initial resonant frequency is tuned by the coil length and spiral pitch and is targeted in the 1-100MHz range for better penetration through soil, water, and tissue. The real and imaginary components of the absorption (S11) and transmission (S21) scattering parameters are measured at a 1-10 cm distance by two loop antenna coils (Fig_b) attached to a vector network analyzer. These parameters shift when the capacitance between the antenna coils change due to variation in dielectric permittivity of the medium on top of the antenna (Fig_c). Changes in skin resistance due to charge near the copper coil also affect the scattering parameter response. We have applied this sensor platform to study the activity of hydrolytic degradation of substrate coatings (e.g. proteases to protein films). The wireless sensors screen effects of substrate composition and enzyme concentration on the overall degradation profile of the film via simple, multiplexed experiments. Furthermore, we show that this can be done in closed systems such as soil. We also present current efforts on transducing protein and cell interactions as well as biofilm growth using resonant sensors. This research will have broad impact in applications of measuring biologic activity where optical line of sight or direct electrical connection are inconvenient or impossible, such as monitoring the contents of a sterile, disposable bioreactor.

BIOT 55

Determining the relationship between T cell activity and affinity of cytomegalovirus-specific TCRs Christopher Stevens1, [email protected], Chad Williams2, Zach Frye1, Jenny Jiang2, Jennifer Maynard1. (1) McKetta Department of Chemical Engineering, The University of Texas at Austin, Austin, Texas, United States (2) Biomedical Engineering, The University of Texas at Austin, Austin, Texas, United States

Human cytomegalovirus (hCMV) is a persistent β-herpes virus that infects 60-90% of the population and remains latent after primary infection though it can reactivate during periods of immunodepression. Infection is usually kept in check by CD8+ cytotoxic T lymphocytes (CTLs) in healthy patients, but it can cause life-threatening diseases in immunocompromised individuals such as hematopoietic stem cell transplant (HSCT) recipients. CMV-specific CTL responses elicited in healthy donors have a more diverse TCR repertoire than in immunocompromised patients, where responses are often dominated by T cells bearing TCRs with closely related or identical sequences between individuals, suggesting these ‘public’ TCRs more effectively induce an immune response. The mechanism by which the TCR/pMHC interaction induces an immune response is not fully understood but is thought to depend on the affinity and/or kinetics of binding. Understanding the role of public vs. private TCRs and their relative affinities in inducing an immune response will aid in the development of vaccines and adoptive cell therapies towards CMV.

To this end we sought to identify a correlation between the affinity and the activity of several public and private TCRs specific to HLA-A*0201-restricted CMV antigen (pp65495-503) isolated from a diverse population of CMV+ patients. Previously, we have developed a robust system for rapid screening of TCR binding and activation using soluble and membrane bound DO11.10 TCR. This system was applied to the immunodominant RA14 TCR by fusing the variable regions to mouse constant domains to ensure correct pairing and transfecting into Jurkat T cell hybridoma. Activation was determined by quantifying IL-2 secretion and CD69 upregulation following co-culture of transfectants with peptide pulsed T2 antigen presenting cells. Affinity was determined using a cellular interaction approach where individual cells were introduced to red blood cells functionalized with pMHC to measure the TCR/pMHC interaction in situ. This ‘2D affinity’ for RA14 was found to be consistent with measurements by SPR and for similar affinity TCRs. Public pp65-specific TCRs generally had higher affinities than their private counterparts yet secreted significantly lower amounts of IL-2 and reduced CD69 upregulation. This suggests that public TCRs are not more effective at inducing an immune response and that a lower ‘optimal’ affinity may be key to maximizing an immune response towards hCMV.

BIOT 56

Simultaneous determination of drug efficacy in single cells using a multiplexed droplet trapping array

Khashayar R Bajgiran, [email protected], Riad Elkhanoufi, Manibarathi Vaithiyanathan, James A. Dorman, Adam T. Melvin. Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana, United States The heterogeneous nature of cancer has resulted in the use of combinatorial therapeutics designed to focus on multiple molecular targets. Monitoring the effectiveness of these tandem therapeutic combinations requires an analytical approach capable of high-throughput single cell screening. This approach provides a deeper understanding of drug treatments efficacies. Traditional approaches require performing replicate measures for each target in series or the incorporation of expensive antibodies to simultaneously track the different inputs. The goal of this work is to combine a suite of spectrally independent rare earth (RE)-doped nanoparticles with a three-input droplet microfluidic platform capable of high-throughput screening of the single cell response to increasing concentration of established drugs. This platform offers fast and accurate screening of encapsulated single cells in 70-micron aqueous droplet chambers that contain both fluorescent viability stains and luminescent RE-doped nanoparticles for droplet tracking. The triple-input microfluidic droplet trapping array was confirmed to achieve ~99% droplet trapping efficiency coupled with ~40-50% single cell encapsulation using GFP-expression HeLa cells. The result is a platform able to analyze approximately 500 cells for each of the three inlet conditions for a total of 1500 cells screened. As a proof of concept of the method, dynamic drug studies were performed to screen three different doses of two different cancer drugs: the proteasome inhibitor Bortezomib against myeloma-derived OPM2 cells and Paclitaxel against the triple negative breast cancer line MDA-MB-231. Image analysis of the encapsulated single cells was performed using a custom Python algorithm capable of distinguishing between nanoparticle trackers and fluorescent cells. The microfluidic platform confirmed that increasing the dose of a drug decreased cellular viability; however, the single cell approach confirmed a heterogeneous response in the population of cells with some cells exhibiting a resistance drug treatment. This platform can offer versatile drug studies, using patients’ samples in order to reach a suitable combined drug treatment that fits the unique needs of each individual patient diagnosed with cancer.

BIOT 57

Small molecule based drug capture materials

Sankarganesh Krishnamoorthy, [email protected], Robert H. Grubbs. Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, California, United States

Direct delivery of chemotherapeutics to cancer cells has improved cancer treatment using methods such as chemoembolization. Unfortunately, some of the administered drugs leave the target cancer cells and enter the systemic circulation, causing serious side effects. To prevent these off-target chemotherapy effects, a variety of drug-filtering materials are being developed. For example, uses of ion exchange resins and DNA embedded on iron oxide nanoparticles have shown promising results. However, a key concern with DNA based materials is possible DNA fragmentation and unknown effects of foreign DNA fragments in the body. Therefore, development of drug filters with biologically benign material is crucial. In the present study, development of selective drug capture materials based on small molecules on metal or metal oxide surfaces will be discussed.

BIOT 58

Electro-responsive poly-graphene medi-patch arm-bands for controlled delivery of STAT-3 inhibitors in anti-stem cell melanoma therapy

Santosh K. Misra, [email protected], Muhammad S. Khan, Prabuddha Mukherjee, Dipanjan Pan. Bioengineering, University of Illinois at Urbana Champaign, Urbana, Illinois, United States

Sustained release of drug formulations using skin patches have been gaining popularity and medical acceptance worldwide. Major challenges for such patches are biocompatibility, drug loading capacity, drug release efficiency, directionality of drug release and finally controlled time dependent biodegradation, which are yet to be handled efficiently. At this very end, we have prepared novel polymer-graphene (PolyGraphene) based drug loaded medi-patch "Arm-bands". Polycaprolactone (PCL) was melted and mixed with graphene (GR) via hydro-thermo-evaporation method to prepare medi-patches with (Nic-PCL and Nic-GR-PCL) or without loaded drug (PCL and GR-PCL), Niclosamide (Nic), a known inhibitor for cancer stem-like cells via STAT-3 (Signal Transducers and Activators of Transcription) inhibition pathway. Furthermore, medi-patch was integrated with a highly sensitive Au based µ-electrode biosensor to study the behavioral response during drug release and study of possible real time monitoring of released drug. The scanning electron microscopic (SEM) investigation revealed the layered structures of prepared medi-patches and raman spectroscopic data showed distinguished features of all components in medi-patches. These medi- patches were found to be responsive in releasing Nic in time dependent manner which could also be followed by our newly designed blue-tooth assisted impedance evaluator. Medi-patch was incubated with C32 (Human skin melanoma) cells for different time points and showed a significant loss in cell population. In vivo studies with mouse model gave similar results. It was also found that GR was playing a major role in controlling the release of the drug. Thus, a highly potential medi-patch for inhibition in stem like cell population among C32 (Human skin melanoma) cells has been prepared with unique ability of tracking and controlling the drug release.

BIOT 59

Precisely targeted photodynamic antitumor therapy by platelet membrane coated nanoparticle

Lulu Xu, [email protected], Feng Gao, Feng Fan, Lihua Yang. University of Science and Technology of China, Hefei, China

Photodynamic therapy ablates target cells with reactive oxygen species (ROS) generated upon light irradiation. Owing to the short lifetime (only tens of seconds) of ROS, photodynamic therapy works only within tens of nanometers around the photosensitizer, suggesting better spatiotemporal control than the closely related photothermal therapy. Clearly, to improve photodynamic therapy efficacy, the key is to precisely deliver the photosensitizer onto/into the target cells. However, current photodynamic systems rely on negative targeting or chemically attached active targeting moieties, which provides limited targeting efficacy and requires extra efforts on surface engineering, respectively. How to achieve photodynamic system that offers precise targeting without requiring extra efforts on surface engineering remains a challenge. Platelet membrane bears over-expressed P-selectin that specifically binds to CD44 up- regulated on the surface of cancer cells, and coating with platelet membrane endows a nanoparticle with long blood circulation lifetime. We hence coat a photodynamic nanoparticle with platelet membrane, in efforts to improve its targeting efficiency and consequent therapeutic efficacy. Results from cell studies and mouse models will be presented. BIOT 60

Eradicating primary tumor and cancer stem cells with prosthetic antigen receptor modified T-cells

Carston R. Wagner, [email protected], Jacob R. Petersburg. 8-174 Weaver- Densford Hall, University Minnesota, Minneapolis, Minnesota, United States

Our laboratory has used chemically self-assembled nanorings (CSANs) to develop prosthetic antigen receptors (PARs) as a non-genetic method for modifying cell surfaces. PARs are formed by fusing a scFv to two E. Coli dihydrofolate reductase (DHFR2) molecules that spontaneously assemble into octomeric chemically self- assembled nanorings (CSANs) upon the addition of a chemical dimerizer, bis- methotrexate (bisMTX). The combination of an anti-CD3 fusion protein with a tumor targeting fusion protein results in the formation of bispecific, multivalent, CSANs that stably bind to CD3+ T-cell surfaces, generating PARs, and selectively target tumor cells. Therefore, we developed anti-Epithelial Cell Adhesion Molecule (EpCAM)/anti-CD3 PARs from bispecific CSANs that selectively target the EpCAM receptor, which is overexpressed on multiple carcinomas and cancer stem cells. The anti-EpCAM/anti- CD3 PARs were found to stably bind T-cells for >4 days. In the presence of EpCAM+ MCF-7 breast cancer cells, PAR-functionalized T-cells were activated and induced cytotoxicity. Using an orthotopic breast cancer xenograft model, tumors were found to be completely eradicated by the anti-EpCAM/anti-CD3 PAR T-cells. Following PAR treatment, the production of IL-2, IFN-γ, IL-6 and TNF-α could be significantly reduced by an infusion of clinically-relevant concentrations of the FDA-approved antibiotic, trimethoprim, signaling pharmacologic PAR deactivation Importantly, the CSANs had no significant immunogenicity in mice and were not shown to initiate naïve T-cell activation without the presence of target EpCAM+ breast cancer cells. In addition, we have developed anti-CD133 targeting CSANs that are capable of targeting T-cells to cancer stem cells. We have shown with the triple negative breast cancer cell line, MDA-MB-231 cells, that combining anti-EpCAM/ant-CD3 PARs with anti-CD133/anti-CD3 PARs can result in enhanced in vitro cell killing. We also demonstrated that in vivo, tumor eradication requires treatment with both nanorings. Collectively, our results demonstrate PAR modified T-cells have the potential to be a viable cancer immunotherapy targeting solid tumors.

BIOT 61

Visualization of a cancer-associated enzyme using near-infrared (NIR) off-on and on-on fluorescent probes in vitro and in vivo

Zhenhua Shen2, [email protected], Robin L. McCarley1. (1) Louisiana State University, Prairieville, Louisiana, United States (2) Chemistry, Louisiana State University, Baton Rouge, Louisiana, United States Enzymatic activatable fluorescent probe development holds great significance in signaling cancer-associated species, unraveling the mystery of cancer onset and progression, as well as their potential use in fluorescence-guided surgery. Human NAD(P)H:quinone oxidoreductase isozyme I (hNQO1), targeted in this research, has upregulated expression in a multitude of solid tumors and close association with tumorigenesis and metastasis processes, which makes it a promising biomarker to develop fluorescent probes. In this research, two near infrared (NIR), wavelength- shiftable fluorescent probes Q3STCy and Q3NTCy were developed to selectively and rapidly report hNQO1 activities in cellulo, in 3-dimensional colorectal tumor mimics, and in a xenograft mouse model. Upon incubation with the developed probes, tumor cell lines with upregulated hNQO1 expression levels exhibited increased fluorescence response with high integrity through monolayer cellular imaging demonstration. They were further applied in a three-dimensional colorectal multicellular spheroids model (1 mm diameter) with spatially heterogeneous hNQO1 activities. The hNQO1 upregulated area showed significant higher fluorescence increase due to the more effective probe activation. In xenograft mouse models, human ovarian cancer-derived metastases as tiny as 0.5 mm in dimension were also successfully visualized by employing the newly developed probes.

BIOT 62

Programmed accumulation of prodrug-converting enzymes in cancer cells

Tiana D. Warren, [email protected], Marc Ostermeier. Chemical and Biomolecular Engineering, Johns Hopkins University, Baltimore, Maryland, United States

Gene-directed enzyme prodrug therapy (GDEPT) is a two-step process designed to make chemotherapy more selective for tumor cells. First, a gene encoding a prodrug- converting enzyme is selectively delivered to tumor cells, then a prodrug is systemically administered to the patient. The prodrug is converted to the toxin only in cells that express the enzyme and harbor the prodrug, thus inducing apoptosis. One limitation of GDEPT is selectively transducing enough cancer cells to express the enzyme. To circumvent this issue, we have developed a protein switch that selectively accumulates in cells experiencing hypoxia by fusing the yeast cytosine deaminase (yCD), a prodrug- converting enzyme, to HIF-1a. HIF-1a is a transcription factor that regulates the cell’s response to low oxygen levels. In normoxia, HIF-1a is hydroxylated at two of its proline residues, which initiates its degradation by the 26S proteasome. However, prolyl hydroxylation is inhibited in hypoxia due to a lack of oxygen, a substrate for the reaction. HIF-1a is also known to accumulate at the core of solid tumors where oxygen levels are low. By properly fusing yCD to HIF-1a, we expected that the fusion protein would be degraded in normoxia via the same pathway as HIF-1a, but accumulate in hypoxia. Higher protein levels of yCD in hypoxia would cause efficient conversion of its prodrug, 5-fluorocytosine (5-FC), to 5-flourouracil, a chemotherapeutic. Here, we designed HIF- 1a Oxygen-dependent degradation domain/Prodrug-converting Enzyme (HOPE) fusions to serve as protein switches. These HOPE fusions exhibited controllable protein accumulation and efficient conversion of 5-FC in mock-hypoxia but not in normoxia. BIOT 63

Tunable, post-translational method for controlling prodrug converting enzymes in cancer cells

Andrew Gaynor, [email protected], Wilfred Chen. Chemical and Biomolecular Engineering, University of Delaware, NEWARK, Delaware, United States

Enzyme prodrug therapies, in which an innocuous prodrug is introduced and later enzymatically converted to the potent active drug by a prodrug converting enzyme (PCE), is a promising direction for treating cancer. Unlike traditional chemotherapies, which offer no ability to target diseased cells specifically, PCEs can be deployed using a variety of targeted delivery mechanisms, including cell-specific protein uptake through targeting peptides and transcriptional control through gene therapies. These strategies, while demonstrating potential, frequently suffer from “leaky” off-target effects which are amplified through the catalytic ability of PCEs.

Degradation tags are a post-translational mechanism for controlling cellular protein levels with much faster response times and lower background levels than transcriptional control. Herein, we propose a novel method for modulating degradation tag activity in mammalian cells by sterically blocking endogenous cellular proteins from acting upon the tags. By fusing a small peptide proximal to a degradation tag, the peptide’s corresponding protein-interaction partner is able to effectively conceal the degradation tag from the proteasome. This can be done using both N- and C-terminal degradation tags. This system is tunable depending on the inherent strength of both the degradation tag and the protein-peptide pair’s interaction. Furthermore, the use of orthogonal protein-peptide interactions potentially allows for constructing post-transcriptional logic gate architectures such that multiple protein cues are required to increase the cellular concentration of a protein of interest. Since our system operates independently of previously described translational or delivery control mechanisms, it could expand the current complexity of cellular computing devices. Looking forward, the system can be adapted to detect one or more cancer-specific protein concentrations as a method for avoiding off-target PCE activity.

BIOT 64

Complete snakebite therapy

Claire F. Komives, [email protected]. Dept of Chemical Engineering, San Jose State University, San Jose, California, United States

Approximately five million episodes of snakebite occur globally each year, according to the WHO. In the United States, over 100,000 dogs and cats are afflicted by snakebite. The WHO has classified snakebite as a neglected tropical disease. Currently available antivenoms are produced in farm animals, mainly horses but also sheep. The venom of local snakes is injected into the animals, causing suffering to the animals, and their blood is later collected and the antibodies are purified to form a cocktail that is able to scavenge the venom toxins from snakebite victims. These antivenoms are mostly distributed by government hospitals but it may be difficult to reach the hospital rapidly after the bite takes place. Once a snake bites the venom toxins rapidly begin to affect tissues and vital organs. Victims also suffer mental trauma from the bite. The gold standard therapy for a snakebite would be a vaccine that enables people to maintain their own human antibodies in their body to begin scavanging the toxins immediately after the bite. The vaccine will also give people assurance of their survival after the bite. This technology is a vaccine for snakebite. Technical details will be presented at the conference as well as strategies for marketing and development.

BIOT 65

Escherichia coli as production vehicle for high value chemicals, glucaric acid and glucuronic acid

Darcy Prather, [email protected]. Kalion, Inc., Milton, Massachusetts, United States

BIOT Tank (Oral)

Kalion is an early stage industrial biotech company providing low-cost access to high purity glucaric acid and other chemicals using traditional, green, fermentation technology.

Technology Differentiation: Biological production that allows for high purity glucaric acid necessary for highest value application and polymer applications Size of Company: less than 2 full time Years in Business: 3 since initial funding: Number of Customers : 0 - Testing with customer Revenue:0

Customer Problem Glucaric acid needs to be of high purity to address pharmaceutical needs. In polymer applications glucaric must be both high purity and low cost. Our fermentation pathway allows us to immediately address the high purity needs of the high costs pharmaceutical markets. Our pathway also allows us address other high value nutraceutical applications. Lastly we will produce low-cost high-purity glucaric for the billion dollar polymer markets. Products & Services We create microbes (E. coli) with a patented pathway that will allow us to produce a range of products at substantially lower costs. The first of these is high-purity, high- value, glucaric acid. The next set of products include glucuronic acid and myo-inositol. The last product will be low-cost, high-purity glucaric acid. By the end of 2017 we expect to have low volumes of our highest value product available for purchase.

Target Market Glucaric acid has been identified as a unique additive to multiple traditional billion-dollar polymers. The high level of impact supports low-cost, high-purity glucaric acid at $10/lb. In pharmaceutical industry, Adderall require high purity glucaric acid that sells for $30/lb so we will start there. Glucuronic acid and myo-inositol are ingredients for many high energy drinks like 5 Hour or Monster energy drinks.

***********Shark Tank Event is different than traditional abstract. Please let us know if you need addition information or different format********

BIOT 66

Next generation biopharmaceutical process development: Stop experimenting. GoSilico

Thiemo Huuk1, [email protected], Teresa C. Beck1, Tobias Hahn1, Juergen Hubbuch2. (1) GoSilico GmbH, Karlsruhe, XX, Germany (2) KIT, Karlsruhe, Germany

Over the last decades, computer simulations evolved into an indispensable tool in a lot of industries, such as automotive, aeronautics, etc. enabling to boost productivity and to speed up innovation cycles. Despite the innovative nature of biopharmaceutical and biotechnological industry, production process development still remains in a pre- digitalized stage, mainly relying on expensive and time-consuming lab experiments and on iterative empirical optimizations. Aside of an increasing economic pressure, this approach reaches its limits in the context of evolving regulatory expectations.

Technology Differentiation: GoSilico provides the biopharmaceutical industry with its simulation software ChromX, allowing for a computer-based, in silico process development. This approach allows to skip up to 95% of the expensive lab experiments and to fulfil the evolving regulatory expectation for a mechanistic process understanding. Customers use GoSilico’s technology in a range of applications from early process development, scale-up and technology transfer to the maintenance of production facilities.

Size of Company: 9 employees (approx. 6 FTE) Years in Business: since 01/2016 Number of Customers: 7 customers

BIOT 67

Zepteon: On demand antiviral antibody production Austin Boesch1, [email protected], Glen Bolton1, Julie Fox2, Michael Diamond2. (1) Zepteon, Inc, Boston, Massachusetts, United States (2) Washington University in St. Louis, St. Louis, Missouri, United States

Zepteon, Inc. is a privately held corporation founded in 2012 that has developed a first in class chromatography resin, Glycap-3A, capable of separating antibodies with specific sugar structures (nonfucosylated) that are more potent against viruses and cancer. The affinity purification resin contains optimally immobilized FcgammaR3A receptors, leading to high antibody equilibrium binding capacities.

Hyperimmune IgG is typically obtained from human donors who have been exposed to a virus and have generated virus specific antibodies. Literature indicates hyperimmune IgG would be far more potent against a range of viruses (Flu, MERS-CoV, Ebola, HPV, West Nile, CHIKV) using optimal IgG glycoforms and subclasses. In this talk, data showing that anti-viral plasma IgG fractionated using the Glycap-3A resin is far more potent against viral infection in mice will be presented.

Zepteon resin is the only technology capable of simultaneously providing more potent IgG glycoforms and subclasses (IgG1 and IgG3) for the $1B USD hyperimmune IgG market. The resin can be used to rapidly obtain therapies after a viral outbreak. The method is faster than the generation, testing and scale up of monoclonal antibody therapies. The technology can enrich the most relevant IgGs from recently infected people to overcome viral strain mutation. In addition polyclonality provides multi-prong approach against viral escape. Zepteon is currently out-licensing the technology to IVIG and hyperimmune manufacturers for up-front, licensing and milestone payments

BIOT 68

Exploring the mulitmodal anion exchange space: Fundamental investigations and applications

Julie Robinson1, [email protected], Hong Chen2, Xuemei He2, Yueping Xu2, Mark A. Snyder2, Steven M. Cramer1. (1) Chemical Engineering, RPI, Troy, New York, United States (2) Bio-Rad Laboratories, Hercules, California, United States

In the past decade, multimodal chromatography has been an area of active research, from both a fundamental perspective and with regards to process development applications. While much of the recent work has focused on multimodal cation exchange (MM CEX) chromatography, the increasing prevalence of non-mAb therapeutics as well as challenging mAb-aggregate separations motivates further development in the multimodal anion exchange (MM AEX) space. The objective of this work was to investigate multimodal anion exchange chromatography, with an eye towards potential applications for both mAb and non-mAb biologics. A library of novel MM AEX prototypes developed by Bio-Rad Laboratories was evaluated using linear salt gradient experiments with model proteins in bind and elute mode. Several of the resins showed unique selectivities compared to the commercially available Capto Adhere. Five of these resins were selected for further evaluation with three antibodies with diverse surface properties. In each resin system, the range of mAb adsorption behavior (i.e. from flow through to fully retained) was classified at various pH conditions. The results of this screen then were used to identify optimal conditions at which to evaluate aggregate removal and recovery for industrial mAb-aggregate samples. This work suggests that a more diverse MM AEX ligand library could not only facilitate mAb- aggregate separations but could also enable MM AEX ligands to be used for finer resolution separations such as the removal of product related variants by expanding the pH operating range of this class of ligands. Expanding the MM AEX library could also potentially have important implications for the purification of non-mAb biologics, particularly low pI proteins, by providing new selectivities compared to the existing commercial resins.

BIOT 69

Evaluation of chromatofocusing and related ion exchange methods as alternatives to protein A chromatography in purification of antibodies

Yang Liu1, [email protected], Sevda Deldari1, Hui Guo2,3, Chittoor N. Rao4, Ronald Bates5, Ryan Swanson4, Zheng Jian Li4, Sanchayita Ghose4, Douglas Frey1. (1) Chemical Engineering, University of Maryland, Baltimore County, Halethorpe, Maryland, United States (2) Chemical Engineering, University of Maryland, Baltimore County, Ellicott City, Maryland, United States (3) Shimadzu Scientific Instruments, Columbia, Maryland, United States (4) Bristol-Myers Squibb, Devens, Massachusetts, United States (5) Bristol-Myers Squibb, East Syracuse, New York, United States

Protein A affinity chromatography is commonly used as the initial capture step in a purification process for monoclonal antibodies. However, the high cost, propensity for ligand leaching, and relatively low loading capacity of this method has led to recent interest in replacing this step. This work investigates the use of both chromatofocusing and ion-exchange chromatography as alternatives to protein A chromatography for the initial capture step in an overall purification process for antibody products. The specific conditions employed for the capture step for the case of chromatofocusing, such as the identities and concentrations of the buffering species used, were selected on a rational basis using a computer-aided design method and a process development tool which incorporated a 3D analytical separation method. Alternative operating conditions were compared experimentally with regard to the product yield, product loading capacity, and product purity. Results of this study indicate that chromatofocusing and ion-exchange chromatography are useful alternatives to a protein A chromatography capture step in many practical cases. This is especially true for the case of chromatofocusing when it is possible to exploit the ability of the method to employ pH gradients to affect the protein adsorption behavior, to create complex gradient shapes that are self-forming inside the column, and to simultaneous focus and separate proteins inside the column.

BIOT 70

A new cation exchange resin designed for the removal of monoclonal antibody aggregates using flow-through frontal chromatography

Matthew Stone, [email protected], Kristen A. Cotoni, Jayson Stoner, Peter Menstell. MilliporeSigma, Arlington, Massachusetts, United States

We report an innovative new cation exchange resin for the downstream purification of monoclonal antibodies having a unique surface chemistry that enables efficient aggregate removal in the flow-through frontal chromatography mode of operation. Operating cation exchange chromatography in the flow-through frontal chromatography mode of operation is advantageous since it allows aggregate removal to be accomplished with significantly smaller columns relative to traditional bind/elute cation exchange chromatography. We have investigated the influence of a variety of different factors (pH, conductivity, etc.) on aggregate removal using this novel cation exchange resin in a flow-through frontal chromatography mode of operation.

BIOT 71

Releasing chromatography from the past: Integrating a new lattice supported bed and resin design to achieve hyperproductive™ affinity capture

Martin Siwak1, [email protected], Jim Van Alstine2. (1) JSR LIFE SCIENCE, Ipswich , Massachusetts, United States (2) JMVA Biotech AB, Stockholm , Sweden As powerful as modern bio-chromatography is there are still constraints in resin design, column design and operating conditions which limit performance. Packing and maintaining column hardware also limit ease of use and manufacturing flexibility. Further linear scalability is not truly achievable due to diminishing wall effects and flow distribution in large diameter columns. Chromassette™, a novel lattice supported modular chromatography device which offers intrinsic bed support through the entire resin volume, significantly reduces the above constraints. Resin designs which may have desirable performance properties that cannot operate in current columns can now be integrated with the supported bed. Chromassette opens the possibility to develop advanced adsorptive materials with large pores for fast binding and small diameters for high capacity with no need to impart typically hydrophobic cross-linking chemistries. Data will be shown on several resin prototypes packed in a supported lattice chromatography device that can achieve 5-10 fold higher productivity, with no sacrifice in yield, purity, elution volume or buffer consumption.

BIOT 72

Control of an engineered self-cleaving affinity tag for purification of recombinant proteins by modulating extein residues

Yamin Fan, [email protected]. Chemical Engineering, The Ohio State University, Columbus, Ohio, United States

Downstream purification development of novel protein therapeutics invovle time- consuming and costly optimization and scale up based on the characteristics of each protein. Thus, it is desirable to create an purification platform for purification of a diverse range of recombinant proteins, but also with the requirment of traceless and tagless features for the purified products. Self-cleaving purification tags provide a robust tool for creating such generic platform and our lab has developed a pH controllable purification system using an engineered split-intein tag derived from Npu DnaE intein from Nostoc punctiforme. This system consists of a covalently immobilized N-intein (NpuN) on to agarose beads and C-intein tag (NpuC) with fusion to the protein of interest. The precursor protein with NpuC tag can be highly selectively captured by the immobilized NpuN fragment on column due to their nano-molar affinity. The intein self-cleavage activity is triggered by a pH shift of the buffer conditions and the purifed protein products can be eluted after a certain period of time of incubation at R.T.. In this work, we characterize the cleavage kinetics of our split-intein system with various combinations of N-exteins and C-exteins of the engineered intein. Previous studies have shown that the extein residues at the C-termius of the intein can strongly affect cleavage kinetics. Through the studies, we identify N-exteins that can lead to enhanced pH sensitivity of the split intein or accelarate the intein cleavage activity. This leads us to better understand the mechanism of the engineered split intein system. We can also use this information to rescue the previously slow cleaving target proteins when using our split intein system by altering the N-exteins of intein to achieve traceless and tagless purification. By modulating the extein residues of the split intein, we can precisely control the cleavage activity during purification for various proteins in our system including biosimilars like Granulocyte colony stimulating factor (GCSF) and Interferon alpha 2b (IFN α2b).

BIOT 73

Robust viral clearance using novel anion exchange technology

Alexei M. Voloshin, [email protected], Narina Stepanova, Angelines Castro Forero. 3M Company, Woodbury, Minnesota, United States

Clearance of viral particles in a bioprocess is a key requirement for product safety assurance as well as for regulatory compliance. Today, the bioprocesses incorporate an entire viral clearance strategy composed of multiple steps, each utilizing an orthogonal principle of viral particle separation. One common part of the viral clearance strategy is Anion exchange chromatography (AEX). This principle is widely used due to its simplicity and ability to couple viral clearance and process related contaminants reduction in one process step. In practice however, the robustness of viral clearance using AEX principle is limited by a number of factors. Since the current crop of chemistries require pH to be 1-2 units above the viral pI for robust clearance (> 4 LRV), AEX technology can only be implemented for high pI molecules, such as classical mAbs. Salt and buffer tolerance is another serious limitation of present AEX technologies, limiting the process parameters to very narrow combinations of salt and buffer conditions. Because the current AEX technologies are sensitive to interfering contaminants load (such as DNA, HCP, partially bind product, insoluble particles), AEX viral clearance strategy can only be deployed at certain parts of the process and has to be carefully sized to account for this sensitivity. Here we present a novel viral clearance approach utilizing AEX principle. We combine a quaternary ammonium (Q) functional non-woven material with guanidinium (Gu) functionalized membrane to overcome the limitations of present AEX technologies. We demonstrate that this strategy can provide robust viral clearance at conductivities exceeding 25 mS/cm, at pH ranges equal to that of pI of the virus, and interfering soluble and insoluble contaminant levels approaching total AEX capacity of the system. This high degree of robustness enables rapid implementation of AEX viral clearance step with little need to reoptimize process conditions or process structure.

BIOT 74

Retooling downstream processing to address or enable an ever-expanding diversity of therapeutic modalities

Charles Haynes3, [email protected], David J. Roush2, Mark A. Snyder1. (1) Bio-Rad Laboratories, Hercules, California, United States (2) Process Development and Engineering, Merck, Sharp and Dohme, Colts Neck, New Jersey, United States (3) Michael Smith Laboratories, University of British Columbia, Vancouver, British Columbia, Canada

Our rapidly improving understanding of biological systems is leading to unparalleled discovery of novel targets whose modulation provides the potential for effective treatment of disease. These targets include both protein-protein and protein-nucleic acid interactions, and modulation of either is often refractory to classic small-molecule therapeutic approaches. This long-recognized challenge has motivated in part the development and expansion of protein-based therapeutics, most notably humanized monoclonal antibodies (mAbs), that have proven effective in modulating important extracellular targets with excellent selectivity and potency. However, mAbs and other recombinant protein therapeutics are generally not effective against intracellular targets due to a variety of factors, including poor membrane permeability. Equally limiting is the fact that current biologic therapeutics typically are not orally available. A host of biologic modalities are therefore being developed to address unmet clinical needs. New, often non-natural biologic and hybrid chemi-biologic molecules (e.g. bioconjugates) offer the potential to successfully target a far broader biological space and to modulate targets in different ways. While these new modalities are expected to provide effective therapies against a host of diseases, their chemical diversity will no doubt challenge, sometimes in seemingly insurmountable ways, current paradigms and platforms for downstream process development, scale up and validated implementation. A concern? Yes, but also a unique opportunity to create purification unit operations in concert with new biologic modalities, and to thereby advanced downstream process development platforms amenable to those modalities. In this talk, I will describe progress within the biotech industry, technology development companies, and my laboratory aimed at addressing this opportunity juxtaposed to the current state.

BIOT 75

Synthetic biology: Putting synthesis into biology

Huimin Zhao, [email protected]. 215 Ral Box C-3 Mc 712, Urbana, Illinois, United States

Synthetic biology is the design of novel or improved biological systems using engineering principles. It is a rapidly growing area with broad applications in medical, chemical, food, and agricultural industries and particularly has become a powerful engine for the exponential growth of the biotechnology industry. In this talk, I will briefly discuss the challenges and opportunities in synthetic biology and highlight our recent work in the development and application of novel foundational synthetic biology tools. Specifically, I will introduce the Illinois Biological Foundry for Advanced Biomanufacturing (iBioFAB) that we have been establishing and discuss its potential biotechnological applications. Examples will include development of high throughput genome-editing tools, programmable artificial restriction enzymes, large-scale discovery of novel natural products for drug discovery, and automated cellular engineering of microbial factories for production of chemicals and fuels. BIOT 76

Application of big data to inform prior knowledge and facilitate process validation risk assessment

Stephanie Wong, [email protected], Kimberly Kaleas, Alicia White, Kathlyn Lazzareschi, Josefine Persson. Genentech, Brisbane, California, United States

Process validation studies are performed to support and demonstrate process robustness and understanding. The study results are also used to inform proper process controls to ensure consistent and robust manufacturing production. Although most processes are optimized to meet program-specific needs, some process step conditions have been consistently implemented across multiple programs and demonstrated similar process performance and effects. This extensive history and experience may provide pertinent information to enable science- and risk-based assessment in a process validation strategy. Through effective knowledge management and application of big data concepts, relevant historical validation knowledge can be further structured to compare overall trends, to inform and support associated risk scores for future process validation studies. A case study will be included that utilized prior validation knowledge to support and streamline process knowledge and validation strategy.

BIOT 77

Leveraging prior knowledge to streamline risk assessment tools and process validation study design

Alicia White, [email protected]. Genentech, Brisbane, California, United States

Traditional risk assessments for affinity chromatography process validation study designs rely heavily on product-specific data to identify potential critical process parameters, which result in large validation study designs evaluating multiple parameters that have consistently been observed to not impact product quality. This case study focuses on utilization of patterns revealed by the extraction and consolidation of prior affinity chromatography knowledge to inform and support a streamlined risk assessment for identification of potential critical process parameters. The resulting study design for this monoclonal antibody focused on process parameters that have historically demonstrated an impact on product quality, which saved resources and time while obtaining robust process understanding of the affinity chromatography unit operation.

BIOT 78

Model evolution: Enabling process knowledge continuum and control Jun Ren, [email protected], Saly Romero-Torres, Chao Shi, Rob Guenard. Biogen, RTP, North Carolina, United States

The primary goal of the PAT framework is to promote process understanding/predictability with the goal that it evolves into process intelligence and vertical/horizontal control. Process predictability relies on deep process knowledge. Knowledge can be derived from first principles or by experience. Modeling can enable different levels of process intelligence - prediction and prescription - depending on the model robustness and the available data package. With regards to data analytics, Biogen is adopting modeling maturity models similar to the ones used by high tech industries such as semiconductors, electronics and AI. The focus of this maturity model is ensuring that a plan for model evolution (and evolutionary intended use) is conceived, and socialized, among SMEs and regulatory agencies early on during process development. This plan is crucial particularly when implementing data driven models that rely on process experience - which is not available at early development or even at early commercial manufacturing stages. A well-planned modeling continuum should allow the pharmaceutical industry to realize the benefits from modeling activities early on, while evolving into more mature prescriptive controllers and release by exemption activities. These efforts are in support of supply chain predictability/reliability and advanced process control (from fault detection to model based optimization).

BIOT 79

Unlocking the potential of big data in a biopharmaceutical manufacturing network

Greg Naugle1, [email protected], Paul Stey2. (1) Process Development, Amgen, Inc., West Greenwich, Rhode Island, United States (2) Brown University, Providence, Rhode Island, United States

Biopharmaceutical manufacturing processes represent a data rich environment. The ability to identify and evaluate weak signals and multivariate correlations on a broad scale and in near real time have been difficult historically. Traditional approaches to interrogating these data have been highly manual, product and facility specific, and reactive. This talk will discuss new approaches to harnessing these data at a portfolio level and using innovative methodologies to convert data to knowledge through exploratory analyses and feature engineering.

BIOT 80

Mining process data to improve environmental impact of biomanufacturing

Kristi Budzinski3, [email protected], Megan Blewis5, Philip Dahlin6, Julia Esparza7, John Gavin4, Sa V. Ho2, Clarice Hutchens2, David Kahn8, Stefan G. Koenig1, Robert Kottmeier2, Jeffrey Millard7, Matt Snyder9, Brad Stanard7, Larry Sun2. (1) MS432a, Genentech, Inc., South San Francisco, California, United States (2) Pfizer, Chesterfield, Missouri, United States (3) Genentech, South San Francisco, California, United States (4) Merck, Kenilworth, New Jersey, United States (5) Amgen, Thousand Oaks, California, United States (6) Johnson and Johnson, New Brunswick, New Jersey, United States (7) MedImmune, Gaithersburg, Maryland, United States (8) Lilly, Indianapolis, Indiana, United States (9) CRB, Plymouth Meeting, Pennsylvania, United States

Within the ACS GCI Pharmaceutical Roundtable (GCIPR), member companies adopted Process Mass Intensity (PMI) as a standard green metric over five years ago. PMI provides a quick and simple method to calculate the material intensity of a process where the sum of all input materials in kg is divided by the total amount of API produced in kg. Biopharmaceuticals (or biologics) are a rapidly growing segment of the pharmaceutical market, as such; the environmental footprint of their manufacture is coming under scrutiny. Biologics are manufactured through a fermentation-based process followed by several chromatography purification processes resulting in significant water use and large quantities of raw materials and consumables. To understand the environmental impact of biologics manufacture, the BioPharma Focus Group, a subset of GCIPR member companies, created a biologics-specific PMI metric. The biologics PMI measures the amount (in kg) of water, raw materials (media, buffer, etc) and consumables (filters, membranes, etc) utilized in the manufacture of biologics (kg API). This presentation will cover the initial PMI benchmarking study of the GCIPR which focused on monoclonal antibodies (mAb) from both commercial and clinical scales. The data highlight the material intensity of standard unit operations and the effect of cell culture methods. Additionally, the development of a similar metric focused the impact of cleaning and sanitation methods will be discussed.

BIOT 81

Application of process data analytics to improve commercial scale mAb manufacturing process

Leo Xu, [email protected]. GSK, King of Prussia, Pennsylvania, United States

Due to the complex nature of biological processing, cell culture titer output often varies from batch to batch in large scale commercial manufacturing. While raw material inputs and key process parameters are well controlled, variation in productivity remains to be understood. Recent advances in process informatics and data historian system have made a lot more process and equipment data much more accessible. We conducted a “big data” analysis on the entire cell culture process including all the data that were captured during the manufacturing process. The analysis has revealed some hidden process parameters which associated with the productivity variation. This led us to further improve process understanding and to identify the opportunity for productivity improvement.

BIOT 82 Application of novel multivariate data analysis (MVDA) techniques to understand and troubleshoot variability in cell culture manufacturing

Lakshmi Cella2, [email protected], John S. Bowers1, Paolo Avalle2, Louis Obando2, Ahmed Abdelsalam2, Krithik Thirupapliyur2, Diego Bonnano2, John Easson1, Suzanne Gough1, Robert Leighty1, Vijay Janakiraman1, Anthony Russo4, Alexander G. Tulloch1, Gargi Maheshwari3. (1) Merck, Kenilworth, New Jersey, United States (2) Merck , Piscataway, New Jersey, United States (3) Merck Co, West Point, Pennsylvania, United States

Cell culture manufacturing processes are inherently complex. They are subject to variability arising from many factors, such as raw materials, operational changes and implementation at multiple sites. Understanding and addressing the root causes for these variations becomes essential to avoid disruptions to the product quality and supply. Multivariate data analysis (MVDA) is a powerful tool for analyzing the “big data” from the manufacturing sites. State of the art techniques were developed to convert discrete offline cell culture data into “continuous” offline data by interpolation. Combining these with the continuous bioreactor data provides much better analysis without unduly biasing any variables. Continuous bioreactor data, collected every five minutes, includes pH, DO, Temperature, Gas flows, bioreactor and feed tank weights. Offline cell culture data, measured daily, includes cell density, viability, pH, DO, pCO2, metabolites, amino acids and trace metals. Following case studies will be presented: Comparing pre and post process/operational changes, addressing trends in quality attributes during and between campaigns and understanding root causes for product quality and productivity differences between sites.

BIOT 83

Understanding upstream variability and new data techniques

Victor Saucedo, [email protected]. PD Engineering, Genentech, South San Francisco, California, United States

Titer in commercial manufacturing can vary and affect supply to patients. Titer decrease is known to be caused by lactate increase. However, lactate increase is not completely understood in commercial operations. Classical Upstream operations are complex batch operations and typical univariate analytical methods are not able to explain the process variation. A variety of new machine learning techniques are gaining ground and acceptance in many applications. Here, we reviewed many techniques to understand the root cause of the process variability. We found that a combination of Partial Least Squares (PLS) and Supporting Vector Machines (SVM) point to some potential causes; however, one important aspect of the result is interpreting the results to get the confidence that they are strong biological hypotheses to drive changes and further investigations. This work shows the results and their interpretation.

BIOT 84 MLProScape: Machine learning guided approach for engineering enzymes faster

Sanjan T. Gupta1,2, [email protected], Evan Glasgow3,2, Brian G. Fox3,2, Parameswaran Ramanathan5, Jennifer Reed4,2. (1) Chemical and Biological Engineering, University of Wisconsin Madison, Madison, Wisconsin, United States (2) Great Lakes Bioenergy Research Center, Madison, Wisconsin, United States (3) Biochemistry, University of Wisconsin Madison, Madison, Wisconsin, United States (5) Electrical and Computer Engineering, University of Wisconsin Madison, Madison, Wisconsin, United States

The field of protein engineering primarily aims at improving the functional properties of a protein such as thermo-stability, binding affinity, and/or catalytic activity and has become a vital step in engineering industrial enzymes.

In this work, we have developed a machine learning (ML) based approach called MLProScape to build an accurate model of protein fitness landscape (ProScape) and then, use it to design synthetic protein designs with superior functional properties. Unlike directed evolution based approaches which require screening of millions of variants, our method requires screening of just a handful of variants (in the order of tens to hundreds) owing to the power of statistical inference. MLProScape consists of three major steps - i) numerically encode a protein sequence using amino acid based physio- chemical properties as features ii) build a machine learning based model using a subset of highly informative features iii) identify synthetic designs with improved characteristics.

To demonstrate the success of our proposed workflow, we have attempted to engineer the catalytic activity of glycoside hydrolases (enzymes that can break down cellulosic and hemi-cellulosic polysaccharides) by using experimentally measured specific activity values corresponding to a diverse set of glycoside hydrolases for training the models. The resulting elastic net regression based models have a high predictive power (with correlation coefficient and R2 values as high as 0.896 and 0.714 respectively between predicted and experimentally measured specific activity values) surpassing previously published ML based enzyme engineering studies. Moreover, the use of position specific features helps us identify positions distal to active site that might play a key role in modulating the activity level. Lastly, this ML based workflow is capable of modeling complex design criteria such as optimizing the protein sequence for hydrolyzing multiple substrates simultaneously as well as to account for other desirable traits such as high stability and better in vivo expression levels.

MLProScape: graphical abstract

BIOT 85

Surface display-enabled selection of bioorthogonally stabilized alpha helices using non-natural amino acid incorporation

Tejas A. Navaratna, [email protected], Lydia Atangcho, Andrew Min, Greg M. Thurber. Chemical Engineering, University of Michigan, Ann Arbor, Michigan, United States

Protein-protein interfaces play a central role in biology, and the majority of these interfaces include an alpha helix, the most common protein secondary structure motif. Such interfaces have posed attractive drug targets due to their importance in a wide range of diseases, in which inhibition, by blocking interactions, or activation, by triggering function, can result in therapeutic applications. However, due to the large size of protein-protein interaction surface areas, small molecules have seen little success in drug development against these targets. Here, we describe an application of bacteria surface display using non-natural amino acid incorporation and bio-orthogonal chemistry for directed evolution of potent stabilized peptide binders.

By incorporating azidohomoalanine (AHA) residues spaced 7 residues apart in chemical synthesis, we and others have demonstrated stabilization of peptides using a bifunctional alkyne-containing linker. Such design allows for the inclusion of fluorophores for imaging applications or pharmacologic enhancers like PEG chains. Our lab and others have demonstrated that click-stabilized peptides have increased protease resistance, subcutaneous bioavailability, and binding affinities. Taking advantage of the ability of methionine auxotrophic E. coli to efficiently incorporate AHA as a methionine surrogate, we constructed libraries of randomized peptides containing two AHA residues for on-surface stabilization. We demonstrate efficient display of AHA incorporated peptides using the eCPX scaffold and robust reaction yields and specificity. Following surface display and chemical double-click stabilization, the resulting peptides show increased protease resistance and are available for surface binding and screening.

To show this method’s potential for diverse applications, we screened a stabilized peptide library with a diversity of 4 x 108 sequences based on the original p53-MDM2 binding motif and obtained several unique sequences demonstrating significantly increased affinity over the native analog. Furthermore, we isolated sequences whose binding was enhanced upon stabilization, indicating the effectiveness of chemical stabilization prior to the selection process. We also screened a naïve library against extracellular targets to isolate stabilized peptide binders for imaging applications. This screening approach demonstrates the ability to affinity mature or select novel sequences with stabilized alpha helix peptide display.

BIOT 86

Three ways to design selective peptide inhibitors of anti-apoptotic Bfl-1

Amy E. Keating, [email protected]. Mass Inst of Tech, Cambridge, Massachusetts, United States

Protein-protein interactions among Bcl-2 family proteins regulate apoptosis by controlling the formation of pores in the mitochondria. Overexpression of anti-apoptotic family members including Bcl-2, Bcl-xL, Bcl-w, Mcl-1 and Bfl-1 is implicated in oncogenesis and in the resistance of many cancers to chemotherapy. Anti-apoptotic function of Bcl-2 can be blocked by the small-molecule drug venetoclax, which is used to treat certain leukemias. However, Bcl-2 protein family members that are not inhibited by venetoclax contribute to cancer cell survival. Potent and selective inhibitors of these cell death resistance factors are needed to advance research and allow for diagnosis and treatment of diverse cancers. Designing tight-binding and selective inhibitors of Bcl- 2 family proteins also poses a fundamental, challenging problem in peptide engineering and a good opportunity to study principles of protein-peptide recognition. We have applied methods that integrate computational structure-based modeling with high- throughput screening to develop peptide-based inhibitors of Bcl-xL, Mcl-1, Bfl-1 and viral Bcl-2 family proteins. In this talk, I will present three distinct protein engineering approaches that we have recently applied to design high-affinity and paralog-selective peptide inhibitors of Bfl-1. I will highlight some of what we have learned about the peptide-binding landscape of the Bcl-2 family.

BIOT 87

Direct measurement of deubiquitinating enzyme activity in intact cells using a protease-resistant, cell permeable peptide-based reporter

Nora Safabakhsh1, [email protected], Jacob Pettigrew2, Ted Gauthier3, Adam T. Melvin1. (1) Chemical Engineering, Louisiana State University, BATON ROUGE, Louisiana, United States (3) Ag Center Biotechnology Lab, Louisiana State University, BATON ROUGE, Louisiana, United States

Tremendous attention has shifted towards implementation of molecularly-targeted therapeutics in cancer treatment. In the case of multiple myeloma, the proteasome inhibitors Bortezomib and Carfilzomib have gained FDA approval; however, patients develop a resistance to these drugs. Recent studies have identified that inhibition of deubiquitinating enzymes (DUBs) have led to increased cancer cell death in bortezomib-resistance cell lines. However, it is still unclear how the inhibition of both DUBs and the proteasome can lead to increased patient survival. To address this question, we developed a reporter of DUB activity that is cell permeable, long-lived, fluorescent, and specific to DUBs. The reporter consists of an N-terminal β-hairpin motif that acts as both a ‘protectide’ to increase intracellular stability and a cell penetrating peptide (CPP) to facilitate the uptake into intact cells. The β-hairpin was conjugated onto the N-terminus of a substrate consisting of the last four amino acids in ubiquitin to facilitate DUB mediated cleavage and a fluorophore AFC. The reporter was characterized in both HeLa and OPM2 cell lysates and intact cells. Reporter performance was compared to an unprotected DUB reporter (Z-LRGG-AMC). The calculated kinetic rates in cell lysates were fitted to the Michaelis-Menten equation by non-linear regression to determine the kinetic rate constants. The Vmax/Km in OPM2 lysates was 0.602*10-3 for RW-LRGG-AFC and 1.378*10-3 for Z-LRGG-AMC. Control experiments with PR-619 DUB inhibitor demonstrated the specificity of the reporter to DUBs. The reporter was internalized by intact cells at concentrations from 200 to 1400 µM as assessed by fluorometry and microscopy. Results demonstrated a rapid uptake of the reporter in intact cells with the enzyme-substrate reaction rates becoming saturated after 5 hours using 1mM peptide. Intracellular peptide stability was assessed using a degradation assay in OPM2 lysates followed by RP-HPLC analysis. Although the kinetics of the Z-LRGG-AMC peptide were slightly better in cell lysates, it was demonstrated that this peptide is not cell permeable, and therefore not compatible with high-throughput screening methods. The novel DUB reporter presented here is cell permeable with rapid uptake kinetics, and superior in terms of intracellular stability and DUB sensitivity. These qualities make it ideal for performing dynamic high-throughput measurements of DUB activity in intact single cells.

BIOT 88

Proteolytic silencing of diverse mammalian protein targets using a functionally remodeled bacterial E3 ubiquitin ligase

Morgan R. Baltz1, [email protected], Erin A. Stephens2, Matthew P. DeLisa1,2. (1) Robert Frederick Smith School of Chemical and Biomolecular Engineering, Cornell University, Ithaca, New York, United States (2) Graduate Field of Biochemistry, Molecular, and Cell Biology, Cornell University, Ithaca, New York, United States

Protein-level silencing technologies hold promise both as a research tool for exploring protein function and as therapeutics for accelerating the turnover of disease-related proteins. To this end, we have previously demonstrated the efficacy of “ubiquibodies”, engineered protein chimeras that combine the activity of E3 ubiquitin ligases with the target specificity of designer binding proteins (DBPs) to selectively and controllably steer target proteins to the ubiquitin proteasome pathway for degradation. Importantly, ubiquibodies are modular - simply swapping DBPs can generate a new ubiquibody with specificity for a different substrate protein. Unfortunately, pre-existing DBPs (scFvs, FN3s, DARPins, etc.) do not exist for many desired cellular proteins. To address this limitation, we have developed a universal ubiquibody based on a bacterial E3 ubiquitin ligase that specifically targets GFP or virtually any mammalian target protein fused to GFP. The resulting ubiquibody exhibits extraordinary silencing efficiency against a wide array of GFP fusion proteins that vary in size, subcellular localization, and method of expression (i.e., transient vs. stable). Because numerous fluorescent protein fusions already exist or can be rapidly created, our GFP-specific ubiquibody is poised to enable proteolytic silencing on a proteome-wide scale without the need for further engineering or optimization.

BIOT 89

Implications of protein sequence variants derived from systematic starvation

Ed Wong, [email protected], Chung-Jr Huang, Zhongqi Zhang. Amgen, Thousand Oaks, California, United States

Many metabolic and biosynthetic pathways can influence protein product attributes during translation and post-translational modification (i.e. sequence variation, glycosylation, oxidation, proteolysis, etc.). For both recombinant and native protein biosynthesis, the resulting protein product attributes are analogous to a historical record of cellular events. Of particular interest are sequence variants caused by amino acid misincorporation. Believed to be universal, amino acid misincorporation is the inherent errant replacement of an amino acid by another. In biological systems, high-level amino acid misincorporation can cause loss of proteostasis and disease through the compounded accumulation of errors. And in biopharmaceutical applications, since protein product function is derived from the product’s attributes, elevated sequence variant frequency has been shown to result in adverse effects such as product instability, structure change, aggregation, and the potential for immunogenicity. Amino acid starvation, fast cell growth, overproduction, and oxidative stress, which are physiological conditions that can be observed in disease states and during biopharmaceuticals production, can all significantly exacerbate mistranslation by orders of magnitude. In an effort to elucidate the major sequence variant mechanisms, to our best knowledge, our work is the first investigation of amino acid misincorporation through systematic amino acid starvation. Our results reveal that one-base mismatches during codon recognition and tRNA misacylation are both common and major mechanisms. Also, different cell lines share common misincorporation patterns suggesting that a conserved set of mechanisms and error propensities exists.

BIOT 90

Design of protein glycosylation sites by high-throughput expression and characterization of polypeptide glycosyltransferases Weston Kightlinger1, [email protected], Liang Lin2, Madisen Rosztoczy2, Matthew P. DeLisa3, Milan Mrksich2, Michael C. Jewett1. (1) Chemical and Biological Engineering, Northwestern University, Chicago, Illinois, United States (2) Biomedical Engineering, Northwestern University , Evanston, Illinois, United States (3) Robert Frederick Smith School of Chemical and Biomolecular Engineering, Cornell University, Ithaca, New York, United States

Protein glycosylation, the post-translational attachment of complex sugar moieties (glycans), plays a crucial role in human health and is a critical quality attribute for protein therapeutic and vaccine production. However, our ability to rationally design and engineer glycoproteins with desired properties remains limited by a lack of high- throughput methods for synthesis and biochemical characterization of glycosyltransferases (GTs), the enzymes that attach glycans to proteins and then elaborate them. New tools are required to rapidly express polypeptide modifying GTs and determine design rules for engineering protein glycosylation sites which can be efficiently and specifically modified with a glycan of interest.

Here we describe a systematic platform for glycosylation sequence characterization and optimization by rapid expression and screening (GlycoSCORES) using cell-free protein synthesis and self-assembled monolayers for matrix-assisted laser desorption/ionization mass spectrometry. We synthesized six N- and O- linked polypeptide-modifying GTs from bacteria and humans and determined their sugar and peptide acceptor specificities at unprecedented depth and throughput using a total of 3,480 unique peptides and 13,903 unique reaction conditions. We then used cytoplasmic N-glycosyltransferase from Actinobacillus pleuropnemoniae (NGT) to demonstrate the use of GlycoSCORES to optimize and design polypeptide acceptor sequences in whole proteins. Optimization of NGT peptide substrates informed the design of small sequence motifs (GlycTags) that successfully directed efficient glycosylation onto the internal loops of three heterologous proteins, including a human immunogloblin (IgG1) Fc domain, in vitro and in the cytoplasm of living E. coli. We found peptide glycosylation levels accurately predict modification of whole proteins, enabling a 3-5 fold greater glycosylation efficiency for GlycTags compared to wild-type glycosylation sequences.

This work provides a broadly applicable and systematic approach to facilitate the rational design of synthetic glycoproteins in bacteria and in vitro for compelling applications in glycoprotein therapeutics, vaccines, and diagnostics.

BIOT 91

Microbiome-virome interactions in bovine rumen: Elucidating metabolic transactions through metabolic modeling

Mohammad Islam1, Wheaton Schroeder2, Samodha Fernando3, Rajib Saha4, [email protected]. (1) Chemical and Biomolecular Engineering, University of Nebraska- Lincoln, Lincoln, Nebraska, United States (3) Animal Science, University of Nebraska- Lincoln, Lincoln, Nebraska, United States

The microbial ecosystem of the rumen has a significant impact on the ruminant health and productivity through breakdown and fermentation of feed into volatile fatty acids. It is well established that bovine dietary variations significantly alter the ruminal microbiome. In addition to the syntrophic, mutualistic, and competitive microbial interactions, viruses impact microbial populations through cell lysis and reprogramming of host metabolism by Auxiliary Metabolic genes (AMGs). However, a comprehensive study to decipher the effect of diet utilization and viral AMGs on the microbiome is still missing. In this work, we investigated rumen microbial and viral community structure and function in varied dietary treatments and the ecological drivers of rumen viruses alongside their bacterial counterparts.

We constructed a simplified rumen community metabolic model, with Prevotella ruminicola, Methanobrevibacter gottschalkii, and Ruminococcus flavefaciens as representative organisms for starch and protein digestion, methane production and fiber digestion, respectively. The genome-scale models for individual organisms were manually curated and integrated into a community model using multi-level optimization frameworks. The community model was used to estimate metabolite secretion profiles and community compositions as a function of diet and host-specific variations. The identified functions of viral AMGs were incorporated into the model as regulatory information. The metabolic hubs and bottlenecks in the community were identified and the metabolite pools for important energy currencies in the community were predicted. Our results show that vi have affected the core rumen microbiome structure and metabolism. Also, the virome dynamics is highly correlated with microbial density and dietary factors. Our model serves to answer key ecological questions of ruminant nutrition through diet-virome-microbiome interactions, discover unidentified metabolite transactions, and promises to develop novel strategies for methane mitigation and increasing nutritional efficiency of domesticated bovine species.

BIOT 92

Evaluation of a cyanobacterial consortium for efficient polysaccharide production

Kunal Poorey2, [email protected], Ryan W. Davis1, Oliver Killian1, Caleb Talbot1, David Smernoff3, Rocco Mancenelli3, Eric Sundstrom4. (1) Biomass Science and Conversion Tech, Sandia National Laboratories, Livermore, California, United States (2) Systems Biology, Sandia National Labs, Livermore, California, United States (3) HelioBioSys, Inc, Woodside, California, United States (4) Advanced Biofuels Process Demonstration Unit, Lawrence Berkeley Laboratory, Emeryville, California, United States

Development of polycultures has been identified as a potential means for overcoming several challenges facing scale-up of algae-based commodities. In this presentation, we will describe findings from our recent studies on the cultivation of a marine cyanobacterial consortium in 1000L algae raceways. In the designed consortium, three distinct cyanobacterial cultivars were combined to provide nitrogen fixation, photoprotection, and high rates of secretion of extracellular polysaccharides in support of a long-term bioproduct ‘milking’ strategy. Each of the cohorts of the cyanobacterial consortium was cultivated individually at lab-scale for assessment of growth rate and other phenotypic variables as a function of light intensity and nitrate concentration in order to identify optimal algae raceway inoculation and maintenance strategies. Following the lab-scale investigations, the consortia were successfully cultivated in pilot- scale algae raceways for >65 days, achieving polysaccharide concentrations of 500-600 mg/L, consisting of a variety of C5 and C6 monosugars. In addition to the remarkable stability of the consortium in open cultivation, measurements of culture density time course indicated insignificantly different log-phase specific growth rates at different levels of nitrate addition, which should have significant techno-economic and sustainability impacts for commercialization. Finally, metagenomic sequencing analysis was performed to characterize species flux within the community, including identification of unintentionally introduced species. Analysis of the metagenomic sequencing data was performed using a newly developed software pipeline, “Metagenomics and Amplicon sequencing pipeline” (MAGPie), which identified the bacterial and eukaryotic cohorts at various time points in the raceway cultures, which corresponded to ~10,000 operational taxonomic units (OTUs) for characterization of the culture microbial ecology. Combining the metagenomic analysis with relevant metadata, including environmental conditions, product yield, and product properties allowed downstream machine learning methods to find patterns of optimal performance. Furthermore, the metagenomics dataset provides recognition of the emergence of deleterious and competing species so that countermeasures can be deployed to prevent culture failure.

BIOT 93

Engineering modular microbial communities for cellulose utilization and bioproduct synthesis

Karolina Z. Kalbarczyk, [email protected], Mattheos Koffas, Cynthia H. Collins. Chemical and Biological Engineering, Rensselaer Polytechnic Institute, Troy, New York, United States

Recent environmental concerns have increased the need to develop systems that can convert biological waste materials to chemicals. Cellulosic biomass can be degraded into sugar monomers and serve as an inexpensive carbon source for microbial chemical production systems. Current approaches for biomass utilization rely on pretreatment processes that include expensive enzymatic degradation steps. Streamlining the cellulose degradation and utilization processes into one system would eliminate the need for costly enzyme purification. To address this challenge, microbial communities can be engineered to synergistically perform complex tasks, dividing the demands among specialized units to improve the efficiency of product formation. Therefore, we aim to build microbial systems with interchangeable modules, with the first module dedicated to cellulose degradation and the second module specialized for bioproduct synthesis. To engineer synthetic communities for the production of biofuels and pharmaceuticals, we are building and testing mixed cultures of Bacillus megaterium and Escherichia coli strains. We will describe our recent progress engineering these two modules and assembling them for activity. Complete degradation of cellulose requires a combination of three cellulase activities – an endoglucanase, a cellobiohydrolase, and a beta-glucosidase. We have developed a cellulose degradation module consisting of Bacillus megaterium strains engineered to express and secrete cellulases. A small library of signal peptides (SPs) was selected to tag each cellulase for secretion from B. megaterium. Cellulase activity against amorphous cellulose was confirmed, and the signal peptide constructs that led to highest levels of secretion and activity were identified for each cellulase. The optimized cellulase secretion strains were characterized in monoculture and coculture, demonstrating synergistic cellulolytic activity. We have further verified that the secreted cellulases can support growth of an E. coli production module via sugars released from cellulose degradation. We will discuss the effect of inoculation ratio, expression conditions, and community compositions on the assembly of a coculture system utilizing cellulose to produce violacein, an antimicrobial pharmaceutical product with anti-tumor properties. The cellulose degradation strains will be further integrated with alternate bioproduct synthesis modules to build new cellulose-based systems.

BIOT 94 Sustainable production of industrially relevant biomonomers: A photosynthetic microbial consortia approach

David N. Carruthers1, [email protected], E. Neil G. Marsh2, Xiaoxia (Nina) Lin1. (1) Chemical Engineering Department, University of Michigan, Ann Arbor, Michigan, United States (2) Chemistry, University of Michigan, Ann Arbor, Michigan, United States

Engineering of synthetic microbial consortia has emerged as a new and powerful biotechnology platform. To date, most microbial consortia have focused on biofuel development, though they also have enormous potential in the production of biobased commodity chemicals. This proposal describes a tripartite system in which three microbes of differentiated specializations can convert sunlight, carbon dioxide, and atmospheric nitrogen into chemical precursors for bulk polymer production. This framework offers a novel opportunity for biobased plastic production without energetically or monetarily expensive nutrient inputs, subsequently providing an attractive, sustainable alternative to fossil fuel analogues. Specifically, Azotobacter vinelandii, a nitrogen-fixing bacterium that secretes ammonia, and Synechococcus elongatus, a photosynthetic cyanobacterium that secretes sucrose form a symbiotic chassis hypothesized to support a third producer strain. Escherichia coli and Corynebacterium glutamicum are two archetypal bacterial species previously modified to produce an enormous array of chemicals. Given that both species can naturally grow on ammonia and sucrose, this will facilitate a tailored tripartite system with drop-in target production. These organisms have demonstrated production of chemical biopolymers well beyond polyhydroxyalkanoates (PHAs) and polylactic acid (PLA), including varied titers of amino acids, which serve as biomonomers for downstream chemocatalytic bioplastic production. This work bridges a fundamental gap between commercial biofermentation and community engineering, potentially alleviating energetic and economic constraints barring market entry of industrial biopolymer development. The consortium will be studied and optimized for the production of industrially relevant biomonomers as illustrated in Figure 1. This extrapolates our current computational and experimental investigation to establish a viable tripartite system for microbial growth. The baseline model will then be used with producer strains for fermentation of amino acid precursors such as L-glutamic acid and L-lysine. Precursors can be readily extracted for chemocatalytic or semisynthetic conversion to biopolymers. Results of this work will be interpreted through a life cycle assessment followed by a techno-economic analysis, which weighs emissions against nonrenewable energy usage and estimates of overall cost at scale, respectively.

BIOT 95

Polysaccharide chemical structure controls on fermentation by gut microbiota, metabolic outputs, and interspecies interactions

Yunus Tuncil, Ming-Hsu Chen, Tianming Yao, Steve Lindemann, [email protected]. Food Science, Purdue University, West Lafayette, Indiana, United States Because diversity in microbial communities is important to the functioning and stability of many microbial ecosystems, such as animal and plant microbiomes, mechanisms to maintain diversity in these environments are important. This is especially true in the human gut microbiome, as low diversity of gut microbiota is linked to multiple acute and chronic disease states, such as metabolic syndrome, type 2 diabetes, inflammatory bowel diseases, and colorectal cancer. High consumption of dietary polysaccharides that are not degraded by human enzymes (termed “dietary fibers”) but are fermented by gut microbes (termed “microbially-available carbohydrates”) is known to sustain gut microbiome diversity. However, it remains unclear to what degree generalizable features of polysaccharide chemical structure influences the composition of the microbial community consuming it and, in turn, its metabolic fate. Here we tested the hypotheses 1) that very fine structural differences govern the outcome of microbial polysaccharide fermentation, and 2) that increasing polysaccharide chemical complexity (e.g. increased molecular weight or degree of branching) can sustain greater microbial diversity. We test these hypotheses by performing in vitro fermentation of polysaccharides with varying structure as sole carbon substrates by human fecal microbiota and measuring alterations in microbial community structure (via 16S amplicon sequencing) and metabolic outcome (e.g. short-chain fatty acids). We found that 1) very fine differences in the structure of arabinoxylans from different wheat cultivars resulted in divergent fermentation outcomes, both in microbial composition and metabolic fate, and 2) that chemical complexity manifests significant differences in microbial ecology over multiple generations. These data suggest that the effects of polysaccharide structural properties may be generalizable across specific polysaccharide structures, intimating the possibility of rational polysaccharide structure design to modulate gut microbiome diversity and function.

BIOT 96

13C metabolic flux analysis in microbial communities: An integrated multi-scale modeling approach

Maciek R. Antoniewicz, [email protected]. University of Delaware, Newark, Delaware, United States

Syntrophy, or cross-feeding, is the co-existence of two or more microbes whereby one feeds off the products of the other. Recently, we have developed an integrated multi- scale modeling approach that allows, for the first time, to dissect interactions in microbial communities using 13C tracers. Specifically, to quantify metabolism and identify cross-feeding interactions we have developed a compartmental multi-scale 13C metabolic flux analysis approach that quantifies metabolic fluxes for multiple cell populations in microbial communities without physical separation of cells or proteins. As a specific example, in this presentation, I will illustrate our investigations of metabolic interactions between several E. coli mutants that are unable to grow on glucose in minimal medium by themselves, but can grow on glucose when cultured together. Using our novel flux analysis tools, we have quantified metabolic interactions (i.e. metabolite cross-feeding) in four distinct synthetic E. coli co-cultures. We also applied adaptive laboratory evolution to elucidate how syntrophic interactions evolve and are strengthened through adaptive co-evolution. Overall, the methods we have developed for studying microbial communities enable a new area of investigations, allowing us and others to dissect complex microbial systems that are of significant importance in biology but cannot be investigated with current tools. More broadly, by better understanding syntrophic relationships at the genetic, molecular, cellular and systems levels we are generating new knowledge on microbial syntrophy that enables us to ensemble synergistic interactions in engineered microbial communities for novel industrial biotechnology applications.

BIOT 97

Electro-active communities for hydrogen production from biomass streams

Alex Lewis2, [email protected], Abhijeet Borole1. (1) Bldg 4505 MS 6226, Oak Ridge Natl Lab, Oak Ridge, Tennessee, United States (2) The University of Tennessee, Knoxville, Knoxville, Tennessee, United States

Microbial communities have great potential as an alternative biocatalyst for conversion of lignocellulosic streams. Microbial communities have evolved to collectively carry out these functions for biomass-derived streams and this functionality can be deployed in bioelectrochemical systems for conversion of biorefinery streams into hydrogen and other products. In the present study, an enriched high-performing community fed with a pyrolysis-derived aqueous phase was analyzed via metagenomics and transcriptomics to characterize the genetic potential of the community for biomass degradation, as well as uncover the active players involved in linking degradation to electron transfer to elucidate functional roles within the community and ecological interactions that drive high performance. An optimized assembly process resulted in the generation of 21 high- quality genomes form the microbial community.Two members of the Deltaproteobacteria and Clostridium families were found to be the dominant strains in the community, making up nearly 50%. The former was demonstrated to possess novel exoelectrogenic behavior for its taxonomy, containing c-type cytochromes such as OmcZ, as well as pilA sequences similar to Geobacter sp. The Clostridium strain contained a substantial number of biomass-degrading genes, with a total of 425 CDS for CAZy related genes. The large amount of Firmicutes in the population totaling 42.5%, and the community as whole containing CAZy CDs spread across different roles in lignocellulosic degradation highlighting selective forces of biomass-derived streams, requiring robust fermenters and cellulolytic microbes to convert the biomass to electrons, which the community demonstrated with high efficiency.

BIOT 98

Metabolic modeling of microbial communities

Siu Hung Joshua Chan, [email protected], Costas Maranas. Chemical Engineering, Pennsylvania State University, University Park, Pennsylvania, United States

Genome-scale metabolic modeling has become widespread for analyzing microbial metabolism. Extending this paradigm to more complex microbial communities is emerging as a promising way to unravel the interactions and biochemical repertoire of these omnipresent systems. Here we present improvements in community metabolic modeling in two aspects. First, we introduce the SteadyCom optimization framework for predicting metabolic flux distributions consistent with the population steady-state requirement originating from ecological considerations. In the absence of this requirement, the faster growing organism will ultimately displace all other microbes in the community. This is particularly important for predicting steady-state microbiota composition as it imposes significant restrictions on the allowable community membership, composition and phenotypes. A significant advantage of SteadyCom is compatibility with flux variability analysis. It is first demonstrated for a community of four E. coli double auxotrophic mutants and is then applied to a nine-species gut microbiota model. By randomizing the uptake rates of microbes, an abundance profile with a good agreement to experimental gut microbiota is inferred. Second, we introduce a systematic procedure for standardizing the biomass reactions in genome-scale metabolic models, which are critical for accurate quantitative predictions especially in community models. The biomass has by definition a molecular weight of 1 g mmol-1 but we identified significant departures after examining 64 published models. The biomass weights of 42 models differed by 5 – 100%. Flux balance analysis revealed >10% differences in growth yields for 12 of the 20 curated models using the original vs. corrected biomass reactions. The discrepancies are accentuated in community simulations. Microbes with underestimated biomass weights are overpredicted whereas microbes with overestimated biomass weights are underpredicted. The observed departures in community composition are disproportionately larger than the discrepancies in the biomass weight estimate. We propose the correction procedure as a standard practice for metabolic reconstructions especially for community modeling.

BIOT 99

Investigation into the thermodynamics of mAb: Ligand interactions in MM chromatography

Ronak B. Gudhka1, [email protected], David J. Roush2, Steven M. Cramer1. (1) Chemical and Biological Engineering, Rensselaer Polytechnic Institute, Troy, New York, United States (2) Process Development and Engineering, Merck, Sharp and Dohme, Colts Neck, New Jersey, United States

Multimodal (MM) Chromatography offers highly selective protein separation systems by taking advantage of hydrophobic, electrostatic and Van der Waals interactions. Due to the complexity of such protein-ligand interactions, there is a need for the molecular level understanding of the nature of proteins binding to different MM chromatographic systems. In addition, it is important to understand the thermodynamics of these interactions, particularly the relative contributions of various entropic and enthalpic components under different conditions. Chromatographic retention data has demonstrated unique selectivity trends between two monoclonal antibodies (mAbs) on various MM CEX resins. In particular, we observed a selectivity reversal between Capto MMC and Nuvia cPrime, which differ in the presentation of functional groups on the chromatographic surface. The objective of this work is to obtain fundamental insights into thermodynamic driving forces involved in the interactions of these mAbs with MM ligands. Isocratic chromatography experiments were carried out with the two mAbs on Capto MMC and Nuvia cPrime at different temperatures and salt concentrations. Interestingly, mAb retention decreased with an increase in temperature, phenomena generally not observed in hydrophobic interaction chromatography. The chromatographic retention data was then evaluated using non-linear Van’t Hoff model to determine the changes associated with enthalpy, entropy, heat capacity and free energy upon adsorption. The governing thermodynamic driving force was observed to shift from enthalpic to entropic at various salt concentrations and temperatures. The isocratic data was also fit to the Melander model to obtain insights into relative electrostatic and hydrophobic contributions towards binding at different temperatures. Further, coarse- grained simulations were used to provide insights into the molecular nature of these interactions. The results in this presentation demonstrate how thermodynamic analysis of chromatographic data can be used in concert with molecular dynamic simulations to more deeply understand the connection between unique protein surface properties and the interactions of proteins to various multimodal surfaces.

BIOT 100

Mechanistic modeling of multi-modal chromatography: Separation of antibody variants using mixed-mode weak cation exchange resins

Felix Wittkopp1, Felix Seelinger1, Romas Skudas2, Michael Schulte2, Christian Dr. Frech1, [email protected]. (1) University of Applied Sciences, Mannheim, Germany (2) Merck KGaA, Darmstadt, Germany

Protein heterogeneity which is associated with mAbs includes charge variants like acidic/basic variants, glycosylation variants, size variants like aggregates, truncated monomers and fragments. Such product variants lead to diverse pharmacokinetics and pharmacodynamics which may affect the stability, efficacy, and potency of the drug. Therefore, they have to be thoroughly profiled and partially removed from the final product. Ion exchange and multi-modal chromatography are widely used methods for the removal of these contaminants. In multi-modal chromatography, often a combination of ionic and hydrophobic functionalities is used to form the ligand. As a consequence, mobile phase pH and additives e.g. salt greatly influence protein binding and elution as well as resolution and recovery rates.

In this presentation, mechanistic modeling studies for mAb charge and size variant separation on prototype mixed mode weak CEX/HIC stationary phases with different type, ligand density and degree of hydrophobicity of functional groups are compared. Two models based on the Donnan equilibrium (DIX) or the Steric Mass Action (SMA) principle were used to predict the elution behavior of these mAb variants. In combination with a simple lumped-rate chromatography model, simulation of the separation of these closely related protein species is possible. Robustness studies for mobile phase variations (pH, salt, flow rate) demonstrate the applicability of the mechanistic models.

BIOT 101

Case of the disappearing protein peak: Gradient elution behavior in HIC with U- shaped retention factor curves

Arch Creasy1, [email protected], Joseph Lomino3, Gregory Barker3, Giorgio Carta2. (1) Chemical Engineering, University of Virginia, Charlottesville, Virginia, United States (2) University of Virginia Che, Charlottesville, Virginia, United States (3) Bristol-Myers Squibb, Hopewell, New Jersey, United States The retention of proteins and other biopolymers in hydrophobic interaction chromatography is described by the solvophobic theory, which predicts that retention is inversely correlated with protein solubility as a function of the kosmostropic salt molalilty. In general, an increase in salt concentration drives protein partitioning to the hydrophobic surface while a decrease reduces it. In some cases, however, protein retention can dramatically increase at very low salt concentrations resulting in a U- shaped retention factor curve. At very low salt concentrations, binding can become very strong and proteins may ultimately unfold and irreversibly bind to the resin for these conditions. During gradient elution, following protein loading at a high kosmostrope concentration where the protein is bound relatively strongly, the kosmostrope concentration is gradually decreased to zero thereby reducing the retention factor and increasing the protein chromatographic velocity. We show that for these conditions, the normalized gradient slope determines whether the protein elutes in the gradient, partially elutes, or is essentially never eluted becoming trapped in the column for conditions where the retention factor increases as the kosmotrope concentration decreases. A local equilibrium analysis of this newly discovered chromatographic behavior is presented along with predictions based on the numerical solution of the general rate model of chromatography. Experimental results are also presented for three different monoclonal antibodies and for lysozyme on Capto Phenyl (High Sub) resin. Two of the mAbs and lysozyme exhibit U-shaped retention factor curves and for each of them, we determine the critical gradient slope beyond where 100% recovery is no longer possible. The remaining mAb does not exhibit this behavior and elutes at virtually any gradient slope. This behavior has important implications for designing gradient elutions since the slope of the gradient directly impacts the recovery of the mAb and the fraction of protein left behind which can eventually foul the column.

BIOT 102

Modified batch isotherm determination method for mechanistic model calibration

Tobias Hahn1, [email protected], Teresa C. Beck1, Thiemo Huuk1, Juergen Hubbuch2. (1) GoSilico GmbH, Karlsruhe, XX, Germany (2) KIT, Karlsruhe, Germany

The fundamental assumption of in-silico scale-up and scale-down of chromatography is that only the fluid dynamics outside the pore system change. Once a molecule enters the pore system, diffusion, adsorption, and desorption are following the same mechanism from a 96-well filter plate up to a production scale column. To obtain adsorption isotherm parameters for column modeling, batch measurements are typically corrected using an “equivalent column volume” factor. In a per-well capacity study of filter plates prepared with a ResiQuot device, considerable well-to-well differences could be found and, most importantly, deviations from the expected equivalent column volume result in wrong predictions of column experiments. To solve this, we present a modified computational method for fitting batch isotherms to mechanistic model equations that relies only on the applied and measured supernatant concentrations. An assumption on the resin amount in the well is not needed anymore. To this, the isotherm equation is reformulated to include the liquid-to-solid ratio (L/S) as model parameter. Using this method, the average L/S in a 96-well plate filled with SP Sepharose FF could be determined from a single isotherm at constant ionic strength. The resulting binding capacity coincides with the average per well measurements, and the isotherm parameters could be used to predict a break-through curve on a 16ml lab- scale column.

BIOT 103

In-depth evaluation of next generation protein A stationary phases

Timothy M. Pabst, [email protected], Johnny Thai, Alan K. Hunter. Purification Process Sciences, MedImmune, Gaithersburg, Maryland, United States

As therapeutic use of monoclonal antibodies and related molecules continues to grow, Protein A chromatography remains the primary mode of purification due to high specificity, platformability, and strong regulatory track record. But as the biopharmaceutical industry faces pressures from biosimilars and demands for lower costs, process developers continue to look for opportunities to improve productivity and lower cost of goods. Some companies are moving towards a continuous processing paradigm, which can increase productivity as a result of parallel processing and high utilization of stationary phase binding capacity. Other organizations may prefer to leverage an existing batch processing infrastructure. Regardless of which strategy is employed, significant productivity improvements can be gained by increasing the binding capacity and utilization of the Protein A stationary phase. In this work, we provide an in-depth evaluation of the next generation of Protein A stationary phases, many of which offer high static and dynamic binding capacities coupled with excellent impurity clearance. Other properties investigated include mass transfer rates, particle size and morphology, and pressure flow relationship.

BIOT 104

PEGylating protein A chromatography media increases selectivity and robustness

Justin B. Weinberg1,2, Todd M. Przybycien2, [email protected]. (1) 101 Edu, Inc., New York, New York, United States (2) Chemical Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania, United States

We explored the hypothesis that covalently attaching poly(ethylene glycol), or PEGylating, protein A chromatography media could increase its selectivity and its robustness by reducing non-specific protein-protein interactions via an excluded volume effect. We modified a commercial protein A (ProA) chromatography resin with 5 and 20 kDa aldehyde-activated PEG and characterized antibody static and binding capacity, mass transport behavior, selectivity and resistance to proteolytic attack as a function of the volume of immobilized PEG. Antibody binding capacity decreased roughly linearly with the volume of immobilized PEG; decreases in static capacity were attributed to binding site blockage while decreases in dynamic binding capacity reflected decreased pore diffusivity and slower binding kinetics for antibodies in addition to the loss in binding sites. Digestion of the PEGylated media with chymotrypsin showed biphasic binding capacity loss kinetics and that the PEGylation chemistry used provides a significant resistance to proteolytic attack, thereby increasing robustness relative to unmodified media. PEGylated media had increased resistance to non-specific protein binding relative to unmodified media, including resistance to “hitchhiker” proteins with a tendency to bind already bound antibody. Challenge of our most heavily modified protein A media with Chinese hamster ovary harvested cell culture fluids gave a 37% increase in antibody selectivity with respect to host cell proteins relative to unmodified media when using the same, standard load, wash, elute, regenerate protocol. This study suggests that the excluded volume effects can be exploited to improve the performance of affinity chromatography media based on immobilized macromolecular ligands.

BIOT 105

Increasing productivity of downstream processes; the advantage of antibody based selectivity in the purification of biologics

Pim Hermans, [email protected]. R&D, BAC BV - Thermo Fisher Scientific, Leiden, Netherlands

Advances in biotherapeutics are generating an increasing range of complex molecules that present unique and often difficult purification challenges. By taking advantage of antibody based selectivity, Camelid heavy-chain antibody fragments (VHHs) have proven to be a reliable immuno affinity chromatography (IAC) solution in the downstream process of biologics. Their unique characteristics, such as low molecular weight, favorable physical-chemical properties, solubility and high expressibility in yeast have enabled the successful introduction of affinity matrices that reveal protein A like features with regard to HCP -, DNA - and viral clearance. The strength and robustness of this antibody scaffold technology has been demonstrated by, for example, purification of adeno-associated viruses (AAVs) in a single-step, by showing selectivity for conformational epitopes or the biologically most active form of a target protein. The succesful development of such highly selective affinity ligands is based on a HTS platform where chromatographic behaviour of ligands is simulated, binding affinity profiles are determined and elution buffer compositions are tested. Ligand candidates meeting the pre-defined criteria are selected for generating prototype affinity matrices and screened under conditions that represent the final manufacturing process. Here we describe the advantage of antibody based selectivty in addressing protein purification challenges by providing purification tools for even the most demanding biotherapeutics.

BIOT 106

Eshmuno® P anti-A and anti-B: Affinity resins for removal of anti-A and anti-B isoagglutinins

Santosh Rahane, [email protected], Matthew Turiano, Matthew Stone, Lloyd Gottlieb. MilliporeSigma, Bedford, Massachusetts, United States

Therapeutics purified from pooled blood plasma contain a wide variety of immunoglobulins (Ig) including the blood typing, anti-A and anti-B Ig, that are responsible for adverse hemolysis reactions. A set of resins, called Eshmuno® P anti-A and Eshmuno® P anti-B, have been developed to specifically target removal of anti-A and anti-B isoagglutinin antibodies from plasma. The resins bind the anti-A/B Ig while allowing the target Ig to flow through. The resins typically show higher than 85% removal of these antibodies from a pool having typical isoagglutinin titers (32/64) at challenging loading levels of desired IgG products. Several loading conditions such as pH, ionic strength were found to have an impact on the removal of anti-A and anti-B isoagglutinin antibodies; however, the removal performance was found to be satisfactory for a wide range of pH (4 – 9) and ionic strength values. Typically, the final product was recovered at yields more than 98% in all of the cases investigated. Additionally, the proprietary chemistries of the base bead synthesis and ligand coupling used for manufacturing of the Eshmuno® P products impart exceptional stability at extreme pH conditions typically used for cleaning and sanitization for over 200 cycles, where 1 cycle corresponds to 1 hour of contact time. Stability was observed under both the acidic and basic cleaning/sanitization solutions such as caustic (0.5 M NaOH) as well as benzyl alcohol containing acidic solutions (for example, PAB). Notably, such stability should allow for the Eshmuno® P resins to be used for a large number of cycles, thereby, improving the effective efficiency and cost of use. The foregoing observations will be discussed in detail. Additionally, a method was established that is independent of anti-A and anti-B titers in a given pool to consistently evaluate the performance of the Eshmuno® P resins. The relationship between the output of this method, and anti-A and anti-B isoagglutinin removal capacity will be discussed.

BIOT 107 Technical capability of Raman spectroscopy for in-line quantification of antibody product quality

Gordon Magill, [email protected], Barbara Chiang, Brian Horvath. Late Stage Cell Culture Development, Genentech, South San Francisco, California, United States

Recent years have shown an increase in the development and application of Raman spectroscopy for process monitoring and control in the biopharmaceutical industry. While Raman spectroscopy has been successfully used to predict cell densities and concentrations of conventional metabolites, limited work has been presented on its capability to measure product-related attributes, including titer and product quality. Here, we present results using Raman spectroscopy to obtain in-line measurements of CHO antibody titer as well as charge, size, and glycan species across a range of antibodies including complex molecular formats. In addition, we discuss the technical challenges in deriving these measurements using multivariate analysis techniques.

BIOT 108

Utilization of computational fluid dynamics (CFD) to guide scale up of cell culture bioreactors

Garima Chaudhary1, [email protected], Thomas Wucherpfennig3, Jan-Erik Schaefer2, Erik Hasenfus4, Johannes Wutz5, Henry Lin1, Daniel Bock1, Marcella Yu1. (1) Cell Culture Process Science, Boehringer-Ingelheim, Fremont, California, United States (2) Early Stage USP Development, Bioprocess Development Biologicals, Boehringer-Ingelheim, Biberach, Germany (3) Late Stage USP Development, Bioprocess Development Biologicals, Boehringer-Ingelheim, Biberach, Germany

Computational fluid dynamics is a valuable tool for characterizing and better understanding the flow regimes in bioreactors at various scales for mammalian cell cultivation. Knowledge of bioreactor characteristics at different scales generated by CFD will aid in developing efficient process scale up and transfer strategies. Boehringer Ingelheim has put tremendous effort in fully characterizing our multiple bioreactors at different scales across multiple sites. This presentation will focus on the equipment in our Fremont facility. In the current work, the first study included single phase CFD simulations to generate power numbers and compare against vendor data for bioreactors at various scales to investigate power input. The power number stability was demonstrated with different agitation/Reynolds number curves. Single phase mixing simulations were also performed for these scales to evaluate mixing time in the bioreactors. Furthermore, efforts toward achieving multiphase simulations to evaluate kLa (for O2 and CO2) for different scales of bioreactors would provide even greater value for supporting the scale-up of our bioreactor systems. All the data generated based on the CFD simulations is being further utilized to build appropriate scale-up criteria at various scales to ensure successful process transfers. BIOT 109

Limitations of subcloning as a tool to characterize homogeneity of a cell population

Nicole Tejeda, [email protected], Hedieh Barkhordarian, Tharmala Tharmalingam, Pheng Yam, Fang Lu, Samih Yaghmour, Trent P. Munro, Chetan Goudar, Jennitte Stevens. Amgen Inc, Thousand Oaks, California, United States

Cloning is a critical step in the generation of manufacturing cell lines. The expectation is that the process of cloning will result in a uniform and homogeneous cell line that will ensure robust product quality over the lifetime of the product. Regulatory guidelines require the sponsors provide assurance of clonality of the production cell line. When such evidence is not available, additional studies are required to further ensure consistent long-term manufacturing of the product. One approach to characterize homogeneity of a cell line is subclone analysis where clones are generated from the original cell line and an evaluation of their similarity is performed. To study the suitability of subclone analysis to provide additional assurance that a production cell line is clonally derived, an antibody producing CHO Master Cell Bank (MCB), which was cloned by a validated FACS method and with a clear documented day 0 image, was characterized. Specifically, this MCB was subcloned and imaged to assure each of the subclones were derived from a single cell. A total of 46 subclones were analyzed for growth, productivity, product quality, as well as copy number and integration site analysis. Despite demonstration of clonality for both the MCB and the subclones, significant diversity in cell growth, protein productivity, and product quality attributes was observed among the 46 subclones. The diversity in protein productivity and quality were reproduced across bioreactor scales, suggesting that albeit different, the subclones were stable populations that varied from the parental clonal cell line. Additionally, while ~2-fold shifts in copy number were seen, no significant integration site changes were observed. Our data suggest subcloning induces changes (genetic or epigenetic) outside the region of the transgene, which result in the subclones exhibiting a wide diversity in cell growth, protein productivity, and product quality. Transcriptomic and genomic characterization studies are underway to further characterize the differences between subclones and the MCB. Importantly, the subclones do keep their individual characteristics as they mature and stabilize, suggesting that the resulting population that grows out of a single cell is stable, but with unique properties. Overall, this work adds to the growing body of work on CHO cell plasticity and suggests that subcloning is not an effective approach to demonstrate homogeneity of a cell bank.

BIOT 110

Assessment of the genomic instability of CHO cells for cell line development

Sofie O'Brien1, [email protected], Arpan Bandyopadhyay2, Conor M. O'Brien2, Wei- Shou Hu2. (1) Biomedical Engineering, University of Minnesota, Minneapolis, Minnesota, United States (2) Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, Minnesota, United States

The genomic structure of Chinese Hamster Ovary (CHO) cells undergoes changes during successive serial propagation in culture. Chromosomal aberrations such as mutations, deletions, duplications, and other structural variants can accumulate within the genome during this process. In constructing a production cell line, it is important that exogenous genes are introduced into a region that is stable in its expression over a long period of cell proliferation, whether they are for therapeutic protein production or cell engineering. Understanding the relationship between structural characteristics and stability is thus important for cell culture engineering.

The genome stability of CHO cell lines was investigated at the macroscopic chromosomal and microscopic gene and sequence levels. Chromosomal and karyotypic variation were apparent at the macroscopic level. In the short span of a few dozen generations, the progenies of single cell clones exhibited a wide distribution of chromosome numbers and karyotypes. At the microscopic level, stability was assessed using a gene-coding region focused comparative genomic hybridization (CGH) microarray and genome sequencing. By analyzing gene copy variations between related cell lines, including subclones and relationships between host and producing cell lines, we identified common segment changes that happen during cell line development and subcloning. Whole genome sequencing data of multiple CHO cell lines was then used to further examine variation at the microscopic level. Structural variants, such as deletions, inversions, and duplications were identified using the tools DELLY2 and LUMPY. Heterogeneity was present within each cell line and visible in the form of genome mosaicism.

This integrated approach combining chromosome number, karyotyping, CGH, and genome sequencing gives us insight into the heterogeneity and instability of CHO cells, allowing us to identify undesirable regions for gene integration, and screen out potentially unstable cell lines.

BIOT 111

Development of high-producing CHO cell lines through target-designed strategy

Dalton Chen, [email protected]. Bioengineering Department, New Taipei City, Taiwan

Productivity and stability are critical for the producing cell lines for manufacturing protein drug. Given that the integration sites of gene of interest (GOI) could contribute remarkable effect on the productivity and stability of GOI expression, we intended to develop a time-saving and less labor-intensive strategy to generate the high-producing cell lines through targeting the active and stable regions of CHO genome.

To identify the active and stable regions located in CHO genome, we generated high producing stable cell line, and employed PCR-based or NGS-based method to analyze the integration sites of the high producing stable cell line. The pair-end reads with one read mapped to the vector and the other read mapped to the CHO reference genome are extracted to identify the integration sites.

In order to verify the expression activity and stability of the integration sites, we employed CRISPR/Cas9 to specifically knock-in the expression vector with antibody gene into CHO genome. Our data showed the cell pool generated by knock-in of expression vector into the IS1 integration site present higher expression titer than cell pools generated by random integration or integrating into other sites . We further cultured the single cell clones derived from this cell pool through screening and selection by Clonepix and limiting dilution. These single cell clones have high expression titer ranging from 254 to 804 mg/L in batch culture of after 6 Days. A single cell clone(376 mg/L in batch culture) can reached 2 g/L in fed-batch culture. The stability analysis showed this clone maintain stable expression of GOI after 60 generation. Here, we demonstrated the generation of stable cell line with high protein expression by CRISPR/Cas9 mediated target integration. The target designed strategy will serve as a platform for therapeutic antibody production.

BIOT 112

Unravelling molecular mechanisms that govern the shift to a lactate consumption metabolic state in CHO cell fed-batch cultures

Michaela Jacobs, [email protected], Pauline Geoffroy, Taylor Kalomeris, Cameron Harrington, Bhanu Chandra Mulukutla. Cell Culture Process Development, Pfizer, Andover, Massachusetts, United States

Currently under legal review. Will be submitted to the session chair as soon as possible.

BIOT 113

Cell characterization using a novel online cell quantitation microscopic technology in comparison with conventional offline methodologies

Jianxin Sun2, [email protected], Vijaya Doddi2, Victor Chang2, Daisie Ogawa1, Todd Luman2, Henry Lin2. (1) Bioprocess Engineering, Boehringer Ingelheim, Fremont, California, United States (2) Cell Culture, Boehringer Ingelheim, Fremont, California, United States

Currently, bioreactors are only sampled once per day or less due to the limitations of labor, bioreactor volume, etc. Daily manual sampling means data unavailable for hours, delaying troubleshooting efforts and identification of problems. The iLine F microscope from Ovizio utilizes Differential Digital Holographic Microscopy (DDHM)- a new quantitative imaging technique that can monitor cell counts and viability in-line at real- time without staining. The microscope captures holograms of the cells travelling through the BioConnect disposable fluidics system, which connects the microscope to the bioreactor. The closed loop set-up minimizes contamination during sampling without sample volume consumption. In this study, the iLine F microscope was tested against conventional cell quantitation methods such as using flow cytometry (different flourescence staining dyes), trypan blue dye exclusion and capacitance probe measurement. While the trypan blue dye exclusion is the most common method for cell quantitation and viability measurements, it tends to report higher viability when compared to fluorescence staining techniques such as by propidium iodide [1]. The capacitance probe utilizes frequency-dependent polarization of dielectric material to detect live cells by measuring capacitance [2]. Capacitance probes have limitations that they cannot detect non-viable cells. In addition, these probes measure biovolume that must be correlated to viable cell concentration which is subject to change over the culture duration. The iLine F microscope, on the other hand, is based on holographic microscopy and can detect at the single cell level. The study shows proof of concept of monitoring cell growth and viability in real-time as well as the performance comparison with other methods. Different cell lines were also tested to confirm the feasibility and robustness.

BIOT 114

RNA-seq data reveals metabolic regulation in Chinese hamster ovary cell culture

Seongkyu Yoon, [email protected], Sha Sha. University of Massachusetts Lowell, Lowell, Massachusetts, United States

Mammalian cells have been known to be undergoing amount of physiological alterations during culture. This is a result of cell responses to extrinsic and intrinsic changes from culture environment. Understanding the alteration from intracellular context is an important way to decipher cell regulations and omics’ approaches have become to be the main way to gain insights beyond the traditional investigation of extracellular culture parameters only. Metabolic flux studies from literature had extensively discussed cellular changes overtime in culture. However, studies of gene expression changes during culture are lacked. An aspect of gene regulation in culture is important because genes become the targets of cell line engineering. In this study, RNAseq was used to carry out a comprehensive examination of the transcriptomic changes of Chinese hamster ovary (CHO) cells across the time progression in batch culture. Genes associated with energy pathway, carbohydrate distribution, oxidative stress, apoptosis and glycosylation related pathways were investigated. From the results, the genes catalyzing energy pathways showed overall decrease during culture. This result complies with the understanding from many metabolic flux analyses. A very distinct characteristic of transcriptome was found at the point when lactate started to be consumed in culture. From a wide range of transcriptome, a series of alteration in the genes associated with stress and glycosylation, as well regulation to the energy pathways was concurrently found. This study provides a view of cellular activity comprehensively progressing in culture; and reveals possible intrinsic correlation between veracious cellular activities. Three CHO cell lines producing different monoclonal antibodies were included in the study. The results showed a good extent of similarity from the results of the three cell lines.

BIOT 115

Engineering modular sensing-actuating membrane fusion machinery from influenza hemagglutinin

Mauricio Valverde1, James Price1, Morgan R. Baltz1, Jeong Lee2, Eric T. Boder1, [email protected]. (1) Department of Chemical and Biomolecular Engineering, University of Tennessee, Knoxville, Tennessee, United States (2) Department of Chemical and Biomolecular Engineering, University of Pennsylvania, Philadelphia, Pennsylvania, United States

The envelope protein of influenza virus, hemagglutinin (HA), represents one of the best studied paradigms among protein machines that facilitate fusion of lipid bilayer membranes. These proteins demonstrate environmentally-triggered conformational changes that activate their fusogenic function. While understanding the structure and function of these systems would benefit efforts to design and engineer regulated membrane fusogens for a range of applications, including drug or non-viral gene delivery, many details regarding both the sensing and actuating functions of these molecular switch proteins remain to be elucidated. Here, we have engineered mutants of fowl plague virus HA that respond to alternative environmental stimuli compared to the wild-type protein. Directed evolution yielded mutations both stabilizing or destabilizing HA against an activating pH change. Structural considerations support a model in which association of HA headgroups in their pre-activation trimeric structure is important for inhibiting the conformational transition. We further created a mutant designed to exhibit conditional binding to a fluorescent ligand (FlAsH) following activation and discovered that FlAsH binding is sufficient to induce the HA conformational transition. Analysis of the pH- or FlAsH-sensing phenotypes of these mutants demonstrated that all mutants recovered by directed evolution, as well as the FlAsH-binding mutant, were unable to induce membrane fusion despite activation of the fusion-associated conformational change; however, co-expressing these mutants with fusion-competent wild-type HA enabled recovery of fusion triggered by the alternative stimuli, indicating that modular fusogenic systems can be created in which HAs engineered for novel sensing functions can be mixed with other HAs that provide the actuating component of the system. Finally, studies of both conformational activation and fusion indicate a significant delay in the kinetics of conformational activation and the associated extent of fusion when cholesterol is depleted, suggesting that membrane properties may play an important role in HA function. Our results motivate further protein engineering efforts to develop membrane fusogens responding to additional environmental stimuli and compel further studies of HA behavior in synthetic systems with defined phospholipid composition. BIOT 116

Yeast as a platform to study adenosine receptor signaling

Abhinav R. Jain2, [email protected], Anne S. Robinson1, Claire McGraw2. (1) Chem and Biomolecular Eng, 300 Lindy Boggs Bldg, Tulane University, New Orleans, Louisiana, United States (2) Chemical and Biomolecular Engineering, Tulane University, New Orleans, Louisiana, United States

G-protein coupled receptors (GPCRs) bind to extracellular molecules and produce an intracellular downstream signaling response inside the cells. Because of their membrane localization and the ability to produce intracellular changes, they are easy target for therapeutics; approximately 40% of drugs in the market target these receptors. Yeast share many functionally exchangeable proteins with humans involved in the GPCR signaling pathway, and provide a relatively easy and inexpensive platform without the complexities of multiple GPCRs, receptor promiscuity and crosstalk that occurs in native mammalian hosts. The adenosine receptor subfamily (A1R, A2aR, A2bR and A3R) of GPCRs bind the natural ligand adenosine, an important energy metabolite produced under stressful conditions and these receptors help carryout its essential functions in the cells. Adenosine receptors are therapeutic targets for Alzheimer’s disease, Parkinson’s disease and many other diseases. This talk will describe research using chimeric and truncated adenosine receptors expressed in yeast to determine the role of the receptor C-terminus for ligand binding and downstream signaling. These observations were validated in HEK cells, and will be used to highlight the importance of yeast as a platform to study GPCR-ligand binding and downstream signaling.

BIOT 117

Engineering mucoadhesive silk

Georgia Petrou1, [email protected], Ronnie Jansson2, Mark Högqvist2, My Hedhammar2, Thomas Crouzier1. (1) Division of Glycoscience , KTH Royal Institute of Technology, Stockholm, Sweden (2) Division of Protein Technology , KTH Royal Institute of Technology, Stockholm, Sweden

Mucoadhesion is defined as the adhesion of a material on the mucus gel covering the mucous membranes. The mechanisms controlling mucoadhesion include nonspecific and specific interactions between the materials and mucins, the heavily glycosylated proteins that form the mucus gel. Mucoadhesive materials can be used to develop mucosal wound dressings, and non- invasive trans-mucosal drug delivery systems. Spider silk is strong, biocompatible, biodegradable, non-toxic and light-weight, thus a good candidate. Here, we recombinantly produce and functionalize partial spider silk proteins (4RepCT) in order to enhance their mucoadhesive properties. The pLys-4RepCT variant is genetically functionalized with six positively charged lysines, aiming to provide nonspecific adhesion from electrostatic interactions with the negatively charged mucins, while the hGal3-4RepCT variant is genetically fused with the residues of the Human Galectin-3 Carbohydrate Recognition Domain which specifically binds the mucin glycans Galβ1-3GlcNAc and Galβ1-4GlcNAc. First we demonstrate that we can form various material types from both silk variants, including coatings, fibers, meshes, and foams. All material types from both variants show enhanced mucoadhesive properties compared to wildtype silk materials, as suggested by the adsorption measurements of both Bovine Submaxillary Mucin and Pig Gastric Mucin. Moreover, we show that the increased mucoadhesion of hGal3-4RepCT is a result of specific glycan-protein interactions. We use both quartz crystal microbalance with dissipation monitoring and biolayer interferometry to characterize the coatings and their interaction with mucins, and fluorescence microscopy imaging and image analysis to characterize the 3D materials. In summary, we achieve to recombinantly produce two mucoadhesive silk variants: pLys-4RepCT and hGal3-4RepCT. Alone, or combined with other biofunctional silk proteins, they constitute useful new building blocks for a range of silk protein based materials for mucosal treatments, including coatings, meshes, and foams.

BIOT 118

Directed evolution of tissue inhibitor of metalloproteinase-1 (TIMP-1) as a selective inhibitor of matrix metalloproteinase-3 (MMP-3)

Maryam Raeeszadeh Sarmazdeh, [email protected], Derek Radisky, Evette Radisky. Cancer Biology, Mayo Clinic, JACKSONVILLE, Florida, United States

Matrix metalloproteinases (MMPs) play a pivotal role in diagnosis and progression of several diseases such as cancer and fibrosis, which makes them promising targets for developing protein therapeutics. The challenge in using MMP inhibitors as drugs in MMP-associated diseases is lack of high selectivity for binding to a specific MMP. Therefore, development of selective MMP inhibitors is a critical goal. Tissue inhibitor of metalloproteinases (TIMPs), as natural MMP inhibitors, offer an ideal scaffold for engineering highly selective MMP-targeted therapeutics. We have previously shown a key role for MMP-3 in the pathogenesis of pulmonary fibrosis.

TIMP-1 is a natural inhibitor MMP-3 with picomolar affinity; it also binds to other MMPs with a spectrum of affinities. I have used directed evolution and yeast surface display toward development of selective MMP-3-targeted biologic agents through protein engineering of TIMP-1. I have screened a library of TIMP-1 mutants generated by random mutagenesis of residues comprising the MMP-interacting loops of TIMP-1 to select for enhanced MMP-3 binding on the yeast surface. Diversified residues have been chosen based on structural studies of TIMP-1/MMP-3 interactions. TIMP-1 mutants isolated after rounds of Fluorescent-Activated Cell Sorting (FACS) showed substantial improvement in MMP-3 binding compared to the wild-type TIMP-1. Mutations found in mutants with enhanced binding affinity toward MMP-3 were further characterized using soluble proteins. I have also developed a counter-selection strategy to screen TIMP-1 mutants that bind selectively to MMP-3 in the presence of MMP-10, an MMP with the most structural similarity to MMP-3. The isolated TIMP-1 mutants after five rounds of competitive screening showed up to 20-fold improvement in binding selectivity, highlighting the significant potential of this approach for development of inhibitors with single-MMP selectivity. This work provides a promising strategy to develop protein therapeutics based on natural enzyme inhibitors, critically improving selectivity to eliminate off-target effects.

BIOT 119

Engineering binding affinity and specificity: Application to antibodies targeting phospho- and acetyl-tau

Dan Li1, [email protected], Lei Wang3, Xudong Yao3, Yongku Cho2,1. (1) Biomedical Engineering, University of Connecticut, Storrs, Connecticut, United States (2) Chemical and Biomolecular Engineering, University of Connecticut, Storrs, Connecticut, United States (3) Chemistry, University of Connecticut, Storrs, Connecticut, United States

Antibodies specific to protein post-translational modifications (PTMs) are an essential tool in basic biology and clinical investigations. Due to the transient and heterogeneous nature of protein PTMs, high specificity and affinity are required in PTM-specific antibodies for sensitive and reproducible detection. In vast majority of PTM-specific antibodies, however, the specificity - in particular their ability to distinguish modified and unmodified forms of a target site - is often not characterized, and whether the affinity meets the required detection sensitivity is unclear. Using yeast surface display, we have shown that the PTM-specificity of antibodies can be quantified in flow cytometry experiments. This approach enabled us to previously show that affinity maturation of a phospho-specific antibody leads to increased cross reactivity towards the non- phosphorylated target site in majority of, but not all affinity improved clones. To efficiently identify PTM-specific antibodies, here we present a robust and quantitative screen that applies selection pressure for enhancements in both affinity and specificity. We will present the application of this approach to identify high affinity and specificity antibodies against disease-specific PTMs of human microtubule associated protein tau (MAPT). Starting from a naïve human single-chain variable fragment (scFv) library, we identified binders to acetylated tau at lysine 174 (ac-K174), which was recently reported as a key PTM in tau-mediated neurodegeneration. We will also present a novel approach to quantify antibody specificity directly in immunocytochemistry experiments, in which antibody specificity is quantified in each pixel of a fluorescence image. Supported by theoretical calculations, we demonstrate that an optimal window of antibody concentration exists in cell labeling experiments, and propose a parameter that can be used as a guideline for high concentration limit for specific imaging of PTMs in general. The new technologies developed here have great potential for validation and improvement of the quality of PTM-specific antibodies and potentially antibodies in general.

BIOT 120 Optimizing a monoclonal antibody for manufacturability, stability, and cost

Julee Floyd, [email protected], Christine Siska, Alison Gillespie, Megan McClure, Yan Brodsky, Jeremy Shaver, Rutilio Clark, Randal Ketchem, Bruce Kerwin. Just Biotherapeutics, Seattle, Washington, United States

One of the major hurdles for producing monoclonal antibodies (mAbs) that are targeted for geographically diverse populations is the need to produce a mAb that is manufacturable, thermally stable, and low cost. By using a combination of in silico molecular design, activity assays, and biophysical tools, we will show how this can be accomplished. First, variants of a parental antibody are defined in silico based on sequence and structural considerations using a combination of molecular design tools. Next, a suite of biophysical and activity assays are used to characterize and identify the most suitable variants. Based on these results, another round of modifications may be designed to enhance the variants that demonstrated measurable improvements. This new set of variants are tested using an expanded suite of biophysical assays to identify the variant most suitable for production. Specifically, we will discuss how the combination of biophysical assays such as differential scanning fluorimetry, chemical unfolding, low pH stability, and others together with activity, helped identify variants with significantly improved manufacturability characteristics. Additionally, these variants demonstrated decreases up to 6-fold in sub-visible particle formation during storage at 40oC. This approach can be used to optimize antibody therapeutics for improved manufacturability, stability, and lower overall costs leading to expanded access to important biotherapeutics.

BIOT 121

Improved library design and selection methods for identifying antibodies with high specificity and solubility

Alec Desai2, [email protected], Lilia Rabia1, Mark Julian1, Kathryn Tiller1, Seth D. Ludwig1, Yulei Zhang2, Peter Tessier2,1. (1) Rensselaer Polytechnic Institute, Ann Arbor, Michigan, United States (2) University of Michigan, Ann Arbor, Michigan, United States

Antibody specificity is arguably the most important and most challenging property to optimize during in vitro antibody selection and affinity maturation. We have sought to develop methods for both designing and sorting antibody libraries that maximize the specificity of the selected variants. First, we have developed a novel library design method (Natural Diversity Mutagenesis) that samples wild-type residues and a small number of the most frequently occurring residues at each position of antibody complementarity-determining regions (CDRs) and/or framework regions based on natural antibody diversity. This mutagenesis approach results in antibody libraries with variants that have a wide range of numbers of mutations, including antibodies with single mutations and others with tens of mutations. Second, we have developed methods for performing strong negative selections using different types of non-specific (peptide and protein) reagents and strong positive selections using the targets of interest. Interestingly, we have identified reagents that are unusually effective at removing non-specific antibody variants during negative selections. Moreover, we find that our library design and selection methods result in the identification of antibody variants with extremely high specificity in addition to high expressibility and solubility. These variants are enriched with specific types of amino acids in their CDRs that are linked to drug-like antibody properties. We will discuss how we are using these and related approaches to improve in vitro isolation of antibodies with properties that rival those of natural antibodies.

BIOT 122

Engineering novel glycophenotypes in bacteria for the production of authentically glycosylated therapeutic proteins

Xiaolu Zheng, [email protected], Matthew P. DeLisa. Robert Frederick Smith School of Chemical and Biomolecular Engineering, Cornell University, Ithaca, New York, United States

N-linked glycosylation plays an important role in protein folding, stability, and function. Hence, glycoengineering could provide a great opportunity to optimize the performance of therapeutic proteins. However, the complexity of the protein glycosylation process in eukaryotes limits its use in glycoengineering applications. The discovery of an N- glycosylation pathway in the bacterium Campylobacter jejuni and its functional reconstitution in Escherichia coli, has made it possible to produce and conveniently work with engineered glycoproteins and gain basic mechanistic insights into the process of glycosylation. A central enzyme in N-glycosylation is the oligosaccharyltransferase (OST), which transfers preassembled glycans to specific asparagine residues of target proteins. The C. jejuni OST, PglB, can transfer a range of diverse glycan structures, but the sequon it recognizes is restricted predominantly to those having a negatively charged residue in the −2 position relative to the asparagine (D/E-X-1-N-X-2-T, where X-1 and X-2 are any amino acids but proline). Since human glycosylation does not have this restriction, our ability to produce authentic human glycoproteins in E. coli is limited. To address this limitation, we aim to relax the sequon specificity of the bacterial OST in order to easily target and N-glycosylate minimal N-X-T acceptor sites in proteins without modification of the targeted site. Towards this goal, we have developed a genetic assay that facilitates high-throughput screening of glycophenotypes in E. coli. Specifically, a combinatorial library of mutant OST sequences was designed using different PglB homologs and subsequently screened for expression, membrane localization, and glycosylation capability. We identified several OST variants that were capable of glycosylating an array of noncanonical acceptor sequences including one in eukaryotic N-glycoproteins. This not only provides a potential platform to produce glycoprotein therapeutics such as full-length antibodies in E. coli but also allows us to better understand the mechanism of this important class of enzymes.

BIOT 123 Development of a 3D co-cultivation process of beta cells and mesenchymal stromal/stem cells for diabetes therapy

Florian Petry1, Denise Salzig1, Tobias Weidner1, Peter M. Czermak1,2, [email protected]. (1) IBPT, University of Applied Sciences Giessen, Giessen, Germany (2) Bioresources, Fraunhofer Institute IME, Giessen, Germany

The number of diabetic patients grows rapidly every year. Diabetes is characterized by failure of the insulin-producing beta cells and cannot be cured, but only treated symptomatically with insulin application. The functionality of beta cells in vivo can be restored with the transplantation of whole pancreatic islets, beta cell pseudoislets or the pancreatic-differentiated induced pluripotent stem cells (iPSCs). However, this therapeutic concept bothers from the facts that beta cell grafts do not survive at the transplantation site and beta cells lose functionality when expanded/cultivated in vitro. Beta cells can be strengthen and remain their functionality by a co-cultivation with human mesenchymal stem/stromal cells (hMSCs). hMSCs are known for their therapeutic properties transmitted by direct cell contact and/or the secretion of trophic factors. Due to their origin from the islets of Langerhans, beta cells prefer a 3D environment in form of cell agglomerates/spheroids. So the development of a 3D co- cultivation process in bioreactors is necessary, which ensures the high amounts of cells needed for cell therapy (106-1010 cells per dose) and fulfills the requirements of good manufacturing practice (GMP) and process analytical technology (PAT). Therefore, we investigated the effects of hMSCs on beta cells in direct and indirect co-cultivation set- ups.

BIOT 124

Chemically defined microcarriers for the expansion of mesenchymal stem cells

John Krutty2, [email protected], Andrew Dias4, William L. Murphy1,2, Padma Gopalan3. (1) University of Wisconsin, Madison, Wisconsin, United States (2) Biomedical Engineering, University of Wisconsin-Madison, Madison, Wisconsin, United States (3) Materials Science and Engineering, University of Wisconsin-Madison, Madison, Wisconsin, United States (4) Orthopedics, University of Wisconsin-Madison, Madison, Wisconsin, United States

Millions of mesenchymal stem cells (MSCs) per kilogram of patient body weight are generally required for clinical applications, underscoring a pressing need for scalable culture systems. To reduce the time and cost associated with the expansion of MSCs, microcarriers that provide a surface for cell adhesion and are subsequently kept in suspension in a bioreactor were developed. Microcarrier culture systems can produce hundreds of millions of cells and show promise for scale-up to meet industrial lot size requirements with lower media and resource requirements than traditional cell culture.

We previously validated the synthetic copolymer (polyethylene glycol methyl ether methacrylate-ran-vinyl azlactone-ran-glycidyl methacrylate; PVG, Figure 1A) based on polyethyleneglycol in 2D, spin-coated films onto flat substrates as a chemically defined template for cell culture. When functionalized with the cell adhesive peptide sequence RGD this chemically defined surface is suitable for MSC adhesion and expansion. We present a novel method for the creation of chemically defined, customizable microcarriers via covalent anchoring of the PVG copolymer to the surface of PS microcarriers and its characterization by spectroscopy. These PVG-coated microcarriers reduce protein adsorption and prevent the adhesion of cells until they are functionalized with RGD peptide. MSCs cultured on these coated microcarriers adhere, expand, and retain their ability to differentiate, indicating maintenance of the stem cell phenotype. We envision these microcarriers that use customizable polymer coatings to lower the currently intractable cost associated with cell expansion and achieve efficient, xeno-free expansion of functional therapeutic cells.

Figure 1: A) Schematic of coating process/polymer chemistry. Polystyrene microcarriers are coated in (ii) poly-L-lysine (PLL) and (iii) polyethylene glycol methyl ether methacrylate-ran-vinyl azlactone-ran-glycidyl methacrylate; P(PEGMEMA-r-VDM-r- GMA); PVG. Desired peptides can be applied to this coating, in this example a peptide functionalized with (iv) Arg-Gly-Asp (RGD) and (v) a scrambled version of the peptide. (B) cells adhered to functionalized microcarriers

BIOT 125

Cell therapies and their impact on the future of medicine

Knut Niss, [email protected]. Mustang Bio, Waltham, Massachusetts, United States

Over the last few years, cell therapies have entered the mainstream of clinical development. While this development has been fuled by the Chimeric Antigen Receptor (CAR) technology other promising cell therapies have emerged. Today, cell therapies are being tested in a wide variety of diseases from diabetes to cancer. With these new therapies new challenges arise from clinical support to manufacturing. This presentation will provide an overview of the history of cell therapies and describe current approaches from the recently approved CAR-T cell approaches to innovative pluripotent stem cell approaches. In addition, an overview of the challenges in manufacturing of cell therapies will be discussed.

BIOT 126

Histone-targeted gene transfer to stimulate chondrogenic MSC differentiation and bone repair

Millicent O. Sullivan, [email protected], Erik Munsell. University of Delaware, Newark, Delaware, United States

The regeneration of damaged skeletal tissue following traumatic injury involves a complex series of cascades that direct the differentiation of mesenchymal stem cells (MSCs). These signals include a series of key osteogenic growth factors that are actively released within the fracture site. The necessity for controlled and localized expression of these factors has highlighted the role gene therapy may play as a promising treatment option for bone tissue regeneration. However, the design of nanocarrier systems that negotiate efficient intracellular trafficking and nuclear delivery represents a significant challenge. Recent investigations have highlighted the roles histone tail sequences play in directing nuclear delivery and retention, as well as activating DNA transcription. We previously established the ability to recapitulate these natural histone tail activities within non-viral gene nanocarriers, improving gene transfer and expression by enabling effective navigation to the nucleus via retrograde vesicular trafficking. In this current work, we demonstrate that histone-targeting leads to a near 4- fold enhancement in osteogenic bone morphogenetic protein 2 (BMP-2) growth factor expression by MSCs over the course of 6 days, as compared with standard polymeric transfection reagents. This improved expression ultimately augmented chondrogenic differentiation, an essential first step in the fracture healing cascade. Most notably, significant enhancements in the expression of cartilage-specific extracellular matrix proteins were triggered by histone-targeted gene transfer, as compared with the response to treatment with equivalent amounts of recombinant BMP-2 protein. In fact, a 100-fold dose increase in recombinant BMP-2 was required to achieve similar levels of chondrogenic gene and protein expression. The enhancements in differentiation achieved using histone-targeted gene transfer were in part enabled by an increase in chondrogenic transcription factor expression, which functioned to drive MSC condensation and differentiation. These novel findings demonstrate the utility of histone- targeted gene transfer strategies to enable substantial reductions in BMP-2 dosing for bone regenerative applications.

BIOT 127

Enzymatically generated hyaluronic acid hydrogel for cytokine therapy of osteoarthritis Sílvia Pérez Rafael1, [email protected], Francesca Perrone2, Eva Ramon1, Tzanko Tzanov1. (1) Chemical Engineering, Universitat Politècnica de Catalunya, Terrassa (PC 08222), Barcelona, Spain (2) Collegio di Ingegneria Biomedica, Politecnico di Torino, Torino, Torino, Italy

Osteoarthritis (OA) is a degenerative disease characterised with pain, stiffness and loss of function in the weight-bearing joints, caused by the inflammatory biological molecules that disturb the balance between the tissue synthesis and degradation. Maintaining the composition of the physiological lubricant, named synovial fluid, by restoration of homeostasis is thus a necessary step to retard the OA progression and provide a suitable environment for tissue remodelling.

Thiolated hyaluronic acid (HA-SH) was enzymatically crosslinked with gallic acid (GA) by the action of laccase as a method to prepare multifunctional hydrogels, designed as delivery systems that enable long-term efficiency of biological entities in cell-based therapies. Owing to its characteristics and mild conditions employed, the encapsulation of very unstable and sensitive cargoes is possible, highlighting its potential as a platform for a variety of biomedical applications. A cytokine cocktail (ACC) and chondrocytes cells were encapsulated together as therapy to delay the disease progression and boost the anabolic pathways for recovery of the damaged cartilage. The morphology, swelling ratio, rheological properties and stability of the hydrogels were characterised and optimised in order to fit the requirements of the therapy. The efficiency of the developed platforms was evaluated in vitro against the major factors governing the OA disease, namely the anti-inflammatory effect, cellular ROS scavenging and the inhibitory capacity of the hydrogels towards deleterious enzymes as MMPs, MPO and hyalurodinase. Finally, experiments to determine the ability of the cells to survive, growth and differentiate in the multifunctional hydrogels were assessed in order to explore their potential in a cell-based therapies and regenerative medicine.

BIOT 128

Decoy TRAIL receptor CD264: A predictor of in vitro regenerative potential for mesenchymal stem cells

Sean Madsen1, [email protected], Katie Russell1, Alan Tucker3, Julie Glowacki2, Bruce Bunnell3, Kim OConnor1. (1) Dept. of Chemical & Biomolecular Engineering, Tulane University, New Orleans, Louisiana, United States (2) School of Medicine, Harvard University, Boston, Massachusetts, United States (3) School of Medicine, Tulane University, Boston, Massachusetts, United States

The regenerative potential of marrow-derived mesenchymal stem cells (MSCs) exhibits significant variation, particularly among older patients. The objective of this study is to identify a cell-surface marker whose expression is predictive of the in vitro proliferation and differentiation potential of MSCs. This study evaluates surface expression of decoy TRAIL receptor CD264, in vitro regenerative potential and metrics of cellular aging for marrow MSCs from 12 donors, 20-60 years old. Male and female donors were age- matched. When CD264(+) cell content was 20% to 35%, MSC cultures from young (20 < age < 40) and older (45 < age < 60) donors proliferated rapidly and differentiated extensively. Older donor MSCs containing < 35% CD264(+) cells had a small size and negligible senescence despite the donor’s advanced chronological age. Above the 35% threshold, CD264 expression inversely correlated with proliferation and differentiation potential. When CD264(+) cell content was 75%, MSCs were enlarged and mostly senescent with severely compromised regenerative potential. There was no correlation of the older donors’ chronological age to either CD264(+) cell content or the regenerative potential of the donor MSCs. CD264 was upregulated after p53 and had a similar expression profile to that of p21 during serial passage of MSCs. No sex-linked differences were detected in this study. The strong inverse correlation of CD264(+) cell content to the in vitro regenerative potential of MSCs has possible application to predict the therapeutic potential of patient MSCs, and to standardize the composition and efficacy of MSC therapies.

BIOT 129

Labeling neural stem cells using trackable ultrasmall iron oxide nanoparticle for cell transplantation therapy

Joe Park, [email protected], Yonghyun Kim, Yuping Bao, Jennifer Sherwood. Chemical and Biological Engineering, University of Alabama, Tuscaloosa, Alabama, United States

Neural stem cell (NSC) transplantation is starting to be used in clinical trials for regenerative rehabilitation in spinal cord injury, Parkinson’s disease, amyotrophic lateral sclerosis, and other neurodegenerative diseases. However, proper tracking of the transplanted NSCs is challenging. Currently, magnetic resonance imaging (MRI) is broadly used after NSC transplantation. Gadolinium (Gd) has been used as a MRI contrast agent because of its high resolution imaging. However, Gd contrast agents cause several adverse reactions such as nephrogenic systemic fibrosis, allergy-like reactions, and Gd retention in tissues. Another disadvantage is that Gd is not trackable in the NSCs for long term while they are migrating to lesions. To overcome these disadvantages of Gd, we labeled NSCs using ultrasmall iron oxide nanoparticles (Fe- USNPs). We were able to grow NSCs as neurospheres followed by the addition of the Fe-USNPs. Cell growth and viability were not hampered by Fe-USNP up to 1 mg during Fe-USNP uptake. Stemness was maintained and verified using stem cell markers and differentiation markers. We were also able to confirm the tracking of Fe-USNPs in NSCs after their uptake by Prussian blue staining and MRI. These early results show a novel method in adapting the Fe-USNP labeling strategy for improved tracking of transplanted NSCs in the clinic in the future.

BIOT 130

Chromatography modeling in purification development: Assessing accuracy of predictions to select phase-appropriate modeling strategies Ferdinand Stueckler1, [email protected], Katharina Doninger1, Jessica Yang2, Christopher Williams2, Till Briskot1. (1) Roche Diagnostics Germany, Munich, Germany (2) Genentech, South San Francisco, California, United States

Modeling, simulation and machine learning techniques nowadays allow for a data-driven analysis of complex processes. These computational modeling techniques can be used to summarize process characteristics from experimental data and offer a framework to also include process knowledge from historical data. The development of biopharmaceutical purification processes can benefit from applying these methods to ensure an efficient and robust DSP design.

The application of mechanistic models for the description of chromatography processes has been shown in various examples and case studies. Model-based techniques allow for analyzing experimental results and help to improve process development by guiding experimental design. Based on model predictions optimal conditions for protein purification can be chosen.

The success of model-based process development depends on the prediction accuracy of in silico simulations. Factors which influence this accuracy are for instance: general model assumptions, quality of the data used for calibrating the model, and model parameter estimation results.

We will therefore discuss phase-appropriate modeling strategies which take into account these factors with respect to their influence on the required prediction accuracy for specific DSP process development stages. For instance at early stage development limited data sets are available. While it remains challenging to establish high accuracy semi-quantitative model predictions have shown to be satisfying in order to guide process development and optimize protein purification processes.

We will highlight applications of this approach for case studies with in-silico generated data and experimental data from multivariate studies. In particular we will discuss statistical and computational methods that were applied to evaluate the model prediction quality. Finally we will also address challenges and gaps which need to be addressed for a successful implementation of data-based technologies as a tool in purification process development.

BIOT 131

Alternative downstream processing strategies to resolve high conductivity issues for the purification of biopharmaceuticals

Elke Prade, [email protected], Stefan Oelmeier, Ingo Gorr. Early Stage Bioprocess Development, Boehringer Ingelheim, Biberach, Germany

An early stage downstream processing (DSP) platform approach provides standardized procedures to ensure a robust and time-effective purification of biopharmaceuticals. In our current protocol, acetate is the primary buffering agent used for Protein A elution and acid treatment. The extensive use of acetate has raised concerns, as it introduces high conductivities into the purification process. The Anion Exchange (AEX) column, responsible for the clearance of impurities such as host cell proteins (HCPs), DNA and endotoxins, is particularly sensitive to ionic strength, which interferes with binding conditions to the resin. Interestingly, the extent of conductivity introduced to the sample seems to be dependent on the molecule type. We evaluated alterations to the platform process in order to ensure efficient AEX binding conditions necessary for DNA and virus safety. Firstly, the buffering system was reassessed by testing various buffers for Protein A elution, acid treatment and neutralization steps with regard to reducing overall conductivity. Variations were applied to the buffering agent as well as the employed molarity. Changes to the buffering system require careful considerations, as various buffering capacities are required throughout the process and need to be ensured by the new buffering agents. Secondly, we evaluated the use of salt-resistant AEX resins in terms of enhancing HCP clearance by an order of magnitude. Both approaches were carried out for a panel of molecule types to investigate the root cause for the differing behavior of molecules throughout the DSP purification process. A stronger understanding of molecule-dependent purification requirements will improve our platform in terms of robustness and efficiency.

BIOT 132

Development of an overload chromatography step for high titer processes

Xin Xin Lin, [email protected], Alex Seay, Chris Williams. Genentech, South San Francisco, California, United States

Fitting high titer, high demand processes within existing fixed manufacturing facilities is a familiar challenge faced by many project development teams. The application of overload chromatography maximizes mass throughput and minimizes resin and buffer costs while effectively removing impurities and product variants at very high load densities, making it an increasingly utilized mode of chromatography operation. Undesired product variants, such as aggregates, bind preferentially to the resin and displace the monomer species as the binding capacity of the resin is exceeded. Compared to more traditional bind-and-elute or flowthrough chromatography steps, the present knowledge base for developing an overload step is smaller. However, this lack of historical knowledge can be overcome by applying existing high-throughput tools and multivariate experimental designs to develop a robust purification step and improve process understanding.

Using the development of an overload cation exchange step for a high titer molecule, this presentation will describe the use of robotic high-throughput screening methods to identify appropriate parameter ranges for study, detail application of multivariate screening designs to understand the impact of varying controllable process inputs, and discuss considerations for streamlined development in preparation for a QbD validation approach. BIOT 133

Two column mAb purification platform: Evaluation of AEX step for aggregate clearance

Abhijeet Shirke, [email protected], Dharmesh Kanani, Mi Jin. Downstream Processing-CMC Biologics, Teva Pharmaceuticals, West Chester, Pennsylvania, United States

A typical purification process for the mAb product involves one AEX (Anion Exchange) flow-through polishing step for its robust virus clearance capability. Additional impurity (e.g. aggregate, HCPs) clearance by flow through AEX benefits towards development two column mAb purification platform. This paper focuses on interrogation of the aggregate clearance capability of AEX flow through step. Different commercially available AEX resins were evaluated for the aggregate clearance from the low pH VI (virus inactivation) neutralized pool of a mAb product. Poros HQ owing to its unique characteristics was found to be the best resin for aggregate clearance. A Robust Poros HQ based Flow through AEX step with both high aggregate clearance and step yield was developed which enabled a two column purification process. Key considerations for the development an efficient AEX flow through step with optimum aggregate clearance and step yields will be discussed in detail.

BIOT 134

Downstream process control of aggregates using light scattering and real time molecular weight

Adrian Gospodarek1, [email protected], Bhumit Patel1, Michael Larkin2, Sophia Kendrick2, Mark Brower1, Douglas D. Richardson1, Nihal Tugcu1. (1) Merck & Co., Inc., Kenilworth, New Jersey, United States (2) Wyatt Technology Corporation, Santa Barbara, California, United States

Aggregate removal can be a challenge in downstream purification processes for many proteins, including monoclonal antibodies (mAbs). During process development, fractions are typically collected during purification with pooling decision made afterwards with offline Size Exclusion Chromatography. This is both laborious and time consuming. In this work, we demonstrate connecting an AKTA Avant purification system to a Wyatt TREOS: an inline multi-angle light scattering (MALS) detector that directly calculates average molecular weight, Mw, using first principles. With only minimal apriori protein characterization, the monomer, dimer, and higher-order oligomer fraction content can then be determined. The TREOS can also send a start/stop fractionation trigger to the Avant based on a preset Mw or percent aggregate criteria. The Mw analysis and fractionation occurs in real-time with only a 15 second delay, eliminating the need for analysis after purification. We applied this unique Process Analytical Technology (PAT) tool for monitoring and control of average Mw in various purification applications. Both Flowthrough and Bind and Elute Hydrophobic Interaction Chromatography (HIC) were performed where only protein within a preset Mw was collected. In the case of Flowthrough HIC, the loading limit can be more aggressive as loading can be set to not stop until a specified Mw (or aggregate level) is observed in the breakthrough. This maximum usage of a resin’s aggregate binding capacity can reduce the cycles and buffers needed for processing batches. Finally, the TREOS was connected to an Avant in a continuous production process to monitor an Anion Exchange Chromatography step for several days. Any deviation from the expected Mw (and thus aggregate content) was controlled by signaling the AKTA Avant to discard the material. The novel Avant- MALS system demonstrated here could be used as a powerful tool for real-time aggregate monitoring during process purification steps.

BIOT 135

In-line Fourier-transform infrared spectroscopy as a versatile process analytical technology for preparative protein chromatography

Adrian Sanden2, [email protected], Steffen Grosshans2, Matthias Rudt2, Nina Brestrich2, Josefine Morgenstern2, Stefan Heissler1, Juergen Hubbuch1. (1) KIT, Karlsruhe, Germany (2) BLT MAB, Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany

Preparative liquid chromatography has been a staple of purification processes for biopharmaceutical products for decades. Current liquid chromatography processes rely heavily on in-line measurements of UV absorption at 280 nm as a surrogate signal to monitor the separation of the target compound from impurities such as product related species, DNA, host cell proteins and media components. Monitoring the UV signal may indicate deviations from expected process characteristics but does not usually yield any highly specific information. While FTIR is slower and less sensitive, it is more selective than UV/Vis as it can detect molecular vibrations of any heteromolecular bonds. Therefore, FTIR can be utilized for monitoring multiple critical quality attributes of the chromatographic process such as the secondary structure of proteins and process related impurities. Here, a custom experimental setup was implemented to perform in-line Fourier Transform Infrared Spectroscopy (FTIR) measurements to evaluate the potential of using in-line FTIR as a PAT tool in downstream processing. The experimental setup consisted of a BioATR II flow cell in a Bruker Tensor 27 attached to an AKTApurifier chromatography system and was used to monitor different cation-exchange chromatography experiments. The recorded data was pre-treated to account for the influence of salt gradients on the spectrum. Partial least square regression was employed as a means of multivariate data analysis (MVDA) to correlate offline analytics of collected fractions to the FTIR 3D-fields. As a first case study, the elution of a protein mixture consisting of lysozyme and IgG 2 was used as a model system to show the two species can be distinguished using the FTIR signal. Second, the separation of PEGylated protein species was monitored, with PEG being an example of a compound widely used in pharmaceuticals that is not UV- active. Finally, Triton X-100 was added to a feed of lysozyme as a simulated process related impurity and it was shown that the species could be observed in the FTIR 3D- field of the effluent. The proposed PAT tool can potentially be used as an alternative option to monitor downstream processes towards a more complete implementation of the PAT initiative by mitigating limitations of currently used techniques.

BIOT 136

Two-peak elution behavior of therapeutic IgG4 on cation exchange chromatography

Zhiqiang Chen1, [email protected], Mian Wang2, Xuankuo Xu1, Sanchayita Ghose1. (1) Biologics Process Development, Bristol-Myers Squibb, Devens, Massachusetts, United States (2) Northeastern University, Boston, Massachusetts, United States

Two-peak elution behavior of therapeutic IgG4 in cation exchange chromatography (CEX) has been systemically investigated in this study. We report a case study in which IgG1 and IgG4 show significantly different chromatographic profiles in salt gradient and as well as isocratic elution in CEX. Process parameters and several other factors affecting the peak splitting phenomena have been evaluated in this study. Data suggests that, whereas pH, column loading, hold time, resin type, and residence time of elution are important, the hydrophobicity of the resin base matrix is likely a key contributing factor of the two-peak elution behaviors for IgG4. In-line fluorescence results indicate that different binding conformations for IgG4 exist during the two-peak elution process. The impact of the two-peak elution behaviors on CEX process development was also investigated. Finally, a hypothesis is proposed to explain the phenomenon observed. The study may provide insights in platform process development, analytical method development as well as therapeutic IgG4 molecular design in early discovery phase.

BIOT 137

Impact of process parameters on a double-peak elution profile of a mAb during CEX chromatography

Leslie Konher1, [email protected], Vishal Manickam1, Mandakini Sharma2, Brian To3, Min Zhu1. (1) Protein Science, Boehringer-Ingelheim, Fremont, California, United States (2) Project Management, Boehringer-Ingelheim, Fremont, California, United States (3) Clinical Supply/MSAT, Boehringer-Ingelheim, Fremont, California, United States

During processing of an IgG4 (mAb A) on a cation exchange resin an unexpected elution behavior was observed. The molecule exhibited a double-peak elution profile when purified via POROS HS50. The phenomenon was observed at both bench and pilot scales, excluding the impact of column size. Elution pools were subsequently fractionated and analyzed by various analytical methods. Higher levels of acidic variants were observed in the early fractions by cIEF, but the results, together with the elution profile suggested that the double-peak profile was not solely due to separation of charged variants. Peak 1 and 2 were analyzed by UPSEC which showed no major aggregation. In addition, the peaks were analyzed by peptide mapping (non-reduced and reduced) and binding assay (Octet), both showed no detectable differences. When peak 1 or peak 2 was isolated and re-injected onto the same POROS HS50 column, both re-exhibited the same double-peak profile, which further supports the hypothesis that separation is not likely due to charge heterogeneity. In order to elucidate which parameters in the process were contributing to this double-peak elution profile, a number of hypotheses were examined, including on-column aggregation, surfaced induced hydrophobic interaction or self-association and histidine protonation. The impact of Hofmeister series salts in equilibration and wash buffers in combination with resin pore structure and size were also examined. The studies suggest that double- peak profiles in CEX chromatography we observed are complex and can be significantly impacted by minor changes in process parameters such as buffer composition, load conditions and resin type; likely through more than one mechanism, and should be investigated to understand.

BIOT 138

Metabolic engineering strategies for producing oleochemicals in bacteria

Brian Pfleger, [email protected]. Department of Chemical and Biological Enginee, University of Wisconsin Madison, Madison, Wisconsin, United States

Finding a sustainable alternative for today’s petrochemical industry is a major challenge facing chemical engineers and society at large. To be sustainable, routes for converting carbon dioxide and light into organic compounds for use as both fuels and chemical building blocks must be identified, understood, and engineered. Advances in metabolic engineering, synthetic biology, and other bioengineering disciplines have expanded the scope of what can be produced in a living organism. As in other engineering disciplines, synthetic biologists want to apply a general understanding of science (e.g. microbiology and biochemistry) to construct complex systems from well-characterized parts (e.g. DNA and protein). Once novel synthetic biological systems (e.g. enzymes for biofuel synthesis) are constructed, they must be engineered to function inside evolving cells without negatively impacting the host’s physiology.

In this talk, I will describe pathways for producing high-value commodity chemicals derived from fatty-acids and how my group and others have combined synthetic biology and systems biology to improve oleochemical production in bacteria using sustainable feedstocks. The talk will describe the critical regulatory points in native fatty acid metabolism, strategies for deregulating the pathway, and alternatives that by-pass it altogether. I will highlight the use of heterologous plant and bacterial enzymes to alter the chain length distribution of products from common long-chain molecules to higher- value medium-chain analogs. I will also highlight strategies that we have used to produce medium-chain fatty alcohols, the highest value compounds in the class, through engineering of thioesterase and thiolase driven pathways. I will conclude with commentary on the remaining barriers to commercializing these technologies and areas where further research investment could prove fruitful.

BIOT 139

MAT & EMBER: CFD modeling framework for optimizing mixing and aeration- agitation strategies in mixing tanks and bioreactors

Ahsan Munir, [email protected], James Fisher, Gregory T. Frank, Tim Kellogg, Pablo Rolandi. Process Development, Amgen Inc., North Andover, Massachusetts, United States

Bioreactors & mixing tanks are the workhorse of the biopharmaceutical industry. It is therefore imperative that both mixing and various gas-liquid interactions in these systems should be well understood and characterized. To help reduce experimentation and provide an in-silico modeling platform for accelerated troubleshooting, design assessment and optimization of process technology transfers, we have developed Computational Fluid Dynamics (CFD)-based modeling frameworks called Mixing Analysis Toolkit (MAT) and Eulerian Multiphase Bioreactor (EMBER) model. MAT is based on single phase turbulent computational fluid dynamics (CFD) methods and computes several important performance parameters for both Newtonian and Non- Newtonian fluids in both batch and flow process conditions. EMBER is an Eulerian- Eulerian multiphase turbulence model coupled with a population balance model (PBM) to predict kLa (mass transfer coefficients).We have successfully used MAT and EMBER for tech transfer, root-cause analysis, risk mitigation, and design assessments during commercial operations at Amgen. In this presentation, we will showcase MAT and EMBER models and discuss few examples were these models were used to compute mixing performance, estimate mass transfer coefficient (kLa), and develop an information-rich in-silico design space for optimizing scale-up/scale-down strategies during process technology transfer. Both MAT & EMBER provides a sound and rigorous framework for early integration in process development activities in order to provide significant opportunities for in-silico process understanding, performance optimization and quality assurance. These models are enabling process engineers and scientists to tackle some of the challenges associated with process technology transfers, increase agility and gain speed.

BIOT 140

Introduction of a flexible 2000L single-use bioreactor platform: Technology transfer and process-fit challenges in downstream processing

Phillip R. Smith, James E. Crawford, David N. Paolella, Antonio R. Ubiera, [email protected]. GlaxoSmithKline, King of Prussia, Pennsylvania, United States With the primary purpose of streamlining technology transfer and enabling accelerated “scale-out” approaches for clinical and commercial manufacturing, GSK is expanding the mAb manufacturing supply chain network to adopt a flexible 2000 L single-use bioreactor platform. The increase in flexibility gained, particularly for multi-product manufacturing across a broad range of products with varying cell culture productivities, has resulted in a re-evaluation of the facility-fit methodology that is applied in downstream equipment selection and unit operation design during product-specific technology transfer. In this presentation, using select examples, we will discuss key challenges and the development of a process-fit methodology for maintaining operational flexibility while minimizing potential clinical product comparability risks that may arise during transfer and scale-out. A key element to be discussed is the introduction of multi-day and accelerated multi-cycle chromatography approaches that selectively incorporate single-use technologies for closed processing while using existing clinical manufacturing equipment and column hardware. The developed approach remains in accordance with classic downstream process scale-up methods and ensures functional equivalence of the process, relative to that expected in commercial production.

BIOT 141

Maintaining product quality across multiple site-to-site transfers for a commercial mAb: Some successes and lessons learned

Scott A. Tobler, [email protected], Alexander G. Tulloch. Biologics Process Development and Commercialization, Merck and Co., Inc., Kenilworth, New Jersey, United States

A monoclonal antibody has been approved and commercialized, which included the process development packages often used to understand the process variability and support the control strategy for the drug substance (DS) manufacturing process. Since its approval, the DS process has been transferred to additional manufacturing sites to ensure product supply. When performing a site-to-site commercial tech transfer, the operational parameter controls should be maintained, and the performance of the process is evaluated at the new site through process and product comparability exercises. In addition, continuous process verification and process monitoring efforts within and across the sites would be set up to ensure consistent performance and flag any process drift. This talk will highlight some successes and lessons learned with respect to performance and product quality for this DS manufacturing process implemented at multiple sites, including comments on what process understanding knowledge was known from the original process characterization package versus learned only through ongoing commercial manufacturing (and tech transfer) history.

BIOT 142

Tech transfer: The challenges and opportunities represented by a rapidly evolving product pipeline and global manufacturing network Patrick Gammell, [email protected]. Amgen, Cambridge, Massachusetts, United States

As with many other organizations Amgen has made significant advances towards the standardization of the tech transfer business process to ensure successful and consistent performance between sending and receiving sites, in particular for antibody based fed batch processes. Today however, the growing diversity of pipeline molecules resulting from a biology first, modality independent approach that leverages multiple manufacturing technology options has led to new unknowns and challenges, in particular, new sources of process-input variation. This presentation will describe the application of prior lessons learned to new modalities and intensified processes. This presentation will also review the challenges associated with new materials and process technologies. Approaches for the systematic assessment of development data, prior platform knowledge and manufacturing process data will be presented as a means to manage and overcome these challenges.

BIOT 143

Challenges overcome and robustness improvements in technology transfer of formulation and filling of a vaccine pre-filled syringe

Robert P. Kasprow, [email protected]. WP37B-212, Merck and Co Inc, West Point, Pennsylvania, United States

A case study is provided for an internal, site-to-site, technology transfer of formulation and filling of a pre-filled syringe image for a commercial vaccine product. Differences between sending and receiving sites (intended improvements or due to facility fit) were formally assessed for impact as part of process development. Manufacturing robustness was enhanced through implementation of single-use components for buffer preparation and fluid handling, on-line dose verification, modified elastomeric tubing, and electronic batch records. A statistically-based approach was used to define an appropriate increased sampling plan and to assess the Process Performance Qualification (PPQ) results to gauge the robustness of the process to withstand hypothetical future shifts in mean and/or variability. Additionally, PPQ results were assessed versus historically- derived comparability ranges as well as against the results of “sister batches” manufactured contemporaneously at the sending site using the same Drug Substance inputs. Challenges experienced in process development and PPQ due to process, equipment, and components will be discussed along with their applicability to other technology transfers.

BIOT 144

Application of small-scale modeling to de-risk filling processes of drug product formulations David M. Saldana1, [email protected], Simon Hanslip2, Paula Chaffin3, Mark Palmer4, Aarti Gidh5, Igor Rusanov5, Rachel Forcino3, Monica Dimps6, Kashappa Goud Desai5, Ian Kemp3. (1) Biopharm & Steriles COE, GSK, King of Prussia, Pennsylvania, United States (2) UKMBACTEC NPI & PT SIBU Technical, GSK, Barnard Castle, United Kingdom (3) R&D Platform Technology & Science, GSK, King of Prussia, Pennsylvania, United States (4) Early Stage Parenteral Device Development, GSK, Ware, United Kingdom (5) Biopharm Product Sciences, GSK, King of Prussia, Pennsylvania, United States (6) Production/Manufacturing Operations, GSK, King of Prussia, Pennsylvania, United States

Unanticipated interactions of product attributes with manufacturing filling process parameters represents a risk to successful commercialization, in particular for high concentration monoclonal antibody (mAb) formulations and other unique drug product formulations. Identifying these interactions ahead of pivotal clinical manufacture presents an opportunity to de-risk commercial process steps early in product development and facilitate the accumulation of process knowledge ahead of product licensure. This presentation will describe the progression of an integrated modeling approach to filling operations that originated from needle clogging observed during liquid filling of a high concentration mAb during late stage tech transfer. A case study in the development of a standardized modeling system aligned with commercial platform processes will be presented. The benefits of this approach are anticipated to lead to aligned process development with platform manufacturing processes, reduced line time at manufacturing facilities for engineering and development studies, as well as increased process knowledge appropriate for control strategy development in support of product licensure.

BIOT 145

Driving change in biomanufacturing through innovation in processes, technologies and operations

Eliana Clark, [email protected]. Biogen, Cambridge, Massachusetts, United States

A critical step in meeting the demand of biologic production worldwide involves implementing disruptive manufacturing technologies, processes and capabilities. This talk will discuss Biogen’s new manufacturing site in Switzerland, due to go online in 2019, including the new processes, operational models and technologies being adopted to drive value through innovation and deliver new medicines in areas such as Alzheimer’s .

BIOT 146

Polymer conjugation to enhance cellulase activity and preserve thermal and functional stability Thaiesha A. Wright2, [email protected], Melissa Lucius2, Benjamin Schmitz2, Kevin Burridge2, Katherine Makaroff2, Jamie Stewart2, Henry Fischesser2, Jerry Shepherd2, Jason Berberich1, Dominik Konkolewicz3, Richard C. Page3. (1) Chem, Paper, Biomed Eng, 64 Eng Bldg, Miami University, Oxford, Ohio, United States (2) Chemistry and Biochemistry, Miami University, Oxford, Ohio, United States (3) Department of Chemistry and Biochemistry, Miami University, Oxford, Ohio, United States

With an increasing demand for energy, several promising options that complement fossil fuels have arisen, specifically cellulosic ethanol. A significant challenge in using cellulose derives from linkages in the cellulose backbone, which cannot be hydrolyzed by yeast and many other fermenting organisms. Therefore, it is imperative to develop stable and efficient cellulase enzymes that can degrade these linkages to produce sugars that are fermentable by yeast and other microbes; as well as the ability of the enzyme to tolerate harsh conditions. Here, we describe the production of bioconjugates between polymers and a thermophilic cellulase, FnCel5a, to create an active cellulase with resistance to heat and chemical denaturation. This enzyme was conjugated with various functional polymers including cationic, anionic, and strongly and weakly hydrogen bonding polymers. The activity of FnCel5a toward a high-molecular-weight carboxymethyl cellulose substrate was enhanced by polymer conjugation, greatest enhancements were suggested to be caused by polymers capability of noncovalent interactions with the substrate. The conjugates were found to have nearly identical thermodynamic stability to the native enzyme, as assessed by free energy, enthalpy, and entropy parameters extracted from differential scanning fluorimetry. Polymers tended to confer comparable tolerance to high concentrations of dimethylformamide, with longer polymers typically enabling higher activity relative to shorter polymers. The new FnCel5a conjugates represent an advance in the production of cellulases that maintain activity at high temperatures or in the presence of denaturing organic solvents.

BIOT 147

Care packages: Developing a cell-free glycoprotein synthesis platform for portable, on-demand antibacterial vaccine production

Jessica C. Stark1, [email protected], Thapakorn Jaroentomeechai2, Mingji Li2, Rachel S. Dubner5, Karen J. Hsu3, Taylor C. Stevenson4, Matthew P. DeLisa2, Michael C. Jewett1. (1) Chemical and Biological Engineering, Northwestern University, Evanston, Illinois, United States (2) Chemical and Biomolecular Engineering, Cornell University, Ithaca, New York, United States (3) Mechanical Engineering, Northwestern University, Evanston, Illinois, United States (4) Biomedical Engineering, Cornell University, Ithaca, New York, United States (5) Biological Sciences, Northwestern University, Evanston, Illinois, United States

The growing threat of antibiotic resistance necessitates new strategies for rapid development and production of antibiotics and vaccines. Cell-free protein synthesis (CFPS) is an emerging technology that promises to enable rapid therapeutic production due to the ability to synthesize proteins in under 24 hours. However, existing CFPS systems are limited in their ability to efficiently co-activate protein synthesis and a key post-translational modification: glycosylation. Glycosylation, or the attachment of sugars to proteins, is required for many classes of protein therapeutics, including a class of antibacterial vaccines called conjugate vaccines. FDA-approved conjugate vaccines targeted against Haemophilus influenzae, Streptococcus pneumoniae, and Neisseria meningitidis have been successfully used to prevent infections and the emergence of drug-resistant strains of these pathogenic bacteria. However, purified conjugate vaccine formulations must be refrigerated, making it challenging, if not impossible, to carry out vaccination campaigns in remote or resource-limited settings. Here, we develop a cell- free glycoprotein synthesis (CFGpS) platform by co-activating in vitro protein synthesis and glycosylation to enable conjugate vaccine production at the point-of-need. First, we demonstrate that E. coli cell lysates can be selectively enriched with active sugar precursors called lipid-linked oligosaccharides (LLOs) and/or the oligosaccharyltransferase (OST) from Campylobacter jejuni via overexpression of heterologous genes in the chassis strain. Next, we show that transcription, translation, and glycosylation can be co-activated to enable one-pot synthesis of glycoproteins in reactions lasting 20 hours. With the ability to produce glycoproteins in vitro, we use the platform to produce antibacterial vaccines directed against the highly infectious pathogenic bacterium Franciscella tularensis. Importantly, we are able to synthesize and glycosylate a variant of maltose binding protein (MBP), which has been previously shown to elicit humoral and cell-based immunity to polysaccharide antigens in mouse models. Finally, we demonstrate that the cell-free platform can be lyophilized for distribution and decentralized production of antibacterial vaccines. The CFGpS platform uniquely enables rapid, in vitro synthesis of glycoproteins inside and outside of the laboratory, which opens the door to production of customizable antibacterial vaccines at the point-of-care.

BIOT 148

Targeting glyco-immune checkpoints for cancer immune therapy

Carolyn R. Bertozzi, [email protected]. Chemistry, Stanford University, Stanford, California, United States

Potentiating the immune system’s ability to recognize and destroy cancer cells is an exciting new frontier in oncology. T cell checkpoint inhibitors, CAR-T cells and cancer vaccines are examples of modalities that have achieved unprecedented clinical success. However, many cancers don’t respond to these treatments, suggesting that as yet untargeted modes of immune suppression are contributing to disease progression. This presentation will focus on our identification of glyco-immune checkpoints that contribute to immune cell suppression in the tumor microenvironment. To intervene in these pathways we developed antibody-enzyme conjugates that remodel the cancer glycocalyx and unleash immune cell activity against cancer cells. Our progress toward development of cancer therapies based on these constructs will be discussed. BIOT 149

Enzymatic construction of EpCAM targeting delivery systems by protein farnesyltransferase

Yi Zhang1, [email protected], Feng Xu1, Jonas Schaefer2, Andreas Plückthun3, Mark D. Distefano1. (1) Department of Chemistry, University of Minnesota, Minneapolis, Minnesota, United States (2) Department of Biochemistry, University of Zurich, Zurich, Switzerland

The design and construction of antibody-drug conjugates for targeted cancer therapy is one of the fastest growing areas of biomedical research. Compared to the conventional chemical labeling of the antibodies on the lysine or cysteine residues, site-specific modification via chemoenzymatic methods to construct homogeneous products is becoming increasingly important. In addition to the usage of antibodies as the targeting moiety, numerous alternative protein scaffolds are being studied, which are usually highly stable, small in size and produced in bacterial cultures. These distinct characteristics may enable them to penetrate deeper into tissues to reach solid tumor sites as well as simplify conjugation to drugs or imaging agents. Here, we described the labeling of an EpCAM-binding Designed Ankyrin Repeat Protein (DARPin) as a targeted delivery system. To ensure site-specific modification, protein farnesyltransferase (PFTase), an enzyme that catalyzes the transfer of isoprenoids to proteins containing a C-terminal tetra-peptide sequence was employed. In contrast to direct chemical labeling approaches, enzymatic protein labeling using PFTase is fast, performed under mild conditions and most importantly, site-specific among other cysteine residues present in the protein. PFTase substrate analogues containing various bioorthogonal functional groups that can be incorporated to proteins of interest have also been developed to meet different applications. The DARPin protein being studied here was engineered to contain a C-terminal CVIA sequence for enzyme recognition. Successful labeling of the DARPin with an aldehyde-containing substrate analogue was confirmed by LCMS. Using an oxime ligation reaction, a DARPin-fluorophore conjugate was constructed. The specific binding to cell surface EpCAM and subsequent internalization of the conjugate were validated using flow cytometry and confocal imaging analysis. No binding was observed when a non-binding DARPin that does not recognize EpCAM was modified in the same way. We are currently preparing DARPin-toxin conjugates and will test their cytotoxicity in cell cultures. Future work will focus on employing these constructs for both therapeutic and imaging applications in mouse models.

BIOT 150

Engineering a blue light inducible SpyCatcher system (BLISS) for in vitro photo- pattering of proteins and optically controlled protein rescue

Emily Hartzell2, [email protected], Justin Terr2, Wilfred Chen1. (1) Chemical Engineering, University of Delaware, Newark, Delaware, United States (2) Chemical and Biomolecular Engineering, University of Delaware, Newark, Delaware, United States

The SpyTag-SpyCatcher protein conjugation system has recently gained popularity due to its fast kinetics and high yield under biologically favorable conditions in both in vitro and in vivo settings. We can further expand the utility of this system by introducing the ability to spatially and temporally control the conjugation event. For intracellular applications, the light-oxygen-voltage 2 domain of Avena sativa (AsLOV2) has been used for modulating the affinity of protein-peptide ligand interactions in response to blue light. We have thus genetically fused the SpyTag to the AsLOV2 domain to create a Blue Light Inducible SpyCatcher System (BLISS). In the dark state, the SpyTag is blocked from reacting with the SpyCatcher, but upon irradiation with blue light, the Jα- helix of the AsLOV2 undocks to expose the SpyTag. We have demonstrated the ability to use light to biologically pattern surfaces with proteins. We have also applied this system for the optogenetic rescue of intracellular proteins, where a protein targeted towards degradation is cleaved from its degron tag in response to the BLISS protein conjugation event. We anticipate the precise control our strategy provides will be a strong tool for making protein microarrays and crafting biomaterial composition as well as controlling therapeutic protein activity in select cells.

BIOT 151

Enhancing the chemical versatility of yeast display to discover potent, specific enzyme inhibitors

Jessica T. Stieglitz, Haixing P. Kehoe, Laura Quinto, Gregory Berumen, Jacob B. Lissoos, James A. Van Deventer, [email protected]. Chemical and Biological Engineering, Tufts University, Medford, Massachusetts, United States

Although cancer-promoting activities of extracellular proteases and peptidases in the tumor microenvironment are well known, the disruption of individual enzymes in this environment remains a fundamental challenge. We are exploring the hypothesis that integrating additional chemical functionality into binding proteins will lead to potent inhibitors that simultaneously exploit the best features of proteins and small molecules. In order to construct, evaluate, and screen protein-small molecule “hybrids,” we have established a noncanonical amino acid-compatible (ncAA-compatible) yeast display platform. To broaden the range of chemical functionality that we can use with our platform, we are using yeast display-based reporter constructs with tags before and after the ncAA insertion site to quantitatively evaluate ncAA incorporation efficiency and fidelity. In multiple systems, we observe stop codon readthrough efficiencies of 10-25 percent and misincorporation of canonical amino acids below 20 percent using flow cytometry-based readouts. This performance is suitable for some applications, but we continue to work on identifying enhancements to broaden the capabilities of the platform. To exploit the system’s existing capabilities, we are investigating strategies for incorporating ncAAs into antibody fragments at positions located close to and within antibody complementarity determining regions. We have identified several positions that tolerate ncAA incorporation and facilitate chemical modification using copper-catalyzed azide-alkyne cycloadditions. Experiments with five antibody variants have uncovered general reactivity trends: 1) conserved sites react similarly in different antibodies; 2) subtle changes in amino acid side chain properties affect chemical reactivity; 3) changes in the structures of small molecules affect reactivity. We have exploited our understanding of these trends to construct “hybrid” structures containing functional groups suitable for disrupting major classes of enzymes including metalloproteinases and serine hydrolases. Efforts to identify antibody-based inhibitors that leverage this expanded range of chemical functionality are ongoing. We anticipate that our platform will lead to inhibitors with enhanced specificity and potency suitable for studying the biology of the tumor microenvironment, and that “hybrids” may also serve as new therapeutic leads.

BIOT 152

Engineered binding proteins as replacements for antibodies in immunoassays

Hadley D. Sikes, [email protected], Eric A. Miller, Ki-Joo Sung. Chemical Engineering, MIT, Cambridge, Massachusetts, United States

Antibodies are unrivaled as molecular recognition agents in the context of the immune system of an organism. However, in the context of an in vitro device, certain opportunities for improvement can be readily identified. Engineering strategies and accomplishments towards replacing antibodies in medical diagnostic tests with engineered binders produced from the reduced charge Sso7d protein scaffold (rcSso7d) will be presented. Differences of these engineered binders from antibodies include twenty-fold smaller size, one-hundred-fold better thermal stability, and the potential to engineer universal compatibility of each clone with the in vitro test format while avoiding the cross-reactivity that is often observed with antibodies. Each of these differences comes with both challenges and opportunities that will be discussed in terms of how to modify the binding molecule so that it functions when i.) coupled to a solid support and ii.) coupled to a label that signifies capture of an analyte molecule.

BIOT 153

Optimization of heterologous mevalonate production in E. coli with a combinatorial promoter library and novel biosensor

Jacob Englaender1, [email protected], Alex W. Murfee1, Kevin Solomon1,2. (1) Agricultural & Biological Engineering, Purdue University, West Lafayette, Indiana, United States (2) Laboratory of Renewable Resources Engineering, Purdue University, West Lafayette, Indiana, United States

Mevalonate, an intermediate in the mevalonate pathway of eukaryotes and archaea, is a precursor in the production of biologically important terpenes, such as the anticancer agent taxol and the antimalarial prostratin. These molecules are naturally produced by plants, where producing large quantities can prove problematic due to challenges in plant culture such as aggregation and oxygen transfer. As a result, more robust bacterial production appears to be a promising avenue for optimization. The upper mevalonate pathway consists of three genes responsible for the conversion of acetyl- CoA to mevalonate. Mevalonate is then converted by the lower pathway to isopentyl pyrophosphate (IPP) and dimethylallyl pyrophosphate (DMAPP), which are used by the cell for terpene and steroid production. Many optimization strategies have been employed to increase mevalonate production across multiple species including both bacteria and yeast by focusing on fermentation optimization and pathway gene homologs. Here, we have modified the previously described ePathOptimize[SKV1] system, in conjunction with a Golden Gate cloning strategy, to construct a combinatorial promoter library, consisting of six promoters of variable strength, to modulate the expression of each of three pathway genes from the upper yeast mevalonate pathway to optimize mevalonate production in E. coli. A novel biosensor was developed to facilitate screening of a large library of 216 promoter gene combinations for high- producing strains. Our biosensor couples a promoter identified from transcriptomic (RNA-Seq) data with a fluorescent reporter protein to facilitate high throughput library screening. This work represents the first utilization of RNA-Seq for the optimization of the mevalonate pathway. The optimization of mevalonate production is an important first step for the production of higher complexity terpenes such as taxol and its precursors in E. coli.

BIOT 154

Exploring non-natural pathways and native efflux systems for improved aromatic chemical production in Escherichia coli

Michael Machas, [email protected], Gavin Kurgan, Xuan Wang, David Nielsen. Arizona State University, Tempe, Arizona, United States

Aromatic chemicals represent promising metabolic engineering targets with a wide range of applications. 2-Phenylethanol (2PE), for example, is a key molecule used in the fragrance and food industries, as well as a potential biofuel additive. A non-natural, alternative route to the established yeast Ehrlich pathway for microbial 2PE production was developed via the modular extension of a previously-engineered styrene pathway. When directly compared using a host strain engineered for L-phenylalanine over- production, preservation of phosphoenolpyruvate, and reduced byproduct formation, final 2PE titers and yields were each increased >50% using this novel “styrene-derived” pathway. After further optimization of induction timing and glucose loading, 2PE titers now approach 2 g/L, nearing the toxicity threshold. End-product toxicity is not unique to 2PE, but rather represents a substantial bottleneck limiting high production metrics for a range of value-added chemicals. To address this in the case of aromatics, tolerance engineering strategies were also explored by exploiting E. coli’s native multidrug resistance (MDR) efflux systems. Although offering the potential to reduce intracellular toxic chemical exposure and increase bioproduct export, MDR efflux pump specificity towards most industrially relevant compounds, including aromatics like 2PE, lacks comprehensive characterization. Towards this aim, a library of known and putative E. coli membrane transporters (primarily composed of MDR pumps) was screened for their ability to confer increased tolerance to a series of aromatic bioproducts. Utilizing a growth-based plasmid enrichment strategy applied at inhibitory concentrations followed by detailed toxicity tests of recovery isolates, effective pumps were identified for several aromatic compounds of interest. Among positive isolates, those of the small multidrug resistance (SMR) family were particularly prevalent. For example, SugE was found to increase tolerance for multiple aromatic chemicals. By this approach, efflux pumps were isolated whose subsequent overexpression improved E. coli growth rates in the presence of toxic compounds by nearly 3-fold.

BIOT 155

Building a bridge between cell-free experimentation and cellular metabolic engineering

Ashty Karim1, [email protected], Quentin Dudley1, Alex Juminaga2, Samantha Crowe1, Jacob Heggestad1, William Grubbe1, Yongbo Yuan2, Rasmus Jensen2, Sean Simpson2, Michael Koepke2, Michael C. Jewett1. (1) Chemical and Biological Engineering, Northwestern University, Evanston, Illinois, United States (2) LanzaTech, Skokie, Illinois, United States

Speeding up design-build-test (DBT) cycles is a fundamental challenge facing metabolic engineering. To address this challenge, we report a new in vitro prototyping and rapid optimization of biosynthetic enzymes approach (termed iPROBE) to inform cellular metabolic engineering. In our approach, cell-free cocktails for synthesizing target small molecules are assembled in a mix-and-match fashion from crude cell lysates selectively enriched with pathway enzymes. This approach reconstructs pathways in two steps where the first step is enzyme synthesis via cell-free protein synthesis and the second step is enzyme utilization via substrate and cofactor addition to activate small molecule synthesis. We demonstrate that iPROBE can quickly study pathway enzyme ratios, tune individual enzymes in the context of a multi-step pathway, screen enzyme variants for high-performance enzymes, and discover enzyme functionalities. The rapid ability to build pathways in vitro using iPROBE allows us to generate large amounts of data describing pathway operation under several operating conditions. However, to date no easy method of analysis provides informative bridging of cell-free data to cellular metabolic engineering. In this work, we address this limitation by developing a quantitative metric that combines titer at reaction completion, rate during the most productive phase of pathway operation, and enzyme expression as measured by protein solubility (TREE score). By reducing the complexity of available cell-free data to one value we can now quickly screen and rank pathways in the cell-free environment and provide useful information for cellular metabolic engineering. We demonstrate iPROBE and the use of the TREE score for the production of 3-hydroxybutyrate and n- butanol in Clostridium, an industrially relevant non-model organism. This work shows that iPROBE can be used for multiple enzymatic pathways and for non-model organisms. We anticipate that iPROBE will facilitate efforts to define, manipulate, and understand metabolic pathways for accelerated DBT cycles in the cell-free environment before engineering organisms.

BIOT 156

Use of an Escherichia coli pyruvate-overproducing platform strain to produce L- valine

Paul A. Adamczyk2, [email protected], Shu Pan4, Xiaolin Zhang3, Jennifer Reed1. (2) Chemical and Biological Engineering, University of Wisconsin - Madison, Madison, Wisconsin, United States (3) Chemical and Biomolecular Engineering, University of Delaware, Newark, Delaware, United States (4) Chemical and Biological Engineering, University of Wisconsin-Madison, Madison, Wisconsin, United States

Bio-catalytically derived, carbon-neutral chemicals are becoming increasingly attractive over traditional catalytic methods, but are impeded by the slow, metabolic engineering design-build-test cycle. To expedite this process, a precursor-abundant strain to industrially relevant chemicals will enable high-throughput engineering of diverse biosynthesis pathways. To this end, the metabolic potential of Escherichia coli was assessed computationally and pyruvate identified as the optimal precursor to produce commercially relevant chemicals. Subsequently, a genome-scale metabolic model of E. coli was used to design strains that achieved astonishing pyruvate yields up to 95% of the theoretical maximum. Next, as a proof-of-concept, L-valine was chosen as a target for overproduction to test the platform capabilities of a pyruvate-producing strain since it is an amino acid widely used as a nutritional supplement in several industries with a global demand of about 500 tons annually. Currently, a pyruvate-producing E. coli strain harboring a plasmid expressing L-valine biosynthetic genes has been engineered to achieve 69% of the theoretical maximum yield, the highest reported for E. coli, as opposed to 39% for wild-type E. coli with the same plasmid, in approximately 22 hours of aerobic culturing in a defined, glucose minimal medium. This study demonstrates that a pyruvate platform strain is, indeed, a versatile tool to rapidly engineer strains to produce various chemicals.

BIOT 157

Tailoring chain-length specificity for oleochemical production in bacteria

Nestor J. Hernandez Lozada2, [email protected], Brian Pfleger1. (1) Department of Chemical and Biological Enginee, University of Wisconsin Madison, Madison, Wisconsin, United States (2) Chemical and Biological Engineering , University of Wisconsin-Madison, Madison, Wisconsin, United States

Microbial production of oleochemicals from simple sugars, such as free fatty acids (FFA) and fatty alcohols (FAOH), has been a subject of research interest in recent years since it could provide with new routes to these important molecules. Oleochemicals, which are currently obtained from oily plants such as coconut and canola, are in high demand and has created tensions between food and fuel industries as well as concerns about the high degree of deforestation. One of the challenges in microbial production of FFA and FAOH is the lack of enzymes that are both highly active and specific towards a particular carbon chain length. As a result, research in this area is often categorized between strains that are either able to produce high titers of FFA and FAOH with mixed chain lengths or strains that have low titers with high chain length specificity. A key step that controls this specificity is the thioesterase, which catalyze the last step in microbial FFA production strategies by hydrolyzing the thioester bond in cellular fatty acyl-ACP producing a FFA. Since this step commits the fatty acyl chain length to its current size it is critical to have a thioesterase that is both highly active and specific towards the desired chain length to drive its production. We developed a screen that relies on high production of a particular chain length of FFA for cell survival. The screen was used to improve thioesterase activity on a library created through random mutagenesis in a low activity plant thioesterase. With this approach we were able to achieve >1 g/L FFA titers with high specificity for the desired chain length. This was then used to assembly the pathway for FAOH production from simple sugars and we were able to obtain >1 g/L of FAOH and high specificity for our desired chain length. These results highlight the potential of fatty acid biosynthesis pathway as a route for the production of FFA and FAOH.

BIOT 158

Understanding lipogeneity in oleaginous yeast: source of reducing equivalents for proliferating cell and non-proliferating cell

Peng Xu, [email protected]. Chemical and biochemcial engineering, University of Maryland, Baltimore County, Baltimore, Maryland, United States

Nitrogen starvation induces lipogenesis in oleaginous yeast. It is generally believed that the depletion of AMP represses TCA metabolic activity, then the overflown citrate is directed to cytosolic acetyl-CoA for lipid biosynthesis. The switch from biomass to lipid synthesis traps the cell at non-proliferating state, where the non-lipid biomass keeps constant and almost all carbon flux is directed towards lipid synthesis. It is true that proliferating cancer cell relies on pentose phosphate pathway to generate both NADPH and ribose for cell dividing. While for non-proliferating oleaginous yeast, the sourcing of NADPH from oxidative pentose phosphate pathway, appears economically unfavorable for cell to survive: it is a waste of carbons to operate pentose pathway without the needs for synthesizing DNA and proliferating. Here we present new evidence and hypothesis that reducing equivalents in non-proliferating oleaginous yeast may come from alternative pathways. These findings will be important for us to understand lipogeneity in oleaginous yeast and may facilitate the development of sustainable biorefinery platforms that upgrade lipoyeast metabolic engineering.

BIOT 159 Transcriptional-sensor based increase in peroxisomal fatty acyl-CoA flux improves fatty alcohol production in Yarrowia lipolytica

Murtaza Shabbir-Hussain, Michael Spagnuolo, [email protected], Mark A. Blenner. Chemical and Biomolecular Engineering, Clemson University, Clemson, South Carolina, United States

Fatty alcohols currently find use in areas such as surfactants, plasticizers, lubricants, fuels, and the cosmetics industry; however, traditional production methods rely on petroleum-derived compounds or the low-yield harvest from plants. Recently, conventional hosts including E. coli and S. cerevisiae have been used for fatty alcohol production with some success. Oleaginous yeast, such as Yarrowia lipolytica, offer significant advantages as a production host for oleochemicals, as the native metabolism already has high flux through its fatty acid biosynthetic pathway; however, efforts to engineer significant fatty alcohol production in this organism have been unsuccessful. Fatty alcohol production in the cytoplasm is hindered by low acyl-CoA availability and competing reactions. As a result, we have targeted fatty alcohol production to the peroxisome, where acyl-CoA flux is directed during beta-oxidation, and where there are comparatively fewer potential competing reactions. Several acyl-CoA reductases from bacterial and mammalian sources were screened. Gene knockouts and overexpressions were investigated with a recently developed peroxisomal fatty acid sensor for Y. lipolytica. Strains with improved peroxisomal fatty alcohol production were found to lack cytoplasmic acyl-CoA synthetase activity. When expressing acyl-CoA reductase, a strain identified through sensor based screening resulted in over 500 mg/L without additional optimization. We will further report on our recent efforts to improve titer and alter the fatty alcohol chain length profile.

BIOT 160

Engineering gut fungal membrane-embedded transporters for metabolic engineering in Saccharomyces cerevisiae

Igor Podolsky1, [email protected], Susanna Seppälä1, Michelle A. O'Malley2. (1) University of California Santa Barbara, Santa Barbara, California, United States (2) Mail Code 5080, University of California, Santa Barbara, Santa Barbara, California, United States

Dwindling fossil fuel resources and food security concerns motivate production of a wide range of chemicals and materials from renewable, abundant sources such as lignocellulosic biomass. Treatment of lignocellulosic biomass releases an array of sugars that are used by microbial cell factories (e.g. Saccharomyces cerevisiae) for biofuel-, biomolecule-, and biomaterials production. Economic productivity on lignocellulosic hydrolysates using S. cerevisiae requires co-utilization of abundant hexose and pentose sugars during fermentation. However, metabolically engineered strains will preferentially utilize D-glucose resulting from poor transport of other relatively abundant sugars (e.g. D-xylose) arising from competitive inhibition. Realizing sugar co- utilization for bioconversion purposes requires additional novel carbohydrate transporter discovery and/or protein engineering efforts. Anaerobic gut fungi produce and utilize an array of sugars from raw lignocellulosic biomass in a highly competitive environment, and therefore they are an attractive platform for transporter discovery. We have identified more than 100 carbohydrate transporter sequences from the assembled transcriptomes of three gut fungal isolates, including families novel to fungal organisms (i.e. ABC and SWEET systems). In this study, putative gut fungal carbohydrate transporters were functionally characterized in S. cerevisiae for the first time. Membrane proteomics were used to validate in silico identification and functional annotation of these transporters. In parallel, recombinant expression in the transporter-deficient strain EBY.VW4000 assayed production compatibility and carbohydrate transport function, and showed that gut fungal transporters can recover transport of select carbohydrates. Novel transporter characteristics were characterized using FRET methods and liquid scintillation counting. Furthermore, proof-of-principle evolutionary protein engineering sought to eliminate glucose transport in these transporters in favor of other hexose and pentose sugars. Future work aims to expand discovery using native consortia metatranscriptomes and to realize selective transport through additional engineering efforts.

BIOT 161

Guanidine-containing polymers as flocculating agents and filtration aids: Harvest solution alternatives to the centrifuge?

Jerald K. Rasmussen, [email protected], Andrew Vail, Tonya D. Bonilla, David H. Brandwein. 3M Center, 3M, Saint Paul, Minnesota, United States

Continuous improvements in upstream processing, leading to dramatically increased IgG titers and cell densities, have exerted increased pressures on downstream processing, and particularly on the initial clarification step(s). Previous work from our group has identified the guanidine group as a salt-tolerant ligand, and has demonstrated its utility for the capture of negatively charged proteins, nucleic acids, bacteria, and viruses. Soluble, guanidine-containing polymers were shown to be useful for the flocculation of bacterial and other cell cultures. The present work was undertaken to explore the potential utility of guanylated polymers to provide simplifications to the harvest step. Two different soluble guanidine-containing polymers, guanylated polyethylenimine (G-PEI) and poly(diacetoneacrylamide guanylhydrazone), were evaluated for their ability to flocculate CHO cells. Factors such as molecular weight of the G-PEI, polymer to cell ratio, settling time of the floc, filterability after flocculation, etc., were probed. In addition, we have looked at development of assay methods to quantitate residual polymer in process streams and at toxicity of the guanylated polymers, as well as at alternative solutions, such as immobilizing the guanylated polymer on a porous substrate. Results of these studies, as well as comparison to a well-known flocculant, pDADMAC, will be presented.

BIOT 162 Clarisolve® mPAA polymer: A stimulus-responsive flocculation polymer for cell culture clarification

Michael J. Susienka1, [email protected], Kristen A. Cotoni2, John P. Amara1, Christopher Gillespie2. (1) BioProcessing R&D, MilliporeSigma, Bedford, Massachusetts, United States (2) Next Generation Processing, MilliporeSigma, Bedford, Massachusetts, United States

Flocculation is a simple method to pre-clarify high-density Chinese hamster ovary (CHO) cell culture feed streams, allowing for an efficient clarification process that would otherwise not be feasible using traditional depth filters. Clarisolve® mPAA polymer is a stimulus-responsive flocculation polymer that may provide a reduced dosing dependency and lower residual polymer levels when compared to other flocculants. Cell culture pre-treatment with Clarisolve® mPAA polymer followed by clarification using Clarisolve® 60HX depth filters enables significantly increased harvest throughputs and allows for a single clarification stage for these high-density feed streams. To demonstrate its effectiveness, CHO cell culture was treated with a range of Clarisolve® mPAA polymer concentrations (0.05, 0.1, and 0.2%) to evaluate the impact on the harvest clarification and a subsequent 3-step chromatography process (Protein A, CEX, and AEX). These results were compared to material prepared using conventional Millistak+® HC depth filtration media. The particle size distribution, depth filtration and sterile filtration capacities, process yield, soluble impurity clearance, and residual polymer levels were determined in order to deliver a comprehensive assessment of the effects of Clarisolve® mPAA polymer pre-treatment upon the clarification and subsequent downstream chromatography steps. Clarisolve® mPAA polymer pre- treatment provided a 2x increase in the depth filtration capacity compared to an untreated benchmark/control (Figure 1). Pre-treatment with Clarisolve® mPAA polymer also provided increased host cell protein (HCP) and DNA clearance during clarification. The downstream purification performance was comparable for all feedstreams evaluated.

Figure 1. (A) Pressure vs. throughput for a CHO cell culture feed stream pre-treated with Clarisolve® mPAA polymer at a range of optimal and sub-optimal doses and filtered using Clarisolve® 60HX filters compared to an untreated/benchmark feed stream filtered using dual- stage filtration with Millistak+® D0HC and X0HC filters. (B) Average depth filter throughput at 20 psi (terminal pressure) for each feed stream.

BIOT 163

Inline flocculation for harvest clarification

Akshat Gupta1, [email protected], Juan Castano2, Elizabeth M. Goodrich1. (1) Applications Engineering, MilliporeSigma, Burlington, Massachusetts, United States (2) MSAT, MilliporeSigma, Burlington, Massachusetts, United States

Inline flocculation of cell culture harvest offers various advantages for clarification unit operation. These advantages include accurate dispensing of flocculant, reduced reliance on large scale mixing systems, potentially reduced number of mixing vessels and reduced cleaning validation requirements. The process is also more suitable for adoption in a continuous process. A critical aspect for designing an inline flocculation system is identifying minimum residence time required for achieve a stable particle distribution of flocculates. This information is used to designing an inline system and define process parameters necessary to provide suitable residence time prior to depth filtration. In this work pDADMAC flocculation of Chinese Hamster ovary cell was investigated. Effect of power to volume ratio on residence time required for attaining uniform flocculate particle distribution was studied using a Focused Beam Reflectance Measurement Probe (FBRM) and results were used to design an inline flocculation system. Inline pDADMAC flocculation was performed and clarification using depth filters were demonstrated. BIOT 164

Considerations for process development and implementation of fully synthetic depth filters

Dogan Ornek1, [email protected], Joshua Souther1, Thomas Parker2. (1) Upstream Process Development, Fujifilm Diosynth Biotechnologies, Cary, North Carolina, United States (2) Process Development Scientist, EMD Millipore Corp, Billerica, Massachusetts, United States

As the transition step between upstream and downstream processing, clarification is an important and often complex unit operation in biopharmaceutical manufacturing. The considerations in clarification increase for a Contract Development and Manufacturing Organization (CDMO), as wide ranges of cell culture conditions, various products, and compressed timelines are typical. Depth filtration has commonly been used for clarification of cell culture and post viral inactivation pools in both clinical and commercial manufacturing; however, CDMO’s and companies throughout the biopharmaceutical industry are experiencing process limitations due to increasing cell culture densities, titers, and complexity as well as a push for cleaner materials of construction, making the use of depth filtration for clarification much more involved. The work presented here evaluates the use of a new generation of depth filters consisting of fully synthetic materials of construction, where a silica filter aid and polyacrylic fibers replace diatomaceous earth (DE) and cellulose typically found in depth filters. Transition to synthetic filtration media offers significant improvement from traditional DE-based depth filters in terms of process performance as well as implementation in cGMP manufacturing of biologics. Performance enhancements in terms of volumetric loading capacities and impurity removal over a wide range of harvests with various cell culture conditions are presented. Additionally, the impact of cleaner materials of construction on manufacturing aspects of biologics is provided. This work demonstrates the ability of synthetic depth filters to simplify process development while increasing ease of use in manufacturing over traditional DE-based filters.

BIOT 165

Aligning acoustic wave separation with concentrated fed batch processing

Michael Collins, [email protected]. R&D, PALL LIFE SCIENCES, WESTBOROUGH, Massachusetts, United States

There is a growing interest within the biopharmaceutical industry towards intensified cell culture techniques that show the potential to achieve significantly higher product titres compared to traditional fed batch cultures. One method of doing this is termed concentrated fed batch, or perfused fed batch. The cell culture operates similarly to a perfusion reactor but utilizes an ultrafiltration(UF) membrane rather than a microfiltration membrane. The use of the UF membrane still allows waste products to be removed and also allows fresh media to enter the bioreactor. This technique can result in very high cell densities (>200x106/ml) and mAb titer (25-30g/l). The use of the UF membrane means that the mAb is retained within the bioreactor during the process as opposed to traditional perfusion whereby the mAb is harvested continuously. The challenge with this approach is harvesting the mAb and achieving high product yield, due to the very high cell density and also significant cellular debris concentrated in the bioreactor due to the tight rating of the UF membrane. Traditional harvesting techniques such as depth filtration become impractical under these conditions, alternative centrifuge approach is largely not viable In this work we investigate Acoustic Wave Separation for cell harvest using the Cadence™ Acoustic Separator. This is designed for the single use clarification of mammalian cell culture, including high density feeds, containing up to 100x106/ml TCD range or higher. In this study we apply the technology to a concentrated fed batch process and show how to achieve high product yield (>90%) through a combination of dilution and cell washing while minimizing product dilution. The new approach was then verified at a leading Biopharm company.

BIOT 166

Evaluation of gravity settling for disposable harvest

Diana Lok1, [email protected], Edward Chan1, Ekta Mahajan2, Rick St John1. (1) Genentech, South San Francisco, California, United States (2) Process Development Engineering MS75B, Genentech, South San Francisco, California, United States

Many companies are looking towards moving to fully disposable manufacturing because of the benefits of avoiding infrastructure for cleaning fixed equipment. A key remaining gap for implementation, however, is an effective single-use harvest option. This talk will review the current disposable harvest options available and explore gravity settling followed by filtration in depth.

Gravity settling is a process by which cells are allowed to settle in a bioreactor prior to harvesting through depth and sterilizing grade filtration. Gravity settling is the simplest single-use harvest option, for PCV up to 6.5% and viability over 75% in a 1kL SUB. More work needs to be performed to demonstrate reproducibility for manufacturing. This feasibility evaluation for utilizing gravity settling included 1) time required for settling large scale 2) product quality impact over time 3) filterability

Gravity settling for a 1kL SUB required a minimum of 5 hours up to 20 hours with the feedstocks tested. This is considered a feasible option considering that no significant product quality changes were observed up to 24 hours at ambient temperature. Filter area requirements were reduced by 8-13x.

BIOT 167 Perfusion membrane selection to enable continuous capture chromatography

Rob Piper, [email protected]. Just Biotherapeutics, Seattle, Washington, United States

Continuous processing of biologics offers the promise of lower capital outlay for manufacturing, improved productivity and reduced cost of goods. A continuous, non- product retentive perfusion process capable of delivering consistent titer and product quality would enable the implementation of a continuous capture step. High cell density cultures can often impact perfusion filter lifetime and product passage resulting in changing titer and product quality over the course of a run. To address this, we evaluated several hollow fiber filter configurations and pumping systems with respect to their impact on product passage and filter and cell culture performance. Additionally, we evaluated the effect of perfusion filter choice on downstream processes through turbidity and impurity levels to determine whether a dedicated clarification step is necessary. Finally, we evaluated the impact of perfusion filters on the ability to retain retrovirus-like particles (RVLPs), which could potentially lower the amount of virus clearance needed for the downstream process.

BIOT 168

Ceramic membrane filtration for biopharms: Potential application for the mammalian harvest process

Amelia Ryan, [email protected], Nidhi Thite, Cindy Jung. GSK, Philadelphia, Pennsylvania, United States

The purification process for mammalian processes typically includes a two-step harvest process: centrifugation followed by a depth filtration. The main goal is to remove intact cells and cell debris from the culture fluid. Difficulties with filtration arise due to decreased cell viability and increased cell number, which can cause a significant decrease in filter capacity and potentially an insufficient removal of cell debris. Also, upstream advances to increase productivity are increasing the number of cells harvested, which further decreases filter capacity. Dynamic cross flow filtration with ceramic membranes is an alternative filtration technique to the typical normal flow and crossflow modes. It uses rotating disk filtration where the ceramic membranes are stacked on a spinning hollow shaft. The high crossflow action is generated by the rotating filter discs and results in improved cleaning and therefore capacity. A benefit to using ceramic membranes is the ability to include a backflush during operation, which forces the cell debris off the ceramic membrane surface, resulting in a further increase in capacity. We have demonstrated high cell number mammalian cell culture can be filtered through a 2 µm / 0.2 µm ceramic filter train using dynamic cross flow filtration without an impact to cell integrity. We can achieve an approximate 4X filter loading using the ceramic membranes compared to the standard depth filters used during harvest. In addition to the increased capacity, we have found that the membranes can be cleaned and reused which can significantly reduce cost of goods and ergonomic handling.

BIOT 169

Overcoming cell line limitation and improving medium in a perfusion process for manufacture of a labile and difficult-to-express enzyme

Jeong Lee, [email protected], Brena Williams, Peter Amaya, Chris Klaver. MedImmune, Clarksburg, Maryland, United States

Manufacture of a labile and difficult-to-express protein often requires a perfusion process for maximum productivity and to avoid loss of activity as a result of long residence time in the bioreactor. The benefit of a perfusion process is further increased by maximizing the duration of the harvest phase which can be extended by maintaining a steady state cell density and specific growth rate through continuously bleeding the culture.

A perfusion process was developed for a recombinant enzyme which showed a very rapid activity reduction under cell culture process conditions and was expressed at a very low level in CHO cells in a fed-batch process. The optimized perfusion process improved both productivity and specific activity. However, challenges related to the cell line properties and large scale operations were identified during process development. The cell line used during process development showed a negative correlation between specific productivity (Qp) and specific growth rate (µ). Therefore, productivity and harvest duration had to be balanced in order to maximize yield per run while also maintaining a minimum titer requirement for scalable purification operations. By controlling the cell specific perfusion rate (CSPR) and implementing a variable bleed rate, the harvest duration was extended while maintaining a minimum Qp. Also, during scale-up, it was discovered that the medium storage condition in combination with certain components in the perfusion medium led to lower productivity at the large scale. Modifications in medium storage conditions and increased level of copper sulfate resulted in a significant improvement in Qp.

Strategies used to address these challenges will be discussed in this presentation along with the resulting improvements in process performance.

BIOT 170

Maintaining product quality from early to late stage upstream process development

Lewis Yung, [email protected]. Genentech, South San Francisco, California, United States Maintaining product quality of a monoclonal antibody through all phases of clinical development creates many challenges. Changes in cell line, cell line host, media and process conditions can all affect product quality attributes as a product moves from phase I to phase III development. All of these factors must be balanced with the desire to maintain protein production levels, particularly for the commercial process in order to maintain low COGs. Product quality attributes such as glycosylation, charge variants, isoform and aggregate profiles can be impacted by a number of these factors. In this poster, a case study will be presented on the efforts to maintain product quality when switching from an NS0 cell line in Phase I to a CHO cell line in pivotal clinical studies for an IgG2 monoclonal antibody. Particularly, the effects of media composition and process conditions on product quality attributes will be examined.

BIOT 171

Improving global access to biotherapeutics

Dean Pettit, [email protected]. Just Biotherapeutics, Seattle, Washington, United States

Global access to biotherapeutics is limited by many factors, including the cost of product development and goods manufactured. This presentation will illustrate considerations for accelerating product development and lowering costs of goods manufactured through the integration of molecular design, process and product design, and manufacturing facility design and operation.

BIOT 172

Production and purification of oncolytic measles viruses

Tanja Grein1, Hauke Dieken1, Daniel Loewe1, Denise Salzig1, Tobias Weidner1, Peter M. Czermak1,2, [email protected]. (1) IBPT, University of Applied Sciences Giessen, Giessen, Germany (2) Bioresources, Fraunhofer Institute IME, Giessen, Germany

Oncolytic measles viruses are identified as a promising and potent candidate for cancer treatments. However, for a successful treatment of patients suffering e.g. on multiple myeloma, more than 1011 of infective virus particles are required. To ensure future therapy, this calls for an innovative bioprocess development that includes the integration of both efficient upstream and downstream processes with a suitable in-process control system to further fulfill GMP-standards. Since the measles virus shows a half-life time of just 1 h at 37°C, the virus production is hampered by the poor virus stability. As consequence, maximal yields of the instable product can only be achieved, if the optimal time of harvest (TOH) can be determined precisely. We integrated impedance spectroscopy in the production process for the determination of the optimal TOH. In 16 independent bioreactor runs, the maximum Measles virus titer was achieved approximately 40 h after the permittivity maximum. Standardization of the USP leads to a reduction of batch-to-batch variations which simplifies the optimization of the production to gain a maximum of infectious MV particles, followed by a highly efficient downstream processing. Purification steps have to lead in a minimum in product loss and a maximum in impurity removal, due to authority regulations. Therefore, the first clarification, based on depth filtration, and the finally chromatographic virus purification were investigated.

BIOT 173

Impact of high extracellular lactate on induced pluripotent stem cell metabolism

Daniel Odenwelder1, [email protected], Sarah W. Harcum2. (1) Bioengineering, Clemson University, Clemson, South Carolina, United States (2) Bioengineering, Clemson University, Clemson, South Carolina, United States

Induced pluripotent stem (iPS) cells hold the potential to drastically improve cell-based therapies in the near future. However, in order for stem cell therapies to become clinically feasible, these cells must be generated in sufficient quantity and quality. This aim will require a comprehensive understanding of how environmental conditions affect iPS cell metabolism and pluripotency. Rapidly proliferating cells, including cancer and iPS cells, consume glucose and secrete lactate at high rates, even in the presence of sufficient oxygen, a process referred to as the Warburg effect. In cancer cell metabolism, lactate accumulation is associated with cancer stem cell-like gene expression, drug-resistance, metastasis, and poor prognosis in breast cancer patients. Yet, there remains an incomplete understanding of the role of lactate in stem cell metabolism and pluripotency. This study examined the impact of extracellular lactate on the metabolic activity and pluripotency of iPS K3 cells grown with sufficient glucose. Extracellular glucose, lactate, and amino acid concentrations were monitored throughout the experiment to determine the extracellular consumption or production fluxes. High extracellular lactate resulted in altered cell metabolism, including a slight decrease in lactate production and glucose consumption. High extracellular lactate also resulted in a significant decrease in pyruvate consumption. In addition, high extracellular lactate resulted in a significant decrease in glutamine consumption. These results suggest decreased metabolic activity through the tricarboxylic acid (TCA) cycle and oxidative phosphorylation. Low mitochondrial metabolic activity is a feature of pluripotent stem cells, with increased TCA cycle and oxidative phosphorylation occurring as cells differentiate. The metabolic role of lactate to iPS cell metabolism and pluripotency will be discussed. The implications of these findings towards understanding iPS cell metabolism and designing cell culture conditions to limit lactate accumulation will be discussed.

BIOT 174

Investing in process development for MSC production in stirred tank bioreactors Tiffany Hood, [email protected], Jane Ring, Tristan Lawson, Megan Pease, Kara Levine, Aletta Schnitzler, Julie Murrell. MilliporeSigma, Bedford, Massachusetts, United States

Industry trends in regenerative medicine highlight a critical need for closed cell culture systems that support scalable manufacturing of cell therapies. Typical static in vitro culture methods are often too cumbersome and inefficient to support commercial scale production of mesenchymal stem/stromal cells (MSCs). Single-use stirred tank bioreactor systems are a platform that address this limitation and have proven effective for large scale microcarrier-based production of adherence-dependent cells. Implementation of optimized process control strategies for parameters including dissolved oxygen (DO) and agitation rate are key to making an efficient transition from planar culture to stirred tank bioreactors. Herein, a stepwise approach to process development for MSCs cultured on microcarriers in a small-scale single-use bioreactor is presented. The first case study examines the impact different gassing methods have on DO control and whether hypoxic growth conditions affect MSC function. The second case study demonstrates the application of Zwietering’s equation for suspension of solids to overcome scaling challenges often associated with microcarrier culture in stirred tanks. Identifying optimal process control strategies for microcarrier-based bioreactor expansion of adherent cells is paramount for the development of a robust cell therapy manufacturing platform.

BIOT 175

Engineering uniform-shear rate bioreactors to mimic bone marrow and lung vasculature niches for the production of platelets ex vivo

Andres Martinez1, [email protected], Richard McMahon2, Marc Horner3, Damien Doser2, William M. Miller1,2. (1) Chemical and Biological Engineering, Northwestern University, Evanston, Illinois, United States (2) Master of Biotechnology Program, Northwestern University, Evanston, Illinois, United States (3) ANSYS, Inc., Evanston, Illinois, United States

Platelet transfusions are entirely derived from volunteer donors. However, these methods are limited by a 5-day shelf life, the potential for contamination, and differences in donor/recipient histocompatibility. These challenges invite the opportunity to generate platelets from blood stem cells ex vivo. Yet, inefficient platelet release from culture- derived megakaryocytes (Mks, precursor cells to platelets) remains a major challenge. We applied computational fluid dynamics (CFD) modeling to assess published slit bioreactors and used those results to develop an improved 7-μm slit bioreactor with well-defined flow patterns and uniform shear profiles (USRB-7µm). The slits mimic fenestrations in endothelial cells lining sinuses in the bone marrow through which Mks extend cytoplasmic projections, called proplatelets (proPLTs), that are sheared off (50- 200 s-1) into platelets. In the USRB-7µm, a center channel flow pushes Mks into 7-μm slits, with shear rates of 5000 s-1. Two outside flows converge at the slits exerting near uniform shear rates (250-350 s-1) to fragment extending proPLTs, similar to the in vivo process. The USRB-7µm permits real-time visualization of proPLT formation and the rapid-release of individual platelet-like-particles (PLPs), which has been observed in vivo, but not previously reported for bioreactors. Collected PLPs exhibited characteristics similar to fresh blood platelets. Surprisingly, using only the center flow without the outside flows led to a 6-fold increase in PLP production. Based on this, we scaled-up the USRB-7µm using only a single flow to carry Mks into high-shear slit regions, mimicking in vivo observations of Mks being processed directly into platelets within the lung capillary bed (>2600 s-1). The new lung-USRB retained uniform shear rates with a 93-fold capture area increase to allow more Mks to be processed into PLPs. The USRB-7µm and lung-USRB will be useful tools for the analysis of proPLT/PLP formation to further understanding of how to increase ex vivo platelet production.

BIOT 176

Assessing multi-scale molecular modeling and biased search algorithms for candidate selection with respect to biophysical stability

Christopher J. Roberts, [email protected]. University of Delaware, Newark, Delaware, United States

Candidate selection for biopharmaceuticals has traditionally focused on clinical attributes such as target selectivity, in vivo half-life, and other features that may inherently lead to poor biophysical properties from the perspective of manufacturability and product development. Prediction of biophysical properties such as solubility, self- association, and aggregation propensity remains an outstanding challenge, particularly if one requires atomic-scale resolution. For purposes of candidate selection, this must be balanced with the need for rapid calculations to allow a large range of candidates to be tested in parallel, as well as to consider both dilute and concentrated protein solutions. Advancements in coarse-grained molecular modeling and algorithms for more rapid / efficient sampling have the potential to make these predictions more routine, while also providing “design rules” that can be implemented without the need for expensive simulations. This presentation focuses on a combination of different levels or scales of molecular modeling, and development of design rules based on protein sequence and three-dimensional structures. This includes first-order effects of the solution environment, with an emphasis on antibodies and other anisotropic protein structures. Comparison between a priori predictions and experimental results for solubility and self-association illustrate lucrative approaches that balance qualitative vs. quantitative accuracy, and computational burden. Outstanding challenges are also highlighted based on a number of current limitations for capturing protein-protein interactions in different solution environments that are routinely encountered during manufacturing and for final formulated products.

BIOT 177

Physiochemical determinants of drug-like monoclonal antibodies Lilia Rabia1, [email protected], Seth D. Ludwig1, Yulei Zhang2, Mark Julian1, Peter Tessier2,1. (1) Rensselaer Polytechnic Institute, Troy, New York, United States (2) University of Michigan, Ann Arbor, Michigan, United States

The success of therapeutic drugs is dependent not only on their functional activities (e.g., ability to kill cancer cells) but also on their physiochemical properties (e.g., solubility, stability and specificity). We have developed guidelines for identifying drug- like monoclonal antibodies based on their chemical compositions and biophysical properties. We have established limits for the levels of several amino acids in the variable regions of antibodies that are able to differentiate between approved and non- approved (phase 2 and 3) antibody drugs. By combining these chemical rules with biophysical rules (e.g., maximum levels of antibody non-specific interactions), we observe even stronger differentiation between approved and non-approved antibody drugs. We expect that our physiochemical guidelines will significantly improve antibody drug development by more reliably identifying antibody candidates early in the discovery process with properties that are suitable for therapeutic applications.

BIOT 178

Developability of antibodies: Viscosity of high concentration formulations

Ronak B. Shah1, [email protected], Emily Carvalho1,2, Andrew J. Ilott1, Eugene Guo1, Peter Soler1, Scott Hart1, Manisha Desai1. (1) Drug Product Science & Technology, Bristol-Myers Squibb, New Brunwick, New Jersey, United States (2) Chemical Engineering, Carnegie Mellon University , Pittsburgh, Pennsylvania, United States

Selection of lead antibody candidates is a highly involved process with many facets, with the ultimate aim of generating highly efficacious and safe molecule, which can be efficiently developed into a product. Development teams are often tasked with prioritizing multiple characteristics to screen for, with a very limited amount of protein. High concentration formulations are being developed to cater to a broad range of indications, as well as improving patient compliance and convenience. Viscosity of such formulations could be a major challenge, limiting the delivery routes and methods of administration. The nature of interactions in high concertation solutions are known to be complex and rather protein specific. Here, we evaluated viscosities of a number of antibodies at > 140 mg/mL across a range of temperatures. Biophysical characterization was carried out using fluorimetry, calorimetry, light scattering and Raman spectroscopy to understand interactions, which influence conformational stability and viscosity at different temperatures. With concentration, impact of protein packing on hydrodynamic radius and conformational stability was analyzed. Computational parameters based on antibody sequence and homology models were used to predict biophysical properties and interactions. Mechanistic insights on interactions driving viscosity would allow developmental teams to select lead candidates with favorable viscosity profiles. BIOT 179

Effects of chemical oxidation on the structure, stability, aggregation, and function of an IgG1 mAb

Dinen Shah, [email protected]. Pharmaceutical Sciences, University of Colorado, AMC campus, Aurora, Colorado, United States

Protein formulations typically require a shelf-life of 18-24 months. During their long periods of storage, oxidation of methionine and tryptophan residues is one of the most common forms of chemical degradation. In this study, we probed the effect of oxidation on the structure, stability, aggregation, and function of a therapeutic monoclonal antibody (mAb-8). We examined whether the extent of protein destabilization, the aggregate formation, and the loss of specific protein activity can be correlated with the site and extent of oxidation. We used three oxidants, hydrogen peroxide (H2O2), t-butyl hydroperoxide (t-BHP), and 2,2'-Azobis (2-amidinopropane) dihydrochloride (AAPH) commonly used in pharmaceutical industry to test the chemical stability of proteins. Both H2O2 and t-BHP preferentially oxidize methionines whereas AAPH oxidizes tryptophans in addition to methionines. The oxidized mAb by the three chemical agents did not show any noticeable changes in its secondary structure, with minor changes in its tertiary structure. All the three oxidized mAbs show a significant decrease in conformational stability compared to the un-oxidized mAb. Both H2O2 and t-BHP selectively 2 2 destabilized the constant domain CH , whereas AAPH destabilized both the CH as well as the variable domain of Fab. Consistently, increased mAb aggregation was found significantly more in the case of AAPH-oxidized mAb, confirming the earlier observation that the variable domain primarily controls the aggregation of this mAb. Loss of mAb effector function upon oxidation also showed a clear correlation with the site of oxidation. A significant decrease in Fc receptor binding was observed in all the three 2 oxidized mAbs, since all the three oxidizing agents modified the CH domain. Since only AAPH modified the Fab region, but not H2O2 and t-BHP, a significant decrease in cell proliferation and antibody-dependent cell cytotoxicity (ADCC) activity was observed only in the case of AAPH-oxidized mAb compared to those modified by H2O2 and t-BHP. Presence of free methionine in the formulation buffer seems to alleviate the effect of oxidation completely for H2O2 and AAPH but not for t-BHP. This study for the first time provides a complete mechanistic understanding of the effect of chemical oxidation on the stability, aggregation, and function of a mAb.

BIOT 180

Investigating a correlation between chemical degradation and gelation behavior observed for a monoclonal Fab

Christopher Petry1, [email protected], Yunqi Zhao2, Li Yi1, Shelly Pizarro1, Aditya Wakankar1, Jun Liu1. (1) Genentech, San Francisco, California, United States (2) Pharmaceutical Chemistry, University of Kansas, Lawrence, Kansas, United States A monoclonal antibody fragment (Fab) undergoes reversible gelation when incubated at temperatures greater than 30°C at high protein concentration (> 100 mg/mL) and pH < 5.5. This behavior appears to be facilitated by isomerization (Asu formation) of a solvent accessible aspartic acid residue. The extent of gelation during incubation at high temperature, as monitored by turbidity and viscosity, was found to correlate with Asu levels, as monitored by ion exchange chromatography (IEC). The rate of gelation was also found to decrease with increasing solution ionic strength, indicating that electrostatic interactions are involved in the gelation mechanism. Changes in the biophysical properties of the protein were observed via circular dichroism (CD), intrinsic Trp fluorescence, and differential scanning calorimetry (DSC) during the gelation process, indicating that Asp isomerization could be inducing a loss in tertiary structure. To further investigate the relationship between Asp isomerization and gelation behavior, a mutant with the labile Asp residue substituted was evaluated and found to improve the physical properties of the protein.

BIOT 181

Impact of freezing and thawing rates on protein aggregation for drug substance long term storage

Ameya Borwankar2, [email protected], Niloufar Parsaei2, Cheng Du2, Nripen Singh1, Sanchayita Ghose1, Zheng Jian Li2. (1) Process Development, Bristol Myers Squibb, Acton, Massachusetts, United States (2) Bristol-Myers Squibb, Devens, Massachusetts, United States

Proteins tend to be more stable at lower temperatures especially when stored frozen below the glass transition temperature of the formulation. Therefore, long term storage and transportation of drug substance for monoclonal antibody therapeutics is often conducted in the frozen state at below -40°C. However, limited information is available on the impact of the freezing and thawing rates on protein aggregation as the freezing and thawing rates are typically not well controlled and are hard to measure. A wide range of approaches to freezing are used in the field with various freezing platforms and containers. This study examines measures the impact of freezing and thawing rates on protein aggregation during freezing and thawing of drug substance for long term storage. A scale down model was developed to mimic the conditions in bulk drug substance bags and the freeze and thaw rates were controlled over a wide range relevant to the manufacturing scale utilizing the same model. Protein aggregation was characterized both in terms of soluble aggregates and particulates. The impact of other parameters such as the protein concentration, buffering species in the formulation, and formulation pH on protein aggregation and its interaction with the freeze and thaw rates was also examined. Several different mAbs were included in the study in order to probe protein specific effects on protein aggregation during freeze and thaw. The results of the study were leveraged to develop a best recommended practice for freezing and thawing of mAbs in the absence of any previous freeze/thaw stability data at the bulk scale.

BIOT 182 High-throughput platform for monitoring protein aggregation kinetics and rapid formulation development

Michael F. Drenski, Harrison P. Rahn, Rick D. Montgomery, [email protected]. Fluence Analytics, New Orleans, Louisiana, United States

Formulations are developed in the biopharmaceutical space as sets of solution conditions that lead to high stability of therapeutic substances such as antibodies, proteins, or peptides. The solution conditions that generally affect the stability of the sample are solvent and buffer choice, ionic strength, pH, and excipient, surfactant, and sample concentrations. In this study we demonstrate a recently developed instrument’s ability to screen a formulation's aggregation propensity with respect to the aforementioned solution conditions in addition to heating and shearing stress. The instrument, named ARGEN, measures aggregation rates of sixteen samples in parallel via static light scattering. Formulations of immunoglobulin G and bovine serum albumin are optimized by minimizing the aggregation rate over each set of conditions and stressors. The generated aggregation rates are applied to the Arrhenius equation to provide a systematic look at protein stability patterns. Additionally, formulations developed by ARGEN are interrogated with GPC analysis after storage at ambient temperature to demonstrate long term stability.

BIOT 183

PLGA-bevacizumab implants for long-acting anti-VEGF efficacy in a rabbit retinal vascular leakage model

Rae Sung Chang1, Jeffrey Jamison3, David A. Antonetti4, Steven P. Schwendeman1,2, [email protected]. (1) Pharmaceutical Sciences, University of Michigan, Ann Arbor, Michigan, United States (2) Biomedical Engineering, University of Michigan, Ann Arbor, Michigan, United States (3) Ophthy-DS, Kalamazoo, Michigan, United States (4) Ophthalmology, University of Michigan, Ann Arbor, Michigan, United States

Wet age-related macular degeneration (AMD) is a major causes of vision loss among the elderly. In wet AMD, new abnormal blood vessels grow under the macula, thus displacing the macula from its normal position, and resulting in rapid central vision loss. Overexpression of vascular endothelial growth factor (VEGF) results in the growth of abnormal blood vessels under the retina, and therefore, anti-VEGF therapy has become a potent treatment option for wet AMD. Administration of the anti-VEGF agents is accomplished by intravitreal injections typically every 4 weeks, but this dosing frequency is problematic for patient convenience and compliance, and repeated intravitreal injections may induce infection, inflammation and hemorrhage. Sustained release formulations of the anti-VEGF agents can potentially reduce administration frequency while maintaining the therapeutic concentration in the vitreous. Therefore, we developed injectable poly(lactic-co-glycolic acid) (PLGA) cylindrical implants (0.8 mm diameter), which sustain the release of stable bevacizumab (Avastin®, anti-VEGF antibody). Protein stabilization was accomplished by optimization of the content of both the original excipients from the Avastin® product and bevacizumab in PLGA 50/50 core implants, and release was controlled by a coating the core implants with the same pure PLGA. The optimized implant formulations achieved high bevacizumab loading (7.6 to 8.2 % w/w) and continuous in vitro release kinetics over six weeks with total cumulative release of 82 ± 8 to 89 ± 4 %. Little changes in monomer content, immunoreactivity, and secondary structure of the released antibodies during the 6-week in vitro release period were observed by size-exclusion high performance liquid chromatography, enzyme- linked immunosorbent assay, and circular dichroism spectroscopy, respectively. Anti- VEGF efficacy of the optimized bevacizumab implant was compared to the same dose of free bevacizumab in the rabbit VEGF-induced retinal leakage model (400 µg dose). Six weeks after intravitreal injection of both formulations, only the bevacizumab implant protected retinal blood vessels while significant leakage was observed in the no- treatment control and free bevacizumab groups (4 rabbit eyes/group). The retinal blood vessels of the implant group were still protective over 8 weeks. Hence, this approach may be useful for development of long-acting anti-VEGF therapy.

BIOT 184

Development and validation of a purification platform for proteins expressed in a CHO cell-free system

Sevda Deldari3, [email protected], Yang Liu4, Hui Guo5, Chandrasekhar Gurramkonda1, david Burgenson2,1, shayan borhani2, Govind Rao1, Douglas Frey2. (1) TRC Building, Ctr for Advanced Sensor Technology, Baltimore, Maryland, United States (2) UMBC, Baltimore, Maryland, United States (3) Chemical Engineering , UMBC, Baltimore, Maryland, United States (4) Chemical Engineering, University of Maryland, Baltimore County, Halethorpe, Maryland, United States (5) Chemical Engineering , UMBC, Ellicott City, Maryland, United States

Efficient protein purification methods are an important element for on-demand production of protein biologics. In this study, the rapid in vitro expression of a protein was accomplished using lyophilized cell-free extracts derived from Chinese hamster ovary (CHO) cells. A CHO-based expression system was chosen because most biologics produced at the industrial scale utilize CHO cells. The purification of proteins produced in a cell-free expression system was performed using a versatile purification platform consisting of ion exchange chromatography or immobilized metal affinity chromatography (IMAC) as an initial capture step. As proof-of-principle, granulocyte colony-stimulating factor (G-CSF) was selected as a model case because it is a well characterized biologic and was the first approved biosimilar introduced in the market. The specific conditions employed for the capture step, such as the volume of the column and the compositions and flow rates of the wash and elution buffers, were optimized to achieve the maximum product yield and purity. In order to more easily develop the purification platform, and also to measure the quality of the final product, high-sensitivity silver-stained sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) with quantitative image analysis was employed to detect impurities down to 10 ppm in the purified product. This method was validated using other analytical methods, such as enzyme-linked immunosorbent assay (ELISA), size exclusion chromatography (SEC) and the Bradford (660 nm) assay. Results of this study indicate that the optimized platform that was investigated achieved > 90% purity of target proteins in only one step. Furthermore, the high-sensitivity silver-stained SDS-PAGE method used provided accurate measurements of the purity and yield after each purification step. Overall, the data presented in this study show the potential of the validated purification methods under investigation when is applied to therapeutic proteins produced in a CHO-based cell-free system.

BIOT 185

Development of a pharmacy-scale, integrated, automated, multi-product platform for on-demand manufacturing of high quality biologics

Laura Crowell1,2, [email protected], Amos Lu1, Alan Stockdale2, William Doherty2, Alexandra Bonnyman2, Kerry Love2, John C. Love1,2. (1) Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts, United States (2) Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology, Cambridge, Massachusetts, United States

The emergence of precision medicine promises improved, more targeted therapies for patients. It creates a significant challenge for manufacturing, however, as conventional manufacturing techniques for high quality biologics historically rely on centralized, large- scale production of drug to supply global markets for broad disease indications. To enable manufacturing capability compatible with precision medicine, we need the ability to produce high quality biologics at a small-scale, on-demand, and with minimal human intervention within pharmacies, hospitals or healthcare clinics. To this end, we have demonstrated a bench-scale, integrated and automated system capable of producing, purifying, and formulating high quality biologic drug in less than 80 hours without intervention. We have paired this system with a complementary, integrated approach for accelerated process development to make new, non-platform products on the system in 3-6 months. Our system includes modules for cultivation, multi-stage chromatography, and ultrafiltration / diafiltration, along with integrated sensors and system controllers. We utilize the yeast Pichia pastoris to capitalize on its fast growth to high cell densities, which facilitates shortened production timelines. Additionally, its small, active genome and high capacity secretory pathway enable accelerated strain development, and its low and consistent host cell protein burden enables rapid purification development. Combined with our production system in fast, at-scale process development and process qualification, we can drastically reduce process development timelines. To date, we have demonstrated the reproducible manufacturing of three high quality biologics on the system: granulocyte-colony stimulating factor (G-CSF), interferon-α2b (IFN-α2b) and human growth hormone (hGH). We believe that reduction in the timelines for process development combined with the ability to produce small volumes of high quality biologics on a fully integrated and automated manufacturing system could increase access to medicines where and when patients are in need. BIOT 186

Using HTPD to link AMBR and purification to assess manufacturability of cell line clones

Ryan Zolyomi4, Hendri Tjandra1, [email protected], Yu-Hsi Chang3, Ashley Hesslein2. (1) Biologics Development, Bayer, Union City, California, United States (2) Bayer HealthCare, Berkeley, California, United States (3) Isolation and Purification, Bayer U.S. LLC, Berkeley, California, United States (4) Biologics Development, Bayer US LLC, Berkeley, California, United States

For biologics in the therapeutic pipeline, cell clone selection is a labor- and time- intensive component of process development. The pressure to shorten development timelines has led to a growing use of Automated Microscale Bioreactors (AMBR) which mimics bench-scale bioreactor systems and can be used to evaluate the manufacturability of various clones in parallel. Traditionally, cell metabolic performance and mAb titer are the primary criteria used to discriminate between clonal candidates and select the top 2-3 performers for the final round of evaluation in 5L-10L bioreactors. Here we describe an integration of the HTPD upstream and downstream workflows to allow faster and more thorough mAb quality characterization. The downstream workflow utilized a Tecan system which automated the removal of cells before affinity chromatography. The purified mAb was analyzed for product quality attributes such as oligosaccharide mapping and charge heterogeneity. The scale down HTPD downstream workflow of AMBR cultures demonstrated a comparability of mAb product qualities to those at bench-scale. Accommodating the AMBR system with a HTPD downstream workflow allowed for an 8-fold times saving by downstream developers to generate purified mAb for product quality analysis. Ultimately, the integration of upstream and downstream HTPD tools led to a rapid screening of clones using more relevant manufacturability criteria/attributes in the selection of final cell clone.

BIOT 187

Development of a novel, high-throughput platform for efficient perfusion-based cell culture process development

Thomas Gagliardi, [email protected]. Bioprocess Development, Shire, Bolton, Massachusetts, United States

Perfusion technology has been successfully used for the commercial production of biotherapeutics, in particular unstable recombinant proteins, for more than a decade. However, there has been a general lack of high-throughput cell culture tools specifically for perfusion-based cell culture processes. Here we have developed a high-throughput cell retention operation for use with the ambr® 15 bioreactor system. Experiments were run in both 24 and 48 reactor configurations for use in media development, clone screening and an assessment of perfusion mimic model variations. Employing an offline process for cell retention and a variable volume model developed with Matlab computational software, the established platform has demonstrated cell culture performance (> 90 E6 cells/mL and >95% viability) and productivity comparable to bench scale bioreactors. The automated, single use, high-throughput platform is a powerful tool that enables us to have rapid and efficient process development of perfusion-based cell culture processes.

BIOT 188

Predicting membrane performance for the concentration and diafiltration of mAb solutions using an ultra scale-down method

Lara Fernandez- Cerezo1,2, [email protected], Jennifer Pollard2, Michael Hoare1, Andrea Rayat1, Alex Chatel1, Gary Lye1. (1) Biochemical Engineering, University College London, London, London, United Kingdom (2) Downstream Process Development & Engineering, Merck & Co., Inc., Kenilworth, New Jersey, United States

The availability of material for experimental studies is a key constraint in process development, especially during the early stages and when using high concentration monoclonal antibody (mAb) solutions. Using a combination of critical flow regime analysis, bioprocess modelling and experimentation at the millimeter scale, ultra scale- down (USD) technologies can significantly reduce the amount of process material, time and cost of process development enabling a more effective process development approach. The ability to predict the performance of membrane separation operations at manufacturing-scale, including the flux profile characteristics and changes in protein structure, will help focus pilot-scale runs during process development. In this study, a USD set-up comprising of a sheared cell unit with a rotating disc is used to predict the performance of a tangential flow filtration (TFF) set-up using a membrane cassette. The USD set-up was designed to reduce effective membrane area, process material and buffer volume requirements of TFF set-up by 520-fold. By characterizing average shear rate using experimental and computational fluid dynamic tools for TFF and USD set-ups respectively, we defined the average shear rate over the membrane surface as the scaling parameter for ultrafiltration/diafiltration (UF/DF) stages. We performed a series of trials at each scale across a typical range of average shear rates, mAb feed concentrations and membrane loadings. The predicted effect of changing shear rate and mAb concentration on flux using the USD set-up matched that found in the TFF set- up during buffer exchange operations. Good agreement in flux performance data at both scales was also found when evaluating membrane loadings, from 200 to 800 g/m2, during UF/DF stages. Future studies will include expanding this ultra scale-down method to new modalities.

BIOT 189

Automated single pass tangential flow filtration (SPTFF) for lab scale continuous bioprocessing Mehdi Ghodbane, [email protected], Nidhi Thite, Cindy Jung. GlaxoSmithKline, King of Prussia, Pennsylvania, United States

Single pass tangential flow filtration (SPTFF) can achieve various objectives within a downstream process. Applications include reduction of process intermediate volumes to enable facility fit, feedstock concentration to improve chromatographic productivity, and continuous final concentration. This unit operation has recently received considerable attention in response to the biopharmaceutical industry’s growing interest in continuous bioprocessing. One advantage of continuous processing is the scaled down equipment required to achieve equivalent facility throughput, resulting in significant reductions in capital and required footprint. Furthermore, the increased automation required for continuous processing has the potential to mitigate the risk of batch rejection due to human error. However, lab scale volumetric flow rates are low. Thus, identifying hardware capable of providing acceptable process control and automation is a challenging task. Previous SPTFF reports have mainly focused on achieving output concentrations required for a given process by altering the number of stages, feed flow rate, and surface areas. This work will focus on the development of a fully automated, lab scale, SPTFF unit operation designed to be integrated into a continuous downstream process. Methods of backpressure and retentate flow control will be examined, focusing of the evaluation of approaches for automated long term operation. In addition, an approach for inline concentration monitoring will be presented. Together, these components will be shown to facilitate complete automation of the unit operation to meet the needs of scaled down continuous bioprocesses.

BIOT 190

Increasing process knowledge using high throughput RoboColumns

Tim Tully2, [email protected], Benjamin Huffman2, Matthew Stork1. (1) Bioprocess Research and Development, Pfizer, Inc, Andover, Massachusetts, United States (2) Pfizer, Inc., St. Louis, Missouri, United States

High throughput screening has become a powerful tool for purification process development. Plate based screens can identify target operating conditions but require lab scale runs for confirmation and testing at target load challenges. RoboColumns offer the ability to test small chromatography columns at conditions more similar to lab scale and manufacturing scale operations. Development efforts can be expedited when 8 RoboColumns are run in parallel and each run generally uses at least 10-fold less material. As many as 16-24 RoboColumn experiments can often be run in a single day compared to 2-3 using lab scale columns.

We have integrated RoboColumns into early and late stage antibody process development. As opposed to the traditional paradigm of moving from HTS to lab scale to pilot scale, we leverage RoboColumn experiments in parallel with lab and pilot runs to build process understanding. RoboColumns have been used to test high load challenge flow through steps, optimize wash and elution buffers for bind and elute steps, and further purify lab scale samples to determine the impact of upstream development on final pool purity. In these cases we increased the number of experiments we were able to perform with limited material and time resources. RoboColumn utilization has consistently provided increased process knowledge and robustness at all stages of process development.

BIOT 191

Innovations in high throughput formulation screening: Promise and reality

Cindy Ren, [email protected], Kimberly Westland, Jeff Yeary, Mohan Boggara, Nitin Rathore. Amgen, Thousand Oaks, California, United States

As pipelines in the biotech industry become more diverse, the IP landscape progressively more competitive and speed evermore critical, it has become increasingly important to develop the means to rapidly and efficiently perform Drug Product development. Formulation development, one of the key aspects of Drug Product, has traditionally been a labor-intensive, multi-year process. Conventional bench-scale (~100 mL) formulation development faces key challenges, including extensive material and analyst time, both of which result in significant timeline requirements and limit the composition space that can practically be evaluated. Wellplate-based screening, on the other hand, which has relied on liquid handler-spiked formulations, has been limited by carryover and low protein concentrations that are often an order of magnitude below the final target. Amgen has developed and implemented several innovative automated capabilities, e.g. plate-based buffer exchange, to overcome these limitations and to establish a new high throughput formulation screening approach. Case studies will be presented demonstrating application of these technologies to tackle various Drug Product challenges, such as minimizing the viscosity of high concentration molecules and exploring large design spaces for novel modalities. New approaches to multivariate study design, automated analytical testing, data analysis and decision-making will be described. At the same time, the hurdles and system limitations that have been identified as well as future directions will be discussed.

BIOT 192

How the Janssen-Biogen partnership enabled delivery of a breakthough therapy for multiple myeloma patients and rapid process optimization to meet market demand

Gene Schaefer, [email protected]. J&J Centocor, Horsham, Pennsylvania, United States

Janssen Biotech Inc. signed an agreement with Genmab on August 30th, 2012, for the rights to develop, manufacture and commercialize Daratumumab (DARZALEX). The drug substance process used to generate material at that time was suitable for Phase 1 supplies, but was not commercially viable and not consistent with the Janssen mAb platform so a conscious decision was made to modify the process to fit the Janssen platform once the cell line was made available to Janssen, approximately three months after the deal had been signed. Daratumumab received Breakthrough Therapy Designation from FDA in May, 2013 which further increased projected material demands. Overall, the following process changes were made in the six months prior to scale up and the start of large scale clinical manufacturing: the basal medium and all feeds were changed to the Janssen platform through a series of cross company, cross site nested design of experiments over a three month period, the downstream purification process was redesigned to introduce commercially robust resins and filters and a more robust control strategy was implemented to maintain control of product quality. The dose and number of patients enrolled in trials which drove a relatively high demand for material early in the program. The first approval for DARZALEX occurred in November, 2015, two months from the date that the Biologics License Application (BLA) was accepted for Priority Review by the FDA and just less than three years from the start of process development. Meeting demands for ongoing clinical programs, a large expanded access program, and projected market demands required very large quantities of product prior to launch. So, although the Gen1 process was reasonably productive, the Janssen Biogen collaboration was again leveraged to rapidly develop and validate a robust second generation process with nearly a three fold increase in bioreactor titer and an intensified downstream process while continuing to meet the complex set of product quality targets determined from the Gen1 product. This work required both companies to work closely and contribute prior knowledge and expertise to the program.

BIOT 193

Nano cassava dregs fabricated via green method

Baofeng Lin, [email protected], Qi-wei Sun, Chuanhui Xu. School of Chemistry and Chemical Engineering, Guangxi University, Nanning, Guangxi, China

The cassava dregs (CDs) are very rich in China. However, most of them are used as feed directly which is a great waste of renewable resources. Based on the sustainable development of materials and the high-value application of CDs, we prepared nano-CDs via green methods in which the CDs were treated by selected additives (NaOH, NaClO and H2O2) combined with three mechanical processes (ball mill, homogeneous, and ultrasonic). CDs nanocrystallization was processed in two stages: (a) Preprocessing stage: Removing the lignin after stirring the CDs in the water and standing for several minutes. At this time, it was possible to obtain 60-300 mesh CDs through sieving. (b) Nanocrystallization stage: Firstly, homogenization of CDs was processed at 80MPa. Then, CDs were ball milled with the additives which was used to improve the effect of nanocrystallization of CDs through reactions. Finally, the CDs were pulverized by the mechanical force ultrasonic. The results showed that the optical treatment condition for CDs was 15% H2O2 (w/w) combined with three mechanical methods. The minimum particle size of NCDs was about 30nm. It was found that the stability of CDs was improved and the sedimentation rate of CDs in suspension became slow.

BIOT 194

Acceleration of native chemical ligation

Shuaijiang Jin, [email protected], Roberto J. Brea, Neal K. Devaraj. Department of Chemistry and Biochemistry, University of California, San Diego, San Diego, California, United States

Native chemical ligation (NCL) is widely used for the synthesis and derivatization of large peptides, small proteins and nucleic acids. This methodology involves the reaction between a peptide containing a C-terminal thioester and a peptide bearing an N- terminal cysteine (Cys) residue, yielding a product with a native amide bond at the ligation site. However, NCL has some important disadvantages, including relatively long reaction times, the necessity of additives and the use of working concentrations in the millimolar range. Therefore, there has been an increasing interest to develop novel NCL-based strategies for the highly efficient formation of functional peptides and proteins. Here we describe a novel NCL methodology to couple peptide segments in a highly chemoselective and specific way that overcomes the limitations of the traditional NCL. We foresee potential applications of this methodology in the synthesis of structurally complex peptides, cyclic peptides and glycopeptides, as well as in the construction and selective modification of functional proteins.

BIOT 195

Nanoparticle mediated release of polymersomes under single pulsed femto second irradiation

Abby R. Robinson1, [email protected], Gina M. Disalvo1, Sean M. O'Malley2, Daniel M. Bubb1,2, Julianne C. Griepenburg2. (1) Chemistry, Rutgers University - Camden, Camden, New Jersey, United States (2) Physics, Rutgers University-Camden, Camden, New Jersey, United States

The self-assembly of amphiphilic di-block copolymers into polymeric vesicles, commonly known as polymersomes, is an area of high interest in research due to the potential applications in the field of drug delivery. Polymersomes are fully synthetic robust vesicles composed of a hydrophobic membrane and a hydrophilic core, providing the ability for stable dual-encapsulation of a variety of molecules. Methods have been developed for triggered encapsulant release using ultrafast, single-pulse irradiation with visible and near infrared light to provide a non-invasive method of achieving spatial and temporal control. We have shown that the incorporation of gold nanoparticles (AuNP) within the vesicle membrane provides wavelength specific vesicle rupture at 532 nm. The wavelength dependence of our polymersome system can be shifted by altering the plasmonic mode of the nanoparticles by way of shape and composition.

BIOT 196

Design and characterization of a benzylguanine-PEG surface for anchoring cellular-derived giant plasma membrane vesicles

Daniel Oseid3, [email protected], Anne S. Robinson1, Julie Albert2, Sierra Lear2, Aomeng Cui2. (1) Chem and Biomolecular Eng, 300 Lindy Boggs Bldg, Tulane University, New Orleans, Louisiana, United States (2) Chemical and Biomolecular Engineering, Tulane University, New Orleans, Louisiana, United States (3) Tulane Brain Institute, Tulane University, New Orleans, Louisiana, United States

Giant plasma membrane vesicles (GPMVs) are large, unilamellar bodies that retain the exact membrane composition of the cells they are released from. In this work, we have isolated GPMVs from Human Embryonic Kidney (HEK) cells overexpressing an N- terminally SNAP-tagged G protein-coupled adenosine receptor. SNAP-tagged proteins rapidly react with benzylguanine derivatives to form an irreversible covalent bond. Thus, we generated a novel benzylguanine-PEG layer using chlorosilane and thiol-ene chemistry on glass for use in anchoring and optical monitoring of fluorescent GMPVs. Surface modification steps were verified using water contact angle and ellipsometry. Generation and characterization of fusion proteins and optical monitoring of GPMV- surface interactions are discussed herein. This system has the potential to be highly versatile, as potentially any protein of interest can be overexpressed in mammalian cells and selectively anchored in GPMVs using our method. This will greatly aid in a variety of downstream analyses, including reagentless optical biosensing.

BIOT 197

Physical properties of cartilage extracellular matrix

Ferenc Horkay1, [email protected], Emilios Dimitriadis2, Iren Horkayne-Szakaly1, Peter J. Basser1. (1) Section on Quantitative Imaging and Tissue Sciences, National Institutes of Health, Bethesda, Maryland, United States (2) NIBIB, National Institutes of Health, Bethesda, Maryland, United States

Cartilage is a composite material composed of relatively small number of cells (chondrocytes) surrounded by the extracellular matrix (ECM). Approximately 70 to 85% of the weight of the tissue is water. The major macromolecular components of the ECM are proteoglycans (PGs) and collagen. Approximately 30 to 35 % of the dry weight of cartilage tissue is composed of PGs., which bind to hyaluronic acid chains forming large aggregates. Collagen is a fibrous protein that makes up 65 to 70 % of the dry weight of the tissue. Type II is the dominant collagen in cartilage, but other types of collagen are also present. Collagen architecture varies with the depth from the articular surface. PG concentration and water content also vary with the depth. Near the articular surface the PG concentration is relatively low, while in this region the water content is high. In deeper regions, the PG concentration is greater, and the water content becomes lower. We studied the physical and chemical properties of cartilage and its polymeric constituents using a number of complementary methods (e.g., osmometry, atomic force microscopy, scattering methods). Combination of these techniques provides insight in the relationship between microstructure and mechanical properties. Our model cartilage was engineered from chondrocytes harvested from chick embryo sternum and cultured on poly(vinyl alcohol) hydrogel scaffolds. Osmotic swelling pressure was measured at different stages of tissue development using a tissue micro-osmometer. Atomic force microscopy was used in tandem to map the local mechanical properties. The concentration of the main biopolymer components was determined by biochemical analysis. The results, which shed light on the role played by the major polymeric constituents of the ECM in cartilage biomechanics and osmotic properties, are discussed.

BIOT 198

Temporal control of encapsulant release from polymersomes upon single pulse irradiation

Gina M. Disalvo1, [email protected], Abby R. Robinson2, Julianne C. Griepenburg3, Daniel M. Bubb3, Sean M. O'Malley3. (1) Chemistry, Rutgers University - Camden, West Berlin, New Jersey, United States (2) Chemistry, Rutgers University- Camden, Cinnaminson, New Jersey, United States (3) Physics, Rutgers University- Camden, Cherry Hill, New Jersey, United States

Polymersomes self-assemble from diblock copolymers into spherical vesicles due to their amphiphilic nature. Previous works have utilized a wide variety of photo-responsive mechanisms to achieve delivery of cargo with the goal of high spatial and temporal resolution. In this work, hydrophobic gold nanoparticles (AuNPs) have been incorporated into the membrane of PBD-b-PEO (polybutadiene(1,2 addition)-b- ethyleneoxide) polymersomes to create a photosensitive carrier ideal for targeted release. The shape and size of the incorporated metal nanoparticles governs release wavelength, providing system tunability due to differences in surface plasmon resonance (SPR). Herein, we have explored the interaction of single pulse laser irradiation on an individual polymersome basis. By using fluorescence microscopy to monitor the intensity of fluorescein isothiocyanate-dextran (FITC-dex), encapsulant release curves can be related to an effective pore size. Single pulse irradiation was generated via femtosecond, picosecond and nanosecond lasers to determine how the pulse energy and pulse width affect the size of the pore formed. This fundamental study demonstrates the ability to control photosensitive polymersomes in a spatial and temporal manner, giving our carriers potential for future use in precision medicine and other targeted delivery systems.

BIOT 199

Near infrared responsive polymeric actuator

Ian Hill1, [email protected], Katerina Nikolaidou2, Jennifer Lu1. (1) School of Engineering, University of California, Merced, Merced, California, United States (2) Physics, University of California, Merced, Merced, California, United States

Materials that reliably produce a mechanoresponse upon low-energy driven stimuli such as near infrared (NIR) (690 - 900 nm) or heating a few degrees above room temperature have significant technological implications. We are investigating two types of thermal mechanoresponsive systems that utilize conformational changes for power- efficient actuation. We have created a dynamic scaffold based on NIR stimuli- responsive polymer composites, i.e. a hydrogel made of poly(N-isopropylacrylamide) (PNIPAM) /graphene. A spatial light modulator coupled with an engineered porous structure to facilitate water diffusion enable spatially and temporally defined mechanical forces to be generated at a fast rate. This new platform can potentially impose force induced by NIR actuation onto cells for influencing stem cell differentiation. We have revealed a new sub-molecular switch, a dibenzocyclooctadiene (DBCOD), a hinge that consists of a flexible eight-membered ring fused with two rigid phenyl rings. Like proteins, the conformational change can be triggered by a low energy stimulus such as NIR photon irradiation or heating a few degrees above room temperature. A polymer system that contains a small amount of DBCODs without any process optimization exhibits an anomalous giant thermal contraction. The thermal contraction value is about 10 times greater than that offered by the second best reported system. A new bilayer based on the DBCOD-containing polymer can offer high-efficient actuation. This opens a pathway for soft robotics applications.

BIOT 200

Heterochiral DNA circuitry: Interfacing L-DNA with endogenous RNAs

Jonathan T. Sczepanski, [email protected]. Chemistry, Texas A&M University, College Station, Texas, United States

The potential for DNA-based nanodevices and circuitry to interface with biological systems through simple Watson–Crick (WC) base pairing has provided the impetus for an intense effort now underway to develop DNA-based devices capable of analyzing and manipulating molecular information in living systems. Although recent work has begun to demonstrate the compatibility of DNA nanodevices with complex biological environments, current DNA-based nanotechnologies still remain poorly suited for intracellular applications. In particular, DNA is rapidly degraded by cellular nucleases, and DNA introduced exogenously into cells is susceptible to unintended interactions with cellular components, including off-target hybridization to native nucleic acids and triggering of the innate immune response. An ideal solution to this problem is the use of L-deoxyribose nucleic acids (L-DNA), which are synthetic enantiomers of native D- nucleic acids. As such, L-DNA has the same physical properties in terms of solubility, hybridization kinetics, and duplex thermal stability as D-DNA, yet L-DNA is completely bioorthogonal (i.e. L-DNA is intrinsically resistant to nuclease degradation and is unable to interact with endogenous proteins and nucleic acids). Despite the profound implications of L-DNA-based nanodevices for biomedical applications, the inability of oligonucleotide enantiomers (D versus L) to form contiguous WC base pairs with each other severely limits integration of L-DNA-based nanodevices with biological systems. In order to overcome this limitation, we developed a novel toehold-mediated strand- displacement methodology for transferring information between orthogonal DNA enantiomers via an achiral intermediary, opening the door for “heterochiral” DNA nanotechnology having fully-interfaced D-DNA and L-DNA components. On the basis of this approach, we show that heterochiral circuits can directly interface endogenous RNAs (e.g. microRNAs) with bioorthogonal L-DNA, suggesting applications in bioengineering and nanomedicine. Our recent efforts to develop heterochiral DNA logic circuits capable of monitoring unique microRNA expression patterns in live cells will be discussed.

BIOT 201

Rosmarinic acid particles with versatile biomedical functions

Mehtap Sahiner2, [email protected], Diane A. Blake3, Nurettin Sahiner1,4. (1) Chemistry, Canakkale Onsekiz Mart University, Canakkale, Turkey (2) Department of Leather Engineering, Ege University, Izmir, Turkey (3) Tulane Univ Health SCI Ctr, New Orleans, Louisiana, United States (4) Physics & Engineering Physics, Tulane University, New Orleans, Louisiana, United States

The bioflavonoid, rosmarinic acid (RA), an antioxidant compound that exists in many plants, including mint, rosemary, and basil, was polymerized and its physical properties and biological activity were evaluated. The poly(rosmarinic acid) (p(RA)) particles were synthesized by employing an emulsion polymerization/crosslinking method using the trimethylolpropane triglycidyl ether (TMPGDE) crosslinker in span 80/gasoline reverse micelle systems. The 3-dimensional, structural, and thermal characterization of the p(RA) particles were carried out using Scanning Electron Microscopy (SEM), Fourier transform infrared spectroscopy (FT-IR), and thermal gravimetric analyzer (TGA), respectively. The zeta potential values of the p(RA) particles at different solution pHs were also determined. The release of monomeric RA from the particles at 37.5 oC was determined at the following pHs: 5.4, 7.4, and 9. The antioxidant properties of p(RA) particles were determined by means of the total phenol capacities employing Folin- Ciocalteau (FC) method, total Trolox equivalent antioxidant capacity (TEAC) and DPPH scavenging capacity techniques. Finally, the biocompatibility of RA and p(RA) particles with mammalian cells was tested with COS-1 cells, a standard fibroblastic cell line derived from African green monkey kidney.

BIOT 202

DNA-based nano-constructs for visual detection of rifampin resistant Mycobacterium tuberculosis

Ryan Connelly, [email protected], Sheila Solarez, Yulia Gerasimova. Chemistry, University of Central Florida, Orlando, Florida, United States

Molecular diagnostics of drug resistant pathogens requires analysis of point mutations in bacterial or viral genomes, which is usually performed by trained professionals and/or by a sophisticated computer algorithm. Here, we developed a DNA-based logic system that autonomously analyzes mutations in the genome of Mycobacterium tuberculosis complex (MTC) bacterial species and communicates the output to a human user as alphanumeric characters read by the naked eye. The five–gate system displays “O” (“no infection”) for the absence of MTC infection; and “P” or “F” for passing or failing a drug susceptibility test, respectively.

This project is supported by the National Institute of Allergy and Infectious Diseases of the National Institutes of Health under award number R21AI123876

Design of G-quadruplex (G4)-based DNA logic gates with visually detected output.

The design and performance of alphanumerical sensor display.

BIOT 203

Creating DNA computer: Five integrated NAND gates with half-adder function

Tatiana Molden2, [email protected], Marcella Grillo1, Dmitry M. Kolpashchikov2. (1) University of Central Florida, Orlando, Florida, United States (2) Chemistry, University of Central Florida, Orlando, Florida, United States DNA is a molecule of choice for building 1st molecular computer due to selectivity and predictability of DNA base pairing. Moreover, a computer made of DNA can find applications in diagnosis and treatment of cancer, infectious and genetic diseases due to its biocompatibility. Operation of modern electronic processors is based on the logic gates - devices that recognize inputs, process them according to a certain Boolean function and produce either high (digital 1) or low (digital 0) binary output. DNA computer can process DNA and RNA sequences as inputs. For instance, AND, NOR and INHIBIT DNA logic gates were successfully designed by us based on assembly of DNA strands in DNA 4 way junction structures. The advantages of this design over alternative DNA logic gates is the possibility to integrate individual gates in DNA circuits using DNA crossover tile, a stable building block used in DNA nanotechnology. This project focuses on design and optimization of DNA-based half-adder, a basic component of any processor required to add two single binary digits. Half-adder processes two inputs and produces two outputs: Sum and Carry. In our experiments, two different oligonucleotide sequences were designated inputs. In the absence of one or both inputs split deoxyribozyme catalytic core was reassembled, resulting in high fluorescent signal upon cleavage of fluorogenic substrate (digital 1). In the presence of both inputs two parts of the catalytic core remained separated resulting in low fluorescence (digital 0). DNA half-adder was designed using five interconnecting NAND gates, placed on a single tile platform to ensure efficient inter-gate communication. We initially optimized the design of individual two-input NAND gates assembled on a DNA tile. Consistent with the NAND logic, high fluorescent output was low only when both inputs were present simultaneously and remained high for all other input combinations. Optimized logic gate was then covalently linked using “click” chemistry and represented a functional unit that could be incorporated into the final assembly. This modular approach allows us to produce each of the functional units separately and then link them to achieve optimal half-adder function, which is the next step in the project. If successful this project will contribute to creation of a DNA computer, a biocompatible device capable of processing DNA and RNA inputs. The next step would include assembly of two half-adders to create a DNA full-adder.

BIOT 204

Translation of a drug delivery system for glaucoma surgery patients: Challenges in bench-to-bedside design

Mitchell Fullerton1,2, [email protected], Brionna King3, Ramesh S. Ayyala4, Vijay T. John5, Diane A. Blake1. (1) Biochemisty and Molecular Biology, Tulane Univ School of Medicine, New Orleans, Louisiana, United States (2) Bioinnovation IGERT, Tulane University, New Olreans, Louisiana, United States (3) BUILD Program, Xavier University of Louisiana , New Orleans, Louisiana, United States (4) Ophthalmology, Tulane Univ School of Medicine, New Orleans, Louisiana, United States (5) Tulane Dept of Chem Bio Engr, New Orleans, Louisiana, United States

Glaucoma is the second leading cause of blindness worldwide with over 80 million cases. Advanced or emergency cases require surgical intervention with an implanted drainage device. All drainage devices have a failure rate of 50% after 5 years due to fibrosis. A drug delivery system has been developed to mitigate fibrosis by releasing 5- fluorouracil (5-FU) and mitomycin C (MMC) into the surgical site for 30 days after implantation. This system uses a unique manufacturing technique to produce a PLGA wafer with a honeycomb morphology that enables dual loading and timed release of the two therapeutics. This case study discusses the challenges of translating this novel drug delivery system to a manufacturing facility including: reproducibility in manufacturing, drug quantification, preclinical tissue culture assays, and specific environmental requirements at the manufacturing facility. An in vitro drug release study was used to ensure that the batch to batch variation was negligible. An assay developed to measure the total drug in each film showed that the amount of 5-FU did not vary from batch to batch. Additionally, when the wafers were cut in halves or fourths, they were homogenous. Tissue culture was used to measure to activity of the MMC in the wafers up to 30 days after manufacture. A high relative humidity (~70%) is required to create the honeycomb morphology characteristic of these wafers. This requirement for high humidity coupled with the unique equipment required for spin coating PLGA required collaboration with a GMP facility to develop a manufacturing station. Many of these studies were conducted during preclinical animal trials to expedite translation to a product that is ready for human trials.

BIOT 205

Expanding the application fields of liquid-handling stations by integrating microfluidics and 3D (bio-)printing

Carsten Radtke, [email protected], Josefine Morgenstern, Nils Hillebrandt, Juergen Hubbuch. Institute of Engineering in Life Sciences, Section IV: Biomolecular Separation Engineering, Karlsruhe Institute of Technology, Karlsruhe, Germany

The high-throughput process development as we know it is well established in academia and industry. For the execution of automated, miniaturized and parallelized unit operations, robotic instrumentation such as liquid-handling stations (LHS) is applied. Through their ability of automated handling of liquids and multiwell plates LHS have emerged to popular tools in bioprocess development. The work required for the investigation of complex interactions in biotechnological processes can be drastically reduced applying screening procedures on LHS. Further improvements of processes or the implementation of additional operations or new concepts on LHS demand the integration of novel technologies. Here, two technologies are investigated on their beneficial linkage to LHS-based process development. The first approach is the integration of microfluidic experimentation. Therefore, the "microfluidics-on-liquid-handling station" system was developed. This system allows the execution of microfluidic experiments on a LHS. The pipetting tips of the LHS can reversibly connect to microfluidic channels molded in silicone chips. Furthermore, integrated devices such as plate readers can be utilized due to the multiwell plate based dimensions of the chips. The applicability of this system was shown in various case studies in the biotechnological field. Besides the application for a screening study of flow through nanoparticle photopolymerization also a microfluidic device for the dilution-free absorption measurement in a high throughput format was developed. The second presented approach is the integration of 3D printing and bioprinting in high- troughput process development. 3D Printing and Bioprinting are among the most promising techniques for various application fields including the biotechnological sector. Nevertheless, the application in the area of downstream processing or in high throughput process development is hardly studied and still undervalued. To underline the applicability of theses techniques in process development, some 3D printing and bioprinting studies are presented. 3D printing is applied for the integration of freshly thoght tools and operations. The implementation of hydrogels in LHS-workflows is shown by the utilization of bioprinting techniques in combination with high-throughput compatible labware. The presented approaches provide novel powerful tools to further extend the spectrum of applications of liquid handling station based process development.

BIOT 206

Peptoid grafting on polysulfone membrane to increase antifouling characteristics: Effect of grafting density and chain length

Neda Mahmoudi2, [email protected], John Moore1, Maria V. Klaus1, Grant Harrison1, Jamie Hestekin1, Shannon L. Servoss1, [email protected]. (1) Chemical Engineering, University of Arkansas, Fayetteville, Arkansas, United States (2) university of Arkansas, Fayetteville, Arkansas, United States

Biofouling is a persistent problem for membranes exposed to blood or other bodily fluids, affecting membrane surface structure and hindering performance. In this study, we demonstrate that polysulfone (PSU) membranes that are grafted with methoxyethyl (NMEG) peptoids resist nonspecific protein adsorption. NMEG was attached to surface via polydopamine (PDA) and the modified surfaces were characterized to determine grafting density, contact angle, strength, pore size, and smoothness. The long-term stability of the peptoid coating was confirmed over six months. Peptoid-grafted PSU has increased hydrophilicity compared to unmodified PSU and comparable mechanical stability. The membranes were shown to resist biofouling using single proteins (bovine serum albumin, lysozyme and fibrinogen) and water flux was shown to decrease slightly. Additionally, the effect of side chain length and grafting density on protein adsorption was evaluated. It was determined that there is an optimal grafting density for reduction of protein adsorption, which was dependent on the length of the peptoid. These studies provides a convenient strategy to improve antifouling, hydrophilicity and hemocompatibility of PSU membranes for use in biomedical and blood-contacting applications.

BIOT 207 Translating unnatural amino acids with computationally-engineered promiscuous EF-Tu variants for broader applications in synthetic polymer synthesis

Vanessa Cox1, [email protected], Mariko F. Matsuura3, Eric A. Gaucher2. (1) Chemistry and Biochemistry, Georgia Institute of Technology, Atlanta, Georgia, United States (2) Biology, Georgia State University , Atlanta, Georgia, United States (3) Biological Sciences, Georgia Institute of Technology, Atlanta, Georgia, United States

An engineered in vitro translation system with the ability to accept building blocks beyond the peptide backbone would result in a novel platform that could generate both bio and synthetic polymers. Applications of this research not only span bioscience but extend into polymer, nano-, and materials science as well. We propose to engineer the translation machinery to create a novel platform for polymer synthesis. This platform will use engineered components of the translation system (e.g., elongation factors and transfer-RNA) to efficiently produce polymers based on a broad range of monomeric building blocks.

The first step toward this goal is to broaden the specificity of the translation machinery to incorporate unnatural amino acids and other monomers. We have demonstrated that computational analysis can be harnessed to design protein variants with diverse phenotypes including recognition and acceptance of unnatural amino acids. Our protein library is designed to probe the association between EF-Tu and aminoacylated tRNAs in order to efficiently deliver tRNAs acylated with unnatural amino acids, as well as other monomers, to the ribosome. We are currently expanding our focus to include engineered ribosomes and synthetases with the goal of designing a novel in vitro translation system that can generate a range of both bio and synthetic polymers.

BIOT 208

Tanshinone as a dual amyloid Inhibitor against the aggregation and toxicity of both amyloid-β and hIAPP

Baiping Ren3, [email protected], Yonglan Liu3, Yanxian Zhang1, Fengyu Yang4, Jie Zheng2. (1) The University of Akron, AKRON, Ohio, United States (2) Chemical and Biomolecular Engineering, The University of Akron, Akron, Ohio, United States (3) Chemical Engineering, University of akron, Akron, Ohio, United States (4) university of akron, Akron, Ohio, United States

Misfolding and aggregation of amyloid peptides are the key pathological events in many neurodegenerative diseases including Alzheimer’s disease (AD) and type 2 diabetes (T2D). Developing amyloid inhibitors to prevent amyloid aggregates is considered as a promising therapeutic strategy for treating these diseases. However, most of the existing amyloid inhibitors only inhibit the aggregation of specific amyloid peptide and show a strong sequence-dependent inhibition capability, making them difficult or impossible to be applicable to other amyloid peptides. Thus, it is both fundamentally and practically important for discovering some ligands that serve as a general amyloid inhibitor to inhibit the aggregation of different amyloid peptides. Herein, considering the common structural and aggregation features, and the correlation of T2D and AD, we discover and demonstrate tanshinones (two compounds of TS1 and TS2) - ingredients from the Chinese herb Danshen (Salvia miltiorrhiza Bunge) - which can impose a dual inhibition activity against the aggregation and toxicity of both full-length Aβ1-42 and hIAPP1-37 peptides using combined experimental and computational approaches. Collective experimental data from ThT, CD, and AFM confirm that both tanshinones show comparable inhibition ability to reduce both Aβ and hIAPP aggregates by inhibiting the fibrillation process and changing the fibrillogenesis pathway, leading to the formation of some amorphous aggregates. More importantly, both tanshinones are capable of disassembling preformed fibrils of Aβ and hIAPP, but TS1 shows the better potency in fibril dissembling than TS2. MTT and LDH assays also show that the tanshinones at very low concentrations of 5 µM can reduce the Aβ- and hIAPP-induced cell toxicity. Molecular dynamics (MD) simulations further reveal that both tanshinones preferentially bind to β-sheets to prevent Aβ-Aβ and hIAPP-hIAPP associations and thus to inhibit their fibril growth, explaining experimental observations. This work discovers that tanshinones act as common inhibitors (very few available today) to inhibit the aggregation of both hIAPP and Aβ, disaggregate preformed hIAPP and Aβ amyloid fibrils, and protect cells from hIAPP- and Aβ-induced toxicity, making them very promising agents against AD, T2D, and probably other amyloid-misfolding diseases.

BIOT 209

Discovering antibody binding signatures in age-related macular degeneration for diagnostic development

Joel Bozekowski, [email protected], Patrick Daugherty. Chemical Engineering, University of California Santa Barbara, Goleta, California, United States

Age-related macular degeneration (AMD) is a leading cause of blindness in older individuals due to deterioration of the macula, a small region of the retina responsible for sharp, central vision. Immune system activation plays a significant role in the development and progression of AMD, but the exact origins and targets of the immune system are not well understood. Generally, a common product of immune activation is the production of antibodies and identification of these antibodies can indicate disease risk, onset, or progression. However, with unknown or uncharacterized immune targets, a high-throughput method is needed for unbiased discovery of antibodies associated with AMD. Here, we utilized random peptide libraries to analyze the collection of peptides that bind to serum antibodies from subjects at various stages of AMD and matched controls from the Age-Related Eye Disease Study by the National Institutes of Health. Computational analysis of the bound peptide sequences enabled the identification of antibody-peptide binding patterns present in AMD subjects and absent from healthy controls. Due to the heterogeneity of the antibody response in complex diseases such as AMD, individual binding patterns demonstrated only 10-20% diagnostic sensitivity in AMD subjects. However, utilizing a set of patterns in parallel as a single diagnostic panel significantly increased the sensitivity while maintaining high specificity in healthy controls. Moreover, binding patterns specific to AMD were detectable at early stages of AMD, suggesting these antibody signatures could lead to the development of biomarkers for improved diagnostics and early detection prior to vision loss. Furthermore, the peptide sequences responsible for antibody binding in AMD subjects could reveal previously unidentified antigens involved in disease onset and progression and inform the development of novel therapies.

BIOT 210

Development of phage strategies to identify affinity peptide ligands against antibody kappa Fab fragments

André Nascimento1,2, [email protected], Akshat Mullerpatan2, Ronit Ghosh2, Maria Raquel Aires-Barros1, Ana Azevedo1, Pankaj Karande2, Steven M. Cramer2. (1) iBB- Institute for Bioengineering and Biosciences, Instituto Superior Técnico, Lisbon, Portugal (2) Rensselaer Polytechnic Institute, Troy, New York, United States

Phage display is a commonly used high-throughput screening technique to identify affinity peptide ligands against a specific target. The appropriate biopanning design can be employed to target a vast library of peptides towards different regions of multidomain proteins. Fab fragments are multimeric biomolecules that are gaining popularity as a class of recombinant protein alternatives to monoclonal antibodies, for a range of different applications. Two different biopanning strategies using protein L magnetic beads, were utilized to develop affinity peptides against both constant and variable regions of Fab fragments. The 80 unique peptide sequences obtained over the two strategies were synthesized and an initial binding screen yielded three high affinity binders (Kd<10 µM) towards a specific recombinantly produced Fab fragment (Fab A). The lead peptides were tested against three different forms of Fab A - recombinant, post papain digestion and intact antibody (mAb). The peptide selected from the "variable" biopanning displayed similar binding affinities for all three forms of the Fab. In contrast, the two candidates shortlisted from the "constant" biopanning possessed significantly different binding affinities across the three Fab forms. These results provide strong preliminary evidence that the "variable" biopanning approach resulted in the discovery of a peptide specifically targeting the variable regions of Fab A. On the other hand, the peptides from the "constant" phage bound closer to the constant region of the Fab fragment – proximal to the hinge region – thereby affecting its binding to the digested Fab fragment and intact antibody. This work demonstrates that it was possible to identify three peptide candidates with high affinity for Fab fragments, that could lead to the creation of a valuable downstream solution for this class of recombinant proteins.

BIOT 211

Monoclonal antibody-derived peptides carrying chemical modifications induce immunogenicity in a transgenic mouse model Bjoern Peters1, [email protected], Y. John Wang2, Antonio Iglesias3, Christian Schoneich1. (1) Pharmaceutical Chemistry, The University of Kansas, Lawrence, Kansas, United States (2) Late Stage Pharmaceutical Development, Genentech, South San Francisco, California, United States (3) Roche Pharmaceutical Research and Early Development, Roche Innovation Center, Basel, Switzerland

The formation of anti-drug antibodies (ADAs) following administration of a therapeutic protein is a constant concern as potential consequences vary widely, from no clinical effects to neutralization of endogenous proteins. Such immune responses have been regularly attributed to the presence of particles and aggregates. Recently, chemical modifications have been identified as important features contributing to the immunogenicity of several protein aggregates. Most interestingly, commonly observed chemical modifications such as Met and Trp oxidation appear to be not directly correlated to ADA formation. The aim of this study was to further substantiate these claims and identify potentially immunogenic chemical modifications. In order to generate a broad range of chemically modified peptides, a humanized monoclonal antibody was photo-irradiated at 254 nm and enzymatically digested with modified trypsin. The protein digest was purified and 6 peptide fractions were collected using reverse-phase chromatography. Purified peptide fractions were mixed with Alum in a one-to-one volume ratio and injected once subcutaneously into wild type (wt) and human IgG 1-tolerant transgenic (tg) mice. Collected serum samples were analyzed by ELISA, and purified peptide fractions were analyzed by mass spectrometry. Out of a total of 6 fractions, peptide fractions 2 and 5 revealed immunogenic response rates of 75% (both 6 out of 8 tg mice). These response rates closely resembled those of the respective wt mice, showing response rates of 78% (peptide fraction 2; 7 out of 9 mice) and 70% (peptide fraction 5; 7 out of 10 mice). Importantly, antigen binding was found to strongly target peptide neo-epitopes. Mass spectrometry data revealed several reaction products in peptide fractions 2 and 5 which were further characterized by MS/MS analysis.

BIOT 212

Recent advances in affinity maturation with the assisted design of antibody and protein therapeutics (ADAPT) platform

Chris R. Corbeil, [email protected], Traian Sulea, Enrico O. Purisima. Human Health Therapeutics, National Research Council of Canada, Montreal, Quebec, Canada

The design of superior biologic therapeutics, including antibodies and engineered proteins, involves optimizing their ability to bind to disease-related molecular targets. Virtual affinity maturation has the potential to quickly focus on the critical hotspot residues without the combinatorial explosion problem of conventional display and library approaches. To achieve this potential, ADAPT (Assisted Design of Antibody and Protein Therapeutics) has been developed as a platform to aid in the selection of mutants that improve/modulate the affinity of antibodies and other biologics. Starting from the crystal structure of an antibody-antigen complex, ADAPT uses a consensus-based protocol which interleaves computational predictions with experimental validation. Multiple rounds of optimization yield mutants with significant increases in binding affinity. The ADAPT protocol significantly enhances the robustness of the design and selection of mutants. Initially the ADAPT computational protocol was validated using a curated antibody-antigen structure-function data set (SiPMAB), demonstrating the power of using a consensus-based protocol. ADAPT was applied to three antibody Fab-antigen systems with over 90% of designed and experimentally validated mutants having higher affinities than the parent. ADAPT was then applied on a single-domain antibody, leading to a 10-fold affinity improvement.

BIOT 213

The effect of a formulation excipient on monoclonal antibody tryptophan oxidation

Jessica Y. Yang1, [email protected], Alex Strahan2, Sreedhara Alavattam3, Y. John Wang3, Bruce Kabakoff1, Cleo Salisbury1. (1) Early Stage Pharmaceutical Development, Genentech, South San Francisco, California, United States (2) Protein Analytical Chemistry, Genentech, South San Francisco, California, United States (3) Late Stage Pharmaceutical Development, Genentech, South San Francisco, California, United States

Monoclonal antibodies (mAbs) intended for therapeutic use are required to be well characterized to ensure that they are potent, safe and stable for extended periods of time. Oxidation of amino acids is one of the major degradation pathways in solution and has presented a significant challenge for formulation development. Unlike Methionine (Met) oxidation, which is relatively common and has been well studied, Tryptophan (Trp) oxidation is less common and is not as well understood. Therefore, formulation strategies to minimize the degradation are not as well established. In the following study, we have applied formulation excipients to three mAbs, each of which has a solvent exposed Trp in the CDR region that has been shown to be susceptible to oxidation. The formulations were placed on stability and Trp oxidation was monitored using various analytical assays. In addition, using one antibody as model molecule, we applied different types of anti-oxidants to evaluate their effects on Trp oxidation during storage. The results showed that one of the buffer excipients tested, arginine, promoted Trp oxidation compared to others.

BIOT 214

Bispecific antibody process development and evaluation of critical parameters during Fab arm exchange

Michael J. Sobkow, [email protected], Daniel Bezila, [email protected], Raphael Bertrand, [email protected], Andrew Detzel, Amin Salehi, Kathleen Van Citters, Pedro J. Alfonso. M1-2, Janssen Pharmaceuticals, Malvern, Pennsylvania, United States

Janssen currently has 10 ongoing bispecific antibody programs in different clinical phases of development to aid in curing diseases from cancer to several immunological based conditions. The production of DuoBody® bispecific molecules is a multistep process: First, two parental antibodies carrying a single point mutation in the CH3 domain of the antibodies are produced. Secondly, a mild reduction step is required to split these parentals into two half antibodies, Fab arms, by reducing the disulfide bonds without further disrupting intra-disulfide chains. Next, the single point mutations in each of the parental Fc regions, within the CH3 domains, by design favor heterodimerisation of these Fab arms over reformation of the parental mAb: the formation the bispecific antibody. Last, the oxidation to reform disulfide bonds between Fab arms is finalized during the UF/DF step upon removal of the reducing agent. Experiments were executed at small scale to determine the critical parameters of the controlled Fab Arm Exchange (cFAE) process. To ensure process robustness, the operating ranges for potential critical control parameters must be determined. A statistical Definitive Screening Design (DSD) and One Factor At a Time (OFAT) study designs were developed to determine the impact of specific operating parameters on the critical quality attributes, and to establish proven acceptable ranges for these parameters.

BIOT 215

Utilizing minimal-volume analytical techniques to assay the structural heterogeneity of IgG1 during clone selection

Phillip Angart2, [email protected], Casey Kohnhorst2, Bingyu Kuang3, Sai Velugula4, Cyrus Agarabi1. (1) FDA, Silver Spring, Maryland, United States (2) CDER, U.S. Food and Drug Administration, Washington, District of Columbia, United States (3) Chemical Engineering, University of Massachusetts Lowell, Lowell, Massachusetts, United States (4) CDER, U.S. Food and Drug Administration, Silver Spring, Maryland, United States

Protein structure, including secondary, tertiary, and quaternary, is a critical quality attribute for biotherapeutics. Variance in the structural conformation of proteins can lead to loss-of-function and aggregation, affecting the potency and immunogenicity of the drug product. In this study, a set of 10 CHO DG44 clones expressing an in-house IgG1 monoclonal antibody were investigated for the production of structural variants. The cell line was generated via stable transfection of two plasmids independently encoding the light and heavy chain, and then selected for using DHFR-/G418 and methotrexate. The top 10 producing clones were selected for further characterization and screening of structural variance to identify potential product problems early in process development. Far-UV circular dichroism (CD), Static Light Scattering (SLS), and SEC-MALS were selected to investigate secondary, tertiary, and quaternary structural perturbations of the secreted IgG1, respectively. These particular techniques were chosen because of the minimal sample required for analysis, which is amenable to analyzing small scale cultures, such as spinner flasks and microbioreactors, used frequently in early stage cell line characterization. Insights from this analysis helped identify clones with structural perturbations. An inflection point in CD spectra was present in seven of the ten clones and had varying degrees of concavity. An investigation found that the population of protein lacking the inflection point had eluted earliest during Protein A chromatography, implicating structural perturbations to the Fc for these IgG1 clones. Using the clone with the highest Tm as the reference spectra, the IgG1 Tm determined by SLS was found to correlate with the degree of CD spectral difference, where Tm increased as the spectral similarity converged. No higher order structures were observed by SEC-MALS. Combining the information produced from these instruments, we were able to eliminate three clones that have the potential to produce misfolded IgG1 products and identify clones that produce a stable product. We have also defined the expected variability at an early stage in process development to aid in the development of product quality goals in support of the Quality by Design initiative.

BIOT 216

Rapid aggregate reduction of bi-specific antibody model by filtration

Joy Adiletta2, Roger Alsop1, [email protected], Carl Breuning1, Larissa Pecore2, Amanda Mak2. (1) Sartorius-Stedim, Bohemia, New York, United States (2) SystImmune, Inc., Redmond, Washington, United States

Removal of protein aggregates from biologic products is critical due to their potential to increase antigenic response and its negative impact on the patients. Aggregates can be formed during upstream operations, and aggregation can also occur during downstream when suboptimal purification parameters are selected. Once formed, aggregate removal and/or reduction becomes challenging for the downstream purification process development team. Chromatography is the most common technology used today for aggregate removal; however, chromatography is time consuming, requires several buffers, method development, and use of costly technology such as purification skids and columns. SystImmune and Sartorius is exploring an alternate technique exploiting the Virosart® Max filter. The Virosart® Max is an optimized triple-layer polyamide 0.1 µm pre-filter specifically designed to remove aggregates and protect the expensive downstream virus removal filter. The Virosart® Max will be operated in a simple flow through mode to demonstrate aggregate reduction in a bi-specific antibody model. This filter based aggregate removal strategy would offer simple buffer preparation, minimal process development, and the option to replace costly chromatographic skids with inexpensive pumping equipment. The work to date is promising and results and methodology will be presented.

BIOT 217

Establishing a high-throughput process development platform for an antibody conjugation process Sebastian Andris1, [email protected], Michaela Wendeler2, Juergen Hubbuch1. (1) Section IV: Biomolecular Separation Engineering, Karlsruhe Institute of Technology - Institute of Engineering in Life Sciences, Karlsruhe, Germany (2) Department of Purification Process Sciences, MedImmune LLC, Gaithersburg, Maryland, United States

Antibody drug conjugates (ADCs) form a relatively new class of biopharmaceuticals which attracts a lot of attention throughout the industry due to its high potential for cancer therapy. Despite continued industrial and academic research and development efforts, clinical success has been limited so far. This can be attributed to the complex nature of the compounds, which are aiming to combine the specificity of a monoclonal antibody (mAb) and the cell-killing capacity of small molecule drugs. Especially for site- specific conjugations, a multi-step process including a buffer exchange is often needed for covalent linkage of antibody and drug via a linker. Despite the range of parameters that have to be investigated, high-throughput methods are scarcely used so far in the field of ADCs.

In this work, a model system, consisting of a mAb and a non-toxic fluorophore as pseudo-drug, was used to develop a high-throughput platform for a site-specific multi- step conjugation process on a liquid-handling station. A high-throughput solid-phase buffer exchange was successfully incorporated for reagent removal by utilization of a batch cation exchange step. For conjugate characterization, a high-throughput compatible reversed phase chromatography method with a runtime of 7 min per sample was developed. The potential of the platform was shown by performing a kinetics case study, where the high-throughput approach was successfully confirmed with results in larger scale. Due to the degree of automatization and parallelization, the platform is capable of reducing process development efforts significantly. This study could promote the use of HTS in conjugation process development, where the wide range of parameters to be screened would make it a standard tool.

BIOT 218

Challenges in developing a FRET assay to measure monoclonal antibody aggregates

Zaenab Oshinbolu1,2, [email protected], Rachana Shah2, Gary Finka2, Mike Molloy2, Daniel G. Bracewell1. (1) Biochemical Engineering, University College London, London, United Kingdom (2) Biopharm Process Research, GlaxoSmithKline, Stevenage, United Kingdom

In upstream processing, there has been significant progress to maximise titres, cell counts and viability. However, there is little understanding to how changes in the upstream conditions impact product profile in the bioreactor. Currently, analytical techniques require samples to be purified prior to measurement. Although purification is possible, the main issue is that key components (e.g. large aggregates) may be removed as a result, ultimately providing data that may not be a true representation of the cell culture. In addition purification is also time-consuming and costly. Hence, there is a need to develop assays/tools that can help characterise aggregate levels in complex multi-component environment earlier in research and development. This can eventually lead to better decision-making for selecting cell lines and candidate molecules.

Fluorescent dyes have been shown to be able to measure aggregates on purified antibodies. However, dyes solely in the cell culture medium do not provide specificity to monoclonal antibody (mAb) aggregates in a complex mixture. Hence, in this study we evaluated the use of two dyes in fluorescence resonance energy transfer (FRET) to increase specificity in measuring mAb aggregates.

The FRET design utilises a donor dye conjugated to a protein that binds mAbs specifically and an acceptor dye which has specificity towards aggregates. To achieve energy transfer, there are a few requirements which are related the Förster's distance equation: shared overlap between the donor emission spectrum and acceptor excitation spectrum (>30%), distance between the donor and acceptor (1-10 nm), quantum yield of the donor and the relative orientation of the donor and acceptor transition dipoles. Achieving FRET is more challenging as both donors and acceptors are free in solution. In addition, if the acceptor interacts with the inner structure of the aggregate there may be steric hindrance towards it interacting with the donor. To fully understand the FRET design, we investigated the spectral overlap and distance requirements.

BIOT 219

Screening and quantification of quality attributes for next generation protein A

Jelena Vasic, Daphne Areskog, Annika Forss, Tomas Bjorkman, [email protected]. GE Healthcare, Uppsala, Sweden

Protein A based affinity chromatography is a well-established robust and generic capture step in monoclonal antibody purification processes. The capture column contains the resin subjected to the least pure material in the downstream process yet it is the most sensitive to stringent cleaning and sanitization because of the protein ligand. This presentation will cover the development and evaluation of an alkali-stabilized protein A chromatography ligand and resin able to withstand 1 M NaOH for repeated cleaning in place cycles, i.e. a cleaning procedure similar to other resins in the process. The screening strategy for ligand comparison aiming at high caustic stability will be presented, and also show an example of surface plasmon resonance (SPR) sensorgram comparison which may be used for quality control of the purified ligand. Finally, a study on the correlation between contact time and sodium hydroxide concentration on resin cleanability and ligand stability will be presented.

BIOT 220

Serum antibody epitope discovery using pattern tiling Michael Paull, [email protected], Patrick Daugherty. Chemical Engineering, UCSB, Goleta, California, United States

The memory of the adaptive immune system ensures that the production of antibodies is sustained even after infections are resolved. Thoroughly characterizing the “immune history” of past antibody targets could be useful for developing diagnostics and vaccines. To address this challenge, we have developed the pattern tiling algorithm, which can identify antibody responses towards multiple past antibody targets from a single initial screen. Starting from a random library, antibody-bound peptides were selected and used to evaluate the enrichment of short patterns. Sequences of possible antigens were tiled into patterns and the enrichment of these patterns was used to find antibody-bound subsequences, known as epitopes. Pattern tiling successfully identified known epitopes targeted by commercial antibodies. Also, epitopes from common antibody targets matched epitopes previously found in the literature. For Rhinovirus, 3 epitopes were found which were bound by 83% of a population of 250 serum samples. Analyzing a list of proteins from all viruses which infect humans revealed the most prevalent viral epitopes, many of which belonged to rhinovirus and Epstein-Barr virus. A similar search on bacterial proteins revealed that common targets were streptococcus and staphylococcus. Pattern tiling currently enables one of the most complete evaluations of immune histories and its further implementation could aid in numerous applications.

BIOT 221

Extreme thermophiles as metabolic engineering platforms: Tailoring the process to the product

Christopher T. Straub, [email protected], Benjamin M. Zeldes, Jonathan M. Conway, James R. Crosby, Jonathan K. Otten, Robert M. Kelly. Chemical and Biomolecular Engineering, North Carolina State University, Cary, North Carolina, United States

Microbial metabolic engineering efforts have focused primarily on model mesophilic organisms, such as Escherichia coli. While genetics are well developed and the organisms well understood, there may be little value added by the model organism’s native metabolism and physiology. Certain extreme thermophiles provide natural advantages, such as ability to grow on atypical substrates, novel carbon dioxide fixation cycles, and survival at extreme conditions akin to conventional chemical processing conditions. Recently developed genetic systems available for these organisms are enabling metabolic engineering efforts that exploit advantageous aspects of their native phenotypes while directing carbon and electron flux to bio-based chemicals and fuels.

Discussed here is the current status of three extreme thermophile metabolic engineering platforms: 1) Caldicellulosiruptor bescii (Topt 78°C) that can extensively ferment lignocellulose at high loadings and utilize C5 and C6 sugars without carbon catabolite repression; 2) Sulfolobus acidocaldarius (Topt 75°C, pH 3.0) whose genome encodes all components of the 3-hydroxypropionate/4-hydroxybutyrate (3HP/4HB) cycle as well as enzymes implicated in elemental sulfur oxidation; and 3) Pyrococcus furiosus (Topt 98°C) a fermentative anaerobe that has already been engineered to produce ethanol, n-butanol, and 3-hydroxypropionate. The biodiversity of these three platforms raises the possibility of tailoring the desired product and process to the choice of host to take full advantage of the native features of each extreme thermophile. Bioprocessing scenarios will be presented for each microorganism in addition to advantages and disadvantages of high temperature bioprocessing.

BIOT 222

Development of an engineered Escherichia coli consortium for bioconversion of multi-substrate biomass streams to commodity chemicals and fuels

Fang Liu1, [email protected], Mary Tran-Gyamfi1, James Jaryenneh2, Xun Zhuang1, Eric Monroe1, Ryan W. Davis1. (1) Department of Biomass Science & Conversion Technologies, Sandia National Laboratories, Livermore, California, United States (2) Department of Systems Biology, Sandia National Laboratories, Livermore, California, United States

Biomass sources such as distillers’ grains with solubles (DGS) from first generation bioethanol production and wastewater cultivated algae provide significant supplies of feedstocks that are rich in both sugars and proteins which can be used as substrates for bio-based commodity chemicals. Here we demonstrate “one-pot” bioconversion of the protein and carbohydrate fractions of DGS and algal hydrolysates into C4 and C5 fusel alcohols using a microbial consortium incorporating two engineered Escherichia coli strains. In the engineered bioconversion system, one E. coli strain was modified for efficient conversion of hexose and pentose sugar to isobutanol; the second E. coli strain was modified for the efficient utilization of the proteins in the hydrolysates to produce mixed C4 and C5 alcohols. The inoculation ratio between the two E. coli strains in the co-culture system was optimized to obtain up to 10.3 g/L and 5.9 g/L titers of total fusel alcohols from DGS and algal hydrolysates, respectively. The bioconversion efficiency of the co-culture was found to be insignificantly different from the efficiencies of fermenting the substrates by each strain alone. Furthermore, a quantitative PCR-based cell quantification method was developed to enumerate the dynamics of each individual bacterial population in the co-culture. The results indicated that despite growth rate difference between the two strains, co-culturing didn’t compromise the growth of each strain. The q-PCR analysis also demonstrated that the fermentation with an appropriate initial inoculation ratio of the two strains is important to achieve a balanced co-culture population which results in higher total fuel titer. The E. coli co-culture strategy eliminates the need for fractionation of hydrolysates and multi-step fermentation for the conversion of multi-substrate biomass streams, significantly reducing the overall process cost and residence time.

BIOT 223 - Research on remediation the NO3 -N polluted groundwater by the synergistic effect of autotrophic and heterotrophic denitrification

Mei Lan, [email protected]. Energy and Environmental Engnieering, Hebei University of Engineering, Handan, Hebei, China

The shortage of the organic carbon source in groundwater has limited the nitrogen removal of heterotrophic denitrification. In this paper, slow carbon releasing material was prepared by straw, polyvinyl alcohol (PVA), cellulose. And five different kinds of amylase were added to control carbon releasing velocity. The prepared material formed the interlocking/disperse-phase structure. PVA served as continuous phase and skeleton, whereas the straw performance as release component. the compressive strength, structural properties and carbon release properties of the material were studied. Then the carbon release material was added to an electrode biofilm reactor to establish the synergistic effect of autotrophic and heterotrophic denitrification to restore the nitrate polluted groundwater. Through the orthogonal test of the ratio of the material, The order of effective extent that effect the carbon releasing was tested and the kinetic equation of the carbon release can be expressed as . The autotrophic reaction and the system of synergistic heterotrophic—autotrophic denitrification was investigated, the influence on denitrification of current ,influent - concentration and HRT were evaluated. NO3 -N removal rate increased first and then - decreased with the increase of current intensity. At same time NO2 - N accumulation - with NO3 -N removal rate was studied. The result of synergistic heterotrophic— autotrophic denitrification showed that the optimum operation parameter was 8hours, 50mA and 30 degrees Celsius respectively. Microorganisms were sequenced using high throughput methods which was performed on two samples taken from the combination system. The results showed that the Coverage index of the two samples were all above 0.999 Shannon and Simpson index value demonstrated that heterotrophic-autotrophic denitrifying bacteria has more abundant bacteria diversity. 12 bacterial phyla were found on contact surface on the carbon materials sample. Among them, the percentage of Proteobacteria is the largest in the sample with 98% in abundance. Pseudomonadaceae and Sphingomonadaceae were the dominant bacteria, the proportion was 48.08% and 22.7% respectively. Pseudomonas have the maximum abundance in the genus which is up to 37.4%.

BIOT 224

Programmable DNA-guided artificial restriction enzymes

Behnam Enghiad, [email protected], Huimin Zhao. Chemical and biomolecular Engineering, University of Illinois at Urbana-Champaign, Champaign, Illinois, United States

Restriction enzymes are bacterial proteins that recognize specific DNA sequences and cut DNA at or near the recognition site. They were discovered in bacteria in the late 1960s and have become the cornerstone of recombinant DNA technology since the early 1970s. So far more than 3600 restriction enzymes with >250 distinct specificities have been characterized, in which >250 are now commercially available for routine use in molecular biology. Most of them (i.e., type II restriction enzymes) only recognize short DNA sequences (4–8 base pairs), which significantly limits their application in recombinant DNA technology.

To address these limitations, we developed a new platform to generate highly active Artificial Restriction Enzymes (AREs) with virtually any sequence specificity and defined sticky ends of varying length based on the prokaryotic Argonaute of the archaeon Pyrococcus furiosus (PfAgo). PfAgo protein is a DNA-guided nuclease (771 amino acids) that targets cognate DNA and is most active at a temperature range from 87 to 99.9 °C. It utilizes small 5′-phosphorylated DNA guides to cleave single stranded DNA targets, and does not utilize RNA as guide or target. Our strategy is to separate two strands of target DNA (circular or linear) by incubating the DNA samples at high temperatures. After the two strands of DNA are partially or completely separated, PfAgo will use two different phosphorylated DNA guides, targeting two strands of DNA at the desired locations, and cleave the strands in separate events. Once each strand of DNA is cleaved at the desired location, the two strands can reanneal to generate desired cleaved dsDNA. We used this platform to generate over 18 AREs for DNA fingerprinting and molecular cloning of PCR-amplified or genomic DNAs. These AREs worked as efficiently as their naturally occurring counterparts, and some of them even did not have any naturally occurring counterparts, demonstrating easy programmability, generality, versatility, and high efficiency for this new technology.

BIOT 225

Spatially organizing biochemistry: Choosing a strategy to translate synthetic biology to the factory

Christopher Jakobson2, [email protected], Marilyn Slininger Lee1, Danielle T. Ercek1, Niall Mangan3. (1) Dept of Chemical and Biological Engineering, Northwestern University, Evanston, Illinois, United States (2) Stanford University School of Medicine, Stanford, California, United States (3) Eng. Sci. and Appl. Mathematics, Northwestern University, Evanston, Illinois, United States

Natural biochemical processes are exquisitely organized in space and time; most industrial bioprocesses lack this kind of organization. There has therefore been consistent interest in recent years in designing methods to organize heterologous biosyntheses inside living cells. Diverse approaches, which have used DNA- and RNA- as well as protein-based organization strategies, have demonstrated the potential of spatial organization to enable and enhance biosynthesis. Despite the explosion of possible methods to organize biosynthesis, there lacks a fundamental understanding of how to choose optimal organization strategies for arbitrary biosynthetic pathways of interest. Building a mechanistic understanding of this problem will be key in applying these technologies broadly and successfully in systems and synthetic biology. We address this gap in understanding using a spatially resolved reaction-diffusion model to quantitatively predict the benefits accrued to a biosynthetic system upon organization. Using an approach we pioneered in understanding the native function of two important bacterial organelles, we develop a mathematical framework that predicts the optimal organizational strategy for a given biosynthetic pathway. We demonstrate that, for two example biochemical pathways, the optimal organizational choice is often distinct and depends both on intrinsic characteristics of the pathway and on extrinsic factors such as culture conditions. Moreover, the choice of objective function (e.g. maximum flux or minimum intermediate leakage) influences the optimal choice of organization strategy. Our explicit comparison of different organization strategies within the same mathematical framework is an important step toward a detailed understanding of the tradeoffs involved in spatial organization both in natural and engineered biochemical systems.

BIOT 226

Portable educational tools for synthetic biology

Jessica C. Stark1, [email protected], Ally Huang3, Thomas C. Ferrante2, Rachel S. Dubner4, Karen J. Hsu5, Peter Q. Nguyen2, Nina Donghia2, Keith Pardee6, James Collins3, Michael C. Jewett1. (1) Chemical and Biological Engineering, Northwestern University, Evanston, Illinois, United States (2) Wyss Institute, Harvard University, Boston, Massachusetts, United States (3) Biological Engineering, MIT, Cambridge, Massachusetts, United States (4) Biological Sciences, Northwestern University, Evanston, Illinois, United States (5) Mechanical Engineering, Northwestern University, Evanston, Illinois, United States (6) Pharmacy, University of Toronto, Toronto, Ontario, Canada

Synthetic biology is the practice of engineering cellular behavior, which is poised to transform society through the manufacture of “green” chemicals and fuels, personalized biological therapies, and cheap and deployable diagnostics. Despite the promise of this field, implementation of synthetic biology curriculum in education has been limited due to the need for refrigeration and specialized equipment to grow, store, and transport cells. We describe here the development of portable, just-add-water educational tools designed for use outside the laboratory by untrained operators. These tools were enabled by high-yielding (≥600 μg mL-1) and visible expression of a fluorescent protein library in freeze-dried cell-free (FD-CF) reactions that can be distributed without refrigeration. We developed two lab modules designed to teach the central dogma of biology and engineering design of biological systems and report data from successful testing with Chicago public school teachers and students. Our educational platform further decreases the cost of FD-CF reactions by two orders of magnitude compared to using commercially available kits, with reaction cost estimated at just ~$0.01 per microliter. This work represents the creation of a “chemistry kit” for synthetic biology, which promises to promote teaching of synthetic biology inside and outside the classroom.

BIOT 227

Characterization and efforts towards trapping a thiyl radical with RlmN I309W

Carlos F. Rivera López1, [email protected], Matthew R. Bauerle2, [email protected], Squire J. Booker2,3, [email protected]. (1) Department of Industrial Biotechnology, University of Puerto Rico, Mayagüez Campus, Orocovis, Puerto Rico, United States (2) Department of Chemistry, The Pennsylvania State University, State College, Pennsylvania, United States (3) Howard Hughes Medical Institute, Chevy Chase, Maryland, United States

RlmN and Cfr are two members of the radical S-adenosylmethionine (RS) superfamily of enzymes and catalyze the methylation of carbons 2 and 8 of adenosine 2503 (A2503) in 23S rRNA. RlmN methylates the C2 position, which has been shown to enhance the translational fidelity of the ribosome. RlmN also methylates A37 of different E. coli tRNAs. Cfr, a homologous enzyme, methylates the C8 position first and then the C2 position. The methylation of C8 by Cfr confers the bacteria resistance to several types of antibiotics, and has been isolated from antibacterial resistant bacteria across the globe. A mechanism for RlmN and Cfr has been proposed, in which a covalent cross- link is formed between the enzyme and RNA. Resolution of the crosslink results in thiyl radical and the enamine tautomer of the methylated product. In order to provide evidence for a thiyl radical, a RlmN I309W variant was generated. Isoleucine 309 is located near the position of the proposed thiyl radical in the active site of RlmN. Upon generation of the thiyl radical in the reaction, the radical may add into the Pi system of the tryptophan and form a cross-link. Here we show that the I309W variant produces methylated product and no radical was detected by EPR, indicating that there was no cross-link formation.

BIOT 228

Marine-derived and cold-adapted recombinant functional tyrosinase, that is functionally available as a monophenol monooxygenase

Eungsu Kang, [email protected], Yoo Seong Choi. Department of Chemical Engineering and Applied Chemistry, Chungnam National University, Daejeon, Korea (the Republic of)

Tyrosinase efficiently catalyzes the ortho-hydroxylation of monophenols and the oxidation of diphenols without any additional cofactors. Although it is of significant interest for the biosynthesis of catechol derivatives, the rapid catechol oxidase activity and inactivation of tyrosinase have hampered its practical utilization as a monophenol monooxygenase. Here, we prepared a functional tyrosinase and relevant mutants that exhibited a distinguished monophenolase/diphenolase activity ratio and an enhanced catalytic efficiency against L-tyrosine. The enzyme was still highly active in iced water, and its activity was well conserved under mild reaction conditions. In vitro DOPA modification and chitosan/gelatin hydrogel formation were achieved by the enzyme taking advantage of these biocatalytic properties. These results demonstrate the strong potential in biomedical and industrial applications.

BIOT 229

Using pH-responsive lignin amphoteric surfactant to recycle cellulase by electrostatic interaction Cheng Cai, Qingqiu Xue, Hongming Lou, [email protected]. South China University of Technology, Guangzhou, China

The recovery of cellulase is a good strategy to solve the problem of large loading and high cost of cellulase during the enzymatic hydrolysis of cellulose. At present, the main methods of cellulase recovery include ultrafiltration, readsorption onto fresh substrates, and enzyme immobilization. In this study, a new method for recovering cellulase by electrostatic adsorption was proposed, and the pH-responsive lignin amphoteric surfactant (pH-LAS) was used as adsorbent. pH-LAS with an isoelectric point of 2.2 was obtained by the quaternization of sodium lignosulfonate (SL). When solution pH dropped from 5.0 to 3.0, more than 95% of pH-LAS was precipitated from acetate buffer. Unlike SL, pH-LAS did not inhibit the enzymatic hydrolysis of Avicel. When the pH of solution which contained 3 g/L pH-LAS and 0.2 g protein/L cellulase was adjusted to 3.2, more than 90% of enzyme activity for Avicel could be recovered by centrifugation. pH-LAS had different recovery capacities for different components of cellulase CTec2, more than 90% of beta glucosidase (β-GI), endoglucanase III (EG III) and EG V could be recovered, and most of Xylanase (Xyn) could be recovered. During the enzymatic hydrolysis of Avicel, using pH-LAS to recycle cellulase by adjusting pH could save about 70% cellulase. This method is not only simple and fast, but also requires no additional equipment, which is an efficient and promising method for cellulase recovery.

BIOT 230

Comparison of substrates for colorimetric assays using peroxidase-like deoxyribozyme PW17

Maithi Tran1, [email protected], Yulia Gerasimova3, Ryan Connelly2. (1) University of Central Florida, Orlando, Florida, United States (2) Chemistry, University of Central Florida, Orlando, Florida, United States

Guanine-rich nucleic acid sequences can fold into G-quadruplex (G4) structures of different topologies. Previous studies have shown that antiparallel G4 structures, when in complex with hemin, act as catalysts for the peroxidation of organic compounds. Therefore, such nucleic acid sequences can be referred to as peroxidase-like deoxyribozymes (pDz) – functional analogues of horseradish peroxidase, a protein enzyme widely used in various colorimetric assays. The catalytic properties of pDz have been explored in designing sensors for the analysis of metal ions, small molecules, and nucleic acids. A number of organic indicators that become colored upon oxidation have been suggested in the literature. Here, we systemically compared a selection of common organic indicators available for colorimetric assays utilizing horseradish peroxidase or pDz. We utilized PW17, a short oligonucleotide that inherently folds into a pDz, as a model to study the peroxidation reaction of 3,3′-diaminobenzidine (DAB), 2,2′- Azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS), 3,3′,5,5′-tetramethylbenzidine (TMB), 2,3-benzopyrrole (Indole), and 3-amino-9-ethylcarbazole (AEC). The results of this study detail the comparative strengths and weaknesses of these substrates for use in peroxidase-based colorimetric assays.

BIOT 231

Investigation into autocatalytic intramolecular isopeptide bond forming reaction sites

Samuel P. Kasson, [email protected], Keiichi Yoshimatsu. Chemistry, Missouri State University, Springfield, Missouri, United States

Autocatalytic intramolecular isopeptide bond forming protein domains have been recently discovered in certain gram-positive bacterial pilus structures. Upon protein splitting of these domains, retention of isopeptide bonding capabilities has been observed, thus forming stable, selective, bio-orthogonal Catcher/Tag systems. One such domain, the CnaB2 domain found in the FbaB pilus structure of Streptococcus pyogenes, has yielded the Catcher/Tag, Protein/Peptide systems termed SpyCatchter and SpyTag. Recent literature has focused on tag optimization, stability and bio- orthogonality evaluation, along with applications in bioconjugation. In order to better understand the nature of the specificity involved in the forming of the SpyCatcher/SpyTag linkage, we have recombinantly expressed SpyCatcher and SpyTag-fused proteins in E.coli, and we are currently investigating the role of amino acid residues on, and around, the reactive lysine residue on SpyCatcher by several different analytical methods.

BIOT 232

Site-specific incorporation of the TEMPO organic catalyst into a thermostable alcohol dehydrogenase produces a selective bio/organo- hybrid catalyst

Walaa Abdallah1, [email protected], Louis Lancaster3, David Hickey2, Shelley D. Minteer2, Ian R. Wheeldon3, Scott Banta1. (1) Columbia University, New York, New York, United States (2) University of Utah, Salt Lake City, Utah, United States (3) University of California, Riverside, Riverside, California, United States

Amber stop codons were introduced at three different positions in the thermostable alcohol dehydrogenase D, AdhD, from Pyrococcus furiosus. This enabled the incorporation of the unnatural amino acid, para-azido phenylalanine, to replace the initiation methionine (M1TAG), the catalytic tyrosine 64 (Y64TAG), and tyrosine 205 near the cofactor binding pocket (Y205TAG). The addition of the unnatural amino acid rendered Y64TAG inactive as its hydroxyl group, which participates in catalysis, was replaced with an azide group. The remaining mutants, M1TAG and Y205TAG, retained AdhD activity with 2,3-butanediol, its preferred substrate, and cofactor, NAD+. The three sites allowed for the site-specific attachment of an alkyne modified TEMPO organic catalyst via strain-promoted cycloaddition. AdhD has a preference for the oxidation of secondary alcohols with a specificity for medium carbon chain lengths while TEMPO prefers to oxidize primary alcohols of varying chain length. The bio/organo- hybrid catalysts demonstrate that enzymatic selectivity can be transferred to an organic catalyst by this approach. And this specificity can be altered by standard protein engineering methods.

BIOT 233

Increase in expression levels, cell lysis, protein refolding and purification efficiency in recombinant Streptokinase using a non Beta Lactum marker for improved patient safety profile in thrombolysis applications for STEMI patients

Debayan Ghosh, [email protected]. President, Epygen Labs FZ LLC, Dubai, United Arab Emirates

For a large part of the world, where patients requiring Thrmbolysis cannot afford an expensive tPA treatment, Streptokinase is still the only viable option. The new recombinant Streptokinase (rSK) clone with a non Beta Lactum marker has been optimized for high level of protein expression using HCD fermentation and intelligent feeding techniques. An efficient Cell Lysis method has been developed by varying OD, and finally by designing an efficient protein refolding mechanism utilizing tools of hydrodynamic optimization. Chromatographic matrices are selected for highest binding efficiencies to deliver high level of purified rSK protein within shortest period of time. The idea behind the exercise is to express high purity rSK proteins at the maximum levels of protein for liter, resulting in a substantial cost reduction. Amongst Cardiovascular patients and number of people suffering strokes in urban and rural parts of India, at least 0.5% are at risk of suffering from myocardial infarction. This gives a potential of 6.0 million cases of thrombolytic treatment in India. Though for urban patients with higher economic capabilities, top preference would be a tPA drug followed by PCI, 60% of urban population and nearly 100% of the rural population are still candidates for streptokinase treatment. It is the key objective of this research to produce this life saving drug in an affordable manner for the millions of patients in South Asia and African region, who need it most.

HIC peaks comparing resins

BIOT 234

Designing synthetic metabolons via dCas9-guided assembly

Emily o. Berckman2, [email protected], Wilfred Chen1. (1) Chemical Engineering, University of Delaware, Newark, Delaware, United States (2) Chemistry and Biochemistry, University of Delaware, Newark, Delaware, United States

In the past few decades, scientists have exploited the natural metabolic processes of microorganisms to achieve synthesis of a many products such as pharmaceuticals, fuel and both bulk and fine chemicals. Metabolic engineering of microorganisms exceeds the current ability of most organic synthetic chemists, but our existing techniques lack the ability to easily optimize cell productivity and product titer for expansion to a variety of enzymatic pathways. Our goal is to move away from traditional approaches of overexpressing rate limiting enzymes and competitive pathway deletion, and towards a strategy in which a cell can sense the accumulation of an intermediate and elicit proximity control over enzyme organization. This proximity enzyme organization, also known as a metabolon, will facilitate in the passage of intermediates from one enzyme to another, thus limiting diffusion rates, enabling fast turnover of toxic intermediates and preventing crosstalk between competing metabolic pathways. Here we have developed a synthetic metabolon using DNA as a target and nuclease-null Cas9 fusion proteins as binding partners. Utilizing DNA as a target allows for easy manipulation of target sequences as well as control of enzyme stoichiometry, which is particularly important for rate limiting reactions. Exploiting the orthogonality of nuclease-null Cas9 proteins and the ability to fuse pathway enzymes to these highly site specific binding proteins allows for not only control over stable multiple binding of nuclease- null Cas9 to the target in a highly predictable manner but also shows the flexibility of the system as it can be applied to a multitude of enzymatic pathways.

BIOT 235

Comparison of transfection methods on yield of recombinant human IgG1 Fc

Evan A. Wells, [email protected], Anne S. Robinson. Chem and Biomolecular Eng, 300 Lindy Boggs Bldg, Tulane University, New Orleans, Louisiana, United States

Monoclonal antibodies (mAbs) offer many advantages for the treatment of chronic and inflammatory diseases over traditional small molecule therapeutics. Indeed, their proven value as therapeutics sparked major interest from pharmaceutical companies for increased development and marketing with the global market for mAbs predicted to reach $1.2 trillion USD by 2020. While unrivaled specificity and a dearth of adverse effects shine as some of the most desirable characteristics, the cost of production, stability issues from manufacturing to delivery, and requisite high concentrations for physiological efficiency are issues which still need consideration. mAbs share many common structural features, so understanding how each part contributes to structural characteristics of the entire molecule may unveil ways to design better therapeutics. The Fc portion of a mAb enables purification via Protein A chromatography and controls antibody effector functions (e.g. antibody dependent cell- mediated cytotoxicity, complement dependent cytotoxicity) depending on glycoform presence and identity. Characterizing both aglycosylated and glycosylated Fc and their contributions to antibody stability requires adequate recombinant protein yield from mammalian cell culture.

Mammalian cell cultures (e.g. CHO and HEK) used for protein production are either stable or transiently transfected, and lipofection and polyfection represent the most common methods of performing the transfection. Here, we determine head-to-head comparison of the yields of recombinant Fc protein from transient/stable and lipid/polymer mediated transfections in 30 mL scale cultures of HEK Freestyle 293 and ATCC HEK293 cells. Heterogeneous populations of stably transfected HEK cells under Zeocin™ antibiotic selection produced 5-10 mg/L yields, but further passages of these cells reduced overall production. The proprietary cationic lipid formulations resulted in the highest overall Fc amounts (5-15 mg/L) in transiently transfected cells, but PEI mediated polyfection (yielding >2 mg/L) could offer greater overall cost efficiency and scalability under similar conditions.

BIOT 236 Engineering stable anaerobic consortia by understanding the genomic basis for syntrophic interactions

Jennifer Brown, [email protected], Xuefeng Peng, Sean Gilmore, John Henske, Michelle A. O'Malley. Chemical Engineering, University of California Santa Barbara, Santa Barbara, California, United States

Waste management and sustainable energy production are two major concerns of modern society. The use of microbial consortia for waste treatment has the potential to address both of these concerns simultaneously, since consortia possess the ability to convert crude biomass into biofuels and bio-based chemicals. The organic fraction of municipal solid waste (OFMSW) is an abundant and inexpensive carbon substrate that can be utilized by microbial systems to generate useful products. Anaerobic consortia containing fungi and methanogenic archaea capable of converting wet waste materials into valuable substances already exist in nature, and have been isolated from the guts of herbivores. Synthetic pairings of fungi and methanogens have been shown to accelerate biomass degradation while producing methane that can be captured and used as fuel. Although bioreactors utilizing undefined natural consortia to digest wet waste and generate biogas have been constructed, the failure rate is high due to consortia instability. The development of biotechnology capable of handling variable input, recovering from environmental disturbances, and producing consistent products is dependent upon engineering stability and robustness into the syntrophic relationship between consortia members. Here, we have used omics techniques to examine what drives stability in biomass-degrading anaerobic consortia. Growth curves were obtained for three synthetic pairings of the methanogen Methanobacterium bryantii with each of the anaerobic fungi Neocallimastix californiae, Caecomyces churrovis, and Piromyces finnis using pressure readings and gas chromatography. All pairings produced measurable amounts of methane when grown on a reed canary grass substrate. Methane production, a consistent indicator of growth that can be used as a metric to define longevity, ceased 14 days after inoculation for all pairings, indicating a distinct trend in consortia longevity. Metabolomics and transcriptomics were used to detect variations in metabolic output and gene expression for the synthetic pairings grown on different substrates in order to evaluate how variable inputs affect product consistency. The ability to engineer stability and robustness into biomass-degrading microbial communities capable of directly converting crude biomass into useful products will result in a reliable process that eliminates the costly, energy-intensive pretreatment step required by traditional bio-based production methods.

BIOT 237 mRNA volatility as a strategy in designing and implementing effective antisense

Christine Endicott, Yongku Cho, Ranjan Srivastava, [email protected]. Chemical and Biomolecular Engineering, University of Connecticut, Storrs, Connecticut, United States Antisense RNA molecules regulate the expression of proteins at the translational level in all branches of life. These molecules hybridize to their target mRNA via Watson-Crick pairing and hydrogen bonding to form a double-stranded RNA complex. This duplex inhibits protein synthesis by preventing ribosome binding or advancement or by recruiting RNases. Antisense can be challenging to design due to mRNA secondary structure formation sequestering regions of the mRNA, preventing antisense binding. However, the mRNA secondary structure volatility may be a physical characteristic that can be used to design effective antisense. Certain regions of the mRNA molecule will experience more frequent hydrogen bond formation and breakage as the molecule fluctuates between various secondary structures. We theorize that being able to predict and target these volatile regions may serve as critical strategy in designing efficacious antisense. We have previously developed GenAVERT, a software platform that predicts mRNA volatility. In this study, we analyzed naturally evolved examples of antisense in prokaryotes. Toxic proteins regulated by antisense showed that the target regions on the mRNA aligned with the regions predicted to be the most volatile. The antisense target region of the mRNA of proteins that had roles in metabolism or transport were found to be in less volatile regions. These results indicated that the antisense had to evolve to be highly effective in the toxin examples because survival depends on tight control of toxic proteins. Metabolic and transport proteins require less stringent, or leaky control, which may be advantageous in adapting to fluctuating nutrient conditions. We also tested our theory in in vitro experiments. GenAVERT was used to predict the volatility of green fluorescent protein mRNA. Antisense RNA was coexpressed in Escherichia coli along with a plasmid carrying the gene for green fluorescent protein. The 113- base pair antisense RNA that was designed based on mRNA volatility showed a 46% decrease in fluorescence compared to uninduced antisense. Antisense RNA of the same length designed by other traditionally-used software only showed a 13% decrease in fluorescence. These results indicate that using volatility as an antisense design tool may lead to enhanced efficiency and the ability to tailor protein expression using RNA antisense.

BIOT 238

Opportunities for facility-enabled science at the DOE Joint Genome Institute (JGI)

Yasuo Yoshikuni, [email protected]. DOE Joint Genome Institute, Walnut Creek, California, United States

Fundamental unsolved problem in genomics is the need for high-throughput approaches to bridge the gap between the availability of DNA sequence data and our ability to assign biological function to it. The DOE JGI’s niche is the development of a diversity of large-scale experimental and computational capabilities to link sequence to biological insights relevant to energy and environmental research. This will range from preparing material and applying functional capabilities prior to genomic analysis to post- sequence processing and manipulations to enable the Institute’s users to carry out sequence-to-function studies that are beyond the capabilities of individual laboratories. Here, I will discuss new sequence-to-function capabilities that have been established at the DOE JGI and some scientific examples enabled with these capabilities (Schwander et. al. Science 2016 and Tsementzi et al. Nature 2016).

BIOT 239

Improving human A1 adenosine receptor expression in yeast towards applications in therapeutics screening and development

Abhinav R. Jain2, [email protected], Anne S. Robinson1. (1) Chem and Biomolecular Eng, 300 Lindy Boggs Bldg, Tulane University, New Orleans, Louisiana, United States (2) Chemical and Biomolecular Engineering, Tulane University, New Orleans, Louisiana, United States

G Protein-Coupled Receptors (GPCRs) are the largest family of membrane proteins characterized by seven transmembrane domains. Biophysical characterization and high-resolution structure determination are routinely used for GPCR drug design and discovery and require heterologous expression and purification of 10-100’s mgs of protein. The adenosine receptor subfamily (A1R, A2aR, A2bR and A3R) of GPCRs bind the natural ligand adenosine, an important energy metabolite and modulate adenosine’s essential signaling. A2aR production is capable of high yields – on the order of milligram protein per liter of culture – making it unusually high compared to other adenosine receptors, or GPCRs in general. Although high-level GPCR expression has facilitated structure determination, to date, heterologous expression of A1R in multiple systems has resulted in low yields of active receptor. Previous studies suggest the C-terminus of the A2aR plays a role in the observed efficient expression and trafficking to the plasma membrane of the receptor. To understand the role of the A2aR C-terminus and to test its efficacy in improving the expression of a subfamily member, we developed A1R chimeras with the A2aR C- terminus using homologous DNA recombination. The role of the C-terminus of the A2aR on chimeric receptor expression, trafficking to the plasma membrane, unfolded protein response, protein turnover rate and radioligand binding relative to wildtype A1R will be discussed. Preliminary results show improved expression and trafficking of the chimeric A1R compared to wild-type receptor. With our engineered chimeric receptors, we hope to improve expression and trafficking of other recombinant GPCRs.

BIOT 240

Aromatic metabolism of 50 non-model oleaginous yeasts

Allison Yaguchi1, [email protected], Kyria Boundy-Mills2, Mark A. Blenner1. (1) Chemical and Biomolecular Engineering, Clemson University, Clemson, South Carolina, United States (2) University of California, Davis, Davis, California, United States

Oleaginous yeasts have long been a target for developing industrial-scale biochemical processes due to their ability to accumulate high amounts of lipids, synthesize complex chemicals and proteins, and robustly metabolize diverse feedstocks. Interest in using lignocellulosic biomass as a feedstock has grown, particularly the lignin fraction for its aromatic composition. Being the second-most abundant biopolymer on earth, lignin is the largest sustainable source of aromatics; however, it is incredibly heterogeneous and recalcitrant. In addition, processing lignocellulosic biomass requires harsh solvents such as acetic acid and produces toxic byproducts such as furfural. Cutaneotrichosporon oleaginosus (ATCC 20509), previously known as Crypotococcus curvatus, is an oleaginous yeast known to metabolize many sugars, such as lactose and xylose, and tolerate harsh processing chemicals. Recently, we discovered this yeast is also capable of tolerating and metabolizing lignin-derived monoaromatics and producing 70% of its dry weight as lipids. This motivated us to explore aromatic metabolism in a large cross- section of oleaginous yeast to determine the uniqueness of this property. This work describes screening of 50 oleaginous yeasts from the Phaff collection for their ability to tolerate and metabolize aromatic compounds commonly found in lignocellulosic hydrolysates. Phylogenetic analysis of the robust growers determined many came from the Rhodotorula, Trichosporon, and Cutaneotrichosporon genera. Key strains were investigated further for their ability to produce compounds of interest, such as β- carotene and lipids, while using aromatics, such as phenol, p-coumaric acid, and ferulic acid, as the sole carbon source. With the development of genetic tools, non-model organisms have the potential to become commercial platforms for valorization of toxic waste streams and production of value-added chemicals.

BIOT 241

Optimizing amino acids in defined media for Pichia pastoris recombinant protein expression

Angel Kuo1,2, [email protected], Catherine B. Matthews1, Kerry Love1, John C. Love1. (1) Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts, United States (2) Chemistry, Wellesley College, Wellesley, Massachusetts, United States

In the biopharmaceutical industry, media for host organisms have been optimized with nutrients like amino acids to maximize recombinant protein production. Because amino acids serve as the building blocks of proteins, as well as intermediates in various metabolic pathways, they are posited to relieve much of the metabolic burden experienced by organisms producing heterologous proteins. Both complex amino acid mixtures and defined amino acids have shown to improve protein titers in various organisms. While there have been some studies on media optimization with amino acids for yeast, few systematic studies of amino acid supplementation have been reported for Pichia pastoris, a widely-used organism for producing biological therapeutics. Commercially available defined media often lacks nutrients, including specific amino acids found in complex media. However, defined media has the potential to simplify downstream protein purification and quality control, and is therefore of great interest. Here, we present a study of the impact of adding specific amino acids on recombinant protein titers. To uncover a general formulation, we studied three P. pastoris strains, each expressing a different heterologous protein. Across all three strains, some amino acids increased protein production, while some amino acids decreased protein production relative to controls lacking amino acids. Given its ability to uncover metabolic pathways associated with a particular phenotypic state, RNA sequencing was performed on a subset of the media conditions to further understand the biological benefits of adding amino acids to the media. Metabolic intermediates of relevant pathways associated with highly productive fermentations, containing either upregulated or downregulated genes depending on the pathway, were further mined for additional gains in heterologous protein expression in Pichia. We believe that this method of amino acid optimization has general application and could be applied to other organisms of interest for improved recombinant protein production.

BIOT 242

Evaluation of L-asparaginase production in flask, tray bioreactor and in-house designed rotary bioreactor using isolated Aspergillus sp

Kruthi Doriya, [email protected], Santhosh K. Devarai. Chemical Engineering, Indian Institute of Technology Hyderabad, Hyderabad, Telangana, India

L-asparaginase is an amido-hydrolytic enzyme that degrades asparagine, an attribute that makes it as anti-neoplastic agent and anti-acrylamide agent in therapeutic and food industry respectively. Several researchers have isolated L-asparaginase from animal, plant, fungal, and bacterial sources. Currently asparaginase from Escherichia coli and Erwinia carotovora are used in the treatment of acute lymphoblastic leukaemia. Toxicity and sensitive reactions caused by these bacterial L-asparaginase shifted focus to eukaryotic microorganisms such as fungi. Therefore, a novel fungi capable of hydrolysing asparagine was isolated and was later identified as Aspergillus sp. Effects of substrate composition and culture conditions were studied using solid-state fermentation at flask level. Simplex centroid experimental design and Box-Behnken design were used during the design and analysis of results. Substrate composition consisting of cotton seed cake (2/3), wheat bran (1/6), and red gram husk (1/6) promoted maximum activity of 12.57 U/mL with optimum conditions of temperature 35 °C, pH-8 and moisture content 70% (v/w). SSF is used traditionally in the production of a wide range of enzymes and metabolites. However, SSF has limited application for large-scale, because of mass and heat transfer glitches. Considering these limitations in conventional SSF reactors, a new design for the SSF has been attempted containing multiple rotating drum bioreactors that are arranged in a horizontal manner, each with its separate feed inlets. The optimized parameters at flask level were transferred to tray and in-house designed rotating drum bioreactor with superior mixing and cooling capability. After 5 days of fermentation in the tray bioreactor, the enzyme activity was 20.58U/gds with 500 g of substrate mixture and 1 cm bed thickness. This demonstration will assist in comparing the L-asparaginase production in flask, tray bioreactor and in-house-designed rotating drum bioreactor.

BIOT 243 High-throughput assessment of a novel, thiol-acrylate hydrogel for tumor spheroid synthesis in a microfluidic device

Nathan Kersker2, [email protected], Wayne Wortmann1, Nora Safabakhsh1, Adam T. Melvin1, John A. Pojman2. (1) Chemical Engineering, Louisiana State University, BATON ROUGE, Louisiana, United States (2) Chemistry, Louisiana State University, Baton Rouge, Louisiana, United States

The extracellular environment plays an important role in regulating cell behavior. The mechanical, structural, and compositional properties of the extracellular matrix can determine the fate of the growing cells. Standard culturing in a flask, fails to represent a native cellular environment. To create this native environment suitable for cells, hydrogels have been previously studied. One of the most extensively studied synthetic hydrogels is a simple poly(ethylene glycol) diacrylate (PEGDA) hydrogel. However, the need for UV polymerization may introduce unwanted changes in cell morphology and viability. The hydrogel being proposed is a form of thiol-acrylate hydrogel that can eliminate this problem and can easily be modified to represent a native cellular environment. To be more specific, it is a PEGDA (700 MW) and Ethoxylated Trimethylolpropane Tri(3-mercaptopropionate) (THIOCURE® ETTMP 1300 MW) hydrogel. This thiol-acrylate hydrogel has been tested by swelling, viability and on-chip microfluidics. The viability testing used common breast cancer lines, MDA-MB-231 and MCF7, while the on-chip trapping experiment used a fluorescent labeled RFP-MDA-MB- 231 cell line. All three test were done over a 7 day period and kept at a temperature of 37.1 °C and pH of 7.4 ± 0.05. Viability testing was done in a 96-well plate that gave the natural spheroid shape that the breast cancer cells are normally seen in.

MDA-MB-231 viability testing with thiol-acrylate hydrogel Day 4

BIOT 244 dsDNA cleavage via deoxyribozyme DNA construct

Martin O'Steen, [email protected], Dmitry M. Kolpashchikov. Chemistry , University of Central Florida, Orlando, Florida, United States

Deoxyribozymes, catalytically active nucleic acid strands, have been selected via in vitro selection to catalyze a variety of reactions. Initially limited to nucleic acid substrates, different selection strategies and technologies have expanded the scope of deoxyribozyme catalysis to other suitable substrates including small molecules and peptides. Even with these developments, the predominant reaction catalyzed by members of the deoxyribozyme family is cleavage of nucleic acid strands. Selection strategy limitations have precluded the development of deoxyribozymes capable of catalyzing the cleavage of secondary structure in biomolecules, including the canonical double helix of DNA. The double stranded helix of DNA presents an exceptional challenge for deoxyribozyme based cleavage in part due to its inherent stability. Here in, we report the development of a DNA based construct that allows for cleavage of double stranded DNA molecules via a deoxyribozyme. Further development of this technology has important implications in both the biomedical and biotechnology fields. BIOT 245

Optimization of probiotic E. coli Nissle 1917 cell-free protein expression system for rapid testing of novel components

Kathryn Beabout1,2, [email protected], Michael Goodson1,2, Svetlana Harbaugh1,3, Jorge L. Chavez1, Nancy Kelley-Loughnane1. (1) 711th Human Performance Wing, Air Force Research Laboratory, Wright-Patterson Air Force Base, Ohio, United States (2) Integrative Health and Performance Sciences, UES, Inc, Dayton, Ohio, United States (3) Henry Jackson Foundation, Bethesda, Maryland, United States

Cell-free protein expression systems (CFPS) can be used as tools for Synthetic Biology with several key advantages over in vivo systems, including the ability to rapidly screen for the functionality of novel genetic components. We are developing and optimizing CFPS from E. coli Nissle 1917, a protective probiotic strain of E. coli, as a tool to rapidly test the functionality of sensors and constructs built for this strain. While preparing E. coli Nissle 1917 extracts we varied several parameters in a protocol established for E. coli BL21* DES, a strain optimized for high protein production and commonly used for generating CFPS. For example, during cultivation of the E. coli Nissle 1917 we selected cells from different growth phases (mid-exponential and late-exponential). Additionally, we optimized the total sonication energy input during lysis, the magnesium concentration of the extract buffer, the cell-free reaction conditions, as well as other parameters. Initial cell-free reactions using the Nissle 1917 extracts and a vector with GFP expressed under the T7 promoter produced modest, but detectable amounts of GFP over background. Optimizing different parameters improved the functionality of the Nissle 1917 system and current work focuses on high throughput testing of relevant components. For example, we plan to engineer E. coli Nissle 1917 to detect stress in the human gut and subsequently produce a response to remediate this condition. A functional Nissle 1917 CFPS provides us with the capability to rapidly test components for this application and others.

BIOT 246

Quantification of the effects of high shear stresses on single circulating tumor cells using a microfluidic device

Grant Landwehr2, [email protected], Andrew Kristof3, Joseph Balhoff2, Jacob Pettigrew2, Sharif Rahman2,1, Adam T. Melvin2. (1) Louisiana State University, Baton Rouge, Louisiana, United States (2) Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana, United States (3) North Carolina State University, Raleigh, North Carolina, United States

An important step in the metastatic progression of cancer is the migration of cancer cells from a primary tumor to distal location through intravasation, in which cancer cells travel through the bloodstream as circulating tumor cells (CTCs). While in the blood vessels, CTCs experience hemodynamic shear stresses at high levels (5-60 dyn/cm2). Previous studies identified that exposing cancer cells to fluid shear stress (FSS) induced a phenotypic change causing the cells to be more metastatic. While a novel finding, this study was limited in its ability to control the FSS used to challenge the cells and was limited to only final measurements at the end of the experiment. The goal of this project is to address this limitation by developing a microfluidic device to isolate and trap single CTCs and then challenge them with varying amounts of FSS. The microfluidics traps were designed with a semi-circular geometry with a space in the middle to allow for optimal trapping and testing of FSS-induced cell deformability. COMSOL simulations were performed to optimize the fluid velocity profile of the device to ensure a high trapping efficiency. The simulations also identified the flow profiles needed to produce FSS on the cells that they typically encounter in the bloodstream. These flow profiles were used to test the phenotypic change in MDA-MB-231 cells trapped in the microfluidic device. Findings demonstrate the ability to efficiently trap single cells in the microfluidic device and the effect of fluid shear stress magnitude and duration on altering size and shape of the cells, allowing for a more accurate determination of cell phenotype. These results will provide new insight into the mechanisms of how CTCs survive in the bloodstream and if changes in FSS results in a more aggressive cancer phenotype.

BIOT 247

Best practice considerations for development and implementation of a mammalian cell culture perfusion process

Patrick McInnis2, Israel Coleman1, Doug Rank2, Elizabeth M. Goodrich1, Michael A. Cunningham1, [email protected]. (1) Applications Engineering, MilliporeSigma, Burlington, Massachusetts, United States (2) Next Generation Processing R&D, MilliporeSigma, Bedford, Massachusetts, United States

Successful production of therapeutic monoclonal antibodies and recombinant proteins is highly dependent on utilizing high-expressing cells, effective cell culture media and optimal upstream processes. Conventional batch and fed-batch modes of bio- production have been constrained by achievable cell density limits and their inability to maximize the expression of recombinant products. Recent process development efforts have focused on evaluating perfusion modes of bioreactor operation in order to increase efficiencies in either achievable cell densities, therapeutic product generation, or both. Implementation of perfusion into upstream process development laboratories necessitates evaluation of many practical factors, including bioreactor and bioreactor control capabilities, assembly and utilization of cell retention devices, and preparation and consumption control of cell culture media and supplements. This study will describe in detail how a 3L-scale bioreactor perfusion platform was evaluated and implemented. A review of scaling considerations and their impact on process economics will be presented. Lessons learned for perfusion implementation and best practice guidance will also be reviewed. In addition, preliminary results will be presented demonstrating the effectiveness of employing a novel cell culture media designed specifically for perfusion applications. BIOT 248

Transitioning from a daily cell bleed perfusion process to a continuous, automated bleed process using a mAb-expressing CHO cell line

Patrick McInnis1, [email protected], Doug Rank2, Michael W. Phillips1. (1) Next Generation Technology and Applications Development, MilliporeSigma, Newton, Massachusetts, United States (2) Next Generation Technology and Applications Development, MilliporeSigma, Bedford, Massachusetts, United States

The biopharmaceutical industry is trending toward continuous bioprocessing due to the potential advantages that it offers over legacy batch processing in regards to facility output, product quality, cost of goods, and ultimately cost per gram of protein product. Historically, perfusion processes have been used for production of unstable molecules, but given the prospective benefits of continuous processing, they are now also being implemented in various applications for stable molecules, such as monoclonal antibodies. Many legacy perfusion processes feature a daily cell bleed as a means to control cell density and therefore extend process duration. While this approach simplifies certain aspects of process characterization, such as culture doubling time, it has significant limitations, including excess media usage, excess product loss, and challenging control of dissolved oxygen and volume. Current long duration perfusion processes operate via continuous bleed, which alleviates some of these challenges. However, the steady cell density can mask changing growth rate, which can lead to altered product quantity and quality. Here we present a case study of how we transitioned from a legacy daily cell bleed perfusion process to a continuous, automated bleed process. We will share and discuss some of the observations and challenges encountered in the continuous bleed mode, such as doubling time determination, death rate determination, variation in cell diameter, importance for fine-tuned volume control, and assessment of consistent cell metabolism. Finally, we will present the metrics with which we evaluate and attempt to solve these challenges, including lactate dehydrogenase release, bleed rate, oxygen demand, and metabolic profile, among others.

BIOT 249

Improvements To CHO cell culture scale-down model across multiple scales

Sean McDermott, [email protected], Andrew Bawn, Nicholas Abu-Absi. Process Sciences, AbbVie Bioresearch Center, Worcester, Massachusetts, United States

The ability to express recombinant antibodies from CHO cells has advanced through developments in cell line selection methods, clone screening, and media formulation. The high oxygen demand of current high performing in-house cell lines was leading to cell culture performance that did not match across the various scales used for cell culture development work. These differences became evident when comparing historical 3 L bioreactor process development studies against runs performed at manufacturing scale (>500 L). Differences in cell culture performance were also evident during the use of cell culture models with volumes at the mL scale, such as spin tubes and multi-well plates, which were utilized for media screening applications. These differences in growth, titer, and PQ across scale signaled that changes needed to be implemented to the scale-down model to create a more efficient and automated system with results that better match manufacturing scale. This model would also be capable of growing alongside industry advancements in cell culture performance, for example, be applied to the development of continuous bioprocessing strategies. Development work began at the 3L scale, testing impeller configurations and dissolved oxygen control strategies that demonstrate alignment of cell culture performance characteristics with manufacturing scale. Under these strategies, parameters, such as total gas flow, dissolved carbon dioxide, and titer, better matched across scales. Work supporting the spin tube as a high-throughput small scale cell culture model will also be presented. Agitation speed, working volume, and system configuration were the parameters under evaluation with this new approach. The correct balance had to be reached that provided for the necessary aeration of the high oxygen demand cell lines without cell damage and reduced performance generated by high agitation conditions. Automated work streams for the spin tube sampling, assay sample preparation, and feed media additions were developed for integration with this cell culture model. Through automation at this scale, the development team will be empowered to increase experimental throughput. Improvement of scale-down models at the 3L and mL scale that can support high performance cell lines will lead to more efficient cell culture development plans that can be easily implemented at the manufacturing scale.

BIOT 250

Investigating the link between pCO2 and protein quality in CHO cell culture

Alexandra Connor, Benjamin Reed, [email protected], Amanda Lewis, Eric Garr. Manufacturing Sciences & Technology, Bristol-Myers Squibb, Devens, Massachusetts, United States

A typical bioprocess to produce a protein therapeutic uses carbonate-buffered media in the cell culture phase, using CO2 to modulate culture pH. Despite this reliance on CO2, dissolved CO2 (pCO2) is rarely directly controlled. Elevated levels of pCO2 have been shown to significantly affect cell growth, protein production, and protein quality in Chinese Hamster Ovary (CHO) cell culture processes, though the mechanism for this impact remains largely unknown. Literature sources have identified pCO2 as influencing intracellular pH, though few of these experiments have been performed in a controlled bioreactor setting under constant culture pH.

This study investigates the effects of both elevated pCO2 at constant culture pH and increased intracellular pH at constant pCO2 on protein quality and productivity in 5-L bioreactors. Specifically, this study aims to determine whether elevated pCO2 under constant culture pH influences intracellular pH, whether modulated intracellular pH at constant pCO2 affects protein quality, and to compare metabolomic profiles and relative transcription of key genes of the two conditions. This work will elucidate the influence of pCO2 on the intracellular machinery of CHO cells, as well as characterize the link between this machinery and outputs of protein titer and quality.

BIOT 251

Culture strategies enhancing the suspension adapted CHO-DXB11-S1 high producing cell line expression

Bor Shiun Chen, [email protected], Hsin-Lin Lu, Dalton Chen, Wei-Kuang Chi. Development Center for Biotechnology, New Taipei City, Taiwan

Bor-Shiun Chen, Hsin-Lin Lu, Dalton Chen, and Wei-Kuang Chi* Bioengineering Department, Biologics Institute, Development Center for Biotechnology No. 101, Lane 169, Kangning St., Xizhi District, New Taipei City, 22180, Taiwan E-mail: [email protected]; [email protected]; [email protected]; [email protected]

In the industry, biologics mass production requires the use of suspension adapted CHO cells and the best culture conditions. Development center for biotechnology (DCB) in Taiwan have adapted CHO-K1, CHO-DXB11 and CHO/dhFr- cell lines to the optimal chemically-defined medium under suspension cultivation with high VCD and low population double time. The adapted cell lines were further applied as the host for therapeutic antibody production. These suspension adapted CHO cell lines and CHO-S cells were transfected with EGFP expression plasmid for the cell transfection studies. In the stable gene expression, the IgG productivity of cell pools and single cell clones in CHO-DXB11-S1 cells was higher than that in CHO-S cells. These results implied that the in-house suspension adapted CHO-DXB11-S1 cells could be a good candidate as the producing host cell line. A series of culture strategies investigations (e.g. osmolality, feeding and epigenetic approach) on cell growth and antibody production for the potential transfected CHO-DXB11-S1 single cell clones were further carried out. These approaches affected both cell growth and monoclonal antibody expression, however not for the cell stability. The monoclonal antibody expression was improved to about 1.3 g/L from 0.9 g/L by drug selection. The antibody titer and specific productivity was further improved by 62.7% and more than 230% respectively by a culture optimization even through 60 generation of adaptive cultivation with medium S2A3. Based on these results, the optimization of culture strategies enhanced monoclonal antibody expression and achieved a material cost reduction of more than 40% and be worth further studies.

BIOT 252

Characterizing CHO cell metabolism in glutamine-limited and supplemented feeds during different stages of exponential growth Brian J. Kirsch1, [email protected], Sandra V. Bennun2, Amy S. Johnson2, Shawn M. Lawrence2, Michael J. Betenbaugh1. (1) Chemical and Biomolecular Engineering, Johns Hopkins University, Baltimore, Maryland, United States (2) Regeneron Pharmaceuticals Inc., Tarrytown, New York, United States

Industrial Chinese Hamster Ovary (CHO) cell strains used to produce biopharmaceuticals can grow in cultures with varying amounts of glutamine, and are characterized by significant changes in their metabolism over a typical fed-batch reactor. In order to understand CHO metabolism during exponential growth, and in particular the role of glutamine, we examined the utilization of 13C-glucose (U and 1,2), 13C-glutamine (U), and 13C-asparagine (U) for CHO cells cultures in limited and supplemented glutamine environments during the early and late exponential phase. Analysis of the intracellular free metabolites revealed differences in the allocation of amino acids to cellular physiology under different nutrient conditions. Likewise, enrichment of cell protein fixed in the biomass indicates different distribution of the 13C from the fed glucose, glutamine, and asparagine following changes in both the glutamine supplementation and time during the exponential phase. 13C metabolic flux analysis (13C MFA) of the data also indicated a change in the distribution of glucose, glutamine, and asparagine into glycolysis versus the TCA cycle for the early and late exponential phase. These findings will increase our understanding of CHO cellular metabolism in different nutrient environments and growth conditions in order to optimize both growth and ultimately protein production for CHO cells in fed-batch cultures.

BIOT 253

Integrating high throughput analytics with early-stage cell line development; a comprehensive quality attribute characterization of CHO clone candidates expressing a model IgG1 antibody

Casey Kohnhorst1, [email protected], Phillip Angart1, Bingyu Kuang2, [email protected], Sai Rashmika Velugula1, David N. Powers1, Cyrus Agarabi1. (1) CDER, U.S. Food and Drug Administration, Silver Spring, Maryland, United States (2) Biomedical Engineering and Biotechnology, University of Massachusetts Lowell, Lowell, Massachusetts, United States

The screening of mammalian cell lines for further development and scale-up in a new biologic manufacturing process has traditionally focused on productivity as the sole metric for clone selection. However, as methods in cell line engineering have improved and high producers are less rare, early screening of product quality, stability, and formulation will become increasingly important in identifying a top performing, low-risk clone. For this study, a CHO DG44 cell line expressing a chimeric IgG1 antibody was used as the model. CHO clones were selected using the DHFR-/MTX system and isolated by limiting dilution. The 10 top-producing CHO clones were determined by an initial ELISA screen and grown in 100 mL spinner flask batch format. Samples were taken daily in order to monitor cell density and viability, nutrients, and secreted IgG concentrations. IgG1 was purified from the cell culture fluid via Protein A chromatography and subjected to a comprehensive quality and stability screening. Stability analysis using SLS and fluorescence spectrometry found that Tagg values of purified IgG1 ranged between 70.2 and 76.6 °C, while Tm values ranged between 68.1 and 69.6 °C. In addition, reducing and non-reducing SDS-PAGE revealed each clone expresses different amounts of a heavy chain variant, while charge variant analysis further distinguished clones as having up to a 20% difference in the proportion of acidic to primary species. Clones favorable in these quality metrics were further analyzed for their glycan profiles using HPLC-MS analysis. Beyond quality metrics, each clone’s specific production/consumption rates for key metabolites and IgG1 were calculated and a DOE study for formulation buffer performed. Taken together, these data were able to clearly differentiate top-producing candidates and readily eliminate clones that would be otherwise acceptable if relying on production value alone. Furthermore, these data elucidate inherent product variability that can exist between clones in a selection pool, which can be used to better inform decisions with regards to quantity versus quality in the early stages of pharmaceutical biologics development.

BIOT 254

Effective CHO lead clone selection for late phase process development in a tight timeline

Christopher L. Oliveira, [email protected], Jianlin Xu, Mariusz Rdultowki, Matthew Rehmann, Andrew Yongky, Bruce Eagan, Michael C. Borys, Zheng Jian Li. Bristol Myers Squibb, Hopkinton, Massachusetts, United States

Since the biopharmaceutical industry is highly competitive, the main purpose of cell culture process development is to develop a highly productive and robust process within a tight clinical development timeline. Thus, sufficient drug substance with a high quality standard can be delivered for clinical trials and unmet patient need. CHO cell line creation and lead clone selection, which often require significant time investment, are critical for cell culture process development and biomanufacturing. Different CHO cell clones show heterogeneity, resulting in different cell characteristics, culture productivity, and protein quality attribute profiles. This can occur even when the clones were created using the same protocol or were subcloned from the same master well. In addition, there are many culture conditions to be optimized for late phase process development and different clones may require different conditions to achieve their high productivity specifically. Therefore, the cell line specific conditions further increase the complexity and timeline for lead clone selection. An additional challenge arises when attempting to change cell line during clinical stages after the FIH campaign. The drug substance quality attributes during the late phase are required to be comparable to the early phase product. In this study, we developed an effective lead clone selection strategy integrated with late phase process development using both high throughput technology and lab bioreactors. After data evaluation of cell culture performance, quality attributes, and genetic testing, we successfully selected a lead clone with favorable titer and quality attributes during late phase process development. In this presentation, we will show two case studies using this strategy for mAb production in a tight timeline. BIOT 255

Evaluation of growth kinetics and antibody quality in mycoplasma-contaminated CHO cell bioreactor cultures

Erica Berilla, [email protected]. US Food and Drug Administration, Silver Spring, Maryland, United States

Mycoplasma contamination events in biomanufacturing can be devastating and expensive. Mycoplasmas can survive in mammalian cell cultures without apparent changes to the culture and can penetrate the 0.2 µm filters often used in the primary clarification of harvested cell culture fluid. Culture cell-based and indicator cell-based assays, the predominant methods used to detect mycoplasma, are highly sensitive but take up to 28 days to complete, and therefore cannot be used for real-time decision making in the manufacturing process. Nucleic acid testing is being endorsed because results are obtainable in just a few hours, but sensitivity is difficult to compare since cell- based methods measure colony forming units and nucleic acid testing measures genome copy number. Due to the high risk of conducting experiments wherein one deliberately spikes mycoplasma into bioreactors, there are many unknowns regarding the limits of detection of mycoplasma species and the upstream effects of a mycoplasma contamination event in the manufacturing setting. Here we studied the ability of mycoplasma to persist in a single-use, rocking bioreactor system containing CHO cells expressing a monoclonal IgG1 antibody. We examined mycoplasma growth and detectability, the effects of mycoplasma on mammalian cell health and metabolism, and effects of mycoplasma on antibody yield and quality. CHO cell growth and antibody titer decreased 48-72 hours after mycoplasma contamination. These data will aid in regulatory decisions concerning mycoplasma risk assessment in biomanufacturing.

BIOT 256

Rapid process monitoring & control in mammalian cell culture using off-gas mass spectrometry analysis

Hai-Yuan Goh1, [email protected], Michael Sulu1, Graham Lewis2, Graham Josland2, Ajoy K. Velayudhan1. (1) Dept of Biochemical Engineering, University College London, London, United Kingdom (2) Thermo Fisher Scientific, UK, United Kingdom

In a concerted effort to investigate the implementation of a process analytical technology (PAT), we have evaluated the applicability of off-gas mass spectrometry (MS) analysis in mammalian cell culture based on the inlet and outlet gases from bioreactors. The limitations of the system & potential applications are also discussed.

Compared to other invasive methods that require physical probes inserted directly into the culture, the MS only requires gas going in and out of the cell culture which means it has minimal impact on the culture itself. With the Prima BT magnetic sector MS (Thermo Fisher Scientific), multiple gas streams can be analyzed. Distance between bioreactors and the MS is also not an issue given a high enough gas flow and a long enough gas tubing connection to the MS. Our data showed that changes in the gas traces on the MS, down to 0.001% mol, can be captured during various events throughout a typical cell culture. The MS data from two replicate 5L benchtop glass bioreactor runs are very comparable, as is the offline data between them; this indicates that the MS data can be used reliably to identify process deviations during bioreactor runs and be used as a convenient way to evaluate batch to batch variation, within predefined specifications, for robust manufacturing. We have also demonstrated that the same MS setup in the benchtop experiments, with minor modifications, can be applied in a 50L single-use bioreactor (SUB). We would also report that there are technical considerations with respect to the feasibility of integrating this system to the various cell culture platforms.

In conclusion, we believe that our work presented here will be of significant relevance to the cell culture community who are keen adopters of PAT and its practical implementation.

BIOT 257

Multiplex characterization of Chinese hamster ovary host cell protein interactions with monoclonal antibodies by yeast surface display

Alexandra Castiel, Colleen McGovern, Kyle M. Doolan, [email protected]. Chemical and Biomolecular Engineering, Lafayette College, Easton, Pennsylvania, United States

Host cell impurities remaining after downstream purification of monoclonal antibodies has the potential to illicit adverse immune responses in patients and limit efficacy. Despite an extensive purification process, Chinese hamster ovary host cell proteins (HCP) have been found to appear with monoclonal antibodies (Mabs) after the purification process regardless of processing conditions. The presence of HCP over a long range of purification conditions suggests the possibility of product association in which HCP may directly interact or bind to Mabs. Due to the complex interaction landscape of many HCPs with product Mab, determination of particular processing conditions that reduce specific HCP-Mab interactions may lead to increased interactions of other HCPs, necessitating multi-dimensional approaches. Yeast surface display is a highly advantageous platform for library studies and fluorescent bar coding approaches which we use to simultaneously measure Mab association with many distinct HCP proteins. Starting with known CHO HCP such as lipoprotein lipase and Cathepsin D, we are developing a high-throughput yeast surface display assay to evaluate the HCP product association landscape over a range of solution conditions.

BIOT 258

Systematic analysis of the effect of trace metal variability on therapeutic protein production in CHO cells Kyle McHugh2, [email protected], Sylvia Sarnik1, Adam Deresienski2, Kathryn Aron2, Barry Drew2, Michael C. Borys2, Zheng Jian Li2. (1) MIT, Cambridge, Massachusetts, United States (2) Bristol-Myers Squibb, Devens, Massachusetts, United States

Trace metals are involved in many cellular processes and have been found to contaminate multiple raw materials used in typical cell culture production processes. As such, understanding how these metals modulate protein expression and quality would provide significant benefits during process development, optimization and manufacturing of therapeutic proteins. A systematic approach was taken to better understand how the levels of specific trace metals may be used to regulate cell culture process parameters and protein quality parameters including N-glycan distribution, charge variant profiles, and protein aggregation. Samples of regular media, reduced- metals media, and additional cell culture additives were tested by mass spectrometry to characterize the levels of trace metal contamination. Design of Experiments in high- throughput Tecan liquid handler production runs was used with a definitive screening design to assess the impact and interactions of 12 different metals on 4 proteins of therapeutic interest. The most significant metal effects were then confirmed in benchtop bioreactors to generate samples for transcriptomic and metabolomic analysis of the molecular mechanisms underlying these effects. Results from the high-throughput screens and follow-up studies will be presented.

BIOT 259

DOE optimization of titer and quality of IgG titer and quality using high- throughput micro-bioreactor system for CHO cell media and supplements

Sai Rashmika Velugula1, [email protected], Abasha Williams2,1, Chih-Jung Hsu1, Nicholas Trunfio1,3, Brittany Chavez1, Cyrus Agarabi1. (1) CDER/OPQ/OBP/DBRR II, US Food & Drug Administration, Silver Spring, Maryland, United States (2) VPPL/VRC/NIAID, NIH, Gaithersburg, Maryland, United States (3) Chemical Engineering, University of Massachusetts, Lowell, Massachusetts, United States

One of the major steps in developing optimal cell culture conditions is understanding media supplementation. Identification of optimal cell culture media and supplements are critical to final production yield and product quality. In this project, we utilized an AMBR (Advanced Micro Bio-Reactor) high-throughput micro-bioreactor system to screen 5 commercially available CHO cell growth media and supplements using Box-Behnken DOE matrixes for each media to determine optimal conditions for our model IgG1 producing CHO cell line. Glutamine, essential amino acids and vitamins were the X- variables and the response Y variable was monoclonal antibody titer. Viable cell density, cell viability and nutrient concentration were recorded and monitored during the cell culture. The harvested IgG concentration was measured using a Protein A tip sensor based assay (Octet Red) system from each culture condition. In this study, with the exception of Media B and D, higher levels of glutamine in media improve titer and specific productivity. Levels of essential amino acids and vitamins vary in their effects depending on individual media. Center points demonstrated a well-controlled bioreactor process within the box-behnken design. We found that for our cell line, optimal supplementation strategy is different for each media type and can impact production and cellular growth profiles.

BIOT 260

Differential responses of CHO cells adapted to limitation of inorganic phosphate, glucose or glutamine

Vishwanathgouda Maralingannavar1,2, [email protected], Dharmeshkumar Parmar1,2, Tejal Pant1, Chetan Gadgil1,2, Venkateswarlu Panchagnula1,2, Mugdha Gadgil1,2. (1) Chemical Engineering and Process Development, CSIR-National Chemical Laboratory, Pune, Maharastra, India (2) CSIR—National Chemical Laboratory Campus, Academy of Scientific and Innovative Research, Pune, India

Adaptation has been widely used in microbial cells as a tool to either engineer cells for biotechnological applications or to understand cellular biochemistry and mechanisms of molecular evolution. Such studies are uncommon in mammalian cells but hold potential for characterising and engineering cellular traits. Here we have characterised phenotypic and metabolic responses of suspension CHO cells adapted to limiting inorganic phosphate, glucose or glutamine concentrations for >50 days and then maintained in nutrient replete conditions. These nutrients were chosen because of their predominant role in central carbon metabolism. Though metabolism of these nutrients are interlinked, long term limitation to these nutrients selected populations with differential phenotypes. Adaptation to phosphate limitation selected cells reminiscent of Warburg phenotype with higher growth and lower specific oxygen consumption. On the other hand, adaptation to glutamine limitation selected cells with an opposite phenotype. Cells adapted to glucose limitation did not show any phenotypic changes. Steady state U-13C glucose labelling experiments suggest that cells adapted to phosphate limitation have a higher pyruvate carboxylase flux. Consistent with this observation, supplementation with aspartate abolished the peak density difference whereas supplementation with serine did not abolish the difference. Experiments to investigate the metabolic changes which result in lower growth in cells adapted to glutamine limitation are underway.

Phenotypic differences in CHO cells adapted to limiting concentrations of inorganic phosphate (Pi), glucose (glc) or glutamine (gln). Peak cell density and specific oxygen uptake rate measured in nutrient replete medium.

BIOT 261

Improved DNA double-strand break repair may reduce production loss in Chinese hamster ovary cells

Xiaolin Zhang1,2, [email protected], Kelvin Lee1,2. (1) Chemical and Biomolecular Engineering, University of Delaware, Newark, Delaware, United States (2) Delaware Biotechnology Institute, University of Delaware, Newark, Delaware, United States

Recombinant protein productivity in Chinese hamster ovary (CHO) cells often decreases over time during long-term cultures. This production instability is a longstanding problem in biopharmaceutical manufacturing processes. The gradual loss of recombinant protein expression can be associated with reduced transgene copy number due to CHO cells’ inherent genome instability. We have previously suggested that the DNA double-strand break (DSB) repair system in CHO may be deficient, leading to both genome and production instabilities. Therefore, increasing DSB repair may provide a method to improve production stability.

In this study, we investigated the impact of improved DSB repair on protein production. Specifically, four production cell lines were established, each stably expressing a heterologous gene that was previously found to enhance DSB repair in CHO cells. A number of stable production clones were cultured over three months without methotrexate (MTX) selection. Growth rates and recombinant protein productivity were measured at initial, early, middle and late stages of the long-term culture. Growth and productivity changes and their correlation with the expression level of heterologous genes will be discussed.

BIOT 262 The interactive effect of trace elements and chelating agents on mammalian cell culture performance in chemically defined media

Yelena Ilin, [email protected], Matej Krajcovic, Shelby Hutchins, Alissa Borshchenko, Lia Tescione, Duane Inlow, Canghai Lu. Process Science, Global MSAT, Sanofi, Framingham, Massachusetts, United States

Trace elements serve as essential cofactors for various biological functions. Mammalian cells require supplementation of several trace elements in chemically defined media to support growth and production of recombinant proteins. The solubility and bioavailability of these trace elements are dependent on the presence of metal-binding compounds, such as phosphate and chelating agents. In this work, we examine the effect of these components on media stability and performance in a fed-batch mammalian cell culture process. Design of experiments was used to evaluate the interaction between a common chelator, phosphate, and metals in the media. Analysis of the results revealed that the ratio of chelator to trace element concentration impacts trace element consumption, cell culture growth, and product quality. It was found that some metals are required to be present at much higher concentrations than their utilization rates, and we hypothesize that the strong metal-chelator interaction affects bioavailability of trace metals in culture. These data can be used to increase media and feed stability and optimize production of therapeutic proteins.

BIOT 263

From 0.5 g/L to 10 g/L: A collective effort to improve upstream production for a fast-track program

Balrina M. Gupta1, [email protected], Linda Hoshan2, John Cacciatore3, Zhimei Du4, Hao Chen5. (1) Upstream Process Development and Engineering, Biologics Process Development and Clinical Manufacturing, Merck & Co., Inc., Kenilworth, New Jersey, United States (2) Upstream Process Development and Engineering, Biologics Process Development and Clinical Manufacturing, Merck &Co., Inc., Kenilworth, New Jersey, United States (3) Biologics Process Development and Commercialization, Global Vaccines and Biologics Commercialization, Merck & Co., Inc., Kenilworth, New Jersey, United States (4) Cell Line Development, Biologics Process Development and Clinical Manufacturing, Merck & Co., Inc., Kenilworth, New Jersey, United States

This poster will highlight the dramatic upstream improvements for an in-licensed monoclonal antibody program under significantly accelerated timelines. The initial process using chemically defined media had a titer around 0.5 g/L for GMP production, a significant drop compared to the performance at bench scale and insufficient to meet the high clinical demand. In four months, the process development team transferred the process to an internal development facility, re-optimized the process to double the titer, confirmed process performance at pilot scale, and successfully transferred an improved process to multiple GMP facilities for clinical resupply. In order to meet further clinical demands, a new cell line and process was developed, which after several rounds of optimization resulted in a fed batch process yielding 10 g/L titer using chemically defined media. This process was successfully scaled up to 500 L. All the accomplishments were achieved by seamless collaboration with different functions within the development teams, as well as the internal and external GMP manufacturing sites.

BIOT 264

Evaluation of perfused high density seed bioreactor step of a legacy biologics process

Danielle Harrison, [email protected], Eric Hodgman, Eric Garr. MS&T Upstream, Bristol-Myers Squibb, Devens, Massachusetts, United States

This work summarizes the implementation of perfusion technology at the seed bioreactor stage to reduce a 14-day fed-batch process down to 10 days, and proposes a large scale stainless facility model to implement perfusion technology in a legacy product. Any reduction in the time it takes a fed batch Chinese Hamster Ovary (CHO) culture process to go from the production step to harvest can improve a manufacturing facility’s cadence. Even with an improved cadence, manufacturing facilities can still experience timeline delays of hours or days in a process. Unlike CHO processes that reach final N-1 viable cell densities (VCD) < 10 million cells/mL, a couple of hours of variability in a high density N-1 timeline can impact the split ratio of the production bioreactor step.

An evaluation that demonstrates how flexible the final VCD of a high density N-1 CHO culture is necessary to understand the impact on the production step. We sought to evaluate the N-1 stage final VCD of a high density N-1 culture in an Fc-fusion CHO cell process with the application of Alternating Tangential Flow (ATF) technology. Using ATF, CHO culture was grown to greater than 10.0 x 106 cells/mL and transferred to the production stage at varying final N-1 VCDs. To further evaluate the impact of perfusion at the N-1 step metabolomic results from an N-1 final VCD varying from 10-40 million cells/mL are summarized alongside production bioreactor step attributes. This work demonstrates the robustness of ATF technology application at the N-1 step and capability to reduce the production bioreactor step duration.

BIOT 265

Considerations for implementation of a virus barrier filter for cell culture media

David Bohonak1, [email protected], Shawn Bates2. (1) Global Manufacturing Sciences and Technology, MilliporeSigma, Burlington, Massachusetts, United States (2) Technology Management, MilliporeSigma, Burlington, Massachusetts, United States Filtration is increasingly used to reduce risk of viral contamination upstream of the bioreactor during bioprocessing. Recently, several filters have been marketed which are specifically designed to retain parvoviruses and other small viruses from cell culture media. Implementation of these filters differs from virus filters used in downstream processing and from sterilizing-grade filters commonly used upstream. In this work, some of the key considerations for implementing virus barrier filters at clinical and commercial scales are examined and specific examples are given. The location of the filter in the process flow is highlighted, including how it may impact the operating mode and sizing of the virus barrier filter, the total number of filters used during media preparation, and the sterile boundary around the bioreactor. Scaling of the filters is also explored, including supporting data, filter sizing methods, and rationalization of an appropriate safety factor. Finally, topics such as filter integrity testing, flushing, and sterilization approaches are discussed. Careful consideration of these factors during filter implementation is essential to enable a consistent and efficient reduction to the risk of viral contamination.

BIOT 266

Liver toxicity comparisons of small molecule drugs in a high-throughput 3D In vitro culture system and in humanized liver mice

Dylan Bruckner2, [email protected], Jeannette M. Connerney3, Jonathan S. Dordick1. (1) Center for Biotechnology, Renselaer Polytechnic Inst, Troy, New York, United States (2) Rensselaer Polytechnic Institute, Troy, New York, United States (3) Taconic Biosciences, Hudson, New York, United States

Predicting drug-induced liver toxicity remains a critical, yet elusive, goal in drug safety studies and drug development. This difficulty is a result of the often weak relationship between in vitro and animal toxicity models, as well as between animal models and human toxicity. To address this gap in drug development, we have employed a high- throughput, three-dimensional cell culture platform containing human and murine primary hepatocytes, and human and murine transformed hepatoma cell lines, to screen a library of 27 small molecule drugs of various mechanisms of action and modes of toxicity. Correlations of in vitro toxicity (as represented by IC50 values) to in vivo mouse toxicity (LD50 values) suggest that this system is mildly representative of in vivo toxicity results. To enhance these in vitro-in vivo correlations, we exploited a humanized mouse model wherein the same human hepatocytes used for in vitro studies were engrafted into TK-NOG mice and in vivo liver toxicity evaluated on a subset of compounds. This humanized mouse model provides highly comprehensive in vitro-in vivo correlations as a result of using identical human hepatocytes in both the in vitro and in vivo systems. In turn, this approach has promise in advancing earlier-stage drug safety assessment with greater predictive value.

BIOT 267 Enhanced biopharm manufacturing facility utilization via the use of larger volume and higher cell density bags to initiate seed expansion

Gregory Laslo, [email protected], Dharmesh Bhanushali. AMT, GSK, King of Prussia, Pennsylvania, United States

Seed expansion in biopharm manufacturing requires numerous shake flask passages starting with vial thaw to inoculate wave bags or bioreactors. This process can take anywhere from 3 to 6 weeks of manufacturing facility time. The time in facility cannot be reduced because traditional working cell banks (or starting vials) are constrained by their volume (1-5mls) and lower typical (15-30 *106 cells/ml) frozen cell concentrations. As the number of commercial Biopharm assets continuous to grow and with limited global manufacturing facility capacity, there is a clear need to increase the throughput for both existing and greenfield manufacturing facilities. By intensifying the concentration and significantly increasing the volume of the frozen cell suspension, seed expansion can be accelerated, whereby, bioreactors or wave bags are directly inoculated from high cell density bags (considered as a process intermediate). This work aims to highlight the successes and challenges with generating and thawing high cell density cryobags including longer term cryobag storage stability studies. In summary, the biopharm manufacturing facility of the future could look very different from the one today with the deployment of high cell density cryobags supporting enhanced biopharm manufacturing facility utilization.

BIOT 268

Characterization and identification of protein components in glanded and glandless cottonseeds

Kandan Sethumadhavan, [email protected], Heping Cao. USDA-ARS, New Orleans, Louisiana, United States

Cottonseeds are classified as glanded or glandless seeds depending on the presence or absence of pigment glands which contain toxic gossypol. Glanded cottonseeds from the most widely grown cotton plant compose of approximately 10% linters, 40% hulls and 50% kernels which contain about 35% of oil and 40% of protein. After oil extraction, commercial cottonseed meal from glanded seeds contains up to 1% residual gossypol, which limits its use of meals primarily to feed ruminants. Gossypol can bind to proteins and makes it more difficult to recover concentrated protein fractions from the meal free of gossypol. Glandless seeds do not contain pigment glands and have only trace amount of gossypol which may be useful for potential utilization of the protein as a food ingredient or as a feed for non-ruminant animals. Developing improved extraction techniques for cottonseed proteins would provide additional markets for the cottonseed meal and improve overall profitability of the seed. The objective of this study was to compare and identify protein components in glanded and glandless cottonseeds. The seeds were treated with liquid nitrogen before being grounded into fine powder. The powders were suspended in aqueous buffer and blended again, resulted in homogenates with yellowish and brown colors from glanded and glandless seeds, respectively. After centrifugation, the top lipid layers were scooped and the middle layer contained proteins. The protein layers were precipitated with ammonium sulfate followed by centrifugation. Protein pellet were suspended in a buffer and desalted with centrifugal concentrator. SDS-PAGE showed that a similar pattern of polypeptide bands from both types of seeds. However, mass spectrometry identified numerous of proteins in both cottonseeds but glandless seeds contained more proteins than glanded seeds. This study suggests that the lack of gossypol gland in glandless seeds altered its protein compositions in the seeds.

BIOT 269

Increasing culture productivity through alternative nutrient supplementation

Krishanu Mathur, [email protected]. Alexion Pharmaceuticals, Cheshire, Connecticut, United States

Low productivity levels in legacy manufacturing processes can be increased through nutrient supplementation and readjustment of operating parameters. A legacy Alexion therapeutic protein process is characterized by low integral of viable cell concentration (IVCC) and product titer, when compared with other commercial CHO-based processes. Earlier efforts made to improve productivity of this upstream process have been met with little or no success. With the target to increase product titer, coupled with availability of new commercial feeds, we revisit this legacy upstream process to evaluate the impact of additional nutrient feeds and delayed temperature shift on culture productivity. Reduction of temperature during a CHO cell culture batch is shown to be beneficial for protein production and in maintaining culture viability. A temperature shift in this upstream process is critical to achieve target sialylation and maximize activity levels. However, by delaying temperature shift, greater integral of viable cell concentration (IVCC) can be achieved, resulting in higher product titer. In addition, supplementation of nutrient feeds to the basal production media, along with continual supplementation during the production run can increase productivity and protein activity. In this work, we screen three chemically defined, commercially available feed supplements with a delayed temperature shift condition. Through this investigation, we report a potential to increase IVCC by 2.7-fold and productivity by 4.5-fold, with acceptable levels of activity, sialyation and process impurities.

BIOT 270

High-throughput single cell screening for expression of recombinant proteins in Escherichia coli and Vibrio natriegens

Joel Eichmann1, Tobias Weidner2, [email protected], Doreen Gerlach3, Peter Czermak1,3. (1) Institute of Bioprocess Engineering and Pharmaceutical Technology (IBPT) , University of Applied Sciences Mittelhessen, Giessen, Germany (2) Life Science Engineering, University of Applied Sciences Mittelhessen, Giessen, Hesse, Germany (3) Bioresources, Fraunhofer Institute for Molecular Biology and Applied Ecology, Giessen, Germany

The gram-negative bacterium Vibrio natriegens is the fastest growing known organism with a doubling time of less than 10 min. A novel engineered strain carrying the T7 expression system now makes V. natriegens suitable and promising expression system for recombinant proteins, analog to E. coli. Although there are manifold combinations of secretion tags, fusion and affinity tags well characterized for E. coli, their functionalty in V. natriegens still has to be shown. As the yield and activity of the protein of interest is always a function of finding the best combination of expression level and fusion tags, these factors have to be adressed. However, most studies focus on only few of these factors at a time. To fill this gap, we established a high-throughput screening platform that allows to study multiple combination of these factors simultaneously. We used a multiplex Golden Gate cloning approach to generate a plasmid library containing 200 combinations of synthetic ribosome binding sites, secretion tags and fusion proteins. In a second cloning step, the library was assembled with a promoter and the protein of interest, tagged with a marker for fluorescent analysis. The resulting library of genetic constructs was expressed in E. coli BL21 and V. natriegens. To identify high producers, single cells were encapsulated by a microfluidic device and sorted by FACS into populations of higher and lower fluorescence intensities. Expression plasmids from each population were isolated, and sequenced using the PacBio SMRT method. Thus, high producing variations of expression vectors can be identified. In sum, this screening not only to identifies high producers out of hundreds of expression constructs, but also gives a wide characterization of recombinant protein expression in a novel host organism.

BIOT 271

Impact of Pluronic® F68 on hollow fiber filter-based perfusion culture performance

Rubin Jiang1, [email protected], Yuetian Chen2, Fengqiang Wang2, Sen Xu1, Hao Chen1. (1) Biologics Process Development & Clinical Manufacturing, Merck & Co., Inc., Kenilworth, New Jersey, United States (2) Biologics & Vaccines Analytical Development, Merck & Co., Inc., Kenilworth, New Jersey, United States

High cell density is an important factor in achieving high bioreactor productivity. To meet the oxygen demand with density at >100 × 106 cells/mL, a frit sparger is often used. In this presentation, the impact of Pluronic® F68 on a perfusion process using a frit sparger will be discussed. The perfusion process was developed using an alternating tangential flow device with a 0.2 µm Polyethersulfone (PES) hollow fiber filter. Pluronic® F68 at 2 g/L was sufficient in preventing cell damage at gas flow rate of ~0.20 vvm from a drilled hole sparger (0.5 mm) but inadequate at ~0.025 vvm from a frit sparger (20 µm). We demonstrated that increasing Pluronic® F68 concentration to 5 g/L prevented cell death at up to ~0.10 vvm from the frit sparger and was able to maintain high cell density at high viability in the range of 60 – 80 × 106 cells/mL. Such positive effect was demonstrated in both 3- and 200-L bioreactors. Supplementing additional Pluronic® F68 was also effective in restoring cell growth/viability from low viability cultures. We will also discuss the impact of increased Pluronic® F68 concentration on target antibody, host cell proteins, and Pluronic® F68 transmissions across the hollow fiber filter.

BIOT 272

High throughput screening of deubiquitinating enzyme activity in single intact cells using a cell permeable peptide-based reporter and a microfluidics droplet trapping array

Nora Safabakhsh1, [email protected], Manibarathi Vaithiyanathan2, Shayan Sombolestani2, Adam T. Melvin1. (1) Chemical Engineering, Louisiana State University, BATON ROUGE, Louisiana, United States (2) Louisiana State University , BATON ROUGE, Louisiana, United States

Advancements in molecularly targeted therapeutics have shown significant promise in treatment of multiple myeloma including the proteasome inhibitors Bortezomib and Carfilzomib. However, challenges in drug efficacy frequently occur due to tumor heterogeneity, drug resistance, and personalized responses. As such, tremendous attention has shifted towards personalized medicine and high throughput biochemical assays for analyzing single cell behavior. In the case of proteasome inhibition, recent studies have found that tandem therapies targeting both the proteasome and deubiquitinating enzymes (DUBs) have allowed cells to overcome drug resistance. DUBs are part of the ubiquitin-proteasome system (UPS) and are responsible for removing the polyubiquitin chain that targets proteins for proteasomal degradation. In this project, a microfluidic droplet trapping array was coupled with a cell permeable, peptide-based reporter to quantify DUB activity in intact single cells. The peptide reporter consists of an N-terminal β-hairpin ‘protectide’ that both confers stability onto the reporter but also acts as a cell penetrating peptide (CPP). The β-hairpin is conjugated to the N-terminus of a DUB recognizing sequence (LRGG) with a C-terminal fluorophore (AFC, 7-amino-trifluoromethyl coumarin). Thus, an increase in intracellular fluorescence corresponds to increased DUB activity. To achieve high-throughput screening of DUB activity and cellular resistance to DUB inhibitors, we utilized a microfluific droplet trapping array. First, we demonstrated that capabilities of the microfluidic device to examine cellular heterogeneity by characterizing CPP uptake in cancer cells. The fluorescent images were processed automatically using a Python algorithm followed by hierarchical cluster analysis using SAS enterprise guide to identify the distinct subpopulations present in each sample. As expected, the microfluidic platform was able to identify highly heterogeneous uptake of CPPs that depended upon the location and number of arginine residues. Once the microfluidic platform was optimized, it was then used to examine DUB activity in model multiple myeloma cell lines using the peptide-based reporter to quantify both heterogeneous enzyme activity and resistance to the DUB inhibitor PR-619. This microfluidic platform has the potential to help with the discovery and diagnosis personalized therapies targeting DUBs in small heterogeneous sample, like patient biopsies.

BIOT 273

Continuous in-line virus inactivation for next generation bioprocessing

Melissa Holstein, [email protected], Lori Mullin, Kristen A. Cotoni, Ronald Tuccelli, John Caulmare, Luc Messier, Christopher Gillespie, Patricia Greenhalgh, Michael W. Phillips. MilliporeSigma, Bedford, Massachusetts, United States

The shift in industry toward connected and continuous monoclonal antibody (mAb) processing has necessitated the development of novel approaches to improve or replace traditional unit operations that rely on hold tanks or operate in bind-elute mode. One such operation is virus inactivation with low pH, a critical virus reduction step in mAb downstream processing. Traditional low pH inactivation operations involve one or more large holding tanks in which product is maintained at a target low pH level for a specific period of time, typically 30-60 minutes. Translating this batch operation to a flow-through continuous process requires careful control of multiple factors to assure effective virus inactivation.

In this presentation, we describe the impact of buffer/mAb composition on the kinetics of virus inactivation. We address incubation chamber design which has implications for system size, processing times, and safety factor. Data demonstrating equivalency between batch and in-line systems is also presented. In-line technology that replaces batch operations and enables effective virus inactivation is expected to play an important role in the development of next generation mAb processing operations.

BIOT 274

De novo small molecule design using deep learning models trained on SMILES strings

Pavan Ghatty, [email protected]. Drug Product Formulations, Amgen Inc., Newbury Park, California, United States

A subset of Machine Learning called Deep Learning (DL) is increasingly being used in diverse fields including recommendation systems (movie/shopping), fraud detection, image classification (cancerous tissues), self-driving vehicles and other humanly intuitive and yet challenging-to-code tasks. Recent applications of DL have been made possible due to a unique combination of algorithmic advances combined with the clever use of video gaming hardware (GPUs) for large scale matrix manipulations. A pre- requisite for DL is the availability of large curated and properly labeled datasets (positive and negative examples).

We have identified a test case for using DL in an early stage small molecule project. Millions of molecules in the Amgen's internal database of compounds were tested for activity which resulted in ~5000 and 300 compounds after first and second round of screening both of which tested for specificity also. To address the program’s need to expand the library of active compounds to increase the chances of finding a chemically synthesizable active compound, we employed DL to generate novel compounds.

The DL model was first trained on millions of compounds and subsequent "fine-tuning" was done on the enriched set of 300 compounds. The resulting model was able to generate chemically valid compounds with specific features of the 300 compounds. This "fine-tuned" DL model was then used to generate novel compounds. In multiple cases of machine generated compounds, characteristic features (chemical groups) found in some of the 300 compounds were mixed with other features which resulted in novel compounds. The chief strength of the DL model lies in its ability to generate chemically valid compounds that borrow features from the parent set of 300 compounds. Thus high frequency of occurrence of functional groups, presumably tied to activity, receive higher weightage in the model.

In the final step, a list of machine generated compounds were filtered for chemical synthesizability and degree of similarity to the enriched set of compounds. This filtered set is being used to arrive at a drug candidate.

BIOT 275

Implementation of process analytical technologies in commercial manufacturing to realize a biologics drug substance enhanced control strategy

Daniel R. Hill1, [email protected], Canping Jiang2. (1) Global Process Analytics, Biogen, Inc., RTP, North Carolina, United States (2) Global Manufacturing Sciences, Biogen, Inc., Solothurn, Switzerland

In recent years, biologics process development teams across the industry have leveraged process analytical technologies (PAT) to improve process understanding and control. This work continues to speed up due, in part, to improvements in data science, process analyzers, multivariate modelling, and regulatory support. As a result, enhanced control strategies are being developed to drive process robustness and consistency, improved efficiency, and real-time quality. This talk will focus on the PAT applications that Biogen is implementing in our Next Generation Manufacturing facility in support of an enhance control strategies and how these technologies were developed/selected to meet reliability, operational, and quality requirements demanded by the commercial environment.

BIOT 276

Enhanced perfusion process development utilizing ambr at clone selection Brena Holman, [email protected], Ken Lee, Jeong Lee. Cell Culture and Fermentation Sciences, Medimmune, Gaithersburg, Maryland, United States

Methods for clone selection have historically revolved around fed batch processes. With an increasing demand for perfusion platforms, these screenings should tailor to the needs of the perfusion process while also remaining cost effective. Data showed that the productivity based ranking order of clones using a perfusion process was different compared to traditional fed-batch ranking. These results indicated that using a fed-batch top ranked clone would not yield the best option for selecting the most productive clone in a perfusion process. The ambr 15 bioreactor system along with repeated drain and fill based on a calculated cell specific medium exchange rate (CSMXR) to mimic cell specific perfusion rate (CSPR) provided the ability to mimic perfusion by keeping similar concentrations of nutrients per cell. This recreated a perfusion-like environment that yielded comparable results to running 3L perfusion bioreactors with alternating tangential flow (ATF) system. The ambr 15 perfusion mimic appropriately predicted capabilities of each of the clones in the 3L perfusion bioreactors, which gave ability to generate clone ranking specifically tailored to perfusion. This shows how different cell responses could be between the two culture methods. This perfusion mimic methodology can increase the number of clones that can be screened over a shorter time frame due to 24-48 available vessels in ambr 15. Furthermore, the implementation of this method for clone selection reduced media demand from about 30L of media per clone in 3L perfusion bioreactor system to 0.350L of media per clone.

BIOT 277

Solvent compatibility and extractables profile for use of Mobius® single-use assemblies in ADC processing

Dana Kinzlmaier, [email protected], Jessica Shea, Sara Bell, Shannon Ryan, Elizabeth M. Goodrich. MilliporeSigma, Burlington, Massachusetts, United States

Antibody Drug Conjugates (ADCs) are a class of biomolecules that has seen rapid growth as an oncology therapeutic. ADCs are comprised of three parts: monoclonal antibody (mAb), linker, and cytotoxic payload, where each component plays a role in therapeutic efficacy. The small molecule linker/payload is a highly potent, toxic agent that requires special consideration for safe handling and containment during conjugation and downstream processing. To address these concerns, many ADC manufacturers are adopting single use technologies, since they provide isolation and eliminate cleaning operations and carryover risk. However, the transition from traditional manufacturing in stainless steel or glass to single-use poses concern, as the conjugation step commonly utilizes aqueous organic solvents. A better understanding of the compatibility with and extractables from materials of construction is needed to enable widespread adoption of single-use technologies in ADC processing. This poster provides an overview of study design and associated results for evaluation of compatibility and extractables using Mobius® single-use assemblies in the presence of 20% DMSO or 20% DMAc. BIOT 278

Implementation of single-use powder delivery systems in continuous bio- manufacturing processes: A case study

Pranav S. Vengsarkar, [email protected], Jungmin Oh, Evon Bolessa, Nandu Deorkar. R&T, Avantor, Bridgewater, New Jersey, United States

Single-use technologies have become prevalent in the biopharmaceutical field due to their low cost, flexibility, reduced utility requirements and ease-of-use. Their primary application in the biopharmaceutical field is in the upstream manufacturing side through the use of disposable filters, bioreactors and membranes. While single use technologies have had impressive progress over the last decade there are still significant needs for these technologies to be applied to pre-upstream processes. Pre-upstream processes are typically manual/laborious in nature and the adoption of single-use systems in them can result into significant process efficiency gains and reduce cross-contamination of materials. Other general issues which can be addressed include GMP regulatory requirements, precise weight dispensing requirements, clumping concerns, quick purity checks and EHS concerns. Specific issues which single-use powder delivery systems can solve include reduced extractable leachables and no contact component identification (Raman ID). Typical applications where such single-use technologies can be implemented include buffer preparation which is critical for on demand continuous manufacturing of drug products. A case study for the application of single-use powder dispensing technologies capable of delivering exact quantities of clump free material for the delivery of buffer prep materials is highlighted as part of this paper.

BIOT 279

Robust platform cell culture processes utilizing the Selexis expression system for decreasing process development timelines and enhancing speed to clinic

Shahid Rameez1, [email protected], Yogender Kumar Gowtham1, Srivatsan Gopalakrishnan1, Ka Carl Zhang1, Sigma S. Mostafa1, Abhinav A. Shukla2. (1) Process Development, KBI Biopharma Inc., Research Triangle Park, North Carolina, United States (2) Process Development and Manufacturing, KBI Biopharma Inc., Research Triangle Park, North Carolina, United States

Performance of mammalian cell culture processes contribute significantly towards the overall success in production and economics of biologics such as recombinant monoclonal antibodies (mAbs) and next generation recombinant proteins. Rapid and effective clinical entry is still a rate limiting item for prospective biopharmaceuticals to enter clinical testing. Chinese hamster ovary (CHO)-based cell culture processes conventionally include cell line generation, media and feed optimization, and process design optimization to increase the productivity and establish acceptable product quality outputs for production of recombinant proteins. This process is complex, time consuming and resource intensive. Significant efforts are ongoing in this area to develop cell culture platforms that combine new state-of the-art technologies for genetic vector design, medium optimization and process design optimization to reduce the overall timeline. At KBI, we have made significant advances in developing cell culture processes that enable high productivity (> 5 g/L in a 14 day fed-batch culture) and meet precise product quality requirements. This has been achieved via implementation of high expression platforms and high-throughput process development workflows that combine to deliver high productivity processes very fast for clinical production. The platform approach includes use of optimally designed proprietary expression vectors (SUREtech vectors that bear Selexis Genetic Elements), high-throughput screening of clones from different rounds of single cell cloning and screening of process conditions using established scale down model miniaturized bioreactors. This is followed by DoE-based experiments to optimize process parameters for high-titer and precise product quality processes. The workflow allows for easy scale up to clinical and commercial scales. We present development of a platform cell culture process for a mAb which was eventually applied to six different mAbs for generation of clinical material at 200 - 2000L scales. The platform allows to produce biopharmaceuticals and support Investigational new drug (IND) application in approximately 9 months after gene synthesis. This approach has been successfully demonstrated at KBI and utilized for over 12 recombinant proteins including the 6 mAbs discussed in this presentation.

BIOT 280

Case study: Virus risk mitigation in cell culture media lab-scale to pilot-scale – virus retentive filtration and cell growth studies

Sherri Dolan1, [email protected], Brian Wong2, Roger Alsop1, Brian Kanoh1, Bjoern Hansmann1, Robert D. Kiss2. (1) Sartorius-Stedim, Bohemia, New York, United States (2) Genentech, South San Francisco, California, United States

Ensuring virus safety is of utmost importance in the Biopharmaceutical industry. Mammalian cell culture process present a unique challenge as these processes are inherently prone to microbial and viral contaminations. The cell culture bioreactor is a perfect environment for the proliferation of these contaminants and the introduction of a very low level contamination can quickly replicate into a major contamination. Past experience has shown that raw materials may be a high risk for introducing viral and bacterial contaminants. Bacterial contaminants can be easily removed by 0.1 or 0.2 μm sterilizing grade membranes, however small viruses (such as Vesivius, MVM) are not removed by these filters. In addition, testing raw materials may not be adequate since low levels of virus contamination may go undetected, hence a mitigation strategy to treat all raw materials which enter the bioreactor for virus removal/inactivation is becoming more popular in the industry. Adventitious contamination events have occurred in the past and may have severe consequences, such as GMP facility contamination, facility shutdown leading to drug shortages, financial losses and lost market share.

Current risk mitigation technologies such as HTST, UV-C, and gamma irradiation are useful, however not always easy to implement and are not cost effective. Size exclusion based filtration is the preferred technology for viral clearance, as it is robust and non- invasive. Current downstream virus retentive membranes do not possess the flux rate or economics when it comes to filtration of cell culture media. Therefore a novel membrane has been developed by Sartorius Stedim to mitigate contamination risks in the bioreactor from chemically defined media and raw materials. The purpose of this study was to determine the capacity of Virosart® Media for chemically defined media as well as to investigate the effect of filtration on subsequent cell growth performance.

BIOT 281

High seed density cell culture processes to reduce manufacturing cycle times

Yogender Kumar Gowtham, [email protected], Shahid Rameez, Srivatsan Gopalakrishnan, Ka Carl Zhang, Sigma S. Mostafa, Abhinav A. Shukla. Process Development, KBI Biopharma, Durham, North Carolina, United States

Mammalian cell culture processes used for protein therapeutic manufacturing are initiated with vial thaw of cell banks, followed by seed train expansion to generate inoculum culminating into the production bioreactor step to generate the required product. Such a cell culture manufacturing process could take anywhere between 4-6 weeks. However, the last decade has witnessed advancement in cell lines and cell culture media enabling improvements in cell growth and productivity. By taking these improvements into account, conventional cell culture processes can be strategically improved by using higher seed densities during seed train expansion and production stage to shorten the overall manufacturing timelines. Here we present case studies that demonstrate the development of high seed-density (HSD) cell culture processes for two Chinese hamster ovary (CHO) cell lines expressing monoclonal antibodies (mAbs). The processes were seeded at high-densities, 2x - 10x higher than typical seed density, for seed train expansion and production phases. The transition to HSD process from a low seed density process offered several advantages. The HSD production processes for two CHO cell lines reached comparable titer yield by 11-12 days, while their respective low seed density process reached the same titer in 14 days. The strategy significantly shortened the overall culture duration, offering potential for increased utilization of the manufacturing/production facility. In one of KBI’s production suites, housing 2x2000L single-use bioreactors, facility utilization increases by 14% and 27% respectively for two or three days decrease in cell culture duration. In addition, HSD processes reduce the residence time of the product in bioreactor, which can modulate key product quality attributes that are impacted by culture duration such as glycosylation. Overall, this approach offers the opportunity to debottleneck the rate limiting step in a typical biopharmaceutical manufacturing facility and to increase manufacturing capacity significantly without requiring an alteration in the operational format.

BIOT 282 Versatility of the continuous chromatography platform

Chia-Yun Sun, [email protected], Rachel Quesenberry. Pall Life Sciences, Arlington, Massachusetts, United States

Interest in continuous processing is spreading within the pharmaceutical industry. The feasibility of continuous processing has now been shown at both the process development and the production scale by early adopters. In this study we focus on questions surrounding the ease of process development for continuous monoclonal antibody (mAb) purification and platform versatility.

We address these questions by challenging the continuous chromatography platform with the purification of four mAbs. The platform comprises three chromatography steps, Protein A capture, anion exchange, and mixed mode cation exchange. Each chromatography step was optimized through a combination of small scale DoE and mathematical modeling. Critical quality attributes such as host cell proteins (HCP) and aggregates are routinely brought to levels below 20 PPM and 1%, respectively. Therefore, the robust purification schemes resulted in products with high yield and purity, including over 4 log reduction of HCP and more than 50% reduction in aggregate protein. Moreover, adapting the continuous chromatography platform to process multiple mAbs proved to be time-efficient, effective and resulted in streamlined and highly productive processes.

Taken together, these experiments allow us to identify a path toward rapid process development for the continuous chromatography platform, which may drive the use of the continuous processing for clinical manufacturing of mAbs.

BIOT 283

Development of a fully automated high-throughput (HT) method to screen cleaning conditions during a resin lifetime study

Kimberly Larson, [email protected], Alvaro Rodriguez, Minni Aswath, Alpana Naresh, Min Zhu. Protein Science, Boehringer Ingelheim Fremont Inc, Fremont, California, United States

Protein A (ProA) chromatography is an affinity-based method that offers high selectivity and robust platform capability. In recent years, the use of alkali-stabilized ProA resin has enabled direct capture of mAb from dirtier feeds as a direct result of stronger cleaning agents that can be effectively used to remove contaminants from a fouled ProA resin. However, these caustic and alkaline conditions need to be optimized in order to minimize loss of ProA ligand. Furthermore, there are no universal conditions that may be implemented, as cell culture and harvest conditions differ at each company. Traditional approaches to resin lifetime studies put a significant burden on both time and material resources. High-throughput and miniaturization of these studies can therefore be a powerful tool to explore wider ranges of conditions. To this end, a fully automated screening method was developed on the TECAN liquid handling system to study resin lifetime. This abstract outlines the process and results obtained for a mAb using MabSelect SureÒ resin and 96-well filter plates. The biggest challenge to using filter plates is gaining consistent results for a sufficient number of cycles. In this study 16 different conditions on one plate were compared to determine cleaning conditions based on low product carry-over and high yields, with a method that delivered reliable results up to 25 cycles. To further screen the cleaning conditions, both fresh and stressed (fouled) resin were utilized to determine which condition best balanced high cleaning- ability and minimum ligand loss, thus maximizing the total number of allowable cleaning cycles. The established HT method can therefore be used to quickly optimize cleaning conditions for any mAb, and the selected conditions can then be verified by running scale-down model columns.

BIOT 284

Advances in affinity purification technologies for monoclonal antibody capture

Krunal Mehta2, [email protected], Anjanesh Venkatesh1, Pranali Shah1, Glen Bolton1, Ryan G. Soderquist2. (1) Amgen, Cambridge, Massachusetts, United States (2) Purification Process Development, Amgen Inc., Longmont, Colorado, United States

Protein A chromatography is very effective in separating impurities from monoclonal antibodies (mAbs) and continues to be used as the capture step in the majority of mAb purification process. The performance of Protein A chromatography, as indicated by capacity, operational flow rate, and productivity, has improved over the past few decades, and has played a significant role in the growth of therapeutic mAb industry. However, the high cost of Protein A resins has required better utilization of its capacity or the use of cheaper and comparable alternatives to improve the downstream process economics. While the improvements in capacities have been limited, achieving higher operational flow rates has led to dramatic improvements in the productivity of the Protein A capture step. Protein A in membrane formats, rather than packed beds, is an emerging technology that enables significantly higher productivity through substantially reduced cycle times. Rapid cycling time enables this technology to be applied in a single column multi-cycle mode, whereby the lifetime of the unit is exhausted over a single batch process, thereby; significantly improving the resin utilization and reducing the resin requirements compared to a conventional single column process, without compromising the throughput of the capture step. The shorter residence time and smaller formats enable acceleration of process development, optimization and characterization activities and reduce load material requirements, and thereby serve as an instrumental tool to accommodate ever-shrinking CMC (Chemistry, Manufacturing and Controls) timelines. In a short time frame, we developed and optimized the Protein A membrane capture step for therapeutic proteins, and demonstrated robust performance. The Protein A membranes delivered consistent product quality, impurity profiles, and chromatograms over a number of cycles, and performed consistently across different scales (scale varied by at least 10-fold). The disposable format of this technology and the ability to perform multiple cycles in a very short time makes Protein A membranes a highly economical, and offers a productive alternative to the conventional packed resin format.

BIOT 285

Displacement chromatography for mAb charge separation

Tomas Bjorkman1, [email protected], Bengt Westerlund1, Anna Graanberg1, Eva Heldin2. (1) GE Healthcare , Uppsala, Sweden (2) GE Helthcare Life Sciences, Uppsala, Sweden

Monoclonal antibodies (mAbs) are large proteins susceptible to several post translational modifications (PTMs) that might cause charge heterogeneity. In the production of biosimilars these modifications play an important role and similarity with the originator molecule must be shown. To accomplish charge distribution similarity the cell culture must be carefully optimized, but still, chromatography polishing is often required. Separation of the different entities using different modes of chromatography with cation and multimodal resins will be shown. Displacement chromatography, a method invented by Tiselius 1943 and further developed by Horváth's group 1981, will also be discussed. During loading a competition occurs between various entities in the sample because of different affinity to the sorbent. Hence, overloading the resin can replace and overcome limitation with gradient separations. Here load and pH was varied on a high capacity cation exchanger and the clearance of acid charge variants and recovery was studied.

BIOT 286

Assessment of production-scale chromatography column performance using direct transition analysis (DTA)

Yuanyuan Cui, [email protected], Zheng Huang, John J. Prior. Sanofi, Westborough, Massachusetts, United States

We present a robust and easy-to-implement chromatography column performance assessment method, called Direct Transition Analysis (DTA). Since conventional Transition Analysis based on Moment Analysis is often challenging to validate and implement in Good Manufacturing Practice (GMP) operations, we developed and implemented a “paper-and-pencil” method for production-scale column performance qualification and monitoring. DTA results were found to be in good agreement of Moment Analysis. We will demonstrate the simplicity of DTA calculation and its successfully implementation in GMP environment through three case studies covering column re-qualification, resin lifetime process validation, and column routine performance monitoring.

BIOT 287 Hydrophobic salt effects in multimodal and ion exchange chromatography

Alejandro Becerra-Arteaga, [email protected], Juan Castano. MilliporeSigma, Billerica, Massachusetts, United States

Multimodal chromatography provides an alternative to traditional ion exchange or hydrophobic interaction adsorbents. These resins are particularly useful for the capture of therapeutic biomolecules under high ionic strength conditions and also provide unique selectivities. However, in many instances the elution and/or cleaning steps require significant optimization to obtain acceptable yields and to maintain the resin performance over time. In this work we evaluated several hydrophobic salts for their potential to act as elution or cleaning agents using a model system. The impact in elution behavior using hydrophobic counterions was also compared to traditional sodium chloride gradients in ion exchange chromatography. The output of this work may provide alternative strategies for elution and cleaning of these chromatographic media.

BIOT 288

High-resolution separation of monoclonal antibody charge variants using a cuboid packed-bed device

Guoqiang Chen2, [email protected], Alisha Gerrior2, Raja Ghosh1. (1) McMaster Chemical Engr Dept, Hamilton, Ontario, Canada (2) Chemical Engineering, McMaster University, Hamilton, Ontario, Canada

Monoclonal antibody molecules frequently undergo chemical modification such as deamination and oxidation, resulting in chemical variants. When such modifications result in changes in isoelectric point, these variants are also referred to as charge variants. Charge variants frequently show altered efficacy, pharmacokinetic properties, and physical properties. However, the difference in charge being subtle, separation based on this property is quite challenging. Charge variants are generally separated at a preparative scale using ion-exchange column chromatography. In large-scale preparative columns, the diameter is frequently comparable, or even larger than the bed height. Dispersion in peripherals, non-uniform packing and severe flow maldistribution in the headers, and consequent poor separation efficiency is a serious problem with such columns. Non-uniformity of flow within preparative columns result in peak broadening and poor resolution, which decreases recovery, increases operating time and buffer usage, and necessitates subsequent concentration steps. Researchers have tried to address the problem of column maldistribution by modification of the headers, better column packing, and by developing parallel flow, curtain flow, and radial flow columns. In our group, we have proposed a radically different approach for reducing flow maldistribution through the use of box-shaped or cuboid packed-bed devices. In this study, a cuboid packed-bed device packed with Capto S ImpAct resin was used to separate acidic and basic variants of monoclonal antibody hIgG1-CD4. The cuboid packed-bed device showed significantly superior separation attributes than its equivalent column. Our process was optimized to obtain high-resolution separation of monoclonal antibody charge variants.

BIOT 289

Development of a chromatography-based hydrophobicity assay for viral particles

Alexander H. Schwartz1, [email protected], Alison A. Walsh2, Cara Romanowski1, Michael S. Burnham2. (1) Viral Clearance, WuXi AppTec Inc., Philadelphia, Pennsylvania, United States (2) WuXi AppTec, Philadelphia, Pennsylvania, United States

For the field of viral safety in biologically derived medicines, viral clearance is a requirement in a risk mitigation strategy, which entails the validation of virus removal in a given downstream process. While there are several unit operations that have historically proven to provide effective virus removal, one process step that is seeing continued evaluation for viral clearance capability is hydrophobic interaction column (HIC) chromatography. Though it has been assumed that a virus’ specific hydrophobic nature plays an important role in the mechanism of viral clearance, HIC chromatography and it’s mechanisms remain poorly understood as the relative hydrophobicity of virus species have been an unknown. In this collaboration, we have developed a chromatographic method to quantify the relative hydrophobicity of several different virus species. Using HIC in a high-salt matrix has allowed us to ascertain the relative hydrophobicity values of three bacteriophage models along with four mammalian viruses as benchmarked to protein standards previously characterized for hydrophobicity, allowing to one day create predictive tool in viral clearance capability.

BIOT 290

Use of high-throughput process development (HTPD) methods to develop a toxicology process: Cation and anion-exchange chromatography development case study

Amber Bill, [email protected]. Purification Development, Genentech, South San Francisco, California, United States

In this study, high-throughput process development (HTPD) methods were used to quickly modify standard cation-exchange (CEX) and anion-exchange (AEX) chromatography conditions for the purification of a monoclonal antibody in the Toxicology phase of development. Initial 96-well plate screening results showed that the antibody eluted late in a standard CEX chromatography gradient elution, resulting in a large pool volume at a high conductivity, which would cause facility fit challenges. CEX and AEX plate screening studies (in addition to PreDictor RoboColumn studies for the AEX step) were used to develop a step elution for the CEX chromatography step and an alternative salt tolerant AEX resin, which reduced in-process pool volumes. The final chromatography processing conditions for both resins were confirmed on small-scale columns and subsequently, the process was implemented at Toxicology production scale (400L). HTPD methods enabled efficient CEX and AEX chromatography development and the purification process can be implemented at manufacturing scale with minimal equipment modifications.

BIOT 291

Effect of sporicidal sanitization with oxidizing agent on protein A resins: Lifetime and purification performance

Anna Graanberg, [email protected], Elin Monie, Tomas Bjorkman, Magnus Wetterhall, Sravani Musunuri, Katarina Oberg. GE Healthcare , Uppsala, Sweden

Bacterial contamination is a pain point in bioproduction which in some cases result in costly investigations to find the root cause and the potential disposal of expensive and underutilized protein A resin. A contamination incident, particularly with spore formers, can in worst case result in long term facility shut down thus leading to loss of revenue, potential impact to public reputation, and most importantly, preventing patients from accessing essential drug therapies. It has been shown that low concentrations of oxidizing agent peracetic acid (PAA) result in more than 6 Log reduction of B. subtilis spores within 15 to 30 minutes contact time. Similar spore Log reduction by conventional sanitization agent, 1.0 M NaOH, can only be achieved after significantly longer contact times. Here, the effect of sporicidal sanitization with peracetic acid (PAA) on the functional and useful lifetime of protein A resins was investigated running more than 100 cycles. Impact on the resin functionality by PAA was verified by characterization of ligand and coupling chemistry using surface plasmon resonance (SPR) and liquid chromatography- mass spectrometry (LC-MS) analysis. Other important aspects such as impact on target mAb quality attributes, impurity clearance, process performance, and clearance of PAA from the column were also evaluated. Different options for sanitization, pro-actively as part of the regular process or, re-actively in response to a bioburden incident will be discussed. Sanitization of a new protein A resin using ≥1.0 M NaOH, will also be discussed in the light of ligand compatibility and sporicidal effect of high NaOH concentrations.

BIOT 292

Increased lifetime of RPC resins in insulin production by clean-up using WorkBeads 40S

Anna Heijbel, [email protected]. Bio-Works, Uppsala, Sweden

Purification of recombinant insulin requires very high purity, often achieved by high- resolution reversed phase chromatography (RPC) based on silica. Impurities from the feed often cause fouling of the silica resins that is difficult to remove since the option of cleaning-in-place using sodium hydroxide is limited. These issues result in shortened lifetime of the RPC columns. We present here that an introduction of an ion exchange chromatography step before RPC removes the bulk of impurities from the feed which significantly will increase the lifetime of the RPC column. By introducing a capture step, using WorkBeads™ 40S (a cation exchanger), a standard purification process involving a combination of two RPC steps with two different buffer systems (to obtain complementary selectivity) was improved. WorkBeads 40S is an agarose based resin with sulfonate ligands which is stable in high concentration of sodium hydroxide thus allowing efficient cleaning-in-place. A feed containing 72.5% pure human insulin precursor (Met-Lys-Human insulin) was applied to the WorkBeads 40S column, followed by the two RPC steps. Each step was optimized to give a yield of 90% while maximizing the purity after each step. The amount of impurities in the applied feed on the first RPC column could be reduced from 27.5% to 12%. This improvement significantly reduced the fouling of the first RPC column and prolongs its lifetime.The final purity for the human insulin precursor was > 99%. A comparison was also done using Capto™ SP ImpRes cation exchange resin in the capture step which resulted in a less pure target protein compared to using WorkBeads 40S.

BIOT 293

Development and evaluation of a novel high capacity protein A chromatography resin with significantly increased alkaline stability

Annika Forss, [email protected], Tomas Bjorkman, Jelena Vasic, Mats Ander. GE Healthcare, Uppsala, Sweden

Monoclonal antibody purification platforms are very well established, and almost all approved mAb purification platforms are using a Protein A capture step. There are however remaining challenges with this technology. Increasing upstream titers can make the Protein A chromatography process step rate-limiting in downstream purification. Protein A chromatography columns are also more prone to bioburden contamination due to heavy impurity load and weak tolerance towards the most efficient cleaning-in-place solutions.

This presentation will cover the development and evaluation of a novel protein A chromatography resin with significantly increased alkaline stability enabled by protein engineering and a high throughput screening technology using surface plasmon resonance. Design considerations for affinity chromatography resins, and how to balance different properties will be discussed. Data on binding capacity, caustic stability and resin performance, in terms of yield and purity, will be also be presented.

BIOT 294

Mixed-mode resin screening for isolation of recombinant thioredoxin from E. coli Ayswarya Ravi3, [email protected], Shengchun Guo4, Zivko L. Nikolov2, Peter Heifetz1. (1) OrPro Therapeutics, Inc., San Diego, California, United States (2) Texas A M Univ, College Sta, Texas, United States (3) Biological and Agricultural Engineering, Texas A and M University, College Station, Texas, United States (4) Texas A&M Univerisity, College Station, Texas, United States

Thioredoxin (Trx) is a 12 kDa, highly conserved redox protein with two thiol groups. The biological functions are diverse and range from DNA synthesis, oxidative-stress relief for regulating apoptosis to immunomodulation. The array of biological functions makes Trx a potential therapeutic protein for treating lung injuries, cardiovascular diseases, cerebral ischemia and cystic fibrosis. This study focuses on developing scalable purification options to obtain therapeutic- grade purity of Trx from E. coli lysates. Initial purity of Trx in the clarified cell lysates was about 20% of total soluble host protein. Further, 2 to 3 fold increase of Trx purity in the lysate could be achieved by isoelectric or salt pretreatment. We found that cell lysis at pH 5 instead of pH 7 was a more convenient way to increase Trx purity two fold without sacrificing Trx recovery. Despite relatively high expression titer and solubility, purification development using typical orthogonal steps for acidic proteins such as anion exchange and HIC did not yield desired purity and recovery of Trx. Low protein hydrophathicity index (- 0.096) and minimal net charge (-0.6 to -6.8) of Trx in pH range from 5 to 8 were some of the reasons for weak performance of ion exchange and HIC. For this reason, salt-tolerant and multimodal resin chemistries, were screened on a 96- well filter plate for optimal binding conditions with minimal co-adsorption or interference from E. coli HCPs. Chromatographic resins CaptoAdhere, HEA HyperCel, PPA HyerCel, MEP HyperCel were screened for selectivity and salt tolerance with clarified cell lysates. Adsorption of Trx on HEA HyperCel demonstrated an increase in purity to 81% and also interacted with Trx in high salt concentrations up to 250 mM NaCl. The process scale-up with elution optimization and binding capacity comparisons will be discussed. Also, the chromatography resin options for intermediate purification step will be presented. This study successfully demonstrates the selectivity and potential use of mixed-mode resin chemistry as an alternate for proteins with weak interactions to single-mode resins.

BIOT 295

Acidic IgM purification using anion exchange and mixed mode chromatography

Jamie C. Greenwood II, Carsten Voss, [email protected], William Rushton, [email protected], Payal Khandelwal. Process Chromatography Applications, Bio-Rad Laboratories, Hercules, California, United States

IgM and IgG antibodies are one of the fastest growing diagnostic and therapeutic candidates. Unlike IgGs, IgM antibodies cannot be purified by affinity chromatography due to their physiochemical properties. We show the purification of acidic IgMs with a new strong anion exchanger (AEX) for the capture and mixed-mode chromatography media for the subsequent polishing. The anion exchange material is optimized to purify larger proteins and shows high binding capacities for IgM molecules (~1000 kDa). The capture step removes process related impurities while the following polishing depletes these impurities further. The described purification strategy is simple and scalable. Initials studies resulted in a product purity of >95% IgM (SDS PAGE). This purity makes the IgM suitable for diagnostic applications.

BIOT 296

Distribution of a unique process derived charged monoclonal antibody species across a cation exchange column

Cate O'Brien, [email protected], Anthony Grecco, Lunedt Gonzalez, Ina Alickolli, Hunter Malanson, Jeremy Pike, Siguang Sui. Alexion Pharmaceuticals Inc, New Haven, Connecticut, United States

Strategies for downstream purification of monoclonal antibodies typically include one or more ion exchange steps, such as cation exchange or anion exchange. Optimization of the cation exchange step may include the utilization of a salt gradient to determine optimal elution conditions in order to remove sufficient process or product related impurities. During the re-development of a monoclonal antibody purification process, a unique process derived charge variant was enriched in fractionated CEX eluate, which provided an effective method to investigate the interplay between this charge variant and disulfide isoform. In this study, we describe the distribution of this unique monoclonal antibody charge variant, and demonstrated an integrated approach to investigate disulfide isoform and charge variant interaction for IgG2 monoclonal antibody process development.

BIOT 297

Implementation of multi-column chromatography for affinity purificaiton

Chase Snyder, [email protected]. Downstream Process Develpment, CMC Biologics, Seattle , Washington, United States

Facilitating cost savings and allowing for manufacturing flexibility are of paramount importance for any organization purifying biologics for pre-clinical or clinical supply. By far the most expensive aspect of manufacturing biologics is affinity capture purification. Often times resin for this process step can be up to 40% of the total cost of goods to manufacturing a single batch of product. Continuous chromatography with multiple columns for affinity purification can mitigate affinity resin costs by futher exploiting the binding capacity of resin regularly underutilized in conventional batch mode chromatography. This technology can be tuned to benefit both commercial and clinical phase processing runs by maximizing either specific productivity or resin lifetime. Thus, this technology enables cost savings for every stage of drug development. We have used multiple column chromatography with several molecules, demonstrating proof of concept for decreasing resin volume demand. Recently, this technology was used to perform a pilot scale run to generate toxicity study material for a client, demonstrating a significant cost savings compared to batch mode chromatography. Along with cost savings, multiple column chromatography can provide flexibility by allowing for a smaller footprint in existing facilities through the use of several small columns, as well as expanding capacity capabilities for larger mass inputs. High titer cell culture processes are becoming more commonplace, which typically is requires procurement of larger column hardware to meet larger mass inputs. Multi-column chromatography can be used as an alternative strategy for high titer processes by using several small columns in tandem. For a multi-product facility processing a wide range of titers, this technology provides much needed flexibility within the same manufacturing footprint.

BIOT 298

Capacity and beyond: Evaluation of a next generation protein A resin

Chris Furcht, [email protected], Felicia Sadikin, Brad Stanley, Engin Ayturk. Biogen, Cambridge, Massachusetts, United States

ProteinA chromatography, an affinity-based purification method that selectively binds antibodies, is generally one of the most expensive steps in a downstream process due to the high cost of ProteinA resin. Decreasing the effective cost of the ProteinA step can be accomplished by increasing the amount of antibody that is processed per column cycle, either by increasing resin capacity or enhancing resin utilization through processing strategies such as continuous chromatography. Additionally, increasing the number of times a column is cycled can drive down the ProteinA cost per batch.

Here, we discuss the evaluation of a next generation ProteinA resin (MabSelect™ PrismA) with increased capacity and base-stability relative to other commercially available resins. Dynamic binding capacity was determined for several monoclonal antibodies on both a current generation ProteinA resin (MabSelect SuRe™ LX) and MabSelect™ PrismA. Across a range of residence times, dynamic binding capacity for MabSelect™ PrismA was noticeably higher for all the antibodies tested. When utilizing MabSelect™ PrismA at an increased column loading, yield and eluate product quality were still found to be comparable to results from MabSelect SuRe™ LX. However, the impact of increased column loading on elution peak width and eluate pool pH were further investigated. Cycling studies with representative cell culture material were also planned to assess whether increased resin lifetime could be achieved with MabSelect™ PrismA by utilizing a more stringent cleaning approach. The results from the aforementioned studies provide evidence for a benefit of implementing MabSelect™ PrismA, which are summarized through cost and productivity analyses for different processing scenarios.

BIOT 299 Experimental and theoretical studies of protein adsorption onto ion exchange and mixed mode chromatographic column packings

Douglas Frey1, [email protected], Mayyada M. El-Sayed3, Payam Rezaei2. (1) UMBC, Baltimore, Maryland, United States (2) Chemical,Biochemical and Environmental engineering, UMBC, Baltimore, Maryland, United States (3) American University in Cairo, Cairo, Egypt

In this work, a phenomenological model for protein adsorption incorporating charge regulation is developed that is based on principles of classical thermodynamics. It is shown that this adsorption model includes as a subcase a charge regulation protein adsorption model developed previously that is based on statistical mechanical perturbation theory. However, the phenomenological model developed here extends these previous adsorption models into several new areas, including to the case of mixed-mode chromatography. To test the model and assess its suitability for correlating adsorption data, ion-exchange (IEX) and mixed-mode (MM) chromatography experiments were used to study the adsorption mechanism for the interaction between proteins and IEX/MM packings. Column experiments using both anion exchange and mixed-mode packings were conducted under different operating conditions of pH and ionic strength. Linear adsorption constants for the protein-column packing interactions were calculated using the measured chromatographic retention times under the different employed conditions. Experimental data were fitted to models to enable prediction of the amount of protein adsorbed and hence elucidate the underlying mechanism of adsorption. To gain further insights into the adsorption mechanism, Monte Carlo Docking simulations were also performed. This work is important in developing computer-aided design methodologies for optimized protein purification.

BIOT 300

Evaluation of five commercial immunoassays for quantification of leaked ligand from a new protein A resin

Elin Monie, [email protected], Tomas Bjorkman, Annika Forss. GE Healthcare, Uppsala, Sweden

There are several advantages with protein A affinity purification of monoclonal antibodies e.g. high robustness, selectivity and purity. A new protein A chromatography resin, MabSelect™ PrismA, has been developed which consists of a new high-flow agarose base matrix in combination with a new alkali-stabilized protein A ligand. Some of the key features include increased binding capacity and the possibility to use up to 1 M sodium hydroxide for cleaning-in-place and sanitization.

One side effect of protein A chromatography is that variable amount of ligand is eluted together with the product and there is a regulatory demand to control and quantify the protein A amounts through-out the purification process. As the MabSelect PrismA ligand is new there are no dedicated commercially available assay for the leakage, hence, five different commercial Protein A immunoassays are evaluated for detection and quantification of the new protein A ligand. In the commercial kits different variants of protein A are included. A comparison between the different assays in respect to quantification using existing standard and the new ligand will be made. Also, a comparison of ligand standard curve quality and dynamic range as well as experimental procedures will be made.

BIOT 301

Packing CHT™ resin in a one meter pack-in-place Pall Resolute® manual column: Understanding the impact of shear on fine generation and column pressure performance

Erin Baragar, [email protected]. Merck and Co., Inc., Philadelphia, Pennsylvania, United States

Transfer of a bioprocess into a commercial facility requires packing of Biorad’s CHT™ ceramic hydroxyapetite resin into a one meter Pall Resolute® chromatography column via a pack-in-place method. This presents a unique challenge: a high flow rate (40 L/min) through the column packing nozzle holes is required to achieve the target packing pressure. The impact of these process conditions on resin fine generation and column performance (pressure) for this shear-sensitive resin is not well understood. To understand this risk, a collaborative study among Merck, Biorad, and Pall was completed. A 30% slurry of Type I 40um CHT™ resin prepared in 0.5M NaOH was immediately passed through a Resolute® pack-in-place nozzle five times at 40 L/min. Samples of the prepared slurry and slurry after one, three, and five passes through the Resolute® nozzle were analyzed. Fines and shifts in particle size distribution were evaluated across multiple methods: Micro-Flow Imaging, environmental scanning electron microscopy, and laser diffraction. Optical microscope imaging of the slurry samples was performed for qualitative analysis. Lab scale 2.5cm diameter columns were packed with resin samples and six column volumes of 0.5M NaOH were passed through each column to evaluate the impact of potential fine generation on column pressure profiles. Across all five methods utilized, there was no clear trend in fine generation with respect to nozzle passes. Results from this study indicate that flow rates up to 40 L/min of a 30% CHT™ slurry through a Pall Resolute® column nozzle does not have a discernable impact on CHT™ fine generation and column pressure performance.

BIOT 302

Particle size and ligand density effects on anion exchange chromatography behaviour for therapeutic protein manufacture

Greta Jasulaityte2, [email protected], Patrick Gilbert1, Hans J. Johansson1, Daniel G. Bracewell2. (1) Research and Development, Purolite, Llantrisant, Wales, United Kingdom (2) Biochemical Engineering, UCL, London, United Kingdom Ion exchange resins for bioprocessing are designed to balance a series of properties, which will define their performance. These include particle size, which affects pressure drop and plate number, intra-particle porosity that directly impacts surface area/capacity and the accessibility of binding sites, and ligand density, which is critical to selectivity. Here, we investigated particle size and ligand density aspects of agarose based anion exchange resins to understand interactions between dynamic binding capacity, protein uptake rate and robustness to fouling. An array of 9 bead types of 45, 65 and 90 µm in diameter and 0.09, 0.17 and 0.24 mmol/mL ionic capacity were generated and studied using bovine serum albumin and immunoglobulin G in chromatographic separations and confocal laser scanning microscopy. Resin with low ligand density (0.09 mmol/mL) provided high binding capacity and a fast protein uptake rate, improved mass transfer and resin lifetime. In contrast, the high ligand density resin (0.24 mmol/mL) suffered from impaired mass transfer effects due to potential steric hindrance and/or electrostatic repulsion as investigated by confocal microscopy. Moreover, this resin lost 20-30 % of its binding capacity after 5 cycles. A novel technique using a Proteostat dye to investigate the extent of fouling was developed. Fresh and used resin samples were stained with the fluorescent dye and imaged under confocal microscopy to determine the extent of protein aggregation. The results confirmed increased protein aggregation levels for high ionic capacity resin compared to the low ionic capacity resin. The study demonstrates that selecting the appropriate combination of ligand density and particle size is crucial for optimal resin design and performance.

BIOT 303

Continuous plasmid DNA purification by periodic counter-current chromatography

Hans Blom, [email protected], Helena Skoglar, Anna Akerblom, Mikael Berg, Linda Persson, Martin Hall, Linda Mathiasson, Helena Nordvarg. GE Healthcare Life Sciences, Uppsala, Sweden

The interest of the biopharmaceutical industry in process intensification is steadily increasing, and some of the main drivers are reduced production costs through increased automation, removal of hold-up steps, reduced equipment footprint, and a shortened process time. For intensification of downstream processes, continuous chromatography has so far mainly been evaluated for the capture of monoclonal antibodies. However, the interest in its use in other applications, such as vaccines, is growing. Vaccine processes are generally far less productive compared with monoclonal antibody processes. The reason for this is mainly that vaccines are not a homogenous target group, and that there is no equivalent to the highly selective protein A capture step used in mAb processes. Due to their size, the binding capacities of most resins for the target entities are also significantly lower compared with protein A resins. For this reason, downstream vaccine processes are often conducted in flow-through mode.

This poster discusses how four-column periodic counter-current chromatography (4C PCC) can be applied to automate chromatographic process steps for vaccines based on plasmid DNA (pDNA). Naked pDNA can be used for direct vaccination through different distribution routes or as a vehicle for introducing lentivirus or adeno-associated virus (AAV) vectors into cells, where both lentiviruses and AAV require several separate plasmids to produce infectious viral particles (virions). A critical step for purification of pDNA from alkaline cell lysates is the removal of RNA. This work shows an example of automated group separation, where the ÄKTA™ pcc chromatography system is used with a continuous feed flow to remove RNA.

BIOT 304

Novel technology for continuous manufacturing of agarose-based, monodisperse, chromatography beads

Hans J. Johansson, [email protected], Patrick Gilbert. Research and Development, Purolite, Llantrisant, Wales, United Kingdom

Today the absolute majority of agarose resins designed for large scale purification are produced by batch emulsification. The resulting beads have, even after extensive sieving, a very broad particle distribution. In this paper, a new technology for continuous manufacturing of agarose beads, “jetting” will be discussed. The beads produced by this process does not require sieving and has a very tight particle distribution compared to beads generated by batch emulsification. The absence of fine particles significantly improves pressure/flow properties and packing performance. In addition, the solvent free emulsification process is more economical and environmentally friendly. Application data from both Protein A and ion exchange resins based on agarose beads produced with the novel jetting technology will be presented

BIOT 305 Enrichment of the components responsible for PS20 degradation

Jie Gu, [email protected], Yinges Yigzaw. Purification, Genentech, South San Francisco, California, United States

Polysorbate 20 (PS20) is a common excipient used in biopharmaceutical formulations to prevent potential protein degradation and aggregation. However, degradation of PS20 in biopharmaceutical formulations after months of storage has been observed and become a major concern due to its potential consequences on generating visible particles in the formulations and shortening the shelf life of the products. The components that degrade PS20 are believed to be the trace amount of impurities that survived through the standard purification processes. Due to their low abundance and unknown properties, the impurities are “invisible” in many analytical methods, adding more challenges in purification process optimization. With the guidance of an activity assay, we demonstrated targeted enrichment of the impurities through chromatography runs, allowing for identification of the impurities by mass spectrometry.

BIOT 306

Insights into the effect of pH on mAb surface properties and selectivity in multimodal chromatography systems

Julie Robinson, [email protected], Steven M. Cramer. Chemical Engineering, RPI, Troy, New York, United States

Recent work in our lab characterized the interactions between a library of multimodal cation exchange (MM CEX) resins and several antibodies with different surface properties. The mAbs showed unique selectivity trends in the various MM CEX systems, and also exhibited different domain contributions, which were attributed to the different distributions of surface charge and hydrophobicity. While some mAbs co-eluted with the (Fab)2 domain, other mAbs co-eluted with the Fc domain. The current work builds on these results by investigating the effect of pH on mAb selectivity and domain contributions. Solution pH is a powerful parameter that can be tuned to produce new selectivities by modulating the charge patches on the antibody surface as well as titrating the ligand itself in weak MM CEX systems. In this work, selectivity trends for three mAbs were determined at a range of pH conditions for four MM CEX resins at two different ligand densities. Antibody behavior was connected to the shifting surface charge and hydrophobicity landscape at the different pH conditions to provide insight into the molecular nature of the mAb-ligand interactions. The results show that pH can be used not only to tune selectivity but can also change the preferred binding regions on the antibodies. This work sets the stage for better understanding the molecular nature of mAb-multimodal ligand interactions and has important implications for the design of MM CEX polishing steps for the removal of product related variants.

BIOT 307 Evaluating wash strategies to remove residual cell culture flocculant interactions with protein A chromatography resin

Kimberlee Sing, [email protected], Michael Peck, Abby Schadock-Hewitt, Nripen Singh, Srinivas Chollangi. Process Development, Bristol-Myers Squibb, Devens, Massachusetts, United States

With rising monoclonal antibody (mAb) titers in harvest, downstream purification is continually challenged to not only capture high process yields, but also remove increasing levels of process and product related impurities. Purifying the mAb through harvest treatment with flocculants and other orthogonal clarification methods, like Protein A (ProA) Chromatography and isoelectric precipitation, have shown to meet drug substance level impurity results. However, since the flocculant is cytotoxic, carryover of residual flocculants remains a concern following ProA as there has been indication that it interacts with the resin. In this study, we investigate varying wash strategies, including pH, salt concentration and excipients, to evaluate how residual flocculant interact with the Protein A chromatography resin and mAb. In general, high ionic strength washes are effective in removing flocculant carryover on the ProA resin. While flocculants show promise in considering alternative orthogonal purification strategies to ProA, like polishing chromatography, ensuring minimal carryover of the residual flocculant to the drug substance is crucial for the safety of the patient.

BIOT 308

MAb aggregate removal with CEX flow-through frontal chromatography

Kristen A. Cotoni, [email protected], Matthew Stone. MilliporeSigma, Bedford, Massachusetts, United States

Removing aggregates during the downstream purification of monoclonal antibody (mAb) therapeutics is imperative since these impurities increase the risk of an immunogenic response and can reduce efficacy. Aggregates are a particularly challenging impurity to remove since they have very similar isoelectric points and hydrophobicities to the monomeric protein and are typically present at much higher percentages relative to other impurities such as HCP, DNA, viruses, etc. Aggregate removal is most commonly performed using bind/elute chromatographic processes. Here, we report a unique CEX resin specifically designed for the removal of aggregates in the flow-through frontal chromatography mode of operation, thereby enabling significantly higher loadings (1000 g/L) relative to bind/elute chromatography (30-80 g/L). In this study, we compare two different 3-step processes consisting of capture chromatography with Eshmuno® A media, followed by CEX bind/elute chromatography with Eshmuno® CPX media or CEX flow-through frontal chromatography with a new CEX media, and finally flow-through chromatography with Eshmuno® Q. We demonstrate comparable aggregate and HCP removal with good recoveries for both processes using two different mAbs. However, the process using the CEX flow-through frontal chromatography was operated at greater than 10 times higher loadings and requires no dilution prior to AEX chromatography, providing a purified product at 2.5 times the concentration obtained with a traditional CEX bind/elute chromatography process.

BIOT 309

Use of a continuous chromatography system for both resin screening and scale- up studies

Linda Mathiasson, [email protected], Linda Persson, [email protected], Mikael Berg, [email protected], Martin Hall, Hans Blom, [email protected], Anders Ljunglöf, Bengt Westerlund, Helena Nordvarg, [email protected], Christel Fenge, [email protected]. GE Healthcare, Uppsala, Sweden

The cost pressure on biopharmaceuticals drives the industry towards exploring process intensification options, while maintaining or improving quality, stability, and manufacturability of the product. Traditionally, biomanufacturing is dominated by batch operations. Lately, however, continuous or connected biomanufacturing is being evaluated. While perfusion bioreactors are approved for use in several upstream processes, downstream processing has remained as a series of separate batch operations with hold-up tanks in between.

Continuous chromatography is applicable in different phases from drug discovery to production of biopharmaceuticals, for example, process development, including screening of suitable resins, and automated production of drug candidates in late research. In this talk, we present results from a case study on a scale-up of a periodic counter-current chromatography (PCC) mAb capture step using one and the same chromatography system, and what this means in terms of required productivity and efficiency. We also present data from a resin lifetime study using PCC. In addition to MabSelect SuRe™ resins, a new protein A resin with improved capacity and alkaline stability compared with its predecessors, was included in the study. Experimental data shows that the new Protein A resin retains a significantly higher relative dynamic binding capacity (DBC) after 50 cleaning-in-place (CIP) cycles with high concentration of NaOH compared with the other tested resins. PCC technology, together with high- capacity and alkaline-stable resins, has the potential to increase productivity, while reducing the size of the equipment needed for clinical and commercial productions.

BIOT 310

Design space development for mAb ProA chromatography and VI

Lunedt Gonzalez, [email protected]. Early Stage Process Development, Alexion Pharmaceuticals, Inc., New Haven, Connecticut, United States

A fractional factorial exploration of the design space for the combination of a monoclonal antibody protein A and viral inactivation steps was performed as part of process intensification efforts. The impact on product related impurities and heterogeneity was assessed by varying column operations including load and elution conditions, and viral inactivation parameters. Results were fit using a mathematical model to determine correlations between operating conditions and product quality attributes. From this analysis a robust operating space for these steps was established to meet requirements for commercialization.

BIOT 311

Improving production efficiency through overload chromatography

Marc Wong2, [email protected], Jason Brown2, Jessica Yang3, David Glover2, Christopher Williams4, Yinges Yigzaw1. (1) Purification, Genentech, South San Francisco, California, United States (2) Purification Development, Genentech, San Francisco, California, United States

This presentation will discuss a case study that highlights the strategy and benefits of cation-exchange (CEX) overload chromatography from a process design and manufacturing perspective. Increasing cell culture titers present a challenge to CEX bind and elute operations which require the process to scale to larger columns, pool vessels, and buffer tanks. Overload cation-exchange chromatography has been shown to have equivalent impurity removal capabilities to bind and elute mode while intensifying the process to improve resin utilization and plant efficiency. The resin is loaded under strong binding conditions beyond the resin dynamic binding capacity. Impurities preferentially bind to the resin while the product breakthrough is collected. Acceptable yield and impurity removal was achieved and significant process efficiencies were realized - higher capacity, lower cost of goods, and resource savings.

BIOT 312

Continuous chromatography solutions and the effect of number of columns on process performance

Mark Allen Pagkaliwangan2, [email protected], Xhorxhi Gjoka3, Rene Gantier4, Mark Schofield1. (1) Biopharm R&D, Pall, North Grafton, Massachusetts, United States (2) Pall Life Sciences, Westborough, Massachusetts, United States

Multi-column chromatography capture processes hold the potential to significantly reduce both operating and capital expenses involved in the manufacture of mAbs while simultaneously improving product quality and consistency. However, transitioning to a well-optimized continuous multi-column process from a batch process can be challenging, due to the different factors that need to be considered, such as feed titer, feed residence time, column scheduling constraints, etc. Here a hybrid experimental and modeling approach was used to optimize four different multi-column chromatography processes, and the effect of process factors on overall operating cost, most notably the number of process columns, were examined. To do so, three chromatography modeling approaches were validated with upwards of 30 experiments to evaluate the accuracy of each model. The best fitting model was then selected to predict process performance for each of the loading methods. A wide range of feed concentrations and residence times were considered, and the response of operating binding capacity, specific productivity, and the number of columns as a function of these process parameters were calculated. Processes that were able to add more columns proved to be up to 65% more productive, especially at feed concentrations above 5 g/L. An investigation of the operating costs shows that discrete column sizing confounds process performance metrics and that the most productive process is not necessarily the most cost effective. However, adding more columns at higher feed concentrations allows for overall cost savings of up to 32%.

BIOT 313

New protein L affinity resins for antibody fragments purification

Masaaki Hanamura, [email protected], Satoshi Nakamura, Jun-ichi Yasuoka, Kiichi Yoshimura. R&D Department Unit 2, JSR Life Sciences, Tsukuba-shi, Ibaraki, Japan

Fragment antibody (Fab, scFv, VHH) biologics are quickly emerging as an important class of next-generation therapeutics. However the manufacturing purification processes lack a robust affinity step comparable to that of the Protein A resins used in most intact antibody processing today. Commercially available Protein L and G chromatography media have affinities to VL and/or CH1 regions of antibody fragments; however, there are limitations with its binding capacity, caustic stability needed for cleaning, and/or limited to low pH for the protein elution step. JSR Life Sciences has developed a novel Protein L media through genetic engineering of a proprietary ligand and optimization of the base matrix. In this presentation, key performance of the prototype will be introduced in comparison with the other products on the market.

BIOT 314

Continuous multi-column chromatography: Methodology for the selection of the most suitable protein A media

Meghan Higson, [email protected], Melissa Holstein, Christopher Gillespie. Next Generation Processing, MilliporeSigma, Bedford, Massachusetts, United States

Continuous multi-column chromatography is being heavily investigated as the biopharmaceutical industry looks for new and improved processes and technologies to intensify current operations. While a variety of Protein A media are available, very few have been specifically designed for this application. Having a comprehensive understanding of how the media performance impacts the process is critical to the selection of an appropriate media to implement in a continuous multi-column chromatography process. In this study, several commercially available Protein A media were evaluated using multiple monoclonal antibodies with standard industry performance criteria including pressure drop and dynamic binding capacity as a function of residence time. Applying this resin evaluation approach could ultimately lead to improved process performance, productivity, resin utilization, as well as provide important insights for the development of the next generation of Protein A resins.

BIOT 315

Purification of cell culture-derived oncolytic measles virus by membrane chromatography

Keven Lothert3, Daniel Loewe4, Tanja Grein5, Peter Czermak2,6, Michael W. Wolff2,1, [email protected]. (1) Bioprocess Engineering, Max Planck Institute for Dynamics of Complex Technical Systems, Magdeburg, Saxony-Anhalt, Germany (3) Institute of Bioprocess Engineering and Pharmaceutical Technology, University of Applied Sciences Mittelhessen, Giessen, Hesse, Germany (6) Project group Bioresources, Fraunhofer Institute for Molecular Biology and Applied Ecology, Giessen, Hesse, Germany

Chromatography is commonly an integral part in downstream processing of biological human therapeutics. In the last 20 years the development of solid phases designed for purification of biological nanoplexes (e.g. viruses, virus like particles (VLPs), phages and polynucleotides) based on monoliths and membranes greatly advanced. The main advantage of these matrices are the fully convective mass transport due to the micron- sized flow channels, the absence of void volumes, and the high mass capacities. Steric exclusion chromatography (SXC) based on cellulose membranes makes full use of these advantages. In addition, no specific ligands are required to capture particles from complex mixtures. Its selectivity is strongly based on the target size. SXC is carried out by mixing the clarified cultivation broth with polyethylene glycol (PEG) and loading it onto a cellulose membrane. Based on the applied PEG molecular weight and concentration, the virus particles are captured without a direct chemical interaction by the mutual steric exclusion of PEG between the target particle and the stationary phase. Smaller impurities i.e. host cell proteins and DNA are washed out. The product is eluted by reducing the polymer (PEG) concentration and can be recovered in a variety of buffers. Here, we compare the application of classical membrane adsorbers with SXC (cellulose membranes) for the purification of oncolytic measles virus in terms of infective virus yield, contaminant depletion, productivity, and process economics.

BIOT 316

High throughput resin selection for the recovery of valuable compounds from food products Monica Moreno Gonzalez1, [email protected], Vasupradha Grish1, Pattra Chuekitkumchorn1, Guilherme Ferreira2, Hilde Wijngaard3, Marcel Ottens1. (1) Delft University of Technology, Delft, Netherlands (2) Food Specialties , DSM, Delft, Netherlands (3) R&D , Unilever, Vlaardingen , Netherlands

Research into the recovery of valuable products from side streams has increased over the past years [1]. Proteins derived from vegetative resources such as oilseed meals (canola, rapeseed, cottonseed, soybeans, and sunflower seed) are considered an alternative to animal proteins for cosmetic applications and for food consumption. However, the purification of vegetable proteins usually requires thermal or physico- chemical processes that can affect their functional properties and the nutritional value of the final product [2]. Adsorption technology is a promising alternative as it can be operated at mild conditions and might also lead to a higher purity of the product. After the oil has been extracted from the meal, a high amount of protein is left behind (~40% protein content) [3]. Proteins and polyphenols can be extracted from the meal and further separated from each other by using, for example, membrane filtration and adsorption. This paper will present a selection of suitable resins for the recovery of polyphenols and for protein purification from oilseed extracts. High throughput experimentation is used to evaluate a set of food grade macroporous polymeric resins (reverse-phase chromatography) for polyphenol recovery. Adsorption isotherms at different conditions and desorption reagents such as an ethanol/water mixture are evaluated. Additionally a set of multimode resins and weak cation exchangers are assessed for protein purification. A resin selection criterion is defined considering capacity, selectivity and ease of desorption. The selected resins will be used for industrial scale adsorptive process design for capture and purification of these valuable side products.

BIOT 317

Increasing continuous purification productivity by combining small particle size chromatography resin with continuous countercurrent tangential chromatography (CCTC)

Oleg Shinkazh, [email protected]. Chromatan corporation, Philadelphia, Pennsylvania, United States

Continuous Countercurrent Tangential Chromatography (CCTC) has been shown to demonstrate significant advantages over column chromatography including higher productivity, lower operational pressure, disposable flow path, and lower resin use. Because the separation in CCTC is performed on a moving slurry, the pressure drop in the CCTC system is significantly lower (<15 psi) compared to a packed column and is independent of resin particle size and rigidity. These features enable a wider design space for development of a high efficiency resin particle that is specifically suited for the CCTC platform.

Purolite and Chromatan co-present data with a new high efficiency resin particle that was jointly developed by the two companies. The new prototype resin has a lower particle size (20-25 microns) compared to a typical prep scale chromatography resin (45-90 microns). The data shows significant improvements in binding capacity, binding kinetics and CCTC system productivity for protein A mAb capture. Using this new resin, the CCTC system has shown productivities of ~140 g of mAb / L resin / hr, which is a 2- 3X improvement over the CCTC productivity using off-the-shelf protein A resins and more than a 10X increase in productivity versus a batch protein A capture column with comparable purification and yield.

BIOT 318

Implementation of directly scalable single-use protein A chromatography for rapid processing from bench to industrial processing

Oliver Hardick2,4, [email protected], Daniel G. Bracewell3, William Lewis1. (1) GSK, Stevenage, United Kingdom (2) Executive, Puridify, London, United Kingdom (3) UCL Dept Biochemical Engr, London, United Kingdom (4) Biochem Eng, UCL, London, United Kingdom

A cellulose adsorbent has been designed to operate in a rapid cycling chromatography mode, with each unit being cycled >100 times in a single working shift. The result is a unit that can be sized such that its lifetime (in terms of cycles) can be exhausted over the batch, with the aim of making single-use operation economically feasible. This gives rise to a new high productivity industrial chromatography technology that can purify 100’s g/L/h mAb, using Protein A and IEX chromatography units that scale directly from process development (PD) to production.

The case study presented covers early stage process development of a three step purification isolating an industrially relevant mAb from CCCF. High throughput PD scale screening was carried out, where process conditions are evaluated. The value that sub second residence times impart upon building process understanding within a matter of hours is demonstrated; with full DoE runs carried out on ProA capture and IEX steps. The 2 min cycle time allows for repeat runs to give confidence in the data. Finally, cycling and process stability data for the chosen performance conditions is transferred to large scale devices to demonstrate direct scalability. Product quality at both scales is analyzed in terms of purity, yield & titre and shown to meet the expected requirements e.g. 3 log reduction in HCP, <10 ppm ProA leaching.

The units can be used at clinical scale in a cost-efficient manner with full utilization of the chromatographic lifetime with a high degree of process robustness having completed scalable PD studies in a fraction of the usual time. Additionally they can be scaled such that production feed material can be processed in a significantly shorter amount of time, this flexibility enables the end-user to choose a processing strategy that best fits their specific needs, or the changing needs of the product throughout its lifecycle. The ability to handle different processing flowrates without affecting the product quality offers a potential for integrated continuous processing with the coupling of independent unit operations.

BIOT 319

CLSM: Process visualization of Eshmuno® A affinity chromatography resin

Oliver Rammo, [email protected], Romas Skudas, [email protected]. BioProcessing, Merck KGaA, Darmstadt, Hessen, Germany

Protein-protein based interaction, regardless its complexity, generated a strong interest in protein purification strategies. Especially the discovery of Protein A and the specific recognition for immunoglobulin G or same in class pharmaceutical molecules, reduced the purification complexity as an universal template approach, resulting in 5 blockbuster drugs within the top 10. And there are more to come taking into account more than 2500 clinical studies ongoing. Regardless the robustness of this method, there remains an open question about how this protein-protein interaction is taking place and what governs the efficiency of the process if the Protein A molecule is bound to the surface having limited accessibility.

We approached this challenge from user perspective, using chromatographic performance as a main criteria. With the combined power of standard protein analysis including finite bath (batch) experiments and the technique of confocal laser scanning microscopy (CLSM), we developed a method to visualize the mass transfer of our affinity Eshmuno® A chromatography resin using fluorescent labeled test molecules, pre and post alkaline treatment. By tracking the fluorescent intensity profile along the radius of a single chromatographic bead, representing the Protein A-antibody interaction, we are able to determine kinetic parameters in static operation mode.

In the poster, we will discuss the outcome of our study, focusing on the bead saturation governed by the enhanced pore diffusion, physical properties from the base support and the effect of alkaline exposure.

BIOT 320

Formation of on-column aggregates during cation exchange chromatography and studies of its elimination

Rachel B. Wollacott, [email protected], Christopher Connelly, Sadettin Ozturk. MassBiologics, Mattapan, Massachusetts, United States

During the scale-up of a monoclonal antibody cation exchange (CEX) purification process on an SP Sepharose FF column, low MAb recovery (70-80%) and a substantial peak during the 1N NaOH sanitization step was observed. This behavior was not seen during initial development work on small-scale columns. After investigation, it was determined that the large sanitization peak was due to on-column aggregation of the antibody. The aggregation was exacerbated by the flow rate of the high salt elution buffer, the buffer system used, and the CEX resin chosen. Due to this aggregation behavior in SP Sepharose FF column, a new CEX process was developed and optimized on a Gigacap S-650M resin. The new process resulted in a process that yielded 95% MAb recovery (99.5% monomer) with twice the binding capacity and lower salt elution conditions when compared to the SP Sepharose FF process

BIOT 321

Use of activated carbon as an adsorbent in polishing step

Romas Skudas, [email protected]. Merck KGaA, Darmstadt, Germany, Darmstadt, Germany

In pursuit of economic biopharmaceutical molecule purification, we evaluated activated carbon and synthetic variants thereof for the flow-through antibody polishing followed by real case studies. In flow-through mode, less abundant impurities bind selectively to the media enabling the reduction of the volumes of buffer and media required . The studies were performed comparing the application of existing IEX chromatographic resins and carbon media in an overlapping operational window from packed carbon columns to carbon-containing depth filter media (e.g. CR40). Our case studies show flow-through purification results in the effective separation of host cell proteins (HCPs), antibody fragments and low molecular weight substances from post Protein A affinity capture solutions. In most cases the mAb recovery was >85% and HCP amount was reduced to <10ppm. Additionally, we were able to show robust and reliable removal of some other critical impurities, such as antibody fragments. Carbon media was found to be effective under a wide range of conditions where AEX and CEX media are typically operated, enabling to link it with one or another without an intermediate solution adjustment. To address the technology’s robustness, investigation covered antibodies originating from different cell cultures as well as different Protein A affinity chromatography capture pools. Our experimental data confirms that this technology can be adapted to standard and new purification templates while maintaining critical product quality attributes and achieving economically efficient biopharmaceutical molecule purification. In some cases, the use of this technology could enable fully continuous operation of mAb purification processes operation by linking the different flow through chromatography media to form a single integrated system.

BIOT 322

Investigation into molecular basis of protein interactions in multimodal chromatography using nuclear magnetic resonance spectroscopy and molecular dynamics simulations

Ronak B. Gudhka1, [email protected], Camille Bilodeau1, Scott A. McCallum2, Steven M. Cramer1. (1) Chemical and Biological Engineering, Rensselaer polytechnic institute, Troy, New York, United States (2) Biotech 2nd Fl MS, Rensselaer Polytechnic Inst, Troy, New York, United States

Multimodal (MM) Chromatography offers highly selective protein separation systems by taking advantage of hydrophobic, electrostatic and Van der Waals interactions. Due to the complexity of such protein-ligand interactions, there is a need for the molecular level understanding of the nature of proteins binding to different MM chromatographic systems. Recent work in our lab has shown the importance of specific domains towards binding for multi-domain molecules such as antibodies. In this work, we employ Nuclear Magnetic Resonance spectroscopy (NMR) to determine the amino acids involved in binding with chromatographic ligands. NMR titration experiments with MM ligands and isotopically enriched Fc domain of a monoclonal antibody were used to determine the relative binding affinities of interaction sites on the protein surface. Further, in order to mimic chromatographic surface, a pseudo solid-state resin system i.e. gold nanoparticles functionalized with Self Assembled Monolayers (SAM) presenting a MM ligand headgroup was used to study the binding. In order to study the co-operativity and avidity effects in these MM chromatographic systems binding was studied with gold nanoparticles functionalized with SAMs presenting different ligand densities of MM ligand headgroup. Further, molecular dynamics simulations with ligands in free solution around the protein were performed to gain additional molecular details on multimodal- ligand protein binding. The results presented in this poster demonstrate that NMR spectroscopy can be employed in concert with chromatography and molecular dynamics simulations to provide important fundamental insights into protein multimodal surface interactions.

BIOT 323

High-throughput screening for purification development of osteospontin-an acidic recombinant protein expressed in E. coli

Shengchun Guo2, [email protected], Ayswarya Ravi1, Stephen Mayfield3, Zivko L. Nikolov2. (1) Biological and Agricultural Engineering, Texas A and M University, College Station, Texas, United States (2) Texas A&M Univerisity, College Station, Texas, United States (3) UC San Diego, San Diego, California, United States

Osteopontin is a structural protein essential for biomineralization such as bone remodeling and recognized as a molecular marker for tissue disorders including cancer due to its role in cell signaling, adhesion and migration in regular conditions and stress responses. Both health as well as diagnostic and prognostic potential of osteopontin have invigorated the search for an efficient and scalable downstream process. Low titer (10 mg/L culture), acidic isoelectric point (pI= 4.3 ), and the lack of well-defined secondary and tertiary structure were some of the challenges that faced purification development of osteopontin from E. coli lysates. Finding an adequate capture step with sufficient specificity and binding capacity that could be paired with an orthogonal purification step was a vitally important objective for this project. The combination of high-throughput screening (HTS) and design of experiments (DOE) allowed us to select the best resin candidates and assemble a two-step purification process. The sorbents screened included strong anion exchange (Capto Q), hydrophobic interaction chromatography (Phenyl Sepharose) and several mixed-modal resins (Capto adhere, HEA HyperCel, PPA HyperCel, MEP HyperCel, and ceramic hydroxyapatite). Plate based (HTS) platform was applied and quickly revealed comprehensive information about the affinity between the ligands and the target protein, specificity of the adsorption, and the binding capacity of the resins. Based on HTS results, the bind/elute processes for all candidate resins were designed, tested and optimized on small scale gravity-flow chromatography columns. The purification factors and osteopontin recoveries achieved by gravity-flow columns were verified on 10 cm packed-bed columns operated by ÄKTA system and analyzed statistically. In terms of purification factor and yield, HEA HyperCel and Capto Q performed significantly better than the rest of the candidate resins. A 2-step purification process assembled with these two resins resulted in a 45-fold purification of osteopontin (>95% purity) from clarified E. coli lysate with greater than 50% yield. The benefits of using HEA HyperCell or CaptoQ as capture or purification step, respectively, will be presented and discussed.

BIOT 324

Investigation into the different mechanisms of interaction in multimodal anion and multimodal cation exchange systems Siddharth Parasnavis1, [email protected], Julie Robinson1, Steven M. Cramer2. (1) Chemical Engineering, RPI, Troy, New York, United States (2) Ricketts Bldg, Rensselaer Polytechnic Inst, Troy, New York, United States

A current limitation of the implementation of multimodal anion exchange (MM AEX) resins in downstream purification processes is the recovery challenge associated with this mode of chromatography. In the literature, this recovery challenge is documented to a greater extent for MM AEX resins as compared to multimodal cation exchange (MM CEX) resins. Since many commercially available MM CEX and MM AEX resins have the same phenyl hydrophobic group, we hypothesize that the difference in resin behavior may be due to the differences in the charge modalities presented on the ligands. In this work, we investigate the mechanisms of protein interactions with MM AEX and MM CEX ligands in order to provide fundamental insight on the observed differences in behavior and recovery. Isocratic experiments were performed at a range of salt concentrations for several antibodies and the results are presented in the form of lnCs vs. lnk¢ plots. The shapes of these plots provide insight about the relative hydrophobic and electrostatic interactions for each protein in the various resin systems investigated. First, we investigate this behavior for Nuvia cPrime and structurally similar Nuvia aPrime prototype resins. To further understand interactions in the MM AEX systems, we evaluate a series of novel MM AEX prototype resins that differed in both charge and hydrophobic modalities. The results from these experiments were then connected to protein surface properties using spatial aggregation propensity (SAP) maps and electrostatic potential (EP) maps to explain the trends. This work provides fundamental insight into protein-ligand interactions in MM chromatographic systems and can be used to inform the design new MM resins with high selectivity and high recovery towards their target biomolecule.

BIOT 325

Fine-tuning your purification process: Using mechanistic modeling of chromatography to optimize the balance between yield, purity, and ease of operation

Tim J. Fattor, [email protected], Stephen Hunt, Jonathan Rocher, Robert J. Todd. KBI Biopharma, Boulder, Colorado, United States

To maintain a competitive position in the biopharmaceutical industry, biotechnology companies need to continually improve the speed of product development and efficiency with which these products are manufactured. With the introduction of biosimilars and increasing pressure by payers to reduce the cost of products, this has become even more important for the industry. As has occurred in other industries, application of advanced computational tools, including mechanistic modeling, to process development provides the opportunity for improved process understanding and therefore the development of more robust and efficient manufacturing processes. With improved understanding around parameters that have the greatest impact, along with an understanding of the magnitude of these impacts, it becomes possible to fine-tune chromatography unit operations to optimize for desired outcomes. Using a targeted set of calibration experiments, we have employed mechanistic modeling to increase our understanding around an intermediate cation exchange purification step. Using this model, we explore process improvements targeting an optimal balance between yield, purity, and manufacturability.

BIOT 326

Comparative study of commercially available protein A chromatography resins and AmsphereTM A3: Qualitative analysis of residual host cell proteins by means of 2D-LC/MS

Tomonori Shiotani1, [email protected], Sachiko Tsuda2, Takashi Tanaka3, Masaaki Hanamura2, Masayoshi Nagaya1. (1) JSR Life Sciences, Sunnyvale, California, United States (2) JSR Life Sciences, Ibaraki, Japan (3) JSR Corporation, Ibaraki, Japan

Process and product-related impurities such as host cell proteins (HCPs) are generally considered as a critical quality attribute for the drug substance because of their potential impact on drug product safety. HCPs are co-produced by the host cell organism as part of their normal cell functions to grow and produce the recombinant therapeutics proteins. Typically, the majority of HCPs in harvested cell culture fluid are removed during capture step with Protein A chromatography resin, and the remaining HCPs are cleared away during polishing step by a combination of ion-exchange chromatography, hydrophobic interaction chromatography, etc. Thus, identification of remaining HCPs after the Protein A capture step can help establish a more robust intermediate and final polish processes to minimize the drug substance impurity levels. In this study, HCPs in eluate samples of Protein A affinity chromatography were evaluated with ELISA (quantitative analysis) and 2D-LC/MS (qualitative analysis). These results were multilaterally compared with JSR's resin, Amsphere A3 and other commercially available resins.

BIOT 327

Single-molecule and isocratic protein competition on ion-exchange adsorbents

Ujwal Patil3, [email protected], Lydia Kisley2,1, Sagar P. Dhamane4,3, Mohan Poongavanam3,6, Andrea Mansur3,1, Sergio Dominguez Medina1, Eliona Kulla1, Jixin Chen1, Lawrence J. Tauzin1, Katerina Kourentzi5, Christy F. Landes1, Richard C. Willson3,5. (1) Chemistry, Rice University, Houston, Texas, United States (2) Beckman Institute, University of Illinois Urbana-Champaign, Urbana, Illinois, United States (3) Biology and Biochemistry, University of Houston, Houston, Texas, United States (4) Drug Substance Technology Engineering, Amgen, Thousand Oaks, California, United States (5) Chemical and Biomolecular Engineering, University of Houston, Houston, Texas, United States (6) Pediatrics-Tropical Medicine, Baylor College of Medicine, Houston, Texas, United States All practical chromatographic processes involve the separation of complex mixtures by competitive adsorption. In this work, we employ single-molecule methods and a novel isocratic competitive mode of column chromatography to study the complex mechanistic details involved in protein chromatography. Using a single-molecule, super-resolution imaging technique called motion-blur Points Accumulation for Imaging in Nanoscale Topography (mbPAINT), we present the direct mapping and kinetic characterization of individual functional sites on thin-film agarose ion-exchange matrices. By extracting single-protein adsorption and desorption kinetics at individual ligands, direct experimental evidence in support of the stochastic theory of chromatography is obtained. We also investigate the influence of ionic strength on the heterogeneity of protein ion-exchange functional adsorption sites, and therefore the heterogeneity of elution profiles. We observe that the number of functional adsorption sites is smaller at high ionic strength and these sites have reduced desorption kinetic heterogeneity; this is in accordance with our earlier reports based on equilibrium adsorption isotherm measurements with cytochrome b5. In an integrated study, we use single molecule spectroscopy and a novel isocratic competitive chromatographic operating regime to study a multicomponent, competitive chromatographic system. We observe that the competitor (insulin) does not change the single molecule kinetics of α-lactalbumin adsorption, but reduces the number of adsorption events on cationic spermine ligands. Overall, we have shown that single molecule super-resolution imaging can advance our understanding of multicomponent competitive ion-exchange adsorptive separations. The results of such studies could be useful in improving ion-exchange chromatographic separations of realistic multicomponent systems.

BIOT 328

In situ analysis of ligand density on the surface of regenerated cellulose membranes

Vibha Bansal1, [email protected], Ezio Fasoli2, Jose O. Sotero Esteva4, Ivan J. Dmochowski3. (1) Dept of Chemistry, Univ of Puerto Rico, Cayey, Puerto Rico, United States (2) Univ of Puerto Rico Chem Dept, Humacao, Puerto Rico, United States (3) Upenn Dept of Chemistry, Philadelphia, Pennsylvania, United States (4) Mathematics, University of Puerto Rico at Humacao, Humacao, Puerto Rico, United States

The goal of this project was to develop an accurate method for determination of ligand density on the surface of para-aminobenzamidine (pABA) linked regenerated cellulose membranes and other chromatography matrices. These membranes have been shown to capture the target serine hydrolase protein selectively from biological mixtures. However, the in-depth analysis of membrane’s protein binding efficiency requires an accurate estimation of the ligand density on membrane surface. Quantification of amidine group in solid phase is challenging since the only known methods for this purpose are either colorimetric or require a disruption of the membrane structure. In this study, we were able to develop a fluorescence based assay for the estimation of ligand density (pABA) on cellulose membrane surface by reacting the immobilized benzamidine moiety with glyoxal and benzaldehyde in alkaline aqueous solution, leading to the formation of the fluorescent probe 2-aryl-4-arylidene-2-imidazolin-5–one (Ex 385 nm, Em 455 nm). The method was found to be linear up to 5 mM amidine concentration. While the pABA fluorescent tagging allowed us to perform a fast and reliable in situ analysis at varying ligand density without disrupting the membrane; imaging the fluorescently labeled membrane using Confocal Laser Scanning Microscopy (CLSM) facilitated the z-stack analysis of the membrane volume. An algorithm for pixel counting on membrane surface was developed to perform the analysis of CLSM images. The z-stack analysis confirmed that the ligand was immobilized uniformly across the membrane volume.

BIOT 329

Optimisation of preparative polishing chromatography for a ternary antibody mixture

Vivien Fischer2,1, [email protected], Richard Kucia-Tran1, William Lewis1, Ajoy Velayudhan2. (1) GSK, Stevenage, United Kingdom (2) Biochemical Engineering, University College London, London, United Kingdom

A combination of empirical and modelling techniques, the adaptive simplex and a local fitting approach were used in tandem to flexibly explore competing operational spaces for the polishing of a monoclonal antibody from a three component system. Antibody related variants in the ternary mixture were present as lumped components making this an industrially relevant and challenging separation. Following the screening of potentially suitable chromatography resins, the cation exchanger Eshmuno CPX was selected and optimised based on a range of different objective functions. Weak partitioning and high initial salt/high loading modes are compared and trade-offs among yield, purity and throughput are discussed and related to practical settings in manufacturing.

BIOT 330

Experimental design for parameter estimation in chromatography

William Heymann, [email protected], Eric von Lieres. Research Center Juelich, Juelich, Germany

Before a chromatography column can be accurately simulated the corresponding model parameters must first be identified. The question then becomes what experiments are best suited to provide numerical identifiability, minimal variance and correlations of these parameters. Most of the required parameters cannot be measured directly, such as column and particle porosities, axial dispersion, pore and film diffusion coefficients, binding capacity, adsorption and desorption rates. Estimating these parameters from measurement data involves goal functions and search algorithms. The standard approach is to couple the sum of squared errors (SSE) between model prediction and measurement data with gradient descent and Newton’s method. However in chromatography modeling the SSE does not always provide numerical identifiability and sufficient accuracy of the sought parameters. We present here the CADET-MATCH framework for parameter estimation based on the Chromatography Analysis and Design Toolkit (CADET). This framework uses composable goals functions that are based on feature detection together with gradient-based as well as gradient-free search algorithms. The framework has been applied successfully on synthetic and industrial data. We compare a variety of goals and optimization techniques and study which are most suitable for designing new experiments.

BIOT 331

Leveraging innovative technologies for cost-effective protein A chromatography operation

Yan Brodsky, [email protected], Eva Gefroh, Megan McClure, Rob Piper, Lisa Connell-Crowley. Just Biotherapeutics, Seattle, Washington, United States

The mission of Just Biotherapeutics is to design and apply innovative technologies to dramatically expand global access to biotherapeutics. This mission will be achieved, in part, by lowering development and production costs. Protein A resin remains one of the most expensive raw materials in a mAb production process. One approach to curb cost is to increase efficiency of resin utilization by taking advantage of multicolumn chromatography. This technology enables operation near maximum resin capacity with higher process flow rates that yields a significant increase in step productivity. As a result, the same volume of harvested cell culture can be processed using ~80% less protein A resin compared with batch chromatography. Multicolumn chromatography technology also promotes savings by facilitating integration of multiple steps into a continuous downstream processing operation. This study was aimed to push efficiency limits of the protein A chromatography step by evaluating competitively priced, novel resins in combination with innovative methods of operation.

BIOT 332

Systemic approach to remove host cell proteins during monoclonal antibody protein A chromatography

Zhen Xia, [email protected], yuan zhou, yuntao wu. Merck Chemicals Co.,Ltd, Shanghai, China

Protein A chromatography is a critical and 'gold-standard' step in the purification of monoclonal antibody (mAb) products. However, the purification of target proteins can be challenging due to the co-purification of a complex group of process related host cell proteins (HCPs). Here, we have evaluated two commercially available protein A chromatography resins, Eshmuno A and ProSep Ultra Plus (PUP) with two different mAbs A and B respectively, in terms of their ability and process conditions to remove HCP. Different intermediate buffers and elution conditions were used to screen the best condition according to the HCP removal during protein A chromatography. We alter intermediate wash pH, buffer salt concentration and different type of additive, including detergent and solvent, and also the elution pH to generate the impurity profile fingerprint for each resin. The results showed the HCP removal capacity raise about 47 % and 111 % for Eshmuno A and PUP respectively result from the optimization process. And the consequence support that increasing elution pH was able to minimize HCP level while maximize target protein yield. It was also found that multi-component intermediate wash buffer was more effective in reducing HCP than the single component buffer which imply the HCP related interaction may not only be hydrophilic type but also hydrophobic or even amphoteric. The methods shown in this study allow for a critical remove of HCP during protein A chromatography which alleviates the burden on subsequent process steps and facilitates the implementation of platform processes. On the whole it provides a better solution for process engineers to optimize rational approaches for the removal of prominent HCPs for monoclonal antibody product.

BIOT 333

Linear scalablity of Planova BioEX and Planova 20N filters

Aesha Shah, [email protected], Brian Buesing, Daniel Strauss, Naokatsu Hirotomi, Tomoyuki Miyabayashi. Asahi Kasei Bioprocess, Glenview, Illinois, United States

The outstanding performance of Planova virus removal filters in manufacturing operations is supported by studies at small scale. Correspondence between small-scale filtration data and manufacturing-scale filter performance is absolutely critical for predicting manufacturing process performance from process development experiments and for assuring the validity of small-scale viral clearance validation studies. The scalability of Planova filters has been demonstrated for several performance metrics, but a comprehensive study demonstrating consistent viral clearance over the range of filter sizes has not been previously exhibited. In this study, Planova 20N and BioEX hollow fibers from each of three spinning series were manufactured into filters with effective surface areas covering the range of commercially available filters (4.0 to 0.001 m2 for Planova 20N and 4.0 to 0.0003 m2 for Planova BioEX). Feed solutions of 1 g/L BSA spiked with PP7 bacteriophage as a model parvovirus were filtered to 200 L/m2, and each Planova filter was evaluated for throughput and viral clearance capability. Overall, average flow rate showed direct proportionality with effective surface area, and all filtrations showed complete clearance with PP7 logarithmic reduction value (LRV) exceeding 4 log10. These results demonstrate the excellent scalability of Planova virus removal filters and support the use of small-scale filters to validate viral clearance at larger scales.

BIOT 334

Evaluation of virus filtration prefilters and load conditions for a highly fouling mAb Holland Keefe, [email protected], Nathaniel Bennett, Sandy Sargis, Alexander Seay. Purification Development, Genentech, South San Francisco, California, United States

Small virus (parvovirus) retentive filtration provides a robust, size based mechanism for removal of viruses. This poster will highlight the work that was done with a highly-fouling mAb to test a variety of prefilter/virus filter configurations at different load conditions. Several different prefilters were tested using feedstocks spanning a range of concentrations and buffer backgrounds. In addition, dynamic light scattering was utilized to calculate the diffusion interaction parameter (kD), which can be correlated to decreases in initial permeability observed at increasing protein concentration. This technique was used to characterize load concentration, pH, and different excipients to inform the virus filtration step development and further characterize the observed fouling of the virus filter. Process throughput was achieved with a specific combination of prefilter and load conditions.

BIOT 335

Synthesis of well-defined and easily functionalized copolymers for the preparation of lipid-disc cell membrane mimetics

Kevin Burridge2, [email protected], Indra Sahu2, Andrew Craig1, Emily Clark2, Sameer Al-Abdul-Wahid2, Madison Dolan2, Carole Dabney-Smith2, Dominik Konkolewicz2, Gary Lorigan2. (1) Chemistry and Biochemsitry, Miami University, Liberty Township, Ohio, United States (2) Department of Chemistry and Biochemistry, Miami University, Oxford, Ohio, United States

Characterization of membrane proteins is challenging due to the difficulty in mimicking the native lipid bilayer with properly folded and functional membrane proteins. Copolymers synthesized from styrene and maleic anhydride (StyMAn) can be hydrolyzed to yield styrene-maleic acid (SMA) copolymers. These facilitate detergent- free preparations of disc-like lipid bilayer mimetics that maintain the structural and dynamic integrity of membrane proteins. Here we report the controlled synthesis and characterization of SMA block copolymers by reversible addition-fragmentation chain transfer (RAFT) polymerization. SMA polymers with different compositions and molecular weights were synthesized and characterized by size exclusion chromatography (SEC). A combination of dynamic light scattering (DLS), solid-state nuclear magnetic resonance (SSNMR) spectroscopy, and transmission electron microscopy (TEM) was used to characterize the size of the nanoparticles created using the SMA copolymers. Using different polymer lengths and ratios of styrene:maleic acid, the smallest size achieved was 10 ± 1 nm for the 3:1 styrene:maleic acid copolymer. Nanoparticle size was unaffected by the molecular weight of the polymer. The maleic anhydride portions of the polymers can be easily functionalized by a variety of nucleophiles; polymers modified in this way are currently being investigated.

RAFT polymerization of styrene and maleic anhydride results in narrow molecular weight distribution copolymers with an alternating styrene-maleic anhydride portion and a long styrene- only portion. The RAFT end-group can be easily removed using excess benzoyl peroxide.

BIOT 336

Nanocellulose-based size exclusion filters as an efficient and convenient virus barrier filter

Levon Manukyan, [email protected], Albert Mihranyan. Nanotechnology and Functional Materials, Uppsala University, Uppsala, Sweden

Sterility of bioreactors in biotherapeutic processing remains an ongoing challenge as several viral and Mycoplasma contaminations have been recorded over the years. Number of technologies such as gamma or UV-C irradiation or high temperature short time treatment may not be always suitable because of side effects, footprint and scalability issues. Virus removal achieved by ultrafiltration is an attractive alternative. Timeframe limitations in cell culture medium production require high flow rates and capacity of the filters. To meet the criteria the larger membrane areas are necessary, what, however, makes filtration economically unfeasible. Nanocellulose-based ultrafiltration may be the most suitable and cost-efficient virus protection approach with minimal impact on cell culture medium. To assess the interaction with media components, serum-free Lonza Power-CHO media, supplemented with ITS (Insulin, Transferrin and Selenium), was processed through the nanocellulose filters at varying transmembrane pressures. Media components were analysed by SDS-PAGE and ICP analysis. The impact of the filters on cell culture performance was further assessed by CHO cells growth using the alamar blue assay, cells morphology and confluence. Also, virus retention of nanocellulose filters was evaluated by plaque forming units (PFU) assay of spiked ΦΧ174 phage as a surrogate for mammalian parvoviruses. This study suggests that nanocellulose-based size exclusion filters can provide good virus retention, high filtration flux, and no interaction with cell culture medium ingredients.

BIOT 337

Exploration of synthetic depth filtration applied to mammalian cell harvest

Pavel Kalashnikov, [email protected]. Process Solutions, MilliporeSigma, Moscow, Russian Federation Depth filtration is typically applied post-bioreactor for clarification in mammalian cells processes. The objective of this study was to explore the performance of newly developed synthetic depth filtration media compared to traditional cellulose and filteraid based media and/or pretreatement. These filters were tested on three different monoclonal antibody cell cultures. The trials show promising results using synthetic media over traditional depth filter by exhibiting a capacity 2 to 4 times superior along with lower flushing volumes and reduced footprint.

BIOT 338

Comparison of DSC cryoporomtery and liquid-liquid porometry for validation of pore size distributions of nanocellulose-based virus removal filter papers

Simon Gustafsson, [email protected], Albert Mihranyan, Laura Roemer. Nanotechnology and Functional Materials, Uppsala University, Uppsala, Sweden

In this work, an evaluation of two pore size distribution methods, namely DSC cryoporometry (DSC-CP) and liquid-liquid porometry (LLP), which are frequently used for probing pore size distributions (PSD) of filters in the wet state is presented for the nanocellulose-based virus removal filter paper, i.e. the so-called mille-feuille filter. The mille-feuille filter is a non-woven, µm-thick filter paper with a narrow and tailorable pore size distribution, highly suitable for removing viruses of all sizes, including the worst case model Minute Virus of Mice (MVM, 18 nm). It is produced by hot-pressing method rather than by phase-inversion, which is otherwise used to manufacture the industrial analogues. Previously, variations of PSD of the filter paper were monitored in the dry state using nitrogen gas sorption, whereas the application of the filter is done in the wet state. Therefore, reliable methods of assessing PSD of the virus removal filters in the wet state are of interest. In this context, two methods are potentially suitable for evaluating the PSD of the filter paper in the wet state, i.e. LLP and DSC-CP. Although both methods originate from the Gibbs equation, DSC-CP is conducted under isobaric conditions, whereas LLP requires isothermal conditions. This work discusses the inherent benefits and drawbacks of each method and presents the outcome of experimental side-by-side evaluations. The presented work is a part of continuous efforts to develop a new class of highly efficient and affordable virus removal filters.

BIOT 339

Investigation of particle size distribution for CHO cell culture media filtration

Takao Ito1, [email protected], Karen Chan2, Yusuke Tomioka1, Albert Cheong2, Yanglin Mok2. (1) Manufacturing Sciences & Technology, MilliporeSigma, Tokyo, Japan (2) Manufacturing Sciences & Technology, MilliporeSigma, Singapore, Singapore Mammalian cell culture media is a mixture of several components of variable range of complexities. Cell culture media used in CHO-based biologic processes are typically sterile filtered to prevent microbial contamination prior to inoculation. Good dissolution of powdered media is critical during the preparation of cell culture media as this will impact sterile filter capacity and cell culture performance in terms of viable cell density and final product titers in manufacturing. Sterile filtration capacity of one cell culture medium formulation can vary from lot to lot and it is understood that the variability is due to its different raw components and non- homogeneity during the hydration process. Although mixing is applied during media hydration, the impact of mixing on particle size distribution is not commonly monitored in process development. In this study, the impact of different mixing conditions on sterile filter capacity of cell culture media commonly used in the biologics industry were evaluated. Particle size distribution was determined to understand the impact of mixing to filtration. The key particle size range for optimal filter performance was found from statistical analysis using standard least squares model.

BIOT 340

Considerations for bulk fill filtration of high concentration drug substance

Thomas Parker, [email protected], Christopher Nieder. Technology Management, MilliporeSigma, Burlington, Massachusetts, United States

Protein therapeutics have traditionally been administered to the patient in a hospital setting; however, with new technologies in bioprocessing, such as high viscosity tangential flow filtration and single pass tangential flow filtration, high protein concentrations are achievable, which could result in a lower dose [by volume] for the patient and the ability to self-administer. With sterile filtration of bulk drug substance, material is at premium value and process control is critical, as the drug will next be filled then given to patients. As such, it is important for sterilizing grade filters to be integral and to give the highest possible product yield. Some of the potential challenges in sterile filtration of high protein concentration bulk drug substance are insoluble impurities [solubility] and molecular interactions at the surface of the sterile filtration membrane [polarization], which could lead to low filtration capacities, oversized filter installations, and low product yield. With the advancements in TFF and the demand for high concentration doses, the challenge of sterile filtration of high concentration MAb could become more common. Many process parameters contribute to the filterability of bulk drug substance, and this work explores these issues by presenting best practices for process control as well as experimental data. It is shown that a poorly designed process can have major issues on high concentration bulk drug substance filtration; however, with proper manufacturing considerations and process development, an economical and simple bulk fill filtration is possible.

BIOT 341 Clarification of a pre-treated mAb-producing harvest using new fully synthetic depth filter technology

Vasily Medvedev2, Youness Cherradi3, [email protected], Sarah Le Merdy1, Tiago Albano2. (1) Process Solutions, MilliporeSigma, Molsheim, France (2) Univercells, Gosselies, Belgium (3) Process Solutions, MilliporeSigma, Overijse, Belgium

Clarification is the first downstream processing step of mammalian cell cultures and is typically performed using depth filtration for its robust linear scalability, ease-of-use and performance. The use of pre-treatment on the harvest can prove to be very beneficial in terms of contaminant clearance but is usually also challenging for the following depth filtration step. The study presented here shows the very high performance of a newly developed synthetic depth filter towards a pre-treated CHO culture expressing a mAb targeting Middle East respiratory syndrome.

BIOT 342

Introduction and removal of β-glucans in a biological purification process: A case study

Adele Pearson1, [email protected], Robert Luo1, Kent E. Goklen2,1. (1) Downstream Process Development, GSK, King of Prussia, Pennsylvania, United States (2) MS UE 0550 PO Box 1539, GlaxoSmithKline, King of Prussia, Pennsylvania, United States

β-glucans are polysaccharides of D-glucose monomers linked by β-glycosidic bonds. Because they pose an immunogenicity risk to patients, drug products should have low levels of β-glucans. They can be introduced into a process through multiple sources: 1) host cell impurity from a yeast cell line; 2) component of yeastolate which may be an additive in production media; 3) leachable from cellulosic depth filters; and 4) potential contaminant in raw materials (e.g. sucrose). In this study, β-glucans were found at unexpectedly high levels in batches of the product of a mAb purification process; levels were higher than seen earlier in the program. The investigation that led to an understanding of the phenomenon and the change that caused it will be presented. In addition, the studies evaluating removal of β-glucans by filtration, including filter screening and sizing, will also be presented.

BIOT 343

Demonstrating robust clearance of beta-glucans in an E. coli purification process

Alexandra Morris, [email protected], Anh Nguyen Dang, Asif Ladiwala, Thanmaya Peram. Purification Development, Genentech, Pacifica, California, United States (1, 3)-β-D-Glucans (beta-glucans), a raw material-related process impurity, are of particular concern in ocular therapeutics due to the potential to illicit inflammatory response in the eye. A multi-pronged beta-glucan removal strategy was developed to demonstrate clearance to low levels. To develop a robust process to clear beta-glucans for an E. coli-derived biomolecule, incoming raw materials and downstream processing steps were assessed for beta-glucan impurities and the subsequent clearance. Bind- and-elute chromatography steps demonstrated clearance of beta-glucan introduced by upstream fermentation sources. Spiking studies were performed to show excess downstream process capability to allow for variation in starting impurity levels from incoming raw materials and operational variability. Additionally, nylon 6,6 membrane- based filtration steps were implemented at various points in the process to remove beta- glucans introduced by sources downstream of the chromatography clearance steps. Beta-glucan spiking studies were performed under worst-case conditions and filtration process parameters were optimized to achieve robust removal of beta-glucans. This holistic clearance strategy resulted in low levels of beta-glucans in the final drug substance.

BIOT 344

Optimization of a multi-level flow rate loading trajectory

Anton Sellberg, [email protected], Niklas Andersson, Anton Löfgren, Bernt Nilsson, [email protected]. Chemical Engineering, Lund University, Lund, Sweden

Chromatography is an important separation process for the biopharma industry. Proteins are selectively separated based on their physical and chemical properties. The main drawback of chromatography is the limited binding capacity of the stationary phase when operated at a constant flow rate. Chromatography capture steps are especially affected by the limited binding capacity since the capture steps need to be able to bind large amount of proteins while keeping the process yield and production rate at acceptable levels. Several engineering approaches, such as CaptureSMB or PCCC, have been developed to address these problems.

The aim of this contribution is to develop an advanced open-loop control approach to improve the capacity of a capture step. To increase the dynamic binding capacity of the column the flow rates used to load the column are chosen to maximize the key performance indicators (KPIs) of the process. By varying the flow rates more proteins can be bound during each capture cycle without having a shallow breakthrough curve. A mechanistic model is calibrated to model the binding of proteins to the stationary phase. The model is then used to optimize the flow rate trajectory used to operate the column.

Process yield, production rate and stationary phase utilization are three conflicting KPIs which together govern the performance of the process. To determine the overall performance of the process these three KPIs are combined to give a total performance indicator. Different flow rate trajectories are obtained depending on the weight assigned to each KPI. This is visualized by a multidimensional pareto front. The flow rate trajectory gives a 10% performance increase, compared to a single flow rate, in regions where process yield have been assigned a large weight.

BIOT 345

Comparison of charge variant methodologies for evaluating protein compatibility with dextrose infusion solutions

Bahar Demirdirek, [email protected], Joseph Valente, Wenkui Lan, Mark Bolgar. Bristol-Myers Squibb, Highland Park, New Jersey, United States

Intravenous (IV) administration of therapeutic proteins may include dilution of the final formulated product into infusion media such as dextrose, 5% in water (D5W) or normal saline. Protein stability must then be evaluated over the intended use-time period of the resultant infusion solution. Matrix effects of D5W require careful consideration since they can impact the stability profile of the protein and have variable effects across different analytical methods. As an example, dextrose (D-glucose) can rapidly react with proteins to form reversible schiff base glycation adducts (Aldimine) that subsequently covert to a more stable glycation products (Ketoamine) over time. Both glycation forms contribute to the charge variant profile of a protein since both are acidic variants compared to the unmodified protein. However, the sensitivity of the reversible glycation adduct to changes in dextrose concentration is often overlooked during analysis, leading to potentially inaccurate results and an incomplete understanding of infusion vehicle compatibility. To demonstrate this point, we monitored the charge variant profiles of several different proteins diluted in D5W using imaged capillary isoelectric focusing (iCIEF), Capillary Zone Electrophoresis (CZE), and cation exchange chromatography (CEX), and found that changes in acidic variants were highly dependent on the selected technique. Furthermore, we have performed force degradation study and exposed the proteins to D5W for up to 15 days. The samples were analyzed by all three techniques before and after offline buffer exchange. We believe these studies highlight the importance of analytical technique selection for investigating protein sensitivity to dextrose and for monitoring protein stability in dextrose containing infusion solutions.

BIOT 346

Use of high-throughput screening technology to develop flow through aggregate removal platforms

Caroline Kittinger1, [email protected], Benjamin D. Gastfriend2, Kelcy Newell1, Ellen O'Connor1, Valeria Riguero1. (1) MedImmune, Gaithersburg, Maryland, United States (2) MedImmune, Gaithersburn, Maryland, United States

Aggregate removal is a challenge facing industrial downstream process development. Typical methods of aggregate removal utilize chromatographic resins in bind and elute mode. These methods are time consuming, labor intensive, and molecule specific. By using an automated robotic liquid handling system to perform high throughput screening, our group tested over one hundred chromatographic resins at various pHs and conductivities to develop aggregate clearing flow through operations for monoclonal and bispecific antibodies. An initial filter plate screening was used to narrow conditions that best removed aggregate. The filter plate results were then scaled-up to 600 µl micro-columns to test dynamic binding and to further optimize run conditions. Results from the micro-columns were then applied to the larger gram scale and tested against multiple molecules of the same sub-class. This screening process from plate to gram scale took one month per molecule, while equivalent experiments performed without the automation technology would have taken over one year. We discovered several flow through conditions that cleared aggregate for each molecular format. Monoclonal antibody aggregate was best removed by particular CHT and Capto adhere conditions, while the bispecific aggregate was favorably cleared by CHT, Poros 50 PI, and Toyopearl NH2. The results of this study has impact to industrial process development from early reagent generation to manufacturing. High purity material can be generated in more efficient ways than traditional bind and elute chromatography.

BIOT 347

Centre-cut separation of biomolecules by stepwise-elution SMB

Chris Wayne, [email protected], Ajoy Velayudhan. Biochemical Engineering, University College London, London, United Kingdom High throughput and low usage of liquid and stationary phases has resulted in the wide use of Simulated Moving Bed (SMB) chromatography in the chemical, food, and pharmaceutical industry, e.g., to produce high-fructose corn syrup and pharmaceutical enantiomers. In the bioprocess area, it has also been demonstrated for the capture chromatography of monoclonal antibodies. However, its application to the polishing chromatography steps of biomacromolecules, in which resolution is essential, has been extremely limited. Polishing chromatography requires gradient or sequential stepwise elution to cope with the ‘all or nothing’ adsorption behaviour of macromolecules. These changes in salt (or other modulator) complicate the design process of a polishing SMB considerably. In this poster, two simple methods of achieving a fully-continuous ternary stepwise-elution SMB separation are contrasted. These two approaches to ternary separations are experimentally demonstrated for the centre-cut separation of ovalbumin variants. This extremely difficult separation is a good representation of product variants in industrially relevant separations. Furthermore, simulation data is presented to highlight the relative advantages and disadvantages of the two approaches to centre-cut stepwise elution SMB.

BIOT 348

Manipulating the selectivity of ceramic hydroxyapatite (CHT) for acidic protein purification

Louisa Vang, [email protected], Xuemei He. BioRad, Hercules, California, United States

Hydroxyapatite is a naturally occurring mixed-mode chromatography media with unique selectivity. Its negatively charged phosphoryl groups (the P sites) interact with the amino residues on protein surface electrostatically, while the calcium ions in the crystal lattice (C sites) form chelation complexes with the acidic ligands on various biological molecules. Ceramic hydroxyapatite (CHT) is the hydrodynamically stable form of hydroxyapatite, which has been widely used for the manufacturing of protein therapeutics in the past three decades.

The purification of acidic proteins with CHT chromatography has been challenging, as the charge characteristics of target proteins are very similar to impurities such as acidic proteins, nucleic acids, and endotoxins from the expression host. We have found that the concentrations of Ca2+ and inorganic phosphate in buffers have dramatic effects on the selectivity of CHT chromatography. By manipulating buffer composition and pH, we were able to modulate the interaction between the target protein molecule and CHT. An acidic recombinant enzyme, CDP-D-Glucose 4, 6-Dehydratase, was therefore purified from the E. coli extract in either flow-through or bind-and-elute mode. The acidic impurity clearance efficiencies of these two chromatography approaches are compared and presented herein.

BIOT 349 Utilizing single use (SU) technology for single pass (SP) TFF operations

Emily Peterson, [email protected]. MSAT, MilliporeSigma, Pepperell, Massachusetts, United States

Single Pass TFF (SPTFF) is an enabling technology for continuous manufacturing. This poster highlights the use of a new holderless single use (SU) TFF capsule to achieve high concentrations in single pass (SP) mode and compares it to the performance of a flat sheet device of the same molecular weight cutoff. This technology is particularly useful in locations where space is limited and/or process time is critical as it eliminates the need for pre/post cleaning while minimizing initial water flush.

BIOT 350

Design and integration of in situ liquid liquid extraction systems for bioacid production

Patrick Saboe2, Lorenz Manker2, Gregg Beckham1, Eric Karp2, [email protected]. (1) NREL, Golden, Colorado, United States (2) National Bioenergy center, National Renewable Energy Laboratory, Golden, Colorado, United States

The economics of a bioprocess are generally driven by the expense of the downstream separation of product from dilute aqueous streams. Continuous separation techniques, such as in situ product recovery (ISPR), are attractive in that they can simultaneously remove and concentrate products from bioreactors, thereby minimizing product inhibition and downstream separation cost. Although there are multiple benefits of ISPR, there are few reports on how to integrate these systems with further downstream processes to recover highly pure product, which is essential for many applications, such as polymer and commodity chemical production. This seminar will cover a liquid-liquid extraction (LLE) based ISPR system integrated with downstream distillation that our group has developed to effectively purify carboxylic acids from dilute aqueous fermentation broth. Discussion will focus on experimental results including measured equilibrium constants of various extractants, a mathematical model describing extraction physics, and an ASPENplus process simulation. The combined work provides a framework that can be used to predict energy input requirements and critical fermentation targets such as pH and titer that must be met in order for this ISPR system to be viable. Overall, our LLE based ISPR system integrated with distillation may provide an approach to 1) enable industrial deployment of new bioprocesses by decreasing industrially targets for titers and 2) process intensification of existing industrial operations via decreased separation costs and increased productivities.

BIOT 351

Di-block copolymer and enzyme conjugated nano layers: A packing density study to design bio-reactive membranes for water purification applications

Jessika Pazol1, [email protected], Eduardo Nicolau2. (1) Chemistry, University of Puerto Rico, Rio Grande, Puerto Rico, United States (2) Chemistry, University of Puerto Rico Rio Piedras, San Juan, Puerto Rico, United States

The proposed study seeks to understand the fundamental aspects related to the packing density of a porous polymeric active nano layer. It is important to create novel membranes for sustainable water treatment as an area of research particularly to mitigate contamination from bacterial origin. In pursuit of that vision, we foster path- breaking scientific investigations via working with bio reactive layers in membrane separation sciences. The porous nano layer results from the self-assembly characteristics of polystyrene-b-poly(4 vinyl pyridine) (PS-b-P4VP) block copolymer (BCP) via two distinct methods. The PS-b-P4VP forms supramolecular systems that had been previously studied in the bulk. More recently, the complexes of this system to produce porous layers has been investigated via phase inversion and as thin films via spin coating technique . Moreover, the use of a small molecule moiety consisting in this case of meta 3-alkyl phenol had been used to understand the conjugated polymer nano layer surface structure and further pores formation. This present work aims to understand the design of ultra-thins conjugated polymeric layers that can exhibit enhanced polymer conjugation ratio properties. The surface morphology will be examined using, among others, atomic force microscopy (AFM) and scanning electron microscopy (SEM) to evaluate morphological information such as pore size and thickness. The prepared films will serve as a model to develop further reactive layer membranes for water purification application. The details for this investigation regarding the methods and analytical techniques will be presented.

BIOT 352

Case study for evaluation of the 3M™ Emphaze™ AEX hybrid purifier involving low isoelectric point target proteins

Joseph Pate1, [email protected], Benjamin Kester1, Claudia Berdugo1, Yunsong Li1, Victor Vinci2. (1) Process Development, Cook Pharmica, Bloomington, Indiana, United States (2) Cook Pharmica, Bloomington, Indiana, United States

The 3M™ Emphaze™ AEX Hybrid Purifier is a novel purification tool that utilizes a Q- functional hydrogel and 0.2 µm membrane to reduce impurities in a biopharmaceutical downstream process. This technology has been reported to provide a 4-log reduction in DNA at the clarification stage, which can greatly improve the efficiency of the Protein A capture step. While the advantages provided by the Emphaze™ AEX Hybrid Purifier are far-reaching, its Q-functionality presents a challenge for target proteins with a low isoelectric point (pI). Proteins under these conditions will usually bind to anion exchange ligands which can negatively impact the functionality of a clarification stage filter. This study involved a Design of Experiment (DoE) based approach to investigate the effects of pH and conductivity on the turbidity, yield, and impurity clearance of the Emphaze™ AEX Hybrid Purifier and the efficiency of the subsequent Protein A capture step. The results were then used to determine optimum filtration conditions of one such low pI fusion protein, and whether elements of this technique can be extended to purification of a monoclonal antibody (mAb) with a higher pI. Filtration conditions for the low pI target protein were identified with a 4-fold greater yield over normal harvest filtration conditions and more than 99.6% removal of HCP post Protein A chromatography. Application of these strategic elements to mAb filtration demonstrated an extended pH range at which post Protein A HCP reduction exceeds 99.9%. For each molecule studied, filtration using the Emphaze™ AEX Hybrid Purifier resulted in greater HCP reduction, higher product purity, and lower turbidity than unfiltered pools. This experimental plan provides a baseline to tailor specific application of the Emphaze™ AEX Hybrid Purifier to each individual protein or antibody.

BIOT 353

Case study: Commercial-scale microbial control validation during centrifuge harvest operations

Karl Saldanha, [email protected]. Process Engineering, Genentech, South San Francisco, California, United States Microbial control during the biologics manufacturing process remains a key Health Authority focus area. In recent years, this expectation has expanded to include demonstration of acceptable levels of control pre-sterilizing grade filtration throughout the recovery and purification process. This is primarily due to the potential for harmful agents (e.g. endotoxins, exotoxins, proteases) resulting from the presence of bioburden in the manufacturing process to adversely impact product quality or patient safety. In this presentation, a case study will be shared based on efforts to validate microbial control at large-scale throughout the (centrifuge-based) harvest operation, the first recovery step following recombinant protein product formation. Study design included evaluation of microbial control during equipment setup, pre-use hold times (to support increased manufacturing flexibility), as well as throughout cell culture fluid (CCF) processing utilizing a worst-case conditions approach. A summary of findings as well as subsequent investigative efforts to more comprehensively understand the system will be provided. In addition, resulting best-practice GMP changes implemented as a result, along with evaluation and considerations for larger-scope continuous improvement initiatives will also be discussed. The key findings and lessons learned from this study highlight challenges for legacy products in which equipment capabilities and historical practices at an aging facility may need to be evaluated on an ongoing basis in order to align with present-day regulatory expectations. However, it also presents an opportunity and motivation to explore newer technologies (e.g. disposables) that enable closed- system processing for future product version upgrades as well as those currently under development within the clinical pipeline.

BIOT 354

Implementation of disposable technologies in support of closed processing and viral segregation for downstream clinical manufacturing

Kimberly-Anne Mattia4, [email protected], Be Vu3, Phillip R. Smith2, Antonio R. Ubiera1. (3) GlaxoSmithKline, King Of Prussia, Pennsylvania, United States (4) GlaxoSmithKline, King of Prussia, Pennsylvania, United States

Disposable technologies offer flexibility, potential improvements in productivity, and enhancements in adventitious agent control for downstream processing. Nevertheless, implementation of such technologies in a clinical manufacturing setting comes with a unique set of challenges, including the potential for extractables and leachables and design of systems to enable closed processing. In this context, a case study on the technical evaluation and introduction of several commercially-available disposable process technologies for clinical manufacturing will be presented. The rationale for selection of each technology, details regarding design and customization, and operational experiences gained in the manufacture of clinical trial material will be described. Additionally, the utility of these technologies for enhancing control of adventitious agents (e.g., through viral segregation) during downstream processing will be discussed.

BIOT 355 Reoxidation of reduced monoclonal antibody using modified downstream processing

Margaret Moynihan, [email protected], Joann Lutje, Pedro J. Alfonso, Pankaj Singh. Large Molecule API, Janssen Pharmaceuticals, Malvern, Pennsylvania, United States

Reduction of disulfide bonds in monoclonal antibodies (mAbs) has been observed in several cases during the scale-up of Chinese Hamster Ovary cell-based products. Literature has linked this reduction to several factors, including cell shearing, temperature, dissolved oxygen content, media composition, reductive enzymes, and harvest duration. In this case study, a pilot scale run resulted in reduced IgG4 mAb product detected by non-reduced gel electrophoresis. This was not seen in reduced scale bioreactors nor in other product quality assays. Replicating this failed batch using the reduced scale model was problematic and required harsh reducing harvest conditions. Additionally, reduced product was challenging to identify with traditional in- process testing, requiring more refined and controlled capillary sodium dodecyl sulfate gel electrophoresis procedure. After successfully reproducing the reduced mAb, process intermediates, including clarified harvest and Protein-A chromatography eluate, were treated and reoxidized from 15% to greater than 95% intact mAb. Furthermore, the reconstituted mAbs maintained the desired product quality attributes of the reference material for control reactors. This work describes small scale time-dependent methods to reduce mAb intrachain disulfide bonds during harvest followed by reoxidation during downstream processing without modification of media or addition of chemical reagents to cell harvest.

BIOT 356

Antisense oligonucleotide purification process: Successes and challenges during scale-up

Robert Gronke, [email protected], Ratnesh Joshi, Hien Nguyen, Yannick Fillon, Kris Ruanjaikaen, Firoz Antia. Technical Development, Biogen, Cambridge, Massachusetts, United States

Biogen's antisense oligonucleotide (ASO) development and manufacturing has been, until recently, carried out at our partner Ionis Pharmaceuticals (e.g., Spinraza). In 2015, this new modality was established within Biogen’ Technical Development group to support the growing ASO portfolio. In 2016, a development lab was established in Cambridge MA for making ASOs at both lab (1/100th) and pilot scale (1/10th MFG scale) using our new platform process. In the fall of 2017, our first full scale GMP batches for the lead clinical development compound was manufactured in the newly built synthesis suite within the oral solid manufacturing facility in North Carolina. A four-step purification process was then carried out in the existing nearby flexible volume manufacturing facility that, up until this point, has been used for manufacturing Biogen’s protein-based parenterals. The ready- to-fill drug substance was then further processed into drug product vials. This was our first test case to demonstrate that Biogen can manufacture ASOs safely, at scale, and achieve high purity and yield. Results will be presented on the scalability of the ASO process from bench to GMP scale, highlighting successes and challenges faced with scale-up of the downstream ASO process.

BIOT 357

Downstream process development and transfer to cGMP manufacturing of novel fusion protein at CMC biologics

Sarah C. Vilardi, [email protected]. Downstream Process Development, CMC Biologics, Inc., Lynnwood, Washington, United States

In the last decade, a new outlook on cancer treatment therapies has emerged thanks to advances in immuno-oncology. By engaging a patient’s own T-cells, scientists have developed novel ways in which the patient’s immune system is the primary source of cancer-fighting material. One such method uses a class of proprietary fusion protein that interacts with tumor malignancies while engaging bypassing T cells, and thus destroying the target tumor.

CMC Biologics is currently in contract to develop and optimize production processes for this protein therapeutic platform at bench scale for transfer and scale-up manufacturing in CMC’s cGMP facility. The downstream purification process of the fusion protein uses a MabSelect Protein-A capture column, low pH viral inactivation, a cation-exchange (CEX) polishing column, an anion-exchange (AEX) polishing column, viral filtration, and ultrafiltration/diafiltration (UF/DF) of the final bulk material. 500-L cGMP production of the first (of four) proteins begin in Q1 of 2018 in Berkeley, CA.

BIOT 358

Use safe light during biologics manufacturing, handling and storage

Cheng Du4, [email protected], Ameya Borwankar3, Angela T. Lewandowski4, Nripen Singh2, Sanchayita Ghose1, Michael C. Borys1, Zheng Jian Li4. (1) Bristol Myers Squibb, Hopkinton, Massachusetts, United States (2) Process Development, Bristol Myers Squibb, Acton, Massachusetts, United States (3) Bristol-Myers Squibb, Devens, Massachusetts, United States (4) Process Development, Bristol-Myers Squibb, Hopkinton, Massachusetts, United States

As macromolecules, biologics are susceptible to light exposure, which induces oxidation of multiple amino acid residues including tryptophan, tyrosine, phenylalanine, cysteine and methionine. Pertaining to safety, efficacy and potency, light-induced oxidation of biologics has been widely studied and necessary precautions need to be taken during biologics manufacturing process, drug substance and products handling and storage. Proteins will degrade to varying extents depending on the protein properties, degradation pathways, formulation compositions and type of light source. In addition to UV light, which has been widely known to degrade proteins, visible light also can mediate protein degradation. Here we examine and identify wavelengths in the visual spectrum (400–700 nm) that can cause monoclonal antibody and histidine buffer degradation. Installation of safe lights which exclude the identified damaging wavelengths from visible spectra in manufacturing and storage areas can provide a balance between lighting requirement for human operators and their safety and conservation of product quality.

BIOT 359

Performance of simultaneous single-pass concentration and diafiltration of protein solution using Pall CadenceTM inline diafiltration module

Blaine Trafton, [email protected], Cameron Matthews, [email protected]. Applications Research and Development, Pall Life Sciences, Westborough, Massachusetts, United States

Concentration and diafiltration of proteins are an essential part of any monoclonal antibody downstream process. They are of particular importance when used as part of a continuous process, as they can be used to intensify chromatography steps as well as formulate bulk drug substance. However, continuous concentration and diafiltration can be complex to implement, requiring multiple filters all needing independent equipment, controls, and monitoring. A method was developed to allow the Cadence™ Inline Diafiltration (ILDF) module to simultaneously concentrate and diafilter a protein solution continuously while maintaining a stable operating pressure without the use of additional filters. A mAb solution was fed into an ILDF module while diafiltration buffer was supplied to the six buffer inlets and the retentate flow restricted to half the feed flow, creating a 2x concentration factor. This 2x concentration serves to pre-treat the membrane surface with a gel-layer of protein that will protect the membrane from fouling at higher concentration factors. After the 2x concentration pre-treatment, the module was able to concentrate the protein solution ≥10x for eight hours, while maintaining three-log buffer exchange (99.9%) from the diafiltration buffer. By simplifying these two unit operations and reducing them to a single filtration operation, the Cadence ILDF module can be integrated as part of any unit operation requiring continuous concentration and/or buffer exchange. These may include pre-protein A capture, between multiple ion exchange chromatography steps, and the final formulation step.

BIOT 360

Accelerating advancement in gene therapy by improving downstream purification of viral vectors

Orjana Terova, [email protected]. Thermo Fisher Scientific, Bedford, Massachusetts, United States The area of gene therapy has been of growing interest due to its potential of correcting the genetic cause of the disease rather than chronically treating symptoms. Viral vectors have shown to be most successful in delivering therapeutic genetic material into target cells and the adeno associated virus (AAV) sub-classes have emerged as the vector of choice for many therapies.

The lack of large scale manufacturing platform technologies for viral vector purification is one of the major challenges the field faces. In order to increase productivity and meet market needs, efficient commercial manufacturing capabilities need to get established. By reducing the number of purification steps and maximizing productivity, affinity chromatography already offers a significant improvement to the downstream process of biomolecules. This technology offers scalability and process consistency, thereby providing a platform solution to the industry.

Herein we outline the benefits of implementing affinity chromatography as platform in the downstream purification of viral vectors, using the CaptureSelect™ technology as the basis of generating high-binding affinity ligands. The methods described will reveal the benefits of affinity chromatography related to specificity, capacity, process yields as well as process scalability for the purification of AAV viral vectors. Specifically, the properties of POROS™ CaptureSelect AAV8, AAV9, and AAVX affinity resins will be discussed as tools to capture and purify a broad range of AAV serotypes. The utilization of these affinity resins will be a significant improvement to the downstream processing, by reducing the purification steps and maximizing productivity, offering scalability and processing consistency.

BIOT 361

Effect of mRNA structure on chromatographic performance

Michael Shamashkin, [email protected], Matthew Scott, Hari Pujar. Moderna Therapeutics, Cambridge, Massachusetts, United States

Modified synthetic messenger RNA (mRNA) represents a new class of biopharmaceuticals currently being tested in clinical trials as vaccines and therapeutics. Unlike therapeutic proteins, which must be expressed in and purified from cell cultures, therapeutic mRNAs are made enzymatically by in vitro transcription from linear DNA templates and then subjected to downstream purification procedures to remove process impurities. During the downstream processing, mRNA undergoes significant structural changes, as Watson-Crick base pairing is affected by both temperature and ionic strength. Denaturation or compaction change the size exclusion behavior, affecting the binding capacity to chromatographic resins. Under certain process conditions and during storage of process intermediates, some mRNA constructs form dimers and higher order structures, as confirmed by SE-HPLC, DLS, and agarose gels. These oligomers were isolated by preparative SEC and shown to be stable at room temperature. Due to the very large size of the mRNA oligomers (> 1 MDa), the binding capacity of porous adsorbents is severely impacted. In addition, the contribution of these dimers and oligomers to product heterogeneity is perceived as a quality issue. Here we show that mRNA oligomerization is fully reversible with a mild heat treatment, which fully restores the monomeric form and the original binding capacity to porous absorbents.

BIOT 362

Impact of multimodal cation-exchange chromatography on impurity separation of non-platform proteins

Engin Ayturk, [email protected], Keith Selvitelli, Matthew Westoby. Biogen, Cambridge, Massachusetts, United States

Unique selectivity of mixed mode chromatography resins has been widely demonstrated to provide noticeable process advantages over traditional chromatographic media. In addition, the hydrophobic nature of mixed-mode resins has been utilized for aggregate removal as a viable polishing step in many post-proteinA platform processes. For non- platform molecules, process development challenges are multi-faceted and often require orthogonal methodologies, especially for the handling of difficult-to-remove impurities. In this context, the effects of hydrophobic and charged moieties on multimodal cation-exchange ligands were examined by studying protein retention behavior on commercially available media, such as CaptoMMC, CaptoMMC ImpRes, Nuvia cPrime and CMM Hypercel. A novel therapeutic protein and an enriched dimer version of the protein were employed to characterize protein – ligand interactions. Binding of these proteins was evaluated as a function of feed, salt concentration and pH in the presence of various mobile phase modulators. Information assembled was applied towards identifying operational conditions that enable sufficient aggregate and impurity removal. In addition to the intermediate stability and virus clearance studies, a polishing step was designed, developed and implemented leading to a successful GMP manufacturing campaign.

BIOT 363

Virus detection using restricted-access adsorbents

Ujwal Patil2, [email protected], Sagar P. Dhamane3,2, Meena Adhikari2, Anna Hagstrom1,5, Ulrich Strych2,4, Katerina Kourentzi5, Richard C. Willson5,2. (1) Purification Process Sciences, MedImmune, CLARKSVILLE, Texas, United States (2) Biology and Biochemistry, University of Houston, Houston, Texas, United States (3) Drug Substance Technology Engineering, Amgen, Thousand Oaks, California, United States (4) Pediatrics-Tropical Medicine, Baylor College of Medicine, Houston, Texas, United States (5) Chemical and Biomolecular Engineering, University of Houston, Houston, Texas, United States

Purification of viruses can take advantage of their large size compared to most contaminants of interest. A novel chromatographic technique for virus purification is supported by media such as Capto Core 700TM, which combines an inert size-selective surface shell with an exclusion limit of 700 kDa, which excludes viruses, with a multimodal octylamine ligand core which captures most proteins. In this work, we have adapted this exclusion technology to virus detection and demonstrated the method using M13 bacteriophage as a model virus. We added enzyme-labeled target-specific antibodies to samples with/without M13 and separated the enzyme-labeled M13 particles on a Capto column. In the absence of M13 target, the antibody-enzyme conjugate is small enough to enter the core and be adsorbed to undetectable levels. In the presence of M13, antibody-enzyme conjugate bound to the virus is excluded from adsorption and is detected in the eluate.

BIOT 364

Characterization of flow mechanics in a scaled down continuous viral inactivation reactor

Matthew Brown1, [email protected], Linus Amarikwa3, Stephanie Parker3, Samet S. Yildirim1, Scott Godfrey1, Raquel Orozco1, Jonathan L. Coffman2. (1) Boehringer Ingelheim, Stanford, California, United States (2) Process Science, Boehringer Ingelheim, Fremont, California, United States (3) Keck Graduate Institute, Claremont, California, United States

Continuous Bioprocessing continues to be at the forefront of therapeutic protein manufacturing innovation. In addition to many challenges facing this technology, appropriate scaled down models are still in development. Continuous low pH viral inactivation remains a significant bottleneck when coupling post Protein A elution to polishing steps. Continuous viral inactivation, utilizing plug flow methodology, converts what was a trivial batch unit operation with uniform static composition into a diverse dynamic system. This study aims at manipulating relevant operational parameters in an attempt to understand their fundamental influence on flow properties and fluid dynamics to enlighten scaled down methodologies.

BIOT 365

Straight-through processing of insulin-like peptides

Peter Tiainen2, [email protected], Anton Löfgren1, Bernt Nilsson1. (1) Chemical Engineering, Lund University, Lund, Sweden (2) Downstream Technologies, Novo Nordisk A/S, Måløv, Denmark

In this poster a flexible and reliable end-to-end purification platform is presented. To demonstrate the versatility of the platform, a process was put up including not only chromatography steps but also a vital enzymatic cleavage reaction. The platform allowed processing of several variants of the insulin-like peptide proving its use in late- stage preclinical research where larger amounts of the variants are needed yet still the format/sequence allows for slight modifications.

The automated and closed setting decreases the need for manual attendance and secures better quality while increasing the throughput of peptides/proteins in our purification labs. To increase the platforms flexibility and robustness even further a call for good and reliable, preferably non-invasive, in-line analytical technologies is made.

A schematic picture of the end-to-end process including chromatography steps together with an enzymatic cleavage reaction

BIOT 366

Surface characterization and identification of optimum mixing conditions for protein solutions

Adam Sokolnicki1, Gayathri Raja1, [email protected], Siyi Liao2. (1) Manufacturing Science and Technology, MilliporeSigma, Billerica, Massachusetts, United States (2) Manufacturing Science and Technology, Merck KgaA, Shanghai, China

Current trends in the bioprocessing industry are driving companies that manufacture mAb and other protein solutions to be mixed at optimum conditions, minimizing the risk of shear and potential degradation. Any air that remains entrapped in a finished mixture can cause structural problems, clouding, discoloration, voids, instability and other undesired qualities depending on the product and end use. Air incorporation was the subject of this study to evaluate and characterize mixing in the Mobius® PowerMIX family. Foam and bubbles arising were quantified through visual observation. Results of the study included generating heat maps to identify mixing conditions that are most gentle to proteins.

BIOT 367

Analytical comparability strategies for biosimilars

Anna Schwendeman, [email protected]. Pharmaceutical Sciences, University of Michigan, Ann Arbor, Michigan, United States Biosimilars promise significant cost savings for patients, but the unavoidable differences between innovator and copycat biologics raise questions regarding product interchangeability. In this study, Remicade and Remsima were examined by native mass spectrometry, ion mobility, and quantitative peptide mapping. The levels of oxidation, deamidation, and mutation of individual amino acids were remarkably similar. We found different levels of C-terminal truncation, soluble protein aggregates, and glycation that all likely have a limited clinical impact. Importantly, we identified more than 25 glycoforms for each product and observed glycoform population differences, with afucosylated glycans accounting for19.7% of Remicade and 13.2% of Remsima glycoforms, which translated into a 2-fold reduction in the level of FcγIIIa receptor binding for Remsima. The same analytical workflows have been applied to characterize other biosimilar mAbs and their reference products. Our mas spectrometry-based analysis provides rapid and robust analytical information vital for biosimilar development. We have demonstrated the utility of our multiple-attribute monitoring workflow using the model mAbs Remicade and Remsima and have provided a template for analysis of future mAb biosimilars.

BIOT 368

Mechanistic understanding of the degassing phenomenon of drug products

Fabrice Schlegel1, [email protected], Chakradhar Padala2. (1) Process Development, Amgen, Cambridge, Massachusetts, United States (2) Process Development, Amgen, Thousand Oaks, California, United States

During the manufacturing process, the drug product often undergoes a hold step for several days. The standard practice is to perform these holds under positive pressure and low temperature to minimize the risk of microbial growth during hold. During filling step, the product comes back to atmospheric pressure and ambient temperature, triggering degassing of the solution. Time pressure filling is observed to be sensitive to degassing due to bubble formation in the tubing. A conditioning step can be implemented as a mitigation prior to filling. The duration of the conditioning step, however, is dependent on the duration and conditions of hold. As a result, there is a need to better understand the degassing phenomenon and to determine the duration of the conditioning step required to achieve equilibration.

A numerical model based on first principles and the finite element method has been implemented to predict the impact of pressure and temperature conditions on drug product degassing. The model solves for the conservation of mass, momentum and energy in tanks and provides the distribution of dissolved O2 and N2 in the product as a function of time.

The model is validated with respect to experimental measurements using water as a drug product. The numerical and experimental results show a great match at batch sizes ranging from 5L to 30L. BIOT 369

Adalimumab like biosimilar purification with ion exchange and mixed-mode resins

Randy Drevland, [email protected]. Bio-Rad Laboratories, Hercules, California, United States

The global biosimilars market is exploding and is expected to hit 10.9 billion USD by 2021. We attempted to purify an Adalimumab like biosimilar with a workflow comprising of ion exchange and mixed mode resins. Using the ion exchanger at the capture step helped overcome the limitations of Protein A capture (low binding capacity and protein leaching). Comparable contaminant removal (HCP and HCDNA) was obtained with parallel workflows consisting of affinity (Protein A) and cation exchange (Nuvia S) capture purification, followed by two ion exchange polishing steps, resulting in greater than 99% purity with both workflows. However, the ion exchange based workflow leads to better process economics, which is an underlying requirement of a biosimilar.

BIOT 370

Effect of low pH transition wash on protein A eluate-mitigation strategy to improve manufacturing filterability

Javier Huerta2, [email protected], Weitong Sun1, Irina Ramos2. (1) BioProcess Engineering, Medimmune, Gaithersburg, Maryland, United States (2) Purification Process Sciences, Medimmune, Gaithersburg, Maryland, United States

A common issue observed for the capture of monoclonal antibodies (mAbs) with protein A resin is the transient generation of high turbidity during product elution. This turbidity is historically attributed to acid induced precipitation of pH sensitive molecules, process- related impurities (e.g. HCP and/or DNA), or liquid-liquid phase separation. The high turbidity can lead to challenges with increased system pressure during elution caused by poor in-line eluate filter capacity. Alternatively, eliminating the in-line eluate filter and instead filtering a pooled protein A eluate might require an additional intermediate vessel and might raise concerns about bioburden control of the process intermediates. In this study, we explored using a low pH wash prior to elution to effectively reduce the turbidity seen in the eluate while maintaining an acceptably high yield. We also investigate the impact of the protein A wash options on product turbidity during the subsequent low pH viral inactivation step. The protein A pH transition wash has a positive result in reducing turbidity during elution and during the viral inactivation step. These results provide insight into implementing a strategy that enables in-line filtration for protein A chromatography and improves filterability during viral inactivation. The overall outcome is a more consistent and predictable large scale manufacturing performance.

BIOT 371 Comparability, host cell protein and facility fit challenges in the upgrade and modernization of the purification process of a non-platform commercially licensed biologics product

William M. O 'Dwyer, [email protected]. Purification development, Genentech Inc., South San Francisco, California, United States

This presentation will describe the challenges encountered and subsequent purification steps developed and implemented into a process version upgrade for a licensed biologics product. Strategic business, quality and safety drivers such as the removal of animal derived hydrolysates and the implementation of modern equipment and technologies, led to the optimization of a decades-old cell culture and purification processes. Novel purification steps were developed and implemented to address facility fit and product comparability challenges. Specifically, parvovirus filtration and chromatography were performed as a linked step, and a separate pool hold step was implemented to ensure mean centering of a product attribute correlated to potency. An additional challenge emerged during large-scale production when a specific host cell protein (HCP) was detected in process pools. An assessment of the purification process led to the identification and implementation of a modification to the capture chromatography step to enhance reduction of this specific HCP. In addition, a second unique orthogonal purification step was developed and implemented which utilized precipitation to remove the HCP while no impacting any other product attributes. Ultimately, a purification process was developed that updated the process to current standards, employed updated viral inactivation and removal technologies, and optimized facility fit. The process was streamlined to reduce manufacturing complexity while achieving process-related impurity removal comparable to that of the original manufacturing process.

BIOT 372

Single use centrifuge (UniFuge) for large and small scale collection,concentration and clarification of mammalian cells, virus fragments and Mabs

David Richardson, [email protected]. SINGLE USE , PSA, El Cajon, California, United States

Advances in cell mammalian cell culture processing techniques have evolved from traditional centrifugation. The Industry has driven these improvements. It has been demonstrated the UniFuge will speed up, close processes, reduce capital equipment cost, and increase titers. The data in this presentation supports the justification of replacing the traditional centrifuges, with a large scale single use centrifuge. The Single use tube bowl centrifuge is named the UniFuge® and has been commercially in use for nearly a decade. The Single use centrifuge, UniFuge® manufactured by Pneumatic Scale Angelus in Clearwater Florida. This presentation will present case studies of infected MRC 5, HEK cells, CHO, Stem cells. The presentation will be compare and contrast various platforms :filtration, disk, rotor and fluidized bed separation devices. The data suggests the UniFuge® as having quicker processing times, while lowering operation costs. In addition, achieving higher cell recovery, lowering NTU centrate values, and increasing cell collection concentrations.

BIOT 373

Development of cost-efficient processes for adenovirus purification

Magnus Bergman, [email protected], Mia Bennemo, Åsa Hagner-McWhirter, Anna Akerblom, Sara Häggblad-Sahlberg, Elisabeth Wallby, Gustaf Ahlén, Mats Lundgren. GE Healthcare, Uppsala, Sweden

Viral vectors such as adenovirus, adeno-associated virus, and lentivirus can be used in various therapeutic and vaccine applications, and are expected to become high-value products. As such, there is an increasing demand for fast, reliable, and cost-efficient solutions for viral vector production. While obtaining a safe and efficacious product should always be the top priority of any process development project, it is also important to include considerations regarding cost-efficiency early in the development process to help find a good balance between technical performance and economic feasibility.

Here, we present a project where multiple approaches for harvest and purification of adenoviruses have been investigated. For each unit operation, modern technologies and solutions have been evaluated against more traditional alternatives in terms of purity, yield, and productivity. In the end, two updated processes based on chromatography were selected for purification of adenovirus: one using modern resins, and the other using a combination of membrane adsorber and resin technologies.

At project start, potential processes were modeled using the BioSolve™ process economic simulation tool (BioPharm Services) to identify process alternatives that were unfavorable from a process economy perspective, and as such, could be excluded from the development work. Throughout the development process, the model was improved and subsequently used to provide decision support. Lastly, the potential benefits and drawbacks of the two selected processes, in relation to the more traditional process alternative, were investigated in various production scenarios.

BIOT 374

High-throughput droplet tracking and automated image analysis in microfluidic device

Manibarathi Vaithiyanathan, [email protected], Khashayar R Bajgiran, Pragathi Darapaneni, James A. Dorman, Adam T. Melvin. Cain Department of Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana, United States Molecularly-targeted therapeutics and personalized medicine have dramatically increased the prognosis of patients suffering from cancer. However, tumor heterogeneity has limited the success of single drug treatment which led to the introduction of multiple therapeutic combinations. As such, multiple therapeutic combinations have required the development of new analytical methods capable of multiplexed high-throughput screening (HTS) technologies capable of examining both concentration and number of therapeutics. Droplet microfluidic devices has garnered significant interest to facilitate high-throughput, single cell analysis. One limitation of droplet generators is their ability to only assess a single input condition like concentration or number of drugs. To address this limitation, rare earth (RE)-doped nanoparticles were coupled with a microfluidic droplet trapping array for a novel approach at droplet tracking to visually assess different doses and drug combinations in the droplets. The β-hexagonal NaYF4 nanoparticles were doped with rare-earth emitters (Eu3+, Tb3+, Dy3+) which have a luminescence emission spectrum in the red, green and blue region upon UV excitation. An advantage of the RE-doped nanoparticles is that their excitation and emission spectra do not overlap with standard fluorophores thus allowing for both droplet tracking and quantification of cell viability or enzyme activity. This spectral independence was observed in (i) GFP-labeled HeLa cell, (ii) RFP-labeled MDA-MB-231 cells, and (iii) pre-killed MDA-MB-231 cells stained with Ethidium homodimer-1. Moreover, the RE-doped nanoparticles were confirmed to be biologically inert. Finally, a Python algorithm was developed to automatically distinguish droplets from cells, count cells in each droplet, and quantify cell viability to facilitate the single cell analysis in a high-throughput manner. The accuracy of this algorithm was compared with manual counting of cells, with a difference of ~2 percent. Thus, this work established the foundation for future applications using this technique in multiplexed droplet tracking to understand tumor heterogeneity through combinatorial therapies. The applications for this algorithm are numerous including fluorescence quantification, droplet tracking, and biomarker detection.

BIOT 375

A new binding model for complex elution behavior of mAb under high loading conditions on cation exchange tentacle resins

Juliane Diedrich, [email protected], William Heymann, w.heymann@fz- juelich.de, Samuel Leweke, Eric von Lieres. Research Center Juelich, Juelich, Germany

Monoclonal antibodies (mAb) are used as therapeutic proteins and therefore underlie strict guidelines. For mAb purification by ion exchange chromatography, tentacle resins are of particular interest due to their high binding capacities and improved ligand accessibility. However, cation exchange tentacle resins like Fractogel are reported to cause complex elution behavior under high loading conditions. Aggregation of mAb and unusual peak shapes are reported, and the causes of these issues are not yet fully understood. We use mechanistic modeling to investigate binding mechanisms that cannot be directly observed by experiments.

A new binding model is proposed that extends the classic SMA model by Brooks and Cramer towards multiple binding stages, while including the impact of ionic strength. Simulations are performed with the Chromatography Analysis and Design Toolkit (CADET) using state of the art mathematical methods. CADET provides comprehensive tools for model-based analysis and optimization of chromatography and related processes.

Four gradient elution experiments under high loading conditions on a tentacle carrying CEX resin are comparatively analyzed using different binding models. The chromatograms show unusual peak shapes and shoulder forming with increasing column load. The new binding model enables a more detailed reproduction of the gradual binding processes on the tentacle resin. The multi-state SMA model allows to quantitatively reproduce all four experiments with one set of parameters. These results strengthen our hypothesis that the observed peak shapes are caused by multiple binding states of mAb on the tentacle ion-exchange resin.

BIOT 376

Development of mathematical models to identify the fouling mechanism in depth filters for cell culture clarification

Dongyoun Jang, Michael Peck, Nripen Singh, Abhiram Arunkumar, [email protected], Sanchayita Ghose, Zheng Jian Li. Product Development, Bristol-Myers Squibb, Devens, Massachusetts, United States

A series of mathematical equations have been developed to model the fouling mechanism of depth filters operated at constant flux during clarification of mammalian cell culture. By modifying and manipulating the Darcy’s Law Equation, equations were derived for each of the main filter fouling mechanisms: complete pore plugging, intermediate pore plugging, standard pore plugging, and cake filtration. Considering that the depth filter media consists of a wide range of pore size distribution, it is believed that the depth filter would foul through more than one fouling mechanism at the same time. Thus, equations were also derived for the following combinations of fouling mechanisms: cake-complete plugging, and cake-intermediate plugging. The model equations were fitted to the experimental pressure profiles of the depth filters, and the results for both primary and secondary depth filters indicate that cake-complete was the dominant fouling mechanism for various feed streams. Correlations from literature were used to determine if constant pressure data can be used to predict performance of primary depth filters for constant flux. Fundamental constants characteristic of each mechanism were interpreted. These mathematical models significantly improved prediction of fouling behavior of depth filters that would enable appropriate selection of filters for different feed streams.

BIOT 377 Process simulations in a clinical scale single-use TFF system examine blend time efficiency

Adam Sokolnicki, [email protected]. MilliporeSigma, Burlington, Massachusetts, United States

Speed and efficiency of clinical scale TFF development is enhanced by single-use systems (SUS). Their design is greatly simplified by eliminating pipework and features for CIP/SIP and optimizing flow paths, enabling ultra-low minimum working volumes under 0.5 L/m2. The current focus on characterizing mixing performance in the MilliporeSigma SU-TFF system, which uses a modified Mobius Mix system as the retentate vessel, will aid end-users with process integration. Low and high VCF process simulations were carried out at low and high viscosity (1, 25 cP) to understand mixing behavior and the time required for a pulse addition of 1 or 5 M salt solution. Blend time was a function of total (pump + impeller) flow rate. Comparing turnover volumes to achieve homogeneity gives a measure of mixing efficiency (# pump + impeller passes). Higher system volumes are more efficient.

BIOT 378

Scalability model for mixing cell culture media in single-use GMP bioprocessing equipment

Anne Hansen, [email protected]. MilliporeSigma, Bedford, Massachusetts, United States

In bioprocess manufacturing, one of the most challenging mixing procedures is the incorporation of powdered media into solution. Media powders tend to float on the surface of the liquid and effective mixing requires a technique to disperse the powder off the surface and into the bulk liquid. In the development of Mobius® Power MIX single- use mixers at MilliporeSigma, it is determined that the presence of a strong vortex is essential for attaining efficient mixing of media powders. By employing novel methods to quantify mixing performance and vortex formation, it is possible to generate scalability models for floating powder mixing. Measurement of mixing time is accomplished through not only traditional methods of tracing conductivity and pH, but also through the measurement of particle count and size over time. Extensive vortex mapping is generated in mixers from 100L to 3000L, at multiple volumes and speeds within each system. The key to creating scalability models is to quantify vortex size according to visual observations of the vortex. Models are made to relate vortex size to key process variables, including power input and liquid height. Using this data to create a vortex factor, it is then possible to create a scalability relationship for floating powder mixing throughout the entire series of the Mobius® Power MIX equipment. Case studies are presented to demonstrate the usefulness of the scalability model to predict mixing performance for a variety of media powders, at various process conditions.

BIOT 379

Increasing efficiency in bioprocessing using available software functions in a process automation approach

Markus Pitkanen2, Annika Morrison4, [email protected], Tomas Bjorkman3, [email protected], Anna Graanberg1, [email protected], Kristin Lenberg5, [email protected], Fredrik Lundström6, [email protected]. (1) GE Healthcare , Uppsala, Sweden (2) RD, GE Healthcare, Uppsala, Sweden (3) R&D, Uppsala, Sweden (4) RD, GE healthcare, Uppsala, Sweden (5) GE Healthcare, Uppsala, Sweden

Markus Pitkänen, Annika Morrison, Anna Grönberg, Tomas Björkman, Kristin Lenberg, Fredrik Lundström

GE Healthcare Bio-Sciences AB, Björkgatan 30, SE-751 84 Uppsala, Sweden Efficiency is a key driver for all processing types, and automation can help provide this to bioprocesses. For manufacturers operating at smaller scales, such as contract manufacturing or research organizations, automation solutions can be a substantial investment. Here, we demonstrate integration of two traditionally separated unit operations, using available system software to control the related technology for parallel processing and a secure hand-over of product. A chromatography polishing step connected to a concentration/diafiltration step was used as model. These process steps are typically situated at the end of the downstream process, where product value is high due to the amount of time and resources already invested. The process steps, including pre- and post-process cleaning and storage, were conducted with minimized manual interactions with the process or the process material. The presented method contributes to a shortened time to market and enables manufacturers to deliver high quality at reduced cost.

BIOT 380

Machine learning approaches for streamlining downstream process development

Derek Lee, [email protected], Ben Tran. Purification Development, Genentech, South San Francisco, California, United States

High-throughput methods are an integral tool for downstream process development to identify promising operation conditions prior to scaling up. With increasing usage of high-throughput techniques in the industry, there is an opportunity to mine and leverage the molecular information collected in these large datasets. While most of our statistical analyses and computational models typically focus on a single molecule’s behavior within a unit operation, there is a tremendous opportunity to leverage our molecule characterization dataset to further streamline downstream process development activities with predictive models that can be applied to multiple molecules. In this work, we have explored utilizing our high-throughput datasets and various machine learning models (such as random forests and deep learning neural networks) in order to construct generalized predictive models for simulating molecule behavior in cation- exchange (CEX) and anion-exchange (AEX) chromatographic operations. These models can then be used to guide molecular assessment and high-throughput experimental design, with the goal of accelerating early-stage downstream process development.

BIOT 381

Role of simulation and scheduling tools in bioprocess development and manufacturing

Dimitrios Petrides, [email protected], Doug Carmichael, Charles A. Siletti. Intelligen, Inc., Boulder, Colorado, United States

The successful scale up and commercialization of biopharmaceuticals is a challenging task that requires collaboration of professionals from many disciplines. Process simulators can facilitate this task by assisting scientists and engineers to answer the following and other related questions: What is the impact of product titer increase on the capacity load of the downstream section, the overall throughput of a plant, and the cost of goods? What changes are required in an existing multi-product facility to accommodate the process of a new product? What is the range of variability that a process can accommodate if it operates under a tight cycle time? What is the impact of single-use systems on the demand for utilities, the environment and the cost of goods? Our experience in addressing the above questions will be presented using industrial examples in which we evaluated alternative technologies for producing therapeutic monoclonal antibodies and vaccines.

BIOT 382

Biomaterial production process intensification and analysis and optimization with process simulation tools

Dimitrios Petrides, [email protected], Doug Carmichael, Charles A. Siletti. Intelligen, Inc., Boulder, Colorado, United States

Analysis is a key element of process intensification and optimization. Process simulators and other computer aids facilitate this task by assisting scientists and engineers to answer the following and other related questions: What process areas are the best targets development effort in terms of return on investment? At what production scale does continuous manufacturing has advantages over batch manufacturing? What is the impact of extensive recycling on the demand for utilities, the environment and the cost of goods? Our experience in addressing the above and other related questions will be presented using an industrial example in which we evaluated alternatives for lysine production via bio-conversion.

BIOT 383

Advanced process optimization - a novel, in-situ real time adaptive control system to accelerate and optimize biologics process development and manufacturing

Greg Emmerson2, Sam Watts2, Simon Saxby2, George E. Barringer1, [email protected]. (1) Stratophase Ltd (USA), W Groton, Massachusetts, United States (2) Stratophase Ltd, Romsey, United Kingdom

An in-line process monitoring control system and method for in-situ, real time monitoring and control of nutrient and carbon source feeding in bioreactors is described. The Ranger responds to the overall metabolic environment of the process under observation and is highly sensitive to any molecular level perturbation in the process media, such as occurs when a biological process is fed nutrients and carbon sources. An automatic, closed loop adaptive feeding protocol that responds to these real time changes in nutrients then maintains proscribed optimum nutrient concentrations. Faster process development cycles, time to clinic, and the promise of better product quality and titer are typical outcomes. The system can actively and independently manage multiple feeds to further optimize bioreactor optimization.

Additonal applications of this technology in media preparation and in downstream processes will be discussed.

The system is applicable to all scales of operation and modalities from process development to commercial production in fed batch, perfusion, and continuous operation. It is compatible with SUBs. This technology is applicable in microbial, fungal, and mammalian cultures.

BIOT 384

Process economics of continuous bioprocessing

Jonathan Hummel1, [email protected], Mark Allen Pagkaliwangan1, Xhorxhi Gjoka1, Terence Davidovits2, Rick Stock2, Thomas Ransohoff2, Rene Gantier1, Mark Schofield1. (1) Biotechnology Process R&D, Pall Life Sciences, Westborough, Massachusetts, United States (2) BioProcess Technology Consultants, Woburn, Massachusetts, United States

The biopharmaceutical industry is evolving in response to changing market conditions, including increasing competition and growing pressures to reduce costs. Single-use technologies and continuous bioprocessing have attracted attention as potential facilitators of cost-optimized manufacturing for monoclonal antibodies. While disposable bioprocessing has been adopted at many scales of manufacturing, continuous bioprocessing has yet to reach the same level of implementation. In this study, the cost of goods for both upstream and downstream processing were modeled. The costs associated with different upstream production formats and strategies were modeled and compared. A particular focus was placed on cell retention for perfusion cell culture to identify the cost differential between acoustic wave separation, alternating tangential flow filtration, and tangential flow filtration. Three downstream manufacturing strategies were also modeled: Pall Life Science’s integrated, continuous bioprocessing platform, a stainless-steel batch process, and a single-use batch process. Evaluation of the models across a broad range of clinical and commercial scenarios revealed that the cost savings gained by switching from stainless-steel to single-use batch processing are often amplified by continuous operation. The continuous platform exhibited the lowest cost of goods across 78% of all scenarios modeled here, with the single-use batch process having the lowest costs in the rest of the cases. The relative savings demonstrated by the continuous process were greatest at the highest feed titers and volumes.

BIOT 385

Copper catalyzed fragmentation of an IgG1 recombinant monoclonal antibody for therapeutic use: using kinetic modeling to fit process to product during technology transfer

Matthew Henry1, [email protected], George Alkire1, Jessica R. Molek2, Alana szkodny3,2, Myrna Monck1, Douglas P. Nesta1. (1) Biopharmaceutical Product Sciences, Glaxosmithkline, King of Prussia, Pennsylvania, United States (2) Biopharm Downstream Process Development, Glaxosmithkline, King of Prussia, Pennsylvania, United States (3) Chemical Engineering, University of Delaware, Wilmington, Delaware, United States

Fragmentation of monoclonal antibodies is a frequently observed degradation pathway that can limit the shelf life of recombinant monoclonal antibodies in biopharmaceutical products. We investigate one cause of fragmentation, copper catalysed hydrolysis of a lys222-thr223 bond in the hinge region of an IgG1 mAb. We also developed sample preparation and handling approaches and applied analytical techniques to overcome the solubility limitations of copper in commercially relevant formulations containing phosphate buffer. Analysis of the kinetics of the degradation pathway over a temperature range of 5-30°C and a copper concentration range of ~50-500 ppb indicated that the fragmentation reaction follows zero-order kinetics and yield products consistent with copper mediated fragmentation. By understanding the fragmentation kinetics, we were able to determine limits on copper impurities that allow for adequate processing time in the drug substance and drug product manufacturing processes while minimizing fragmentation. Furthermore, we investigate the kinetics of copper removal from the mAb using EDTA as a chelator for copper in solution. By understanding the copper removal kinetics, we were able to determine the necessary contact times and EDTA levels to ensure reduction of mAb bound copper to acceptable levels during drug substance manufacturing.

BIOT 386

Use of modelization and scale-down models to support drug product development and transfer to manufacturing facilities: Challenges and benefits

Mohamed Belkacem, [email protected], Carine De Kesel. GSK Vaccines, Brussels, Belgium

Vaccines are composed of large molecules with complex structures. Vaccine antigens are defined by their structurally derived product quality attributes. The process conditions used during manufacture may influence or impact those product quality attributes. The ability to predict and model the effects of process conditions on product quality prior to GMP manufacture is critical to the success of a technology transfer. Process development must, therefore, allow generating extensive knowledge about the impact of process conditions on product characteristics, in-depth experience with all of the operations at both the laboratory and commercial scale, and a deep understanding of how laboratory results translate to production performance.

In that sense, expertise in process modelling is crucial for the successful development of vaccines processes. Effective scale-down models are designed to accurately predict the performance of production-scale runs, and thus allow rapid development and optimization of processes in preparation for scale up and transfer to the manufacturing facility. Once a process is transferred to manufacturing, the scale-down model is further used to support and rationalize the validation strategy and to assess the impact of process changes.

This being said, the modelling and accurate scale-down definition of unit operations is not always obvious and straightforward. But once achieved, the benefits of such approach are multiple and in particular with regards to savings in terms of products, costs, time, ressources and commercial capacity.

Using concrete examples of vaccine drug product and adjuvant formulation, the proposed presentation will show case-studies where Quality by Design approaches including scale-down modelling and design space definition were used to support not only the process development but also the full Process Performance Qualification (PPQ) strategy definition encompassing product impact, homogeneity validation and holding time validation. It will show how these approaches have increased the overall success of new product transfer in commercial facilities and will present some of the challenges encountered in the scale-down definition and the mitigation strategies implemented to ensure successful implementation. BIOT 387

Enhancing multivariate calibration model reproducibility used for the online monitoring of upstream processes in continuous biomanufacturing

Nicholas Trunfio4,6, [email protected], Brittany Chavez2, Sai Rashmika Velugula3, Seongkyu Yoon5, Cyrus Agarabi1. (1) FDA, Silver Spring, Maryland, United States (2) 72/2303, FDA/CDER, Silver Spring, Maryland, United States (3) US Food & Drug Administration, Silver Spring, Maryland, United States (4) Chemical Engineering, University Of Massachusetts, Billerica, Massachusetts, United States (5) University of Massachusetts Lowell, Lowell, Massachusetts, United States (6) OBP/OPQ/CDER, US Food and Drug Administration, Silver Spring, Maryland, United States

The complex mixtures present in biomanufacturing processes have traditionally required slow and expensive experimental assays, as well as time consuming and complicated analyses to be characterized properly. Multivariate Data Analysis (MVDA) can be integrated with spectroscopy to uniquely solve both of these problems simultaneously. Spectroscopic data has been generated in real-time, eliminating the need for offline assays; and MVDA has been used to rapidly analyze the data in a straightforward manner. Prior experiments have shown that this paradigm can be used offline to characterize the raw materials that are used to supplement cell culture media. However, online models that reliably quantify extracellular component concentrations in continuous bioprocesses require additional considerations. Even when the components’ absorbance properties are well understood, cellular metabolism ensures that nutrient and product profiles vary collinearly with one another. This work explored supplementation strategies that break this collinearity to ensure that proper multivariate calibration models are constructed, instead of soft sensor models whose performance is inconsistent due to their reliance on component concentration collinearity for accurate predictions. This allows for more robust corrective action to be taken. Furthermore, the advantages of training multivariate calibration models from continuous bioprocesses’ data, whose steady-state operation allows for more robust and complete design space coverage relative to batch processes, are explored as a way to guide ongoing and future research in this area.

Disclaimers: This article reflects the views of the authors and should not be construed to represent official FDA’s views or policies.

BIOT 388

Development of QbD expertise among scientists in a biopharm product and process development organization

Siddharth Parimal3, [email protected], Phillip R. Smith4, Steven Weisser3, Gerald J. Terfloth2, Kent E. Goklen1. (3) BPDS, GlaxoSmithKline, King of Prussia, Pennsylvania, United States (4) BPDS, GlaxoSmithKline, King Of Prussia, Pennsylvania, United States

Consistent application of up-to-date Quality by Design (QbD) principles and practice (e.g., enhanced development) is essential in the Biopharm industry as regulatory expectations and new technology drive continuous improvement in product understanding and manufacturing processes. However, in many organizations the ultimate QbD practitioners are Scientists assigned to a single project, and it can be challenging for them to allocate time towards acquiring the most current QbD resources and tools for implementation in development projects. This can lead to inconsistencies across the organization when different elements of regulatory filings for different projects are compared (e.g., experimental design and technical risk assessment approaches, etc.). GSK’s Biopharm Process Development & Supply organization has implemented a Cross-Functional Technical Team model to address this gap by providing a development opportunity for Scientists to enhance their understanding of QbD principles and embed harmonized QbD practice within their technical lines. This poster will describe the formation and experiences of this QbD Cross-Functional Technical Team, composed of Scientists as well as other QbD practitioners across the upstream, downstream, analytical, and formulation development technical lines. In particular, the success of this team includes enhanced understanding of QbD principles and providing guidance for consistent application across multiple programs. Examples will be presented which highlight the impact of this concerted and aligned QbD approach on improving process development focused on achieving scientific rigor under increasingly demanding CMC timelines. To conclude, feedback from team members will also be shared, many of whom have developed QbD expertise and are currently involved with continuous improvement of tools, templates and methods for QbD implementation within GSK’s Biopharm organization.

BIOT 389

Near-infrared (NIR) spectroscopic analysis of water in lyophilized products for process efficiency improvements

Adam Hopkins, [email protected]. Metrohm USA, Riverview, Florida, United States

The quality, efficacy, and shelf-life of lyophilized pharmaceutical products and ingredients depends on water content. Routine Karl Fischer (KF) titration is the standard method for measuring residual water content to monitor manufacturing, ensure samples are within specification, and to optimize the freeze-drying process. However, routine application of this technique is not without drawbacks, most notably the use of hazardous chemicals and the destruction of samples. Near-infrared (NIR) spectroscopy is well-suited to measuring moisture, and can do so without sample preparation or chemical consumption. This talk explores how to best link NIR spectroscopy to KF results for lyophilized materials and how this method can be used to monitor the lyophilization chamber performance, improve process efficiency, and tighten product specifications.

BIOT 390

Best practices when developing a point of use pressure decay test

Alexandra Steele, [email protected]. EMD Millipore, Burlington, Massachusetts, United States

In the past 20 years, single use technologies have become ubiquitous in biotech. However, concerns remain surrounding the proper implementation of Point of Use (POU) testing and its utility to identify defects. The effectiveness of the test varies greatly with limited understanding and a lack of consistency in execution. The following outlines considerations and best practices for developing a POU test method with equipment found in a working laboratory. The aim is to further the understanding of how different criteria/methods used for implementing and developing the test method can affect the results. Thus allowing an end user to implement a well understood and reliable POU test.

BIOT 391

At-line mass spectrometry-based process analytical technology for linking nutrient consumption and bioreactor product quality

David N. Powers, [email protected], Nicholas Trunfio, Erica Berilla, Brittany Chavez, Casey Kohnhorst, Sai Velugula, Cyrus Agarabi. CDER, U.S. Food and Drug Administration, Silver Spring, Maryland, United States

Current understanding of nutrient level changes in cell culture media and their influence on bioreactor-produced antibody quality and quantity is unclear. Nutrients such as amino acids are necessary for optimal production from cultured Chinese hamster ovary (CHO) cells that have become the preferred platform for the manufacturing of therapeutic proteins such as monoclonal antibodies. To this end, we present the results of an at-line Process Analytical Technology (PAT) approach that allows for rapid measurement of bioreactor amino acid levels throughout a cell culture process. This method is capable of quantitating amino acids from crude bioreactor media in their native underivatized state within 25 minutes, including major energy sources such as glutamine and more subtle components such as cystine. We demonstrate a manual feedback loop using these at-line results which replenishes depleted amino acids and caused increases in viable cell density growth rates. Our results demonstrate the feasibility of PAT tools in an industrial setting to quickly assess bioreactor nutrient levels in bioprocessing and support the development of advanced feeding strategies to improve product yield and quality outcomes.

BIOT 392 EDTA determination in bovine serum albumin

Natasha Gupta, [email protected], Charles Felice, Bala Ramanathan, Kevin Pipkins, John Cunningham. Janssen R&D, Malvern, Pennsylvania, United States

EDTA is an impurity often found in Bovine Serum Albumin (BSA), which is commonly used as a component in cell culture medium. Here we develop an accurate method to measure EDTA concentration in BSA samples than can be adapted for use as an in- process control. This method uses UV detection to determine the amount of EDTA in a BSA sample. The assay was first developed for manual use and transferred to an automated instrument to decrease operator error and provide increased accuracy. The method was validated and transferred for control of EDTA in the BSA manufacturing process.

BIOT 393

In-line cell culture analytics for high-throughput process development: An evaluation of the integrated FLEX2 bioanalyzer

Peter Liu, [email protected], Matthew Manahan, Linda Hoshan, Jennifer Pollard. BioProcess Development, Merck, Kenilworth, New Jersey, United States

Automation efforts in cell culture development are integral to designing control strategies, media formulations, and cell lines for highly productive processes. Microbioreactor systems, such as Sartorius’ ambr platform, provide automated process control and liquid handling, but in-process analytics have remained a bottleneck, with most analytical solutions requiring heavy maintenance, manual intervention, and tedious sample preparation by scientists. In this work, we share our experience testing the in- line integration of the ambr15 system with Nova Biomedical’s BioProfile FLEX2, a low- maintenance one-stop-shop bioanalyzer for cell culture metabolites, cell density, pH, and dissolved gasses. Throughout our evaluation of over 1000 full-panel assay runs, the FLEX2 delivered metabolite and pH data very comparable to current industry- standard analyzers. This data was communicated to the ambr15 software automatically, where it was used effectively to calculate feed volumes and calibrate online sensors. If implemented into our small-scale development workflow, the FLEX2 can reduce the time and resources necessary to support feedback control by 25% each week. The ambr-FLEX2 integration allows flexibility in scientists’ workflow, supports the process analytical technology (PAT) mission, and represents a significant step forward in cell culture automation.

BIOT 394

Detection and characterization of mycoplasma contamination in cell culture with application of nucleic acid testing assays Talia Faison, [email protected], Casey Kohnhorst, Sai Velugula, Erica Berilla, Scott Lute, Cyrus Agarabi, Kurt A. Brorson, Julie Wang, Sarah Johnson. CDER, U.S. Food and Drug Administration, Silver Spring, Maryland, United States

Mycoplasma are a type of bacteria that lack rigid cell walls and are occasional cell culture contaminants. In a biotechnology setting, mycoplasma could pose a threat to patient safety if undetected contaminants from the production culture remain in the final product after downstream manufacturing processes. In this study, we tracked the growth behavior of mycoplasma in co-culture with biomolecule producing CHO cells and evaluated their impact in a simulated bioreactor contamination. In addition, we investigated a new nucleic acid testing (NAT) method to detect common mycoplasma contaminant species for application in biotechnology manufacturing. Mycoplasma contamination in a bioreactor stimulates detectable changes to CHO behavior in terms of growth/viability and rates of nutrient consumption/metabolite generation approximately 48 hours after introduction. NAT detection of mycoplasma tracks consistently with established growth kinetics in early culture: at no point did the Points To Consider test detect activity where the NAT didn’t. It does not accurately reflect mycoplasma viability in late culture, as residual nucleic acids persist after the contaminants died. This argues that the NAT assay provides a better assurance of product safety in that it more sensitively detects contaminations even after they burn out.

BIOT 395

Purification of PLBL2 critical reagent for ELISA development

Lisa Wong, [email protected], Pat McKay, Jayme Franklin. Purification Development, Genentech, South San Francisco, California, United States

PLBL2 is an endogenous CHO-derived host cell protein that can co-purify with several biological Drug Substances. Demonstrated clearance of PLBL2 and assays for monitoring removal through the Drug Substance processes were required. Standard CHO HCP ELISA showed minimal detection of PLBL2 in purification process pools and therefore a specific PLBL2 ELISA was needed to monitor this impurity in downstream purification processes. PLBL2 critical reagent was expressed in CHO cells and purified using a multi-step purification process. The PLBL2 critical reagent was characterized by N-terminal sequencing, mass spectrometry, SDS-PAGE, and SEC-HPLC and used to develop a specific PLBL2 ELISA. The PLBL2 critical reagent serves as a reference standard and an assay control for the PLBL2 ELISA. This poster demonstrates the purification process developed to isolate PLBL2 for ELISA development.

BIOT 396

In silico design and in vivo selection of riboswitches for cell-based detection of explosive compounds Svetlana Harbaugh1,4, [email protected], Yaroslav Chushak1,4, Jennifer Martin1, Howard Salis2, Michael Goodson1,3, Jorge L. Chavez1, Nancy Kelley- Loughnane1. (1) 711th Human Performance Wing, Airman Systems Directorate, Air Force Research Laboratory, Wright-Patterson AFB, Ohio, United States (2) Pennsylvania State University, University Park, Pennsylvania, United States (3) UES, Inc., Dayton, Ohio, United States (4) Henry M. Jackson Foundation, Bethesda, Maryland, United States

Synthetic biology combines engineering-driven approaches with fundamental biological components to design predictable functions in biological systems. This rapidly growing field utilizes the capabilities of cells by directing their natural functions and adding non- natural functions to complete specific tasks. Our team is using regulatory RNAs, or riboswitches, for the development of highly sensitive, specific cell-based detection systems. Riboswitches are composed of two functional domains: an aptamer domain that binds to a specific ligand; and an expression platform that controls the expression of a downstream gene via conformational changes that are induced by ligand binding to the aptamer domain. Using different approaches, we have developed several synthetic riboswitches that control protein translation in E. coli cells in the presence of 2,4- dinitrotoluene (DNT) and 2,4,6-trinitrotoluene (TNT). In our first approach, a riboswitch library was constructed by incorporating up to 30 degenerate bases between an in vitro selected aptamer and a ribosome binding site. Screening was performed by placing the riboswitch library upstream of the Tobacco Etch Virus (TEV) protease coding sequence in one plasmid, a second plasmid encoded a FRET-based fusion protein linked with a peptide containing the TEV protease cleavage site. The synthesized protease cleaved the linker in the FRET-based construct, causing a change in the fluorescence intensity of cells. The selected riboswitch exhibited a 10-fold increase in fluorescence in the presence of 500 µM DNT compared to the system without the analyte. In another approach, we applied computational RNA design to convert an in vitro selected TNT aptamer into a functional riboswitch coupled with dual color recombinase-based reporter system. This was based on a recently developed statistical thermodynamic model that predicts the sequence-structure-function relationship for translation-regulating riboswitches that activate gene expression inside cells. Two selected riboswitches demonstrated a 19-fold increase in fluorescence in the presence of 100 µM TNT.

BIOT 397

At-line investigation of capacitance for assessing biomass of mammalian cell culture

Kevin Ramer1, [email protected], Juhi Fernandes1, An Zhang1, Jayme Currie1,2. (1) Biogen, RTP, North Carolina, United States (2) NC State University, Raleigh, North Carolina, United States

Capacitance has been used to monitor “biomass” in cell culture and fermentation processes. While it is generally accepted that capacitance measurements are related to viable cell mass, the molecular species responsible for carrying capacitive charge have not been identified. The absence of a mechanistic understanding precludes the ability to predict capacitive behavior across systems or methods. We show that culture samples obtained from bioreactors can be used to predict or confirm on-line capacitance values. The accuracy of our at-line monitoring method depends on proper control of vessel size, temperature, and mixing for the capacitance equipment; and timing is especially important due to the dynamic nature of the cultured cells. We compared the dynamic situation of viable cultured cells with a stabilized capacitance system comprised of formaldehyde-treated cells suspended in standard cell culture media. The stabilized cells are non-viable, but possess a comparable capacitance to viable cell cultures (at the same density of particulates per unit volume). Further, the stabilized system allowed confirmation of critical at-line parameters that had been identified using viable cell cultures. The formaldehyde-fixed cells are shown to function as calibration standards for testing various capacitance measuring equipment and/or methods. Finally, the combination of the at-line equipment and stabilized capacitive cells enabled studies to test the sensitivity of capacitance (rather than cellular responses) to system perturbations. Based on sensitivity to detergents, pH and temperature, the identity of the capacitive charge carriers appears consistent with molecules integral to cellular membranes. Capacitance signals are therefore subject to changes in membrane composition, but are not necessarily dependent on viability of the cell culture being assessed.

BIOT 398

Protein and cell - surface interactions and layer properties using MP-SPR

Annika Jarvinen1, [email protected], Niko Granqvist1, Johana Kuncova- Kallio2. (1) BioNavis , San Diego, California, United States (2) BioNavis Ltd, Ylojarvi (Tampere-region), Finland

Surface Plasmon Resonance (SPR) is an excellent label-free method to measure molecular interactions, such as drug and protein affinity and binding kinetics. Multi- Parametric Surface Plasmon Resonance (MP-SPR) extends traditional SPR with angular scanning optical arrangement and multi-wavelength measurement, expanding the applicability from molecular interactions to layer characterization and surface-layer interaction analysis. MP-SPR measurements were performed to determine adsorption of proteins and cells on surfaces in real-time measurements and for biosensor development.

Adsorption of AD-MSC cells on micrometers thick hydroxyapatite coating was detected to be faster and clearly stronger when compared to binding on reference surface (Vilardell et al. 2016). Binding of cancer cells (MCF7) on target peptide was determined for biosensor development (Etayash et al. 2015). Interactions of C3b protein and 100% serum samples on liposomes were measured to optimize surfaces of drug delivery nanocarrier (Kari et al. 2016). PEG coating on nanocarrier reduced C3b binding causing smaller affinity and adsorbed mass in MP-SPR measurements. Liposomes without PEG produced stronger surface-induced activation of the complement system and thicker and denser corona layer formation on nanocarrier. Antifouling properties of polyethyleneglycol (PEG) based polymer brushes were determined against serum samples, revealing maximum of 99% resistance (Emilsson et al. 2015). The MP-SPR method has shown good correlation with reference methods used in the studies such as QCM, Confocal microscopy, AFM and Raman, and with previous literature values for similar systems.

Very high sensitivity of the MP-SPR method makes is unequalled for interaction studies. Beneficially same measurement provides also layer thickness and refractive index information for complete understanding of the interfaces.

Label-free Cancer Cell Detection and Cell Adhesion on Implant Material Surface using MP-SPR

BIOT 399

Deoxyribozyme cascade for visual detection of Zika virus RNA

Adam J. Reed1, [email protected], Ryan Connelly1, Allison Williams1, Hyeryun Choe3, Yulia Gerasimova2. (1) Chemistry, University of Central Florida, Ft Myers, Florida, United States (2) Chemistry, University of Central Florida, Orlando, Florida, United States (3) Immunology and Microbiology, The Scripps Research Institute, Jupiter, Florida, United States

With the capability of causing major birth defects, such as microcephaly, posterior eye damage, and club foot, the Zika virus is currently a major concern for public health. A cheap and efficient method of detection is crucial for early treatment as well as tracking, and potentially reducing, the spread of the virus. Here, a sensor is proposed utilizing nucleic acid-based nanoconstructs to detect the viral genome. The sensor is comprised of two deoxyribozymes (Dz): a phosphodiesterase-like split Dz and a peroxidase-like G- quadruplex (G4) Dz. The split phosphodiesterase-like Dz hybridizes to a conserved fragment of the Zika virus genome, which re-forms the Dz catalytic core and enables cleaving a G4 Dz-containing oligonucleotide substrate. The G4 Dz, originally sequestered in a stem-loop of the substrate, is released upon substrate cleavage and, in the presence of hemin, catalyzes peroxidation of a colorless organic indicator, yielding a colored product visible to the naked eye. While concentrations of Zika virus RNA in clinical samples are below the capabilities of these sensors, pairing the system with isothermal amplification allows for a point-of-care compatible method for Zika virus diagnostics. This work is supported by the Florida Department of Health, Biomedical research Program (7ZK33).

BIOT 400

Using engineered zinc finger proteins to detect pathogen-specific DNA

Caleb Sedlak1, [email protected], Moon-Soo Kim2. (1) Chemistry , Western Kentucky University , ELIZABETHTOWN, Kentucky, United States (2) Department of Chemistry, TCCW455, Western Kentucky University, Bowling Green, Kentucky, United States

Zinc Finger Proteins (ZFPs) are one of the most common DNA-binding domains. ZFPs can be engineered to bind to specific genes on a double stranded DNA. Developing a rapid and reliable method for detecting specific pathogens would be greatly beneficial to modern biomedicine as well as more resource-limiting areas. A pair of ZFPs was used in a two-step process to first capture the target DNA and then apply the second detection probe ZFP labelled with a fluorescent molecule. A stx2 gene was chosen as a target DNA, which encodes for Shiga toxin a food born pathogen E.coli O157. The ZFP array takes the capture ZFP probe and immobilizes it on an acrylamide gel surface. After target DNA was added, the detection probe a ZFP labeled with a fluoresces molecule was applied to the bound complex of the capture probe and target DNA. At the final step, fluorescence intensity was measured to compare the signals between target DNA and non-target DNA.

BIOT 401

Forensic study: Differentiating and analyzing variances using microbiota to investigate home invasions

Candace Young1, [email protected], Doris Martin1, Crystal Porter2, Valerie Goss2, Jarrad Hampton - Marcell3. (1) Chicago State University, Chicago, Illinois, United States (2) Chemistry, Chicago State University, Park Forest, Illinois, United States (3) Biosciences Division, Argonne National Laboratory, Lemont, Illinois, United States

Microbial communities, or microbiomes, are known to aggregate in the human gut. However, recent studies have shown that bacteria can also be found on the surface of the human skin and can leave an individual-specific signature. Through a genomic system, scientist have found a way to assist in the identification of a home invader by studying the microbiome of the suspected culprit’s skin. This bacterial signature will allow researchers to build a database of dominate bacteria found in the most common places amongst humans to further assist in efforts to solve home invasions. If the human skin microbiome can identify an individual, then we can identify whether an intruder was present during a home invasion. In a home invasion stimulation, the microbiome of homes with and without the presence of a “burglar” were compared to identify disturbances and variances microbial communities. The analyzed data may assist in efforts to prove that microbial signatures of suspected individuals can be used as trace evidence, as an alternative to DNA.

BIOT 402

Visual split peroxidase-like deoxyribozyme probes for the detection of rifampin- resistance in Mycobacterium tuberculosis Charles Verduzco, [email protected], Yulia Gerasimova, Ryan Connelly. Chemistry, University of Central Florida, Orlando, Florida, United States

There has been a growing concern in the medical and pharmaceutical industries regarding the increase in drug-resistant bacteria and viruses. Molecular adaptations, such as single nucleotide polymorphisms (SNPs) in genetic structure, can lead to unconventional protein folding rendering various drug therapies ineffective. Detection of these mutations is critical, and methods that can be used at the point of care settings are urgently needed. Recently, visual assays based on a split G-quadruplex (G4) peroxidase-like deoxyribozyme have been suggested for drug-susceptibility testing of bacterial pathogens. In these assays, the G-rich sequence of the deoxyribozyme is split into two subunits, which form a G4 structure only in the presence of a nucleic acid target. This structure forms a complex with hemin which catalyzes the hydrogen peroxide-mediated oxidation of an organic molecule (e.g. ABTS), producing a noticeable color change. Two strategies to split the four GGG repeats of the deoxyribozyme have been used – symmetrical splitting into the subunits containing two GGG repeats each, and asymmetrical splitting into 3:1 G-triplets. These two designs, while both effective in SNP differentiation, have never been directly compared under the same conditions. In this study, we used both symmetrical and asymmetrical approaches to design split peroxidase-like deoxyribozyme probes targeting the same fragment of the rpoB gene in Mycobacterium tuberculosis. We compared the detection limits and selectivity of both probes to find the best probe design to enable efficient and reliable differentiation of targets corresponding to rifampin-sensitive or resistant M. tuberculosis.

BIOT 403

Advances in the characterization of protein catalyzed capture (PCC) agents for robust biosensing

Curtis Jones, [email protected], Matthew Coppock. Biotechnology, United States Army Research Laboratory, Rockville, Maryland, United States

A major challenge in the deployment of fieldable biosensors lies with the use of monoclonal antibodies, the current gold standard, as the fundamental biorecognition elements. The poor biological and thermal stability, and extensive discovery and manufacturing processes, make monoclonal antibodies a poor choice for on demand reagents in biological assays. Peptide-based, Protein Catalyzed Capture (PCC) agents provide a promising alternative to antibodies due to their high biological and thermal stability, ease of development, and performance tunability. Utilizing a combinatorial peptide library and an iterative in situ click chemistry epitope screening process, the developed peptide-based antibody alternatives can provide nanomolar detection in different types of biological assays; and the rapid discovery and analytical processes deliver a quick response for the detection of new and emerging target proteins. We have been working towards the rapid evaluation of peptide candidates by using the Luminex Immunoassay instrument to determine the PCC agents with optimal binding characteristics for the target epitope. This multiplex assay allows for screening of up to 100 different peptide-based reagents, making it ideal for analysis of multiple binding reagents simultaneously. Surface plasmon resonance (SPR) is then used to evaluate the rate kinetics of the candidates in real time, and determine the best performing bioreagents for the application of interest.

BIOT 404

The fluorescence quenching ability of graphene oxide as a platform for pathogenic double-stranded DNA sensing utilizing engineered zinc finger protein

Dat Thinh Ha, [email protected]. Chemistry, Western Kentucky University, Bowling Green, Kentucky, United States

Two-dimensional graphene oxide (GO) possesses unique electronic, thermal, and mechanical properties. The quenching ability of GO creates novel methods for detection of biomolecules. A zinc finger is a DNA-binding domain that can recognize three nucleotides. Multiple fingers can be linked together to form zinc finger proteins (ZFPs) that recognize extended DNA sequences. Fluorescence dye-labeled ZFPs can bind to GO via stacking interactions of aromatic and hydrophobic residues in conjunction with hydrogen bonding interaction between hydroxyl or carboxyl groups of GO and hydroxyl or amine groups of the proteins. They can come in close proximity with GO, which acts as a quencher due to fluorescence resonance energy transfer (FRET). Thus, in the absence of target DNA, fluorescence signal of dye-labeled ZFP is quenched. When target DNA binds to ZFPs, the bound complex is released from GO, which causes the fluorescence signal to be restored. Here, fluorescence quenching of dye-labeled ZFP by GO has enabled an effective system with high sensitivity and specificity for the detection of pathogenic dsDNA.

Figure 1: Schematic diagram of ZFP array with GO.

BIOT 405 Direct detection of double-stranded DNA of a pathogenic strain of STEC (Shiga toxin-producing Escherichia coli) using engineered zinc finger protein immobilized on paramagnetic beads

Jiyoung Shim, [email protected], Barnabas Kim, Moon-Soo Kim. Chemistry, Western Kentucky University, Bowling Green, Kentucky, United States

The facile and specific DNA detection methods are in great need for clinical (i.e., point- of-care (POC) testing) and biomedical applications. The detection system of low volume of pathogen-specific DNA sequence has been demonstrated by immobilizing zinc finger protein (ZFP) on small paramagnetic beads (2-3 mm in diameter). The small diameter beads provide a larger surface area to volume ratio, which would help to detect more target DNA due to the increasing amount of ZFP immobilized on the beads. The ZFP- based detection system allows for direct detection of pathogen-specific double-stranded DNA (dsDNA) without target DNA amplification. Here, ZFP (i.e., Stx2-268) is engineered to detect Shiga toxin (i.e., stx2 gene) which is released by Escherichia coli O157. The fluorescence-based assay has been applied to monitor the immobilized bound complex of ZFP and Alexa 488-labeled pathogenic dsDNA. The ZFP-based detection of pathogenic dsDNA has demonstrated a sufficient sensitivity, suggesting a potential to be developed into a simple and reliable method for pathogen detection.

BIOT 406

Strategies for controlled bacterial assembly resulting in activation of a quorum- sensing circuit

Mark T. Kozlowski1, [email protected], Bradley R. Silverman1, Christopher P. Johnstone1, David A. Tirrell2. (1) Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, California, United States (2) MC 210-41, Caltech, Pasadena, California, United States Bacterial communities show promise in a wide range of applications, such as environmental remediation and multi-step biochemical synthesis. Here we present methods by which bacteria can be assembled in two distinct architectures through controlled surface display of associating proteins (SpyTag and SpyCatcher or proteins containing coiled-coil peptides including SynZip 17 and SynZip 18). We can tune the size of the aggregates by addition of a “stoichiometric excess” of one type of cell in an interacting pair, as well as by modulating the amount of protein displayed on the surface. We further demonstrate reversibility of the SynZip-mediated interactions, permanence of the SpyTag-SpyCatcher mediated clustering, and construction of an orthogonal assembly system using two different promoter families. We have used these systems to make simple aggregates as well as more complex core-shell structures, and we trigger quorum sensing behavior by cells within these aggregates at low global cell densities. In sum, the techniques we have developed provide a new level of control of genetically-mediated bacterial clustering, and suggest potential applications in the construction of complex bioreactor systems.

BIOT 407

Real time detection of NASBA product by Split DNA aptameric sensor

Nanami Kikuchi1, [email protected], Dmitry M. Kolpashchikov2. (1) Chemistry, University of Central Florida, Orlando, Florida, United States (2) University of Central Florida, Orlando, Florida, United States

Sequence specific fluorescence sensors such as Molecular Beacon (MB) and Taqman Probe havs been widely used for real time detection of sequence specific nucleic acid, small molecules and proteins. However, limitiation of these are high background from the incomplete quenching of the fluorophore. Our recent development - split dapoxyl aptamer (SDA) probe is a highly selective label-free probe for nucleic acid detection. SDA probe consists of two DNA strands that hybridize to a target DNA or RNA sequence and form a binding site for dapoxyl, an water-soluble fluorescent solvatochromic dye with excellent photostability. When bound to the aptamer, the rotational conformational changes of Dapoxyl dye is restricted, allowing 722-fold fluorescence enhancement and a substantially low Kd value of 7.6 ± 1.2 nM. SDA offers high selectivity towards single nucleotide substitutions (SNS) at ambient temperatures and can potentially be used for real time detection of NASBA product.

BIOT 408

Design and fabrication of a triple-input microfluidic droplet trapping array towards multiplexed cancer diagnostics

Riad Elkhanoufi2, [email protected], Khashayar R Bajgiran1, Manibarathi Vaithiyanathan2, James A. Dorman3, Adam T. Melvin2. (1) Cain Department of Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana, United States (2) Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana, United States

The use of personalized medicine in the treatment of cancer has garnered significant interest in recent year; however, many challenges arise due to the heterogeneous nature of cancer and high patient to patient variability. In order to overcome this limitation, combinations of multiple therapeutics have been considered, particularly in targeting drug-resistant tumor cells. Such an approach requires new technology capable of high-throughput screening (HTS) coupled with the analysis of individual cellular behavior. Droplet microfluidic devices is an ideal candidate to achieve this; however, they are often limited to a single input and are not capable of assessing multiple drug combinations simultaneously. The goal of this work was to address this issue by designing and fabricating a triple-input microfluidic droplet trapping device to perform high-throughput single cell analysis in samples treated with various concentration of known drugs. The multi-input microfluidic device was designed using AutoCAD to incorporate three flow-focusing junctions followed by a large downstream trapping array. The device inputs include both aqueous and oil (2 w/w % fluorosurfactant) streams from three distinct sources that are capable of generating 70 µm droplets with a 40-50% single cell encapsulation efficiency. The downstream trapping array has a capacity of 4,000 droplets, which are separated from the bulk solution due to the difference in densities between the aqueous and oil phases. This array allows for time-dependent single cell analysis without the need for expensive high-speed cameras. To ensure an equal distribution (both spatially and quantity) of droplets in the trapping array, a mixing- baffle system was included. The devices were fabricated via a two-step and cost- efficient fabrication process incorporating a soft-lithography wafer fabrication process followed by a polydimethylsiloxane (PDMS) replication process. To demonstrate the feasibility of the proposed triple-input microfluidic device, photo luminescent rare earth doped nanoparticles developed in our lab were utilized as an optical method for droplet tracking to follow the droplets generated at the three different inputs. The work described here is a first step towards a fully characterized multi-input microfluidic droplet trapping device capable of performing high-throughput screening of individual cellular behavior under a variety of cancer therapeutics simultaneously in a time and cost- efficient manner.

BIOT 409

Identifying peptide affinity ligands to demonstrate selectivity between dermal fibroblasts and endothelial cells

Ronit Ghosh1, [email protected], Akshat Mullerpatan1, Tania Baltazar1,2, Carolina Catarino1, André Nascimento1,2, Pankaj Karande1, Steven M. Cramer1. (1) Rensselaer Polytechnic Institute, Troy, New York, United States (2) iBB- Institute for Bioengineering and Biosciences, Instituto Superior Técnico, Lisbon, Portugal

The increasing applications of cell-based regenerative therapies have spiked interest in cell sorting, purification and detection. There are several challenges associated with the development of effective cell separation techniques for optimal purity, yield and scalability.Flow cytometry and Magnetic Activated Cell Sorting (MACS) are currently the most widely used methods for medical and diagnostic purposes. However, these are often expensive and lack scalability.In order to tackle issues with prior techniques, we propose a scheme to identify ligands for the selective targeting of desired cell types. Affinity ligands are ubiquitously used for the purification and detection of a wide range of biomolecules. Among them, peptides are gaining significant popularity. The present study focuses on a phage biopanning strategy to discover peptide ligands from a commercially available phage library, to discriminate between dermal fibroblasts and endothelial cells. Positive screening against fibroblasts was followed by negative screening against endothelial cells. Lead peptide candidates were shortlisted through a secondary enzyme linked immunosorbent assay. Finally, assays were used to determine the binding of lead peptides to dermal fibroblasts. As a negative control, binding was also tested against endothelial cells to ensure specificity of targeting.The peptide ligands identified will be tested for applications such as cell detection, sorting and purification. This method of discovering a high affinity peptide ligand against fibroblasts can help create a platform to target other cell types or specific cell features.

BIOT 410

Split deoxyribozyme probe for efficient interrogation of highly structured RNA targets

Sheila Solarez1, [email protected], Yulia Gerasimova2. (1) University Of Central Florida, Orlando, Florida, United States (2) University of Central Florida, Orlando, Florida, United States

Detection of RNA targets using conventional hybridization probes, such as molecular beacon (MB) probes, is often challenging due to a stable secondary/tertiary structure of the target, which is difficult to characterize with a structurally constrained probe. In addition, the use of MB probes is limited in complex biological samples or cell lysates due to interactions between the probe and the matrix. To overcome these problems, a split deoxyribozyme (BiDz) probe has been suggested. The design of the BiDz is composed of two strands of DNA; each is comprised of half of the catalytic core of a phosphodieserase-like deoxyribozyme, an analyte-binding region, and a substrate- binding region. In the presence of the target nucleic acid sequence, both analyte- binding sections will bind, allowing BiDz to cleave a substrate and produce measurable fluorescence. This approach allows greater sensitivity in analyte selection, but also greater ease in unwinding a tightly packed secondary or tertiary structure that a standard MB probe may be incapable of analyzing. In this proof-of-principle study, we used yeast tRNAPhe as a model highly structured target. We demonstrated that the target was efficiently recognized by the BiDz probe at ambient temperature without the need of a prior annealing step. Comparatively, the MB probe targeting the least stable fragment of the tRNA target was unable to unwind stable secondary and tertiary structure of the target RNA at ambient temperature. The BiDz probe was further characterized in terms of its limit of detection and selectivity. The results obtained in this study highlight the power of BiDz probes for the detection of highly structured RNA analytes, which can benefit diagnostics, fundamental molecular biology research, and therapeutic fields.

BIOT 411

Developing a genetically-encoded insulin sensor for live imaging: Application to Alzheimer’s disease

Syed S. Raza1,2, [email protected], Wenbiao Chen2. (1) Chemistry , Vanderbilt University , Nashville, Tennessee, United States (2) Molecular Physiology and Biophysics, Vanderbilt University, Nashville, Tennessee, United States

Proper insulin signaling in the brain has been found to be of critical importance. In both human and animal models, insulin has been shown to be important in the stability of neuronal growth, metabolic activity, and cytoskeleton growth. These activities are damaged in Alzheimer’s Disease (AD), which is characterized, most notably, by neuronal degeneration, amyloid plaque formation, and dementia. Impaired insulin signaling, also known as insulin resistance, has a well documented relationship with AD whereas normal insulin signaling may inhibit the formation of neurofibrillary tangles and subsequently AD. While there is much work to be done to understand the mechanisms connecting insulin signaling and AD, there is a clear relationship between insulin signaling in the brain and maintenance of neuronal function. A fluorescent indicator of insulin levels will be an essential tool to help understand the relationship of insulin levels and function in tissues and cells of interest. We are developing a fluorescent “insulin sensor” to quantify insulin levels using live zebrafish as a model. The approach we take to engineer this insulin sensor is to insert the conformation-sensitive circularly permutated EGFP (cpEGFP) into sites of the zebrafish insulin receptor that are predicted to encounter a conformational change upon ligand binding. These include, but not limited to, sites in the cysteine-rich (CR) region, L2 region, and αCT domain of the receptor protein. The goal is to identify a construct that displays minimal alteration of ligand affinity but robust ligand-induced change of fluorescence. We have identified residues in the L2 region of the insulin receptor as potential sites that can offer robust and dynamic changes of fluorescence upon insulin binding. If successful, the sensor should allow observations of insulin levels in all stages of zebrafish development and conditions of over-nutrition, insulin resistance, and disease.

BIOT 412

SBAP-linked bioconjugation of oligonucleotides to quantum dots

Barrett Eichler, Troy Hollinsworth, [email protected], Austin Kroeger, [email protected]. Chemistry, Augustana University, Sioux Falls, South Dakota, United States Malignant prostate cancer cells can be detected by targeting surface proteins using bioconjugated quantum dots (QDs). Previous attempts to conjugate QDs with oligonucleotides were unsuccessful, so a project was initiated to create a reliable conjugation protocol. Initial success using succinimidyl-3-(bromoacetamido)propionate (SBAP) as a heterobifunctional crosslinker between a thiolated oligonucleotide and an amine-coated QD suggested further investigation was needed into the optimization of the coupling process. Studies were conducted to investigate optimal conjugation incubation time, oligonucleotide-QD ratio, and effects of oligonucleotide length. Results of these studies will be presented.

BIOT 413

Mutation detection of RNA in extracellular vesicles with a cationic lipoplex nanoparticle biochip

Yan Sheng, [email protected], Jiaming Hu, Ly J. Lee. The Ohio State University, Columbus, Ohio, United States Detection of specific RNAs in circulating cancer cells and cancer cell-derived extracellular vesicles (EVs) has high potential to serve as a non-invasive method for early cancer diagnosis. However, accurate in-situ identification of cellular and EV mRNAs with single-point mutation is challenging.

Herein, we present a unique toehold-initiated molecular beacon (Ti-MB) with a stable hairpin structure, fast hybridization kinetics and single mismatch specificity. Ti-MBs encapsulated in cationic lipoplex nanoparticles (CLNs) that are tethered on a gold coated glass slide can capture cells and EVs and detect encapsulated target RNAs in a single step. This CLN-Ti-MB biochip could sensitively quantify miR-21 single strand DNA oligo mutants in artificial extracellular vesicles and identify single-point mutations in KRAS mutant pancreatic cancer cell derived EVs with high specificity, not achievable by conventional molecular probes. Four pancreatic cancer cell lines, HUT78, PaCa-2, AsPC-1 and PANC03.27 known as a wild-type (WT), G12C, G12D and G12V mutation respectively, were used as the model systems. EVs collected from cell culture medium were directly applied to the CLN-Ti-MB tethered biochips. Compare to KRASWT, the fluorescence signals of KRASG12C, KRASG12D and KRASG12V with CLN-Ti-MBWT were reduced to 18.7%, 25.8% and 27.6%, respectively. By using CLN-Ti-MBG12C, the fluorescence signals of KRASWT, KRASG12D and KRASG12V only reached 9.1%, 18.6% and 12.8% of that of KRASG12C. Similar results for KRASG12D and KRASG12V. These results indicate that our CLN-Ti-MB design is unique that it is capable of direct identification of single-point mutations in living cells and their secreted EVs. Moreover, we show that CLN-Ti-MB biochip with single-point mutation detection capability could selectively and sensitively identify KRAS mutant mRNAs in human serum EVs, distinguishing pancreatic cancer patients with different mutations, which plays a key role on the future targeted drug therapy.

Fig. Principle of CLN-Ti-MB biochip.

BIOT 414

Highly sensitive programmable RNA detection using fluorescent RNA aptamer Yuichi Furuhata1, [email protected], Mizuki Kobayahshi3,2, Ryo Maruyama2, Yusuke Sato4, Kurumi Makino3, Tatsuo Michiue2, Hiroharu Yui3, Seiichi Nishizawa4, Keitaro Yoshimoto2,5. (1) Department of Computational Biology and Medical Sciences, Graduate School of Frontier Sciences, The University of Tokyo, Meguro-ku, Tokyo, Japan (2) Department of Life Sciences, Graduate School of Arts and Sciences, The University of Tokyo, Meguro-ku, Tokyo, Japan (3) Department of Chemistry, Faculty of Science, Tokyo University of Science, Shinjuku-ku, Tokyo, Japan (4) Department of Chemistry, Faculty of Science, Tohoku University, Sendai, Miyagi, Japan (5) JST, PRESTO, The University of Tokyo, Meguro-ku, Tokyo, Japan

RNAs play an important role in various cellular processes and can be used as biomarkers for disease diagnosis. Hence, it has become important to visualize and detect endogenous RNAs in living cells for understanding the diverse cellular mechanisms and developing methods for disease diagnosis and medical treatment. For the construction of low-cost and easy-to-use endogenous RNA detection probe, a chemically unmodified RNA aptamer that binds to a pro-fluorophore and activates its quantum yield is desirable. Here, we focused on Broccoli, one of the brightest variant of Spinach, which forms a complex with DFHBI-1T and activates its fluorescence. For the construction of Broccoli-based RNA detection probe, the structural characterization was performed to identify a contributory region for the stability of Broccoli. Broccoli was predicted to have DFHBI-1T recognition region composed of G-quadruplex structure sandwiched between two stems. Next, Broccoli-based probe composed of destabilized Broccoli and recognition sequences complementary to target RNA was designed based on predicted structure (Figure). Stem structure bridging Broccoli and target RNA was optimized based on thermodynamic stability. After optimization, Broccoli-based probe could exhibit up to 48-fold fluorescence activation in response to target RNA. Interestingly, this optimized stem structure could be applied to the construction of Baby Spinach-based probe, a minimized variant of Spinach. A resulting probe exhibited 140- fold fluorescence activation and could detect a very small amount of target RNA as low as 5 nM.

Schematic illustration for the design of fluorescent aptamer-based RNA detection probe. A probe consists of destabilized fluorescent aptamer and target recognition sequences complementary to target RNA. The resulting probe is stabilized upon binding with its target RNA and form a complex with DFHBI-1T, resulting in fluorescence activation.

BIOT 415

Characterization of nitric oxide flux as a function of time and temperature: A durability review

Mark M. Jeakle, [email protected]. Department of Surgery , University of Michigan, Ann Arbor, Michigan, United States

Nitric Oxide (NO) has previously been shown to interrupt the intrinsic coagulation pathway in vivo. Our laboratory’s prior research has shown that diamines can serve as an NO donor in vivo, and reduce or eliminate the need for classical systemic heparinizing used for an extracorporeal circuit. Therefore we have studied what the optimal storage conditions would be for polyvinyl chlorate circuits coated with a diamine and carbosil layer that releases NO at significant flux. NO flux rate is a key part in identifying the quantity of NO released by a compound to its environment and a good way to indicate effective NO availability in the blood. Previous studies have indicated that a coating of NO-donating diamine is effective at limiting coagulation in Lapine circuit models, however it is unknown how long these coated tubes can be stored for and maintain these high NO flux rates. To assess the best storage conditions for active dibutylhexanediamine (DBHD) coatings with integrated acid complexes, three conditions were assessed: 20oC, -20oC, and 37oC. These samples are assessed at multiple time points for their NO flux rate using a Sievers 280i Nitric Oxide Analyzer. Average net flux for coated tubing is 28.7 at 20oC, 34.7 at -20oC, and 34.5 at 37oC 32 days from fabrication. Average peak NO flux was 79.7 at 20oC, 105.7 at -20oC, and 69.4 at 37oC at a 32 days from fabrication. This shows that the best place to store our tubes would be in a freezer at 20o C to preserve peak NO flux, however net NO flux rate between 37oc and -20oC appear to be similar (p>0.80).

BIOT 416

Kinetic characterization and differentiation of atrazine catabolism via Pseudomonas sp. strain ADP biofilms and planktonic cells

Michael Delcau2, [email protected], Tonya L. Peeples1. (1) Chem Biochem Eng, University of Iowa, Iowa City, Iowa, United States (2) Chemical and Biochemical Engineering, University of Iowa, Iowa City, Iowa, United States

Atrazine is an herbicide used to prevent the emergence of broad-leaved and grassy weeds in a wide variety of crops. It is also the second-most applied herbicide in the U.S. with nearly 76 million pounds applied on an annual basis. Due to its classification as an endocrine disruptor compound and uncertain toxicity, the environmental protection agency established a maximum contaminant level of 3 parts per billion of atrazine in drinking water. Bioremediation, the act of exploiting microbes’ natural capacity to use recalcitrant compounds for growth and metabolism, is considered as an environmentally conscience and cost-effective method to degrade atrazine. Pseudomonas sp. ADP, a model strain for atrazine degradation, metabolizes atrazine via a six-step enzymatic pathway to yield carbon dioxide and ammonia in the complete mineralization. Each pathway enzyme, AtzA-AtzF, is encoded by a corresponding gene, AtzA-AtzF. To further facilitate atrazine degradation, the use of biofilms is considered due to their unique physical and genetic properties capable of efficiently metabolizing contaminants. Pseudomonas sp. ADP biofilms were grown for 10-days in a drip-flow reactor under a consistent feed of media containing 30 ppm atrazine. In parallel, planktonic cells were grown in shake flasks for 10-days in similar conditions. Samples were retrieved every 12 hours for HPLC analysis. Atrazine and two downstream metabolites were quantified to differentiate the degradation kinetics between the conditions of biofilm and planktonic cells. In the planktonic cells, atrazine immediately decreased, while the appearance of the downstream metabolites slowly increased. Conversely, the degradation kinetics in biofilms demonstrated a more complex pattern. However, atrazine decreased at a greater rate overall in the biofilm mode of growth, suggesting it may be more efficient to use biofilms for decreasing the concentration of atrazine compared to planktonic cells. These results can be translated to bioremediation systems for alternative recalcitrant compounds in the environment.

BIOT 417

Microfluidic device to characterize the effect of orthogonal chemical gradients on 3D cancer cell migration

Joshua M. Campbell, [email protected], Sharif Rahman, Adam T. Melvin. Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana, United States

Once a primary tumor has metastasized to another location in a cancer patient, there is approximately a 20% chance of survival. In order to prevent cancer metastasis in the future, it is necessary to understand how cancer cells migrate from a primary tumor to the circulatory system and from the circulatory system to a secondary metastatic site. Within the tumor environment, cancer cells are subjected to competing gradients including chemicals (chemotaxis), extracellular matrix proteins (hapotaxis), and variations in matrix stiffness (durotaxis) which results in the directed migration of cells. Highly metastatic strains of breast cancer, such as the triple negative cell line MDA-MB- 231, will chemotax towards certain chemoattractants such as SDF-1α and EGF. To explore this phenomenon, we have fabricated a microfluidic device that creates a “flow- free” chemical gradient within a chamber of cells seeded in a 3D collagen matrix. The chemical gradient is formed by the diffusion of chemoattractant within an agarose hydrogel using a three-channel microfluidic device composed of PDMS imprinted with three parallel channels including a source channel, a middle flow-free cell channel, and a sink channel. While the chemotactic response of MDA-MB-231 cells to gradients of either SDF-1α or EGF have been characterized, there is little known about how these cells respond to competing gradients or combinations of these chemoattractants. To create an environment with multiple gradients that better replicates in vivo conditions, we designed a second microfluidic device capable of inducing parallel or orthogonal chemical gradients by modifying the initial design to include two source channels, two sink channels, and a middle flow-free cell channel in the shape of a square. This device has the capability to study how cancer cells respond to multiple gradients as opposed to singular gradients. The gradients produced by the device have been simulated using COMSOL, and preliminary results will show MDA-MB-231 cells migrating within the orthogonal device.

BIOT 418

Effects of superparamagnetic Fe3O4 nanoparticles and magnetic fields on oral biofilms

Jane Nguyen, [email protected]. Chemical & Biological Engineering, University of New Mexico, Albuquerque, New Mexico, United States

The CDC approximates 80% of the adult population in America to suffer from periodontitis, an irreversible disease that stems from gingivitis. Gingivitis is caused by poor oral hygiene and plaque production through primary and secondary colonizing bacteria. Without proper oral care, plaque will harden under the gums to cause irritation, resulting in gingivitis. Therefore, it is imperative to prevent the aggregation of primary and secondary colonizers. Attempts have been made to incorporate metal oxide nanoparticles into dental care, but limitations are found in the need to develop nontoxic nanoparticles that display antimicrobial potency. To address this, saliva containing mixed bacterial species was collected from adult individuals to form biofilm. The control contained biofilm in phosphate-buffered saline (PBS). The experimental samples subjected the biofilm to a two-fold concentration gradient of superparamagnetic iron oxide nanoparticles suspended in PBS. The control and experimental samples were exposed to two magnetic field treatments, static and alternating, induced by molybdenum magnets. A static magnetic field strength of 4.44 kGs was induced from the bottom of the samples for 6 hours. The alternating field treatment exposed the samples to a magnet with a field strength of 4.44 kGs placed below the samples and a magnet with a field strength of 0.12 kGs placed above. Both magnets alternated for 30 minutes for 6 hours. An XTT reduction analysis indicated that both magnetic field treatments significantly suppressed biofilm metabolic activity. Fluorescent confocal laser scanning microscopy displayed the static magnetic treatment to be more effective in disrupting the biofilm than the alternating treatment. This difference was more prominent when nanoparticles were used. A Mann-Whitney U-test yielded a p-value of less than 0.05, which concludes that the nanoparticles disrupted the bacterial biofilm significantly but their ability to kill bacteria was minimal. Future studies will apply the nanoparticles to an antimicrobial mouth rinse. Two strains of bacteria will be used to simulate the formation of oral biofilm: Streptococcus gordonii, a primary colonizer, and Fusobacterium nucleatum, a secondary colonizer. The biofilm samples will be subject to a two-fold concentration nanoparticle concentration in mouthrinse and a time-dependent matrix study will be conducted to observe the effect on oral bacteria after sixty seconds with ten seconds increments.

BIOT 419

Microfluidic co-culture of breast cancer cells and adipose stem cells

Sharif Rahman2, [email protected], Katie A Render1, Joshua M. Campbell2, Elizabeth C Martin1, Adam T. Melvin2. (1) Biological and Agricultural Engineering, Louisiana State University, Baton Rouge, Louisiana, United States (2) Cain Department of Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana, United States

Breast cancer is currently the second leading cause of cancer deaths in women in the United States. In the tumor microenvironment, adipose stem cells (ASCs) growing in close proximity to breast cancer cells can induce phenotypic and genotypic changes in both cell types, which can increase breast tissue density and result in poorer patient prognosis. However, the precise mechanism by which the two different types of cells influence each other is not clearly understood. In this study, a microfluidic device was developed to investigate this phenomenon by simultaneously co-culturing both MDA- MB-231 cells and ASC. The microfluidic device consists of two separate layers of polydimethylsiloxane (PDMS) and agarose enclosed in a Plexiglas chamber. The fluidic channels were imprinted into the top PDMS layer using standard soft lithography and PDMS replication techniques. The device dimensions consisted of four 600 μm wide parallel channels with a total operation length of 10 mm and channel height of 150 μm. The two center channels were designed to be 200 µm apart which allows for facile diffusion of external chemical cues generated by the two different cell types as confirmed by COMSOL simulation. Prior to experimentation, the bottom agarose layer was pre-treated with poly-d-lysine for 1 h at 25°C to facilitate cellular attachment within the device. Initial experiments consisted of single cell culturing at 370C in both channels for up to 7 days to obtain baseline measurements of cell growth and phenotype as well as confirming that the device itself did not impede cell viability. Subsequent experiments were performed with both species followed by post-experimental analysis. An advantage to this approach is that the device can easily be disassembled to allow for terminal analysis on the cells including immunostaining and quantitative polymerase chain reaction (qPCR). MDA-MB-231 cells were successfully cultured in the device resulting in an increase in cell density analogous to growth in a culture flask. COMSOL simulations were performed to approximate the diffusion of cytokines and growth factors across two ‘flow-free’ culture chambers through agarose to estimate the time required for communication between cell types to occur. This microfluidic device supports the co- culture of breast cancer cells for 5-7 days.

BIOT 420

Examine how oscillating patterns of chemical gradients break the spatial range limitations of conventional chemotaxis Sharif Rahman1, [email protected], Joshua M. Campbell1, Ian C Schneider2, Adam T. Melvin1. (1) Cain Department of Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana, United States (2) Department of Chemical and Biological Engineering, Iowa State University, Ames, Iowa, United States

Cellular responses to the spatial and temporal pulsing of chemical gradients are common in biological systems which affect the directed migration of cells via a chemotactic response. Traditional approached to study cellular chemotaxis utilize stable chemical gradients; however, in vivo conditions result in both competing and oscillating gradients which the cell must interpret as it migrates. Here, a microfluidic approach is utilized to generate stable and tunable chemical gradients to elicit a dynamic cellular response during 3D chemotaxis. Usually, cells will lose the ability to sense external gradients under high concentrations of chemoattractant. Here, long-range chemotaxis is possible by the temporal pulsing of spatial gradients by establishing and eliminating the gradient within the migratory channel. The microfluidic device consists of two layers of polydimethylsiloxane (PDMS) and agarose enclosed in plexiglass chamber where the main design principal for the device is the high permeability of the agarose layer to facilitate chemical diffusion perpendicular to the direction of flow. The device consists of three parallel channels: a center channel for cell culture between top and bottom channels acting as the source and sink solutions. A constant source solution with chemoattractant flowing through the top channel induces a linear chemical gradient perpendicularly across cell culture channel. The oscillation-type chemical gradient can be generated keeping the source solution flow ‘on’ and ‘off’ for certain time. The flow of both the source and sink can also be altered over time to build and erase gradients allowing for temporal control over the gradient within the device. The cells are seeded into the center channel in a collagen hydrogel to examine the 3D migration of the cells in a system closer to the tumor microenvironment. Two different growth factors were used (SDF-1α and EGF) to elicit a tactic response in the highly motile MDA-MB-231 triple negative breast cancer cells. The chemotactic index and directional migration of the cells were determined using COMSOL simulations to model the chemical gradient and custom MATLAB code to track the cells. Preliminary results have found that the population of breast cancer cells response more strongly to SDF-1α; however, the chemotactic response of the cells is highly heterogeneous with cells responding different to the pulsing chemical gradients.

BIOT 421

Developing a high affinity, dynamic scaffold toolkit for control of intracellular metabolic flux

Alexander A. Mitkas2, [email protected], Wilfred Chen1. (1) Chemical Engineering, University of Delaware, Newark, Delaware, United States (2) Chemical and Biomolecular Engineering, University of Delaware, Newark, Delaware, United States

It is common practice in modern biotechnology to introduce nonnative enzymatic pathways in platform organisms such as E. coli to produce biochemical molecules ranging from specialty chemicals to biofuels. Traditionally, product titer is maximized by fine-tuning the nonnative enzymes’ activity and expression levels; however, in many instances the work can be time consuming and fruitless. Enzyme clustering is an alternative approach that has demonstrably helped increase nonnative pathway titers. However, without dynamic control of when the enzymes' cluster forms, unwanted metabolic imbalances within the cell can often result. To address the issues with these optimization techniques and provide an alternative way to improve nonnative pathway productivity, a high affinity, dynamic scaffold toolkit for the control of metabolic flux was designed. The toolkit building blocks are small RNAs and protein components taken from the CRISPR/Cas Type I systems. These Cas proteins bind to the small RNAs with high affinity and sequence specificity which ensures that the scaffold only dissociates based on a selected cue, in the form of RNAs expressed when certain intracellular conditions are satisfied through toehold-mediated strand displacement. Nonnative proteins are clustered and flux through the pathway is increased when conditions for production of the target molecule are favorable. When resources are depleted, the scaffold disassembles, thus reducing flux through the pathway and allowing the cell to return to its original conditions. Using the split luciferase reporter system, a significant increase in the luminescence of the system has been demonstrated to occur only when all the correct components of the scaffold are simultaneously expressed. The scaffold will also be applied to increase the productivity and specificity of a nonnative pathway. The implementation of the scaffold toolkit will allow for intracellular dynamic process control while also providing an alternative approach for increasing nonnative pathway productivity.

BIOT 422

Alcoholic liver damage prevention of Moringa oleifera extract

Jae Gyu Park1, [email protected], Chang Geon Kim2, Seong Taek Oh3, Sukkum Chang4, Kyung Jong Shin1. (1) Advanced Bio Convergence Center, Pohang TechnoPark, Pohang, Korea (the Republic of) (2) Department of Life Science, Yeungnam University, Gyeongsan, Korea (the Republic of) (3) College of Pharmacy, Yeungnam University, Gyeongsan, Korea (the Republic of) (4) Department of Genetic Engineering, Daegu University, Gyeongsan, Korea (the Republic of)

Alcoholic steatohepatitis has caused alcohol consumption and the liver damage is due to diet habits of alcohol administration. We investigated the abilities of Molinga oleifera and their components to inhibit liver accumulation of triglyceride. Molinga oleifera leaf was extracted by 70% EtOH and evaluated for alcoholic liver damage prevention. Molinga oleifera leaf extract showed interesting prevention activity. As a results, flavonoids were found from Molinga oleifera extract by HPLC, indicated that, among 7 components, the content of flavonoid was 8.2mg of 3-O-caffeoylquinic acid, 5.0mg of 4- O-caffeoylquinic acid, 6.3mg of multiflorin B and 8.1mg of Quercetin-3-O-ß-D-glucoside. Prevention of alcoholic steatohepatitis was evaluated by histologic data from C57B/L6 mice and liver function was detected by radio isotope 99mTc from SPECT/CT.

BIOT 423

Probing protein-ice interactions using high resolution synchrotron X-ray diffraction

Bakul Bhatnagar3, [email protected], Xin Wen4, [email protected], Boris Zakharov6, Alexander Fisyuk5, Fawziya Karim3, Kevin J. Quach4, Iurii Seretkin7, Andrew Fitch8, Elena Boldyreva2, Evgenyi Shalaev1. (1) Allergan plc, Irvine, California, United States (2) NSU Inst Solid State Chemistry, Novosibirsk, Russian Federation (3) Pfizer, Andover, Massachusetts, United States (4) Dept Chem Biochem, Cal State La, Los Angeles, California, United States (5) Organic Chemistry, Omsk F.M. Dostoevsky State University, Omsk, Russian Federation (6) Department of Reactivity of Solids, Institute of Solid State Chemistry and Mech. , Novosibirsk, Russian Federation (7) Institute of Geology and Mineralogy RAS, Novosibirsk, Russian Federation (8) The European Synchrotron Radiation Facility (ESRF), Grenoble, France

Freezing is a ubiquitous process in nature and pharmaceutical manufacturing processes. In the present study, several classes of proteins (isoelectric points, pIs 4.7 to 11.4) were investigated, including an insect antifreeze protein (AFP), model proteins (recombinant human albumin, lysozyme), and a pharmaceutical protein (monoclonal antibody). AFPs are found in many cold-adapted organisms and are well-known for their ability to inhibit ice growth through binding to specific ice crystal surfaces. AFPs from insects generally exhibit higher antifreeze activity than those from other organisms. However, there have been limited studies on insect AFP-ice interactions, which impedes our understanding of AFP-ice interaction mechanisms and limits their application.

We present a new method for investigation of protein-ice interaction, which is based on measuring characteristic features of X-ray diffraction (XRD) patterns of hexagonal ice (Ih). Protein-ice interaction can be direct (sorption of protein on ice crystals), and / or indirect (partitioning of protein molecules into the quasi-liquid layer), thereby exhibiting different “signatures” in the Ih XRD patterns. The experiments were performed using high-resolution synchrotron X-ray diffraction (SXRD) at ID22 beamline at the ESRF, which is optimized to eliminate the instrumental broadening of diffraction lines and reduce the preferred orientation effects, thereby facilitating the quantitative analysis of the X-ray diffraction data.

SXRD patterns of ice were obtained at 100 and 228 K in the absence and presence of the proteins. AFP inhibited growth of ice Ih crystals in directions corresponding to hkl 002, 110, 200, and 105, suggesting a direct AFP-ice interaction. Peak analyses also revealed that the AFP decreased ice crystal size and increased strain. Preliminary data analysis suggests a possible contribution of the pI of the protein on phase behavior of Ih. In this presentation, we present a new approach to investigate the protein-ice system, which could improve understanding of ice-protein interactions (AFPs, non- AFPs), and provide insights into degradation and stabilization mechanisms of frozen therapeutic proteins.

BIOT 424

Isolation and characterization of the EphA2 cytoplasmic domains

Jaden Shirkey1, [email protected], Paloma Gil-Rodriguez2, Matthias Buck2. (1) Biological Sciences, Ferris State University, Coldwater, Michigan, United States (2) Physiology and Biophysics, Case Western Reserve University, Cleveland, Ohio, United States

Eph Receptors, a type of Tyrosine Receptor Kinase, play an integral role in cell-cell interactions. These receptors, having membrane-bound ligands, are heavily involved in processes such as of cell migration, axon guidance, and angiogenesis. As such, they are closely correlated with many aspects of certain cancers. The intracellular portion of these transmembrane receptors contain three major domains, including a linker region, a kinase domain, and a Sterile Alpha Motif (SAM) that is thought to be involved in protein-protein interactions. An understanding of Eph structure and function could serve as a potential therapeutic target for some types of cancer. In an attempt to do such, various segments of the intracellular domain of the EphA2 receptor were expressed in BL21 Competent E. Coli. While two vectors were initially attempted in the transformation (pET-28a and pET32-LIC), higher success was found with pET32-LIC vector, which coded for a fused thioredoxin tag (in addition to a His- Tag). Purification of the expressed proteins was successful, utilizing Ni-affinity and gel- filtration chromatography. Upon isolation of the both the full intracellular portion, and a truncated segment (lacking the SAM domain), kinase activity was determined. Finally, HSQC-NMR was used to identify potential protein-protein interactions involving the SAM domain and/or the kinase domain. While some observations were made, more analysis must be conducted to solidify any correlation.

BIOT 425

Confocal Raman microscopy investigation of cytochrome c-phospholipid interactions which induce permeability in mixed phospholipid vesicle membranes

Jay P. Kitt3, [email protected], David Bryce2, Shelley D. Minteer2, Joel M. Harris1. (1) Univ of Utah Chem Dept, Salt Lake City, Utah, United States (2) University of Utah, Salt Lake City, Utah, United States (3) Chemistry, University of Utah, Salt Lake City, Utah, United States

Permeabilization of the outer mitochondrial membrane (OMM) is a key step in apoptosis leading to release of cytochrome c (CytC) activating the apoptotic caspase cascade and resulting in cell-death. It has been suggested that CytC participates in the permeabilization process through interaction with cardiolipin (CL), an inner mitochondrial membrane phospholipid which translocates to the OMM preceding CytC release. Two mechanisms of permeabilization have been proposed: One is electrostatic, wherein positively-charged CytC interacts with negatively (-2) charged CL inducing curvature and pore-formation. The other is CytC insertion into the membrane where CytC unfolds and interacts with CL to induce an inverse phospholipid phase and pore formation. In this work, CytC-induced permeabilization of phospholipid vesicle membranes of varying composition is investigated via Raman spectroscopy. Vesicles comprising 80mol% POPC (16:0-18:1 phosphocholine, neutral) and 20mol% POPG (16:0-18:1 phosphoglycerol, -1 charge) or 90mol% POPC and 10mol% CL were formed in a solution of membrane-impermeable small molecule tracer nitrobenzenesulfonate (NBS). Vesicles were optically trapped, translated into CytC solutions and monitored for NBS leakage. Leakage was observed in POPG and CL vesicles, supporting the electrostatic interaction hypothesis. Interestingly, disappearance of resonance Raman bands associated with Met-80 coordination of the heme is observed in both cases indicating protein unfolding. However, in CL, distinct changes are observed in the phospholipid acyl-chain C=C modes which are not observed in POPG, suggesting interaction of CytC with CL occurs deeper in the membrane. Additionally, small- molecule permeabilization can occur without pore formation, typically through boundaries between ordered and disordered phases; this type of permeabilization would not allow the much larger CytC to traverse the membrane. To determine the nature of permeabilization in each case, the same experiments were conducted where CytC was used as the “tracer.” CytC-filled vesicles were translated into buffer and monitored for leakage. CytC leakage was only observed in CL vesicles, indicating pores that are sufficiently large for CytC translocation were formed. Despite protein unfolding in both cases, only CL membranes allow for CytC leakage which proceeds through large pore formation induced through interaction with CytC deep in the CL membrane in support of a membrane-insertion hypothesis.

BIOT 426 Contribution of the C-terminus to A2AR activity and stability

Kirsten Swonger2, [email protected], Anne S. Robinson1. (1) Chem and Biomolecular Eng, 300 Lindy Boggs Bldg, Tulane University, New Orleans, Louisiana, United States (2) Chemical & Biomolecular Engineering, Tulane University, New Orleans, Louisiana, United States

G-protein coupled receptors (GPCRs) are seven-transmembrane domain membrane proteins that are drug targets for a number of conditions including pain, hypertension, and schizophrenia. The adenosine A2A receptor (A2AR) is a class A GPCR that is expressed widely throughout the body and is a therapeutic target for many diseases including neurodegenerative, inflammatory, and heart diseases. A2AR is well characterized and has been crystallized in many active conformations to better understand the conformational changes that occur during ligand binding to the receptor. Presently, all crystal structures of A2AR have used a truncated variant, A2AΔ316R, that lacks 96 amino acids of the flexible C-terminus to facilitate crystallization. Our lab has shown that the full-length C-terminus was critical for activating downstream signaling cascades, indicating that A2AΔ316R does not maintain all active receptor functions of the full-length receptor. Here, we use ligand binding to further characterize full-length A2AR, A2AΔ316R, and Rag23, a thermostable A2AΔ316R variant with five point mutations designed to favor agonist binding. FITC-APEC, a fluorescent agonist, was used to quantify ligand binding via fluorescence anisotropy. We used ligand binding to determine equilibrium binding, stability at increasing temperatures, competition with agonists and antagonists, and kinetic rates and find that the C-terminus plays a significant role in providing stability to the native receptor.

BIOT 427

Excipient choice for protection against metal-catalyzed oxidation: Impact of chelator on stability of biologic drug products

Mary E. Krause1, [email protected], Thomas Haby1, Sanket Patke1, Mark Bolgar1, Masano Huang1, Lydia Breckenridge2, Mehrnaz Khossravi1. (1) Drug Product Science and Technology, Bristol-Myers Squibb, New Brunswick, New Jersey, United States (2) Chemical and Synthetic Development, Bristol-Myers Squibb, New Brunswick, New Jersey, United States

Metal-catalyzed oxidation (MCO) of biologic drug products can cause aggregation, formation of low molecular weight species, charge variants, and site-specific oxidation or oxidative carbonylation of various amino acid residues. MCO can result from binding of redox-active metals to the protein itself or to buffers and other excipients in solution. The resulting metal complexes can lead to Fenton chemistry, which generates hydroxyl radicals and/or iron IV oxo species that cause protein oxidation and degradation. Because metal ions can be introduced into drug products from many sources, including water, excipients, and leaching from stainless steel during manufacture, chelators are often added to formulations in order to mitigate the risk of metal-catalyzed oxidation. The chelators EDTA and DTPA are commonly used; however, the structure and properties of the resulting metal-EDTA and metal-DTPA complexes are different. The primary coordination sphere around the metal center dictates reactivity, where metal- EDTA complexes are particularly reactive. Thus, while DTPA is able to fully protect many BMS compounds against metal-catalyzed oxidation, EDTA performance is very protein specific and may not fully protect against MCO. Here, we compare EDTA and DTPA performance against MCO for five BMS compounds under accelerated stability conditions.

BIOT 428

Synthesis of new β-amino acids derivatives as anti trypanocidal agent and docking & Inhibition study on trans-sialidase enzyme

Muhammad Kashif1, [email protected], Muhammad Ashfaq2, Gildardo Reveria1. (1) Instituto Politecnico Nacional, Reynosa, Mexico (2) Chemistry, The Islamia University, Bahawalpur Pakistan , Bahawalpur, Pakistan

The trans-sialidase enzyme plays a key role to enhance the virulence of trypanosoma cruzi for Chagas disease, by transfer of sialic acid from host to parasite surface. In this work we designed and synthesised novel β-amino acids derivatives. The molecular docking study showed, the compounds M-11, N-11 has the highest binding energies (- 11.1 kcal/mol) as compared to reference DANA (-7.8 kcal/mol), the natural ligand of trans-sialidase. The 3D and 2D analysis of these compounds in catalytic pocket of the enzyme, showed the hydrogen bond, pi-pi stacking, pi anion, hydrophobic & van der waals interaction with all important amino acid residues (Arg35, Arg245, Arg314, Tyr119, Trp312, Tyr342, Glu230 and Asp59). In vitro trypanocidal activity were evaluated, on trypomastigotes of NINOA & INC-5 strains, which showed that compounds with electron donating groups support the NINOA, with maximum lysis 85% and electron withdrawing group showed the higest lysis about 80% lysis in INC-5. Furthermore, compound 11in series M, N, & O showed the 65-68 % lysis in both strains which is higher than reference drugs nifurtimox and benznidazole (28-49%). The enzyme inhibition assay of selected compounds was determined by using high performance ion exchange chromatography with pulse amperometric detection. Compound M-11 & M-4 showed the 86.9% & 82.5% enzyme inhibition. A good correlation between docking and enzyme inhibition was also found.

BIOT 429

Folding analysis of bovine pancreatic trypsin inhibitor (BPTI) with aromatic thiols and disulfides in vitro

Na Zhang, [email protected], Ram Marahatta, Watson J. Lees. FIU, MIAMI, Florida, United States Many recombinant therapeutic proteins are produced in bacteria, but the efficient expression of native proteins can be limited by the minimal ability of bacteria to form disulfide bonds in vivo. As almost all therapeutic proteins contain disulfide bonds to stabilize their native structure, it is often necessary to employ in vitro oxidative folding process to form the native disulfide bonds. However, the in vitro folding of disulfide containing proteins is usually slow and low yielding. Traditionally, redox buffers containing small molecule thiols and disulfides, such as glutathione and glutathione disulfide (GSH/GSSG) or reduced and oxidized dithiothreitol (DTTox/DTTred), are used to facilitate the in vitro oxidative protein folding process. Aromatic disulfides are small molecules that were designed based on the physical properties of the active site of protein disulfide isomerase (PDI), which catalyzes the folding of disulfide containing proteins in eukaryotes. Oxidative folding of disulfide containing proteins have confirmed that aromatic thiols and disulfides increased both the folding yield and rate compared to traditional redox buffers, such as GSH/GSSG. In this study, bovine pancreatic trypsin inhibitor (BPTI) was used as a model system to analyze the oxidative folding process with aromatic thiols and disulfides. Aromatic thiols with an elongated alkyl group on the aromatic ring are expected to increase interactions with the hydrophobic core of disulfide containing proteins during folding, allowing more facile access to buried disulfide bonds. Therefore, the buried thiols in the hydrophobic core of protein folding intermediates can be easily accessed by aromatic small molecules, and the intermediates can then be oxidized and rearranged to the native form of the protein more efficiently.

BIOT 430

Prediction and characterization of the equilibrium ensembles of TGF-β3 using bias exchange metadynamics

Richa Singh, [email protected], Gaurav Goel. Chemical Engineering, IIT Delhi, New Delhi, Delhi, India

TGF-β3 (transforming growth factor-β) is a therapeutically important protein belonging to the family of multifunctional cytokines. It has high exposed hydrophobic surface which promotes its aggregation at physiological pH affecting its refolding yield. Aggregation mechanism of TGF-β3 can be predicted with the knowledge of unfolding intermediates which are short-lived therefore difficult to capture using spectroscopic techniques. To this end, we have used well-tempered bias exchange metadynamic simulation (WT- BEMD) of monomeric TGF-β3 in typical refolding buffer conditions: 20% v/v dimethyl sulfoxide (DMSO), 30 mM (CMC = 4-11 mM) 3-((3-cholamidopropyl) dimethyl ammonio)-1-propanesulfonate (CHAPS), pH 9.5 and temperature 300 K. An 8-replica simulation was performed of which 7 were biased along seven different CVs and the last replica was kept unbiased. Besides typically used CVs, we have used two additional CVs to facilitate the folding of this large protein: a CV based native contacts and a CV based on NMR chemical shifts. A bin based clustering method was employed on a total of 8 μs trajectory coupled with Markovian kinetic transition network model. Subsequently, we determined 3 long-lived metastable ensembles of TGF-β3 of which the lowest free energy ensemble shows significant similarity to the reported solution structure of TGF-β3. The backbone chemical shifts in the most stable structure of this ensemble has a high correlation (0.7) with the reported values for the solution structure of TGF-β3. We have also analyzed interactions between CHAPS and various metastable states of TGF-β3 to elucidate the mechanism for improved refolding yields on addition of this surfactant. Preferential binding coefficient indicates selective binding of CHAPS molecules to: (i) residues belonging to the disrupted alpha helices and (ii) the hydrophobic cavity between the hairpin loop (residue 92-95) and the extended loop regions (residue 25-32) of TGF-β3. There was a considerable disruption of secondary structure in the other two metastable ensembles indicating the presence of potential intermediates or denatured states. Differences in solvation of these intermediates further the understanding of positive effects of CHAPS on refolding. Finally, knowledge of these unfolding intermediates along-with the native solution structure can be used to design small molecule excipients to inhibit aggregation and/or increase refolding yields.

BIOT 431

In silico study of dimer association of insulin during aggregation

Rit Pratik P. Mishra3, [email protected], Richa Singh1, Tirumala Rao3, Gaurav Goel2. (1) Chemical Engineering, IIT Delhi, New Delhi, Delhi, India (2) Chemical Engineering, Indian Institute of Technology (IIT), Delhi, New Delhi, Delhi, India (3) Chemical Engineering, Indian Institute of technology Delhi, New Delhi, Delhi, India

Insulin, a therapeutic protein, can aggregate during manufacture and is considered detrimental to its quality. Early identification aggregation inducing factors would be an essential aspect concerning the design of mitigation strategies during manufacturing processes. The timescales and lengthscales relevant for aggregation makes the use of accurate all-atom (AA) explicit solvent simulations for prediction of aggregation thermodynamics and/or kinetics very difficult. Various coarse grained models have been used to access such timescales for protein folding studies but such models are biased towards native interactions and hence, not suitable for studying protein association leading to aggregation which is mainly driven by non-native interactions between aggregation prone species. We have used a combination of all atomic and MARTINI coarse grained model to study the dimerization process of insulin separated by .5 nm. The best rigid body docked complex between folded and partially folded insulin molecules was found through HexServer, MD cluster analysis and mmpbsa ranking scheme and was used for our study. The molecules were allowed to interact and associate for 50ns MD runs with constraints applied on the side chains to prevent the structures and interacting surfaces of the proteins from changing much. At the end of CG runs, the dimer formed was converted to AA model and allowed to run for further 20ns to allow for the changes previously prevented by the CG restraints. The final structures show fairly good similarity in terms of the interacting residues with the initial rigid docked complex. The final dimer also showed similarities with the final dimer obtained through AA 200ns MD runs. The protein association rates of the various final dimers calculated through transition-state theory model are also close to each other suggesting that our method can be further extended to study aggregation pathways in much smaller computational time

BIOT 432

Divalent metal ion-independent cell adhesion events mediated by E-cadherin- binding DNA aptamer that forms parallel type G-quadruplex with three long loops

Ryo Maruyama1, [email protected], Toru Yoshitomi1, Fumiya Wayama1, Koji Wakui1, Koki Makabe2, Hitoshi Furusho3, Keitaro Yoshimoto1,4. (1) Department of Life Sciences, Graduate School of Arts and Sciences, The University of Tokyo, Tokyo, Japan (2) Graduate School of Science and Engineering, Yamagata University, Yamagata, Japan (3) Chemical General Division, Nissan Chemical Industries, Ltd., Chiba, Japan (4) JST, PRESTO, Tokyo, Japan

Nucleic acid aptamers are single-stranded DNA or RNA ligands that recognize their target and selected from random oligonucleotide libraries through a screening method with systematic evolution of ligands by exponential enrichment (SELEX). As cell-cell and cell-matrix interactions regulate many cellular functions including spreading, migration, proliferation, and differentiation, aptamers that trigger cell adhesion may be crucial ligand molecules in the field of cell biology; however, aptamers that can trigger cell adhesion have not yet been studied. Here, as a target of aptamer that induce cell adhesion, we focus on E-cadherin that is the principle cell-cell adhesive molecule of epithelia. Cadherins are a large family of transmembrane or membrane-associated glycoproteins that mediate specific cell-cell interaction at intercellular junctions in a calcium-dependent manner, functioning as key molecules in the morphogenesis of a variety of organs. In this study, DNA aptamers that selectively and strongly bind E- cadherin in a calcium-independent manner were discovered using SELEX with next generation sequencing. Among the aptamer sequences identified, interestingly, a unique parallel-type G-quadruplex-structured sequence with the Kd values of 0.6 nM was found and triggered divalent metal ion-independent cell adhesion successfully (Figure 1).

Figure 1. Phase contrast microscopy images of A549 on aptamer modified surface. Aptamer- modified plates were incubated with A549 (3×104 cells/well) for 24 hour (left) and washed 1 mM EDTA-DPBS (right).

BIOT 433

Extraction of protein thermodynamic parameters from high-throughput differential scanning fluorimetry assays

Thaiesha A. Wright1, [email protected], Jamie Stewart1, Dominik Konkolewicz2, Richard C. Page2. (1) Chemistry and Biochemistry, Miami University, Oxford, Ohio, United States (2) Department of Chemistry and Biochemistry, Miami University, Oxford, Ohio, United States

Thermodynamic properties of protein unfolding have been extensively studied; however, these methods used have required significant preparation time and high protein concentrations. Here, we present a facile, simple, and high-throughput differential scanning fluorimetry (DSF) method that enables a complete set of thermodynamic parameters of protein unfolding to be extracted. This method assumes a two-state, reversible protein unfolding mechanism and provides the capacity to quickly analyze the biophysical mechanisms of changes in protein stability and to more thoroughly characterize the effect of additives, inhibitors, or pH. We show the utility of the DSF method by analyzing the thermal denaturation of lysozyme, carbonic anhydrase, chymotrypsin, horseradish peroxidase, and cellulase. Compared to similar biophysical analyses by circular dichroism, DSF provides greater than 24-fold reduction in experimental time. This study opens to door to rapid characterization of protein stability on low concentration protein samples.

BIOT 434

DNA cleavage via prokaryote argonaute mediated by an uncharacterized N- terminal domain

Kok Z. Lee1, [email protected], Arren Liu2, Archana Kikla2, Kevin Solomon1. (1) Agricultural & Biological Engineering, Purdue University, West Lafayette, Indiana, United States (2) Biology, Purdue University, West Lafayette, Indiana, United States

Gene editing tools harness prokaryotic defense mechanisms, such as CRISPR and Argonaute, which protect the host from invading nucleic acids. Unlike CRISPR, prokaryotic Argonautes (pAgos) have recently been demonstrated to cleave single and double-stranded nucleic acids without regard for an adjacent sequence motif at gene loci via both targeted guide-dependent and non-specific guide-independent mechanisms. However, little is known about the protein domains that mediate the transition between guide-dependent and guide-independent mechanisms, or how they may be altered to control cleavage specificity. Thus, we heterologously expressed a mesophilic pAgo and engineered mutants in E. coli as a model to derive mechanistic insight. To better understand the mechanism of our model pAgo, we created an in vivo survival assay that targets either essential or non-essential genes to determine whether the pAgo interacts with DNA or RNA. Results from the survival assay reveal that targeted pAgo guide-dependent activity reduces survival via a DNA interaction. Further experiments suggest this may involve DNA cleavage as in vitro incubation of purified pAgo with plasmid DNA cleaves the DNA in the absence of a targeting guide. We also observe similar guide-independent cleavage activity in vivo; strains expressing pAgo without guide display reduced plasmid yield that is highly mutated. Based on our protein structure modeling, the model pAgo has a well-conserved catalytic tetrad, which is required for nucleic acid cleavage. Interestingly, we found that the pAgo also possesses an additional uncharacterized N-terminal domain, which is absent in all well- characterized pAgos. Deletion of this domain abolishes all cleavage activity in vivo suggesting a critical role in pAgo function. Similar point mutations of conserved residues also reduced pAgo activity. Collectively, our work provides insight into the structure and function of our model pAgo, while providing a roadmap for its future exploitation as a novel tool for gene editing and genome engineering.

BIOT 435

Insights from molecular dynamics simulations into the structure and dynamics of ITPA mutants

Yao A. Houndonougbo, [email protected]. Chemistry biochemistry, Eastern Washington Univ, Cheney, Washington, United States

Inosine triphosphatase (ITPA) is an enzyme that catalyzes the conversion of non- canonical purine nucleotides inosine triphosphate(ITP), deoxy ITP (dITP), and xanthosine triphosphate(XTP) to the corresponding nucleoside monophosphate and pyrophosphate. Deficient ITPA activity can cause the accumulation of ITP, dITP, and XTP, which may be introduced into RNA and DNA resulting in possible genetic mutation and instability. The ITPA variants P-32 and R-178 are associated to epileptic encephalopathy and purine analog drug toxicity. The human ITPA is a homodimer which consists of central beta-sheet with two lobes having mixed alpha-beta structures. In this study, we used molecular dynamics (MD) simulation to understand the effect of P32 and R178 mutations on the structure of human ITPA. The results for more 300 ns simulations show the influence of point mutation on the 3D structure and flexibility of ITPA. The analysis of the simulation trajectories also revealed the detail conformation change of the ITPA mutant. This study is an important starting point for protein engineering design of ITPA.

BIOT 436

Understanding co-solutes effects on viscosity and protein interactions in highly concentrated monoclonal antibodies through protein structure and dynamics

Jessica Hung2, [email protected], Barton Dear2, Wade Zeno1, Carl Karouta2, Maria Nieto2, Logan Wilks2, Ayush Sharma2, Jeanne Stachowiak1, Thomas Truskett2, Keith P. Johnston2. (1) Biomedical Engineering, University of Texas at Austin, Austin, Texas, United States (2) Department of Chemical Engineering, University of Texas at Austin, Austin, Texas, United States

Highly concentrated (> 200 mg/mL) monoclonal antibody (mAb) formulations with low viscosities and high stability are strongly desired for subcutaneous drug delivery for the treatment of diseases such as cancer. Small-molecule co-solutes such as arginine and various electrolytes have in some cases been shown to greatly reduce the viscosity of concentrated mAbs. However, the mechanism by which co-solutes modify mAb viscosity is not well understood. Herein, we investigate the effects of several amino acids and inorganic ionic co-solutes on the viscosity of mAbs up to 250 mg/mL. We relate the viscosities to measurable changes in the mAb protein-protein interactions (PPI), structure, and self-association behavior as assessed through small-volume high- throughput techniques such as dynamic light scattering (DLS) and static light scattering (SLS). We interpret the mAb diffusion and light scattering behavior in the context of thermodynamic and hydrodynamic interactions, in order to develop a predictive fundamental understanding of co-solute effects on the mAb viscosity and stability. We demonstrate that the viscosity reduction is correlated with the disruption of mAb self- association and attractive PPI, as directly probed by SLS and DLS. Finally, we demonstrate that these changes in the PPI and viscosity are a function of the co-solute structure.

BIOT 437

Facile preparation of covalent C-to-C linked nanobody fusions by freezing

Berlin Zang, [email protected], Lingyun Jia, Jun Ren, Qiang Peng. School of Life Science and Biotechnology, Dalian University of Technology, Dalian, Liaoning, China

Nanobody, a unique type of antibody only have the 1/10 size of the conventional antibody, have been well developed for various applications in the past 20 years. Bi/Multi-valent or multispecific nanobodies have better performance on antigen binding and are able to target various epitopes. In this regard, most of the nanobody drugs in clinical trials are multivalent or multispecific. Bi/Multi-valent or multi-specific nanobodies are usually prepared by technology of recombinant expression. However, since the antigen-binding region of nanobody is close to the N-terminus, the resulting linking of C- terminal with N-terminal could easily affect the activity of the subsequent nanobody. In this work, we generated aldehyde group at specific site in C-terminal of nanobody via formylglycine-generating enzyme (FGE) catalysis, and then added bis- hydrazide/aminooxy- PEG molecules to the aldehyde-nanobodies to prepare the covalent C-C linked nanobodies. We observed that the dimerization-efficiency was increased to about 50% under freezing condition. Then we found the KD of the bivalent nanobody was increased by 20 times which proved that the bivalent nanobodies with natural structures are able to maintain the activity and function of the antibody to a large extent. More importantly, the method provides a facile way for the synthesis of bivalent or multivalent nanobody.

Schemes illustrating our approach to prepare C-to-C linked nanobody fusions by freezing

BIOT 438

Fluoroalcohol-induced coacervation in membrane proteomics

Rion R. Rion, [email protected], Armin Oloumi, Mohammadmehdi Azizi, Amir Koolivand, Morteza Khaledi. Chemistry and Biochemistry, University of Texas at Arlington, Rockwall, Texas, United States

Coacervation is a form of organized self-assembly where amphiphilic molecules aggregate and form two separate phases in aqueous-based media. The two-phase system consists of an amphiphile-rich phase called coacervate which is in equilibrium with a second aqueous-rich phase containing a small level of amphiphile. Formation of the two-phase system can be induced by perfluoro alcohols and acids under various conditions. The combination of an amphiphile, tetra-n-butylammonium bromide (TBA), and a fluorinated alcohol, tetrafluoroethylene (TFE), under various pH and/or ion concentrations, were studied. The coacervate systems were used to extract model hydrophilic and hydrophobic proteins. The systems of interest for membrane proteomic applications are those that show higher enrichment and selectivity towards hydrophobic proteins (higher grand average of hydropathy). Of the systems observed, TBA /TFE showed a higher membrane protein selectivity in the coacervate phase. This system was utilized to extract and enrich proteins from baker’s yeast, Saccharomyces cerevisiae, a well-defined organism in the study of proteomics. A mixture containing TBA/TFE and the crude yeast lysate was vortexed and centrifuged. The two phases formed were separated and filter aided sample preparation was applied to individual phases prior to downstream proteomic analysis by liquid chromatography-mass spectrometry. An approximately 24% increase in intrinsic membrane protein identification was observed. In addition, more hydrophobic proteins were identified in the coacervate phase. Such results can be applied in the diagnostic study of gene mutation diseases, such as breast cancer, where the overexpression of a membrane protein is a key observation.

BIOT 439

Remodeling of a human microbiome model

Amala M. Bhagwat1,2, [email protected], Cynthia H. Collins1,2, Jonathan S. Dordick1,2. (1) Chemical and Biological Engineering, Rensselaer Polytechnic Institute, Troy, New York, United States (2) Center for Biotechnology and Interdisciplinary studies, Rensselaer Polytechnic Institute, Troy, New York, United States

Conventional treatment strategies against human bacterial infections have involved widespread use of broad spectrum antibiotics, which has resulted in emerging antibiotic resistance of human pathogens. Consequently, there is an emergent need to replace conventional treatment strategies with selective pathogen removal strategies to maintain the health of natural microbial communities within the myriad microbiomes of the human body. While metabolomic, transcriptomic and genomic approaches have given us depth in understanding the composition of human microbiome, a fundamental understanding of microbial dynamics and response of a natural microbiome to a treatment strategy remains poorly understood. We have begun to develop a strategy that involves the response of a human skin microbiome to selective pathogen removal using cell lytic enzymes. We employed the bacteriocin lysostaphin and the bacteriophage derived endolysin PlyPH at different concentrations to perform selective removal of Staphylococcus aureus and Bacillus cereus, respectively. In a co-culture of S. aureus and B. cereus, which are potential pathogens in the skin microbiome, we evaluated treatment selectivity and the longer-term dynamics and recovery of community members post-target pathogen removal in a tunable defined growth medium and a more relevant human sweat like environment. Since S. aureus and B. cereus qualify as transient and opportunistic pathogens for a human skin microbiome, we also incorporated a commensal member, Micrococcus luteus to develop a comprehensive microbiome model. With a representative three-member skin microbiome, we demonstrated the effectiveness of a cell lytic enzyme antimicrobial strategy in keeping the commensal microbial community unharmed.

BIOT 440

Phosphoproteomic and transcriptomic analysis of cell wall stress response in Aspergillus nidulans reveals novel cell wall integrity signaling proteins

Cynthia Chelius3, [email protected], Karthik Boppidi3, Stephen Lincoln4, Jyothi Kumar5, Samantha Reese5, Simin Hossain3, Donnel Thomas3, Kelsi Lawson3, Jessica Ramsey3, Alexis Ramsey3, Ranjan Srivastava2, Steven Harris6, Mark Marten1. (1) Engineering Bldg 314, UMBC, Chem., Biochem. Environ. Engr., Baltimore, Maryland, United States (2) Univ of Connecticut, Storrs Mansfield, Connecticut, United States (3) Chemical and Biochemical Engineering, University of Maryland Baltimore County, Baltimore, Maryland, United States (4) Chemical and Biomolecular Engineering, University of Connecticut, Storrs, Connecticut, United States (5) Plant Pathology, University of Nebraska - Lincoln, Lincoln, Nebraska, United States

Phosphoproteomic analysis provides a global cellular perspective on eukaryotic protein phosphorylation. This post-transcriptional modification often mediates signal transduction, resulting in activation of transcription factors and altered gene expression. Resultant changes in gene expression can be monitored via a transcriptomic analysis. Our lab uses these tools to better understand signal transduction involved in gene regulatory networks. As a model system, we are studying the fungal cell-wall integrity signaling (CWIS) pathway in Aspergillus nidulans, responsible for cell wall biogenesis and repair. This pathway is induced upon cell wall perturbation, activating a MAPK cascade (BckA-MkkA-MpkA), which ultimately regulates cell-wall glucan synthesis. However, other cell wall components are regulated by an, as yet unknown, alternative pathway. To make new connections between various pathway components, we used phosphoproteomic and transcriptomic analysis to understand the role of a key pathway kinase (mpkA). We find CWIS to be involved with both expected (morphogenesis) and unexpected (iron metabolism) cellular processes. To develop a kinetic model of signal transduction, we induced CWIS by chemically perturbing the cell wall. We then carried out phosphoproteomic analysis on samples collected over a short time scale (10 minutes), and transcriptomic analysis on samples collected over a moderate time scale (120 minutes). Dynamic phosphorylation profiles of individual phospho-sites provide insight regarding the order in which sites are phosphorylated (network structure), and assist in differentiating between primary responses, crosstalk, and promiscuous phosphorylation. Novel cell-wall repair related phospho-sites, and their downstream effectors, will be discussed.

BIOT 441

Effects of buffer condition on the hydrodynamic radius of biomolecules Fan Rong, [email protected], Namila Khereid, S. Ranil Wickramasinghe, Xianghong Qian, [email protected]. Biomedical Engineering, University of Arkansas, Fayetteville, Arkansas, United States

Similar to charge and hydrophobicity, hydrodynamic radius of a biomolecule plays an important role in its binding, elution and transport. During downstream processing of therapeutics such as the purification of protein products derived from human and/or animal cell lines, size based virus filtration is often used for virus clearance to minimize the risk of viral contamination. Virus particles are excluded/rejected by virus filters via size-exclusion mechanism. Since protein conformation is dictated by the forces experienced, buffer conditions are expected to play an important role in the size and conformation of a particular protein. As virus capsids are proteins, virus particles are expected to exhibit variable sizes at different buffer conditions. This size variation could affect its rejection or breakthrough during virus filtration. Dynamic light scattering (DLS) is a powerful tool to determine the diffusion coefficient and the hydrodynamic size of a biomolecule in aqueous solution. DLS measurements were performed to investigate the change in hydrodynamic radius of model proteins and MVM virus under different buffer conditions. Our results indicate that hydrodynamic radius of model proteins are strongly affected by the buffer pH and salt concentration.

BIOT 442

Downstream process impact of reducing impurities during harvest operations using charged hybrid filters

Brian Kluck1, [email protected], Chi Tran1, alexei voloshin2, Evren Ozcam2, Bryan Dransart1, Rajesh Krishnan1. (1) Biologics Process Development, Gilead Sciences, Oceanside, California, United States (2) Separation and Purification Sciences, 3M Company, St Paul, Minnesota, United States

One of the main goals in downstream processing of recombinant monoclonal antibodies is the reduction of impurities present in the cell culture material, including genomic DNA, host cell proteins, aggregates, endotoxin, and potential virus. Genomic DNA, in particular, exists as chromatin complexes associated with various host cell proteins. Current harvest filtration steps do not effectively remove chromatin, which can non- specifically associate with the protein of interest and Protein A chromatography resin. These associations can result in variable levels of HCP and DNA in the Protein A eluate leading to further variability in precipitation and impurity levels upon pH adjustment. Optimized pH adjustment and depth filtration steps are required to ensure robust impurity clearance. To mitigate these risks and to streamline harvest and downstream operations across production scales, a positively charged 3M Emphaze™ AEX hybrid purifier was tested during harvest of a CHO-expressed mAb to evaluate the benefits of chromatin removal. Performance of the purifier was studied by monitoring filtration pressure, turbidity, residual HCP, DNA, and chromatin complex. Scalable process conditions for harvest filtration operations using the 3M Emphaze™ AEX hybrid purifier resulted in complete removal of genomic DNA prior to the Protein A chromatography step.

BIOT 443

Evaluation of the effects on HCP clearance and other impurities of mPAA and all- synthetic depth filters (X0SP)

Hoang C. Nguyen1, [email protected], Yinying Tao1, Warren R. Emery1, Monica K. Paik2. (1) Purification Development, Eli Lilly & Company, Indianapolis, Indiana, United States (2) Engineering Technology, Eli Lilly & Company, Indianapolis, Indiana, United States

An all synthetic depth filter X0SP (comprised of silica and a charged polymeric component) and partially benzylated poly(allylamine) (mPAA) were evaluated within a typical monoclonal antibody (mAb) purification process to compare their effectiveness in process and/or product related impurity removals. We have demonstrated that X0SP filter and mPAA both contributed to the significant removal of process as well as product related impurities. However, differences in impurity removal were observed between mPAA and X0SP. These differences are discussed in the context of process development and optimization.

BIOT 444

Mechanisms of DNA retention on depth filters

Ohnmar Khanal1, [email protected], Nripen Singh2, Steven J. Traylor2, Xuankuo Xu2, Sanchayita Ghose2, Abraham M. Lenhoff1. (1) University of Delaware, Newark, Delaware, United States (2) Biologics Process Development, Bristol-Myers Squibb, Hopkinton, Massachusetts, United States

Depth filtration is a commonly used bioprocessing unit operation for harvest clarification. The multiple components comprising a depth filter may impart to it additional functionality beyond simply that of particulate filtration, as a result of which the mechanisms by which various impurities such as proteins, nucleic acids and cellular debris are removed are difficult to decouple. In this study, we probe the mechanisms by which nucleic acids are retained on depth filter media. Mimicking the usage of the depth filter capsules in large-scale clarification, we use small depth filter discs for flow experiments. Through non-destructive direct covalent labeling of DNA and visualization of the extent of penetration of the retained DNA on depth filters using confocal microscopy, we are able to resolve retention by adsorption – driven primarily by electrostatic interactions – from retention by size-based filtration. The effects of DNA properties, such as the length, are also considered, because the characteristics of DNA as a process impurity are not fully understood. Our findings can be leveraged in further improving the design of depth filter and in the design of various filtration media for DNA purification. BIOT 445

Mechanistic modeling of the loss of protein sieving due to internal and external fouling of microfilters

Alex Apostolidis, [email protected], Glen Bolton. Process Development , Amgen, Cambridge, Massachusetts, United States

Fed-batch and perfusion cell culture processes used to produce therapeutic proteins can use microfilters for product harvest. In this study, new explicit mathematical models of sieving loss due to internal membrane fouling, external membrane fouling or a combination of the two were generated. The models accounted for membrane and cake structures and hindered solute transport. Internal membrane fouling was assumed to occur due to the accumulation of foulant on either membrane pore walls (pore-retention model) or membrane fibers (fiber-retention model). External cake fouling was assumed to occur either by the growth of a single incompressible cake layer (cake-growth) or by the accumulation of a number of independent cake layers (cake-series). The pore- retention model was combined with either the cake-series or cake-growth models to obtain models that describe internal and external fouling occurring either simultaneously or sequentially. The sequential pore-retnetion followed by cake-growth model provided a good fit of sieving decline literature data during beer microfiltration. The cake-series and cake-growth models provided good fits of sieving decline data during the microfiltration of a perfusion cell culture. The new models provide insights into the mechanisms of fouling that result in sieving decay.

BIOT 446

Effect of zinc chloride and PEG concentrations on the critical flux during tangential and normal flow microfiltration of BSA precipitates

Zhao Li, [email protected], Andrew L. Zydney. Dept of Chemical Engineering, Penn State University, University Park, Pennsylvania, United States

There is renewed interest in the possibility of using precipitation for initial capture of high value therapeutic proteins as part of an integrated continuous downstream process. These precipitates can be continuously washed using tangential flow filtration (TFF), with long term operation achieved by operating the membrane modules below the critical filtrate flux for fouling. Our hypothesis was that the critical flux for the precipitated protein would be a function of the properties of the precipitate as determined by the precipitation conditions. We evaluated the critical flux using a flux-stepping procedure for model precipitates of bovine serum albumin generated using a combination of a crosslinking agent (zinc chloride) and an excluded volume precipitant (polyethylene glycol). Data were obtained using both 0.2 µm pore size hollow fiber and flat sheet (disk) polyethersulfone microfiltration membranes over a range of filtration conditions. Independent measurements were obtained for the viscosity of the precipitated protein solutions, the precipitate size distribution (by laser diffraction), the yield of protein in the precipitate (after centrifugation), and the resistance of the protein deposit. The critical flux in the hollow fiber module varied with shear rate to approximately the 1/3 power, consistent with predictions of the classical polarization model. The critical flux increased significantly with increasing zinc chloride concentration, going from 60 2 2 L/m /h for a 2 mM ZnCl2 solution to 200 L/m /h for an 8 mM ZnCl2 solution. The critical flux in the stirred cell followed the same trend as a function of ZnCl2 concentration, but with much lower values of the critical flux due to the different device hydrodynamics. The critical flux in the hollow fiber module achieved a maximum value at an intermediate polyethylene glycol (PEG) concentration of around 6 weight percent at a fixed ZnCl2 concentration of 3.5 mM. The effects of ZnCl2 and PEG were interpreted in terms of the changes in effective size and viscosity of the protein precipitates. Additional insights were obtained from the compressibility of the cake layer formed after filtration of the protein precipitate in a stirred cell. These results provide important insights into the development of effective tangential flow filtration systems for processing large quantities of precipitated protein as would be required for large scale continuous protein purification by precipitation.

BIOT 447

UF/DF processes for high concentration mAb formulations: Reconciling process development and evolving trends

Abhiram Arunkumar, [email protected], Nripen Singh, Sanchayita Ghose, Zheng Jian Li. Process Development, Bristol-Myers Squibb Company, Framingham, Massachusetts, United States

Ultrafiltration/Diafiltration (UF/DF) is the unit operation in place for concentrating the protein solution and placing it in the final formulation buffer. The biotechnology industry is moving towards high protein concentrations (>200 mg/mL) leading to several challenges in the UF/DF process. In addition to product quality changes, there is no accurate scale-down model for the final concentration and recovery flush strategy. This leads to a viscious cycle of over-concentrating with increasing concentration targets to prevent dilution. The present work will discuss development of a standardized UF/DF development and recovery flush scale-down model for high concentration formulations for three different mAbs having different isoelectric points and different viscosities at 200 mg/mL in Histidine Sucrose buffer. A scalable, viscosity-independent recovery flush strategy to achieve minimum dilution and maximum yield will be discussed. Other challenges pertaining to high concentration formulations like reverse filtration during TFF and excipient partitioning are also discussed. Finally, evolving technologies like single pass TFF and single-use TFF are discussed for high concentration formulations and compared against conventional TFF. This is the first work to systematically discuss the practical considerations in developing high concentration UF/DF processes along with evolving technologies to overcome challenges in high concentration for mAbs that have a range of viscosities at 200 mg/mL.

BIOT 448

Evaluation of single-use TFF technologies for high concentration operations

Veronica Adams1, [email protected], Thomas Parker2. (1) Biogen, Raleigh, North Carolina, United States (2) MilliporeSigma, Burlington, Massachusetts, United States

With increasing numbers of early phase and orphan disease molecules, the industry has started investing in manufacturing processes that emphasize speed and flexibility. One component of such manufacturing processes is single-use technologies. A gap in single-use technologies is high concentration TFF operations, where single-use options have not kept pace with reusable cassettes. Many manufacturers struggle to achieve target concentrations in reasonable process times while maintaining high product recovery when utilizing single-use TFF systems. This presentation will summarize data evaluating MilliporeSigma’s Pellicon® Capsule for single-use TFF operations as compared to Biogen’s current single-use Sius TangenX filters, as well as reusable Pellicon® 3 TFF filters. The Pellicon® Capsule simplifies installation compared to other TFF formats and potentially offers a simpler scale up model from Biogen’s single-use manufacturing facility to Biogen’s larger scale 2k and 15k traditional facilities. Experimental and modeling data for both UFDF and SPTFF mode will be presented, and the impact of this disposable filter will focus on increasing final concentrations in Biogen’s single-use facility from 50 g/L to 150 g/L by changing single-use TFF formats.

BIOT 449

Understanding the sterile filtration of nanosuspensions

Kalliopi Zourna, [email protected], John H. Welsh. Biotechnology, Pall Corp, Portsmouth, Hampshire, United Kingdom

In pharmaceuticals, a nanosuspension is a colloidal dispersion of a drug, usually in an aqueous vehicle. Oil in water nanosuspensions such as liposomes and emulsions are used for topical oral and injectable drug delivery. Typically, the diameter of particles in a nanosuspension is less than 1 μm.

Different manufacturing techniques have been developed to produce nanosuspensions used in sterile injectable drugs. These must be aseptically prepared prior to administration. Due to their properties and propensity to chemical and physical degradation, sterilization of injectable nanosuspensions prior to filling can be challenging. For nanosuspensions with a particulate species of smaller than 200 nm, sterilizing grade filtration can be employed. While in many cases it is effective, bacterial challenge filter validation studies with injectable nanosuspensions have shown a higher likelihood of bacterial recovery compared with other parenteral drug types. Furthermore, the adsorptive impact of a sterilizing grade filter on nanosuspensions has been reported. Consequently, a greater understanding of the sterile filtration of nanosuspensions is required so that appropriate decisions are made concerning the selection and validation of sterilising grade filters for such applications and to ensure the best selection of sterile filter technology and highest degree of process safety.

In the reported work, liposome (Lipoid S) and emulsion (MF59) formulations were manufactured under standard methods for use in sterile injectable drug formulations. Using various techniques, they were characterised in terms of size and size-distribution, zeta potential, viscosity, pH, optical density. Here we present (a) the manufacture of nanosuspensions and their scale up using integrated Pall Allegro Single Use technologies (b) the effect of different filtration conditions/parameters (pre-filtration, pressure/differential pressure) and nanosuspension characteristics (size/size distribution) on sterile filtration and (c) liposome transmission studies and outcomes for different sterilising grade membranes.

BIOT 450

Barrier tissue mimics for drug development

Michael L. Shuler, [email protected]. Cornell University, Ithaca, New York, United States

Microphysiological (or “Body-on-a-Chip”) systems hold great promise to improve the selection of which drugs to take into human clinical trials. An increase in success from order of 10% to 30% would provide society a significant increase in available drugs and potentially reduce the cost of drugs. Particularly important will be barrier tissues such as the gastrointestinal (GI) tract, and the Blood Brain Barrier (BBB), skin, lung, and kidney. The skin, GI tract, and lungs control entry of drugs into the body. The BBB is critical in treating brain disorders and is often the site of undesirable side effects when drugs are used to treat systemic conditions. A common cause of drug failure is nephrotoxicity. Thus models of barrier tissues that accurately predict transport of drugs across the barrier and potential disruption of the barrier are critical to selection of drugs and important constituents of whole body microphysiological models.

BIOT 451

A synthetic biology platform for flexible and on-demand drug manufacturing

Jicong Cao, [email protected], Pablo Perez-Pinera, Ky Lowenhaupt, Ming-Ru Wu, Timothy Lu. Massachusetts Institute of Technology, Cambridge, Massachusetts, United States

The incapability of delivering drugs on demand to people living in remote areas has aroused concern. Current on-site drug manufacturing systems can produce one type of drug at a time, which is unable to meet the requirement of multiple drugs for different patients in timely fashion. Except for the addition of a number of facilities, we developed a novel strategy to produce multiple biologics in Pichia pastoris through a consolidated process consisting of strain development and on-demand drug expression and purification. The ability to produce one or multiple therapeutic proteins of interest at defined ratios simultaneously in a single batch can significantly reduce the number of strains and facilities required for on-site biologics production, thus lowering time and expense.

BIOT 452

Upstream optimization and non-chromatographic downstream strategy for production and purification of antimicrobial peptide in E. coli

Mathias Joachim1, [email protected], Tobias Weidner1, Doreen Gerlach2, Peter Czermak2,1. (1) Life Science Engineering, University of Applied Sciences Mittelhessen, Giessen, Hesse, Germany (2) Bioresources, Fraunhofer Institute for Molecular Biology and Applied Ecology, Giessen, Germany

Standard production processes for antimicrobial peptides (AMPs) comprise of batch cultivations in upstream and chromatographic purification in downstream. In spite of limitations in scale-up and high costs, chromatography still remains industrial gold standard. To overcome this bottleneck, alternative downstream strategies using aggregating tags and membrane filtration are a promising approach. However, for making use of the full potential of such processes, they have to be engineered from scratch. Therefore, a unique combination of a thioredoxin-tag (trxA), elastin-like-polypeptides (ELPs) and a self splicing intein (ΔI-CM) was assembled using golden gate cloning, to facilitate a soluble AMP expression process in combination with a non-chromatographic downstream process. A tailor made chemically defined media with optimized trace elements increased growth and recombinant production of the ELP-tagged AMP up to 167%. The trxA acts as a solubility enhancer and facilitates the release into the extracellular space by simple osmotic shock procedure. This method proofed to be superior to chemical lysis or mechanical cell disruption by a 2x-fold increase in product per host cell protein (HCP) yield. Reversible temperature dependent phase transition of the ELPs enables membrane filtration steps for purification of the desired product from impurities and, after cleavage by self-splicing inteins, to separate the amp from the tags itself. The interaction of thioredoxin-, ELP- and intein-tag facilitates scalability and economic feasibility, rendering cell disruption apparatuses, chromatographic columns and enzymatic digestion obsolete. Integration of process analytical technology (PAT) and a design of experiments (DoE) based upstream optimization of inductor strength, induction temperature and duration generates a production process with high potential for application in pharmaceutical industry in regard of GMP-compliance. BIOT 453

Innovating biopharmaceutical manufacturing for an age of acceleration

John C. Love, [email protected]. Chemical Engineering, Koch Institute at MIT, Cambridge, Massachusetts, United States

The landscape of knowledge about human disease is rapidly expanding. Advances in our abilities to learn directly from clinical samples about diseases using new technologies and computational tools are accelerating these insights. Such precision expands our ability to identify rare diseases, stratify patients by indication, and improve the efficiency of clinical studies to establish efficacy of products. This transformation in how we develop and assess new drugs will further accentuate the need for efficiency, speed, and agility in our manufacturing strategies of high-quality recombinant biopharmaceuticals. This talk will consider the evolution of manufacturing strategies in related industries where shifts in market size and product diversity have led to changes in the underlying manufacturing networks, and examine the opportunities for the design of facilities and process engineering to keep pace with small-volume production. It will highlight a few examples and lessons learned from developing small-scale automated manufacturing systems for high-quality recombinant protein drugs, including how a deep ‘-omic level’ understanding of the biology of manufacturing (e.g, the host itself) accelerates predictable process development for both upstream and downstream operations. In total, holistic integration of knowledge about biology and engineering can facilitate rapid process development and speed manufacturing of high-quality products at scales relevant for a variety of use cases. Finally, the potential impact will be discussed for an open “Maker”-style platform approach to realize recombinant drugs anywhere in order to accelerate access to new medicines.

BIOT 454

Developing an end to end bio-manufacturing platform for the new paradigm of SAM (Self Amplifying mRNA vaccines)

Varnika Roy1, [email protected], Marcin Bugno1, Diana Chinchilla-Olszar1, Kelly Forney-Stevens1, Min (Mandy) Xie1, Laura Harrington1, Chris Baldwin1, Nicolas Delahaye2, Jeffrey Ulmer2, Kunal Aggarwal1, Derek O'Hagan1. (1) Vaccines Technical Research and Development, GSK, Rockville, Maryland, United States (2) Vaccines Research and Development, GSK, Rockville, Maryland, United States

The SAM (Self Amplifying) platform aims to revolutionize the way we prevent and treat infectious diseases by developing the next generation of best-in-class mRNA vaccines. The SAM platform represents a fully synthetic RNA vaccine technology combining a self-amplifying RNA backbone and a non-viral delivery system. While naked mRNA by itself is immunogenic, facilitated delivery is required to attain full potency. The delivery system is a cationic nanoemulsion (CNE), which binds to the SAM RNA, enables its delivery intra-cellularly and prevents degradation and thereby substantially increases the potency of the vaccine. The product is a two vial presentation one for the RNA drug product (DP) and the second for the CNE DP mixed at the patient’s bedside to yield the reconstituted vaccine. Unique challenges such as RNAase contamination, control strategy for novel raw-materials in an in-vitro transcription reaction to yield RNA and various others were encountered from a manufacturing perspective when advancing this novel platform. These challenges and mitigation strategies implemented to overcome them will be discussed. The end to end bio-manufacturing platform developed with this SAM platform presents a disruptive innovation to simplify vaccine discovery and development by providing an agile and robust CMC model for producing mRNA vaccines rather than specific antigen development which hinders a platform CMC approach.

BIOT 455

Techno-economic analysis of semicontinuous production of recombinant butyrylcholinesterase in transgenic rice cell suspension cultures

Jasmine Corbin1, [email protected], Karen McDonald1,2, Somen Nandi1,2. (1) Chemical Engineering, University of California, Davis, Davis, California, United States (2) Global HealthShare Initiative, University of California, Davis, Davis, California, United States

The human enzyme butyrylcholinesterase (BChE) has been extensively studied as a therapeutic and prophylactic treatment against organophosphate poisoning. However, its widespread use has been limited by the prohibitive costs of production and purification from outdated human blood sources. Our group has demonstrated production and purification of active, rice recombinant BChE (rrBChE) from a metabolically-regulated transgenic rice cell suspension culture. This system promises to be a cost-effective source of rrBChE due to 1) lower media costs for plant cell culture compared with mammalian cell culture; 2) a semicontinuous operational strategy that allows for independent optimization of growth and production phases, reducing the need for long seed trains, and minimizing turn-around time, CIP and SIP, chemicals, and energy; and 3) a downstream processing scheme that produces a therapeutic-quality molecule using common unit operations including tangential flow filtration (TFF), diethylaminoethanol (DEAE) anion-exchange, and affinity chromatography.

To assess the impact of these potential cost-saving measures at the manufacturing scale, we scaled-up the process in silico using SuperPro Designer® 9.0 simulation software (Intelligen, NJ). While scale-up of traditional biomanufacturing platforms such as mammalian cells grown in bioreactors has been well-studied, limited work has been done to build that knowledge base for plant-based expression systems. Data obtained from laboratory-scale process optimization experiments provided fundamental insights into the underlying process design. We selected equipment and operational conditions that would enable this specific process to reach realistic production goals for rrBChE. The model was used to evaluate total capital investment, annual operating cost, and cost of goods sold as a function of rrBChE expression levels, facility production capacity, and design and scheduling of the semicontinuous process. The model can be used to identify process bottlenecks, and be modified or adapted to assess the profitability of alternative designs, implement different assumptions, and guide process development and further optimization.

BIOT 456

Use of a carbon footprint calculation to assess the environment impact of drug substance manufacture in GSK biopharm

Phillip R. Smith1, [email protected], Matt Snyder2, Kurtus Kahle2, Kevin Tomko2, Alexander Adams1, Mary Schaad1, Daniel D'Aquila1. (1) GlaxoSmithKline, King of Prussia, Pennsylvania, United States (2) Clark, Richardson and Biskup Consulting Engineers, Inc., St. Louis, Missouri, United States

In line with an overall industry trend, GSK Biopharm is implementing sustainability initiatives as part of corporate responsibility goals. Assessment of early- and mid-phase manufacturing activities affords an opportunity to review and reduce environmental impact throughout the product and process lifecycle, including commercial manufacturing. Accordingly, the GSK Upper Merion Biopharm Pilot Plant is actively assessing environmental impact by calculating both the process mass intensity (PMI) and Carbon Footprint of clinical trial material production. PMI calculations representative of the Biopharm Industry have previously been conducted and reported by the ACS Green Chemistry Institute Pharmaceutical Roundtable, Biopharm Focus Group. These have identified opportunities to mitigate environmental impact such as reducing water usage in Biopharm manufacturing operations. However, this methodology does not enable a comprehensive understanding of the corresponding energy utilization and associated Carbon Footprint. This presentation will illustrate how the Carbon Footprint calculation determines the total amount of carbon released to the atmosphere due to manufacturing activities (including all process reagents, components, and energy usage). Results of the calculation for multiple Biopharm manufacturing processes, including both early- and late-phase processes at multiple process scales, will be presented. Highlights of this analysis will include an illustration of the dramatic impact of bioreactor titer on environmental impact, and that raw materials and single-use components are minor contributors to the total carbon footprint. These highlights identify opportunities to reduce the environmental impact of future commercial Biopharm manufacturing activities and help outline a path to carbon neutrality for GSK Corporate activities. Additionally, the relative strengths and weaknesses of the PMI and Carbon Footprint environmental impact calculations will be discussed.

BIOT 457

Next-generation vaccine facility: Merging high productivity technologies for robust and cost-effective manufacturing Renaud Jacquemart1, [email protected], Jose Castillo2, Chris Yallop3. (1) Natrix Separations, Millipore Sigma, BURLINGTON, Ontario, Canada (2) Univercells, Brussels, Belgium (3) Batavia Biosciences, Leiden, Netherlands

Vaccination greatly reduces disease, disability, death and inequity worldwide. Vaccines even permitted total eradication of Small Pox. However, most vaccines available today are still manufactured with the original process technologies in complex and capital- intensive facilities. The labor-intensive, low-yield and low throughput processes, and the lack of manufacturing flexibility diminish affordability for developing countries, forcing manufacturers to address the problem by supplying these markets with subsidized products and/or non-sustainable prices. Advancements in vaccines manufacturing would benefit both global health and the industry. This presentation will focus on a new holistic approach that integrates tested strategies for cost efficient manufacturing of vaccines, linking innovations in cell line development, bioreactor design, membrane chromatography and flexible facility design for a 500-fold increase in overall productivity. This end-to-end, next generation facility merging high productivity technologies to achieve simplified process architecture, increased robustness and augmented yield can be replicated anywhere to make production in developing countries in need of vaccines a reality. It will contribute to supplying annually 40M doses of vaccines at <$0.15/dose from a $10M facility, and could therefore support eradication of poliomyelitis in the next 10 years.

Industrial production at lab scale with isolator-based micro-facility for *Simplified infrastructure and dramatic decrease of CAPEX, the biggest factor driving reduction in cost/dose *Simplified operations for a robust platform that can be replicated and/or quickly deployed for in-region manufacturing

BIOT 458

Controlled EGFR ligand display on cancer suicide enzymes for targeted intracellular delivery

Rachel Lieser, [email protected], Millicent O. Sullivan, Wilfred Chen. Chemical and Biomolecular Engineering, University of Delaware, Newark, Delaware, United States

Proteins have emerged as a new class of therapeutics in the past several decades due to breakthroughs in molecular engineering that allow researchers to customize proteins for a multitude of clinical applications. However, despite great interest and investment in advanced protein therapeutics, most FDA-approved proteins have extracellular targets, despite the existence of multiple diseases that could be treated through intracellular protein therapies. Engineering efforts to address delivery limitations often rely on modifying proteins through direct conjugation of polymers and peptides using reactive residues on naturally occurring amino acids. The key shortcoming of this method is the inability to modify a specific site within a protein, which can significantly reduce pharmacological action. Additionally, such approaches do not offer control over variables such as ligand clustering, which is an important determinant of targeting efficacy.

Unnatural amino acid (UAA) incorporation provides a method to modify proteins with site-specific, biorthogonal reactive moieties through nonsense codon replacement. This method of modification not only allows site-specific functionalization of therapeutics for improved bioactivity, but also allows exploration of the arrangement of delivery molecules on proteins for optimal intracellular delivery. Previous work has demonstrated the ability to incorporate biorthogonal chemistries into model proteins through UAA incorporation, enabling modification with simple ‘click’ chemistry techniques. In our work, we demonstrate application of this approach for conjugation of epidermal growth factor receptor (EGFR) targeting peptides in fluorescent proteins, using varying EGFR peptide arrangements to control cellular internalization in inflammatory breast cancer (IBC) cells. Furthermore, we demonstrate the ability to adapt this system for delivery of a suicide enzyme to enable IBC-targeted cell death through prodrug application and conversion. Through this approach, we have identified an ideal ligand display for enzyme delivery to IBC cells. Future work will refine the efficacy of the approach via incorporation of endosomal escaping peptides and hydrophilic polymers, enabling tailorable intracellular protein delivery in vivo.

BIOT 459 Enhanced cellular delivery of enzymes using the cowpea chlorotic mottle virus

Mark de Ruiter, [email protected], Jeroen J. Cornelissen. Laboratory for Biomolecular Nanotechnology, University of Twente, Enschede, Netherlands

In this work active enzymes are delivered into cells by encapsulating them inside the protein capsid of the Cowpea Chlorotic Mottle Virus (CCMV). The enzymes are initially functionalized using negatively charged polymers, to impose sufficient negative charge to induce assembly of the capsid proteins of CCMV around the enzyme. This results in stable, icosahedral and monodisperse particles with a diameter of around 22nm. These, so called nano-reactors, allow reactant diffusion across the protein coat and show retention of catalytic activity. The activity of the particles seems to depend largely on the size of the substrate, charge and on the type of enzyme that is encapsulated. Because viruses are inherently designed by nature to get its cargo inside cells, we evaluated the uptake, activity and stability of these particles in cancer cells. These virus like particles are shown to be taken up easily and have superior Intracellular enzymatic activity compared to non-encapsulated enzymes. This system is non-toxic and the use of a plant virus here minimizes the dangers associated with the use of animal viruses. Furthermore, the exterior groups of the virus can be easily modified with various functional moieties, creating a multivalent platform for targeting specific cell lines. This system is shown to be useful for the delivery and protection of enzymes. Which is useful in their potential future applications. It could for example be used in treating enzyme deficiency diseases or in fabricating active drugs ‘on site’ in targeted cells.

A virus-based enzyme nanoreactor

BIOT 460

Novel microfluidic system for encapsulation of equine endothelial colony forming cells for local cell delivery Yuan Tian1, [email protected], Wen J. Seeto1, Randolph Winter2, Fred Caldwell2, Anne Wooldridge2, Elizabeth A. Lipke1. (1) Chemical Engineering, Auburn University, Auburn, Alabama, United States (2) Clinical Sciences, Auburn University, Auburn, Alabama, United States

Neovascularization is critical in restoring blood flow to ischemic tissue in numerous disease states across species. Endothelial colony forming cells (ECFCs) are highly proliferative and can participate in new blood vessel formation, making them a good source for cell-based therapy. However, cells that are injected directly may have low viability and poor retention at the targeted injection site due to the high shear forces and mechanical wash out. In this study, a custom-built microfluidic encapsulation system was developed for rapid production of ECFC-laden poly(ethylene glycol)-fibrinogen (PF) hydrogel microspheres. This system successfully encapsulated ECFCs at a cell density (10 million cells/mL) that was clinically relevant for cell delivery through injection without sacrificing uniformity and reproducibility. Encapsulated ECFCs maintained high cell viability (>95%), robust proliferative capability, and cellular phenotype. Ex vivo and in vivo cell delivery studies were performed by injecting microspheres containing autologous equine ECFCs subcutaneously into distal limb full thickness wounds of adult horses. The cells were labeled with fluorescent nanodots prior to encapsulation for long- term tracking. One week after injection, biopsies of the wounds were analyzed using confocal microscopy. Migration of encapsulated ECFCs into the surrounding host tissue was observed, indicating successful retention and survival of the injected ECFCs and demonstrating feasibility for injectable cell delivery.

ECFC-laden microspheres with high cell density (10 million cells/mL) were highly uniform in size and shape.

BIOT 461

Peptoid-based coatings as artificial extracellular matrix for increased differentiation of human embryonic stem cells into neural cells

Jesse Roberts1, [email protected], German Perez1, Michael Borrelli2, Shannon L. Servoss1. (1) Chemical Engineering, University of Arkansas, Fayetteville, Arkansas, United States (2) University of Arkansas for Medical Sciences, Little Rock, Arkansas, United States

Neural cells differentiated from human embryonic stem cells (hESCs) provide a compelling component for tissue engineering and regenerative medicine. The extracellular matrix (ECM) plays a major role in the differentiation and proliferation of neural cells. The hESCs are extremely sensitive to properties of the matrix in that even small changes to the chemical composition, physical makeup, or topography of the ECM can drastically effect neural cell differentiation. Promising studies have focused on the development of scaffolds with nanoscale features that can incorporate neural differentiation enhancers. In this work, we have created surfaces of peptoid microspheres as well as gold nanorods coated with peptoid. Preliminary studies show that the peptoid coatings enhance differentiation to neurons. This work is being extended to determine the mechanism of action, as well as incorporation of other hESC differentiation modifiers such as retinoic acid.

BIOT 462

D-VITylation: Harnessing the biology of vitamin D to improve the pharmacokinetic properties of peptides and small proteins

Daniel B. Hall, [email protected], Ahmet S. Vakkasoglu, Caroline W. Hill, Laura M. Hales, Tarik M. Soliman. Extend Biosciences, South Easton, Massachusetts, United States

Peptides can be potent and specific therapeutics, yet their small size leads to their rapid removal from the bloodstream by the kidneys, which severely limits their therapeutic potential. Although conjugation of small proteins to large polymers typically results in longer residence times, peptide-polymer conjugates often have a significant loss of biological activity due to steric hindrance. Here, we improve the pharmacokinetics of peptide therapeutics by harnessing the biology of vitamin D. We show that attachment of a vitamin D-based “carrier” molecule (D-VITylation™) protects the conjugated peptide or protein from renal clearance by virtue of reversible binding to the serum-circulating vitamin D binding protein (DBP), without compromising bioactivity. By varying the conjugation site on vitamin D, the binding to DBP varies, with higher affinity constructs corresponding to a longer plasma half-life in vivo. We demonstrated the half-life extending potential of D-VITylation on two natural peptide hormones, ghrelin and insulin, and one small protein, FGF21. We observed increases in plasma half-life and decreases in clearance ranging from 10-100 fold copared to that of unmodified peptide. As proof-of-principle, we demonstrate that a vitamin D-insulin conjugate lowers blood glucose levels for nearly 24 hours in a type 1 diabetic rat model.

BIOT 463

Bioengineered injectable thermogel for intravitreal protein controlled release to the retina

Vianney Delplace2, [email protected], En Leh Tsai3, Arturo Ortin-Martinez3, Malgosia Pakulska2, Valerie Wallace3, Molly S. Shoichet1. (1) Chemical Engineering Applied Chemistry, University of Toronto, Toronto, Ontario, Canada (2) Chemical Engineering & Applied Chemistry, University of Toronto , Toronto, Ontario, Canada (3) Ophthalmology and Vision Science, Krembil Research Institute, Toronto, Ontario, Canada

Blindness and visual impairment are largely caused by irreversible loss of light sensitive photoreceptors and retinal pigment epithelium cells in the retina. Some pro-survival therapeutic proteins, such as ciliary neurotrophic factor (CNTF), can improve retinal cell survival in the damaged retina, but require controlled delivery systems due to their short half-lives and rapid clearance. To achieve local controlled release, we designed an affinity-based system that sustains the release of bioactive proteins from an injectable, thermogel. In this strategy, we designed a hyaluronan-methylcellulose (HAMC) polymer blend, with shear-thinning and inverse thermogelling properties, that allows injection through fine gauge needles, forms a gel at body temperature, and is stable for at least 10 days in vitro. Furthermore, HAMC is cytocompatible with photoreceptors in vitro, confirming its appropriateness for intravitreal injections. To control protein release, we exploit the interaction of a well-known intracellular protein, Src homology 3 (SH3), with its binding peptides. Specifically, by (i) expressing fusion proteins of SH3 with the protein of interest and (ii) chemically modifying the HAMC hydrogel with the corresponding binding peptides, the release can be computationally predicted and designed. In this study, an SH3-CNTF fusion protein was successfully expressed and CNTF bioactivity was confirmed in vitro with a cell-based assay, showing a similar response to commercially available CNTF. The controlled release of SH3-CNTF from peptide-modified HAMC was confirmed over 10 days in vitro. A bioactivity assay on the released samples showed sustained activity over a week, validating SH3-CNTF as a candidate for local affinity release to the retina. Intravitreal injections of the bioengineered hydrogel were investigated to evaluate the in vivo protein activity and the anticipated protective effects of its sustained release on the retina. This system lays the foundation for future in vivo studies where combining cell and biomolecule delivery will enhance cell survival after transplantation. BIOT 464

Wavelength-specific, plasmonic nanoparticle mediated rupture of polymersomes using ultrafast single-pulse irradiation

Julianne C. Griepenburg2, [email protected], Abby R. Robinson1, Gina M. Disalvo1, Sean M. O'Malley2, Daniel M. Bubb2,1. (1) Chemistry, Rutgers University- Camden, Camden, New Jersey, United States (2) Physics, Rutgers University-Camden, Camden, New Jersey, United States

Polymersomes are robust vesicles that are self-assembled from amphiphilic diblock copolymers. They are of tremendous interest in the field of drug delivery due to their ability to stably encapsulate molecules within both the hydrophobic membrane and hydrophilic lumen of the vesicle. In this study, light-stimulated release of hydrophilic encapsulants has been achieved through the incorporation of plasmonic nanoparticles, facilitating disruption of the membrane upon ultrafast, single-pulse irradiation. Cargo release can be controlled ranging from complete vesicle rupture and instantaneous release, to membrane pore formation and effusion. Single vesicle release kinetics were determined by monitoring temporal fluorescence intensity from single vesicles and determined to be related to pulse energy and nanoparticle location.

BIOT 465

Cheminformatics-driven discovery of polymeric micelle formulations of poorly soluble drugs

Eugene Muratov3, [email protected], Vinicius M. Alves1, Duhyeong Hwang3, Marina Sokolsky3, Elena Lebed4, Natasha Vinod3, Carolina Andrade1, Alexander Kabanov3, Alexander Tropsha2. (1) Faculty of Pharmacy, Federal University of Goias, Goiania, Goias, Brazil (2) Univ of North Carolina, Chapel Hill, North Carolina, United States (3) Medicinal Chem Natural Products, University of North Carolina, Chapel Hill, North Carolina, United States (4) Odessa National Medical University, Odessa, Ukraine

Many drugs fail therapeutic development because of poor aqueous solubility. Polymeric micelle formulations were shown to alleviate this problem for some but not all such drugs. We conceived a novel computer-aided strategy for rational design of formulations for poorly soluble drugs. Novel descriptors of drug-polymer complexes were developed and employed to build models predicting drug loading efficiency (LE) and loading capacity (LE). Using these models for virtual screening of drug libraries, we selected eight drugs (four positive and four negative hits). Three positive and three negative hits were confirmed experimentally (75% prediction accuracy). Ironically, the negative hit in this context also had the desired micelle solubility. Podophyllotoxin and simvastatin, with LE of 95 and 87 and LC of 43 and 41, respectively, were among the top five compounds ever studied by us. The success of the strategy described herein suggests its broad utility for designing other drug delivery systems. BIOT 466

MinGenome: Top-down synthesis of genome minimized strains for bioproduction

Lin Wang, [email protected], Costas Maranas. Penn State, University Park, Pennsylvania, United States

Genome minimized strains are obtained by removing genome segments associated with genes or processes either detrimental or unneeded under bioproduction conditions. They offer advantages as production chassis [1] by reducing transcriptional cost, eliminating competing functions and limiting unwanted regulatory interactions. Existing approaches for identifying stretches of DNA to remove are largely ad hoc based on information on presumably dispensable regions through experimentally determined non- essential genes and comparative genomics. As more sophisticated genome editing tools (e.g., CRISPR) are becoming commonplace, the need for a computational aid that will help successively minimize genomes consistent with a set of performance criteria beyond simply growth rate is becoming more pressing. Here we introduce a versatile genome reduction algorithm MinGenome that implements a mixed integer linear program (MILP) to iteratively identify the largest dispensable contiguous sequences without affecting the organism’s growth or other desirable traits. Known essential genes or genes that cause significant fitness or performance loss are flagged and their deletion is thus prohibited. MinGenome also preserves needed transcription factors and promoter regions ensuring that retained genes will be properly transcribed while also avoiding the simultaneous deletion of synthetic lethal pairs. The potential benefit of removing even larger contiguous stretches of DNA if only one or two essential genes (to be re-inserted elsewhere) are within the deleted sequence is explored. We apply the algorithm to design a minimized E. coli strain and a minimized B. subtilis strain and find that we are able to recapitulate the long deletions identified in previous experimental studies [2] and discover alternative combinations of deletions which have not yet been explored in vivo. MinGenome provides a pragmatic approach for constructing bioproduction chassis as genome reduction proceeds only up to the point that the minimized strain reaches growth and production yield and rate goals while alleviating the efforts to find the “true” minimal genome.

BIOT 467

MODCELL: A prototype for modular cell engineering

Cong T. Trinh, [email protected]. Chemical and Biomolecular Engineering, University of Tennessee, Knoxville, Knoxville, Tennessee, United States

To produce a target chemical at high yields, titers, and productivities, various combinations of available genetic parts for expression system can result in a large number of microbial cell factories generated for characterization. This engineering approach will become increasingly laborious and expensive when seeking to develop optimal strains for production of a large space of biochemicals due to extensive screening. Our recent theoretical development of modular cell (MODCELL) design principles can offer a promising solution for rapid generation of modular optimal production strains in a systematic manner using a modular cell and exchangeable production modules. Here, we present validation of the MODCELL design and development of a modular cell engineering prototype. We demonstrate i) a modular (chassis) cell is required to tightly couple with a production module (e.g., alcohols and esters), ii) degree of coupling between the modular cell and production modules can be modulated to enhance product synthesis, iii) a modular cell can be used as a host to select and discover functions of hypothetical proteins, and iv) metabolic pathway evolution based on growth selection could enhance growth and product rates of the weakly coupled cells. We envision that MODCELL provides a powerful platform for rapid development of optimal production strains in a plug-and-play fashion to produce a large space of biochemicals.

BIOT 468

Single-pot glycoprotein biosynthesis using a cell-free transcription-translation system enriched with glycosylation machinery

Thapakorn Jaroentomeechai1, [email protected], Jessica C. Stark2, Michael C. Jewett3, Matthew P. DeLisa1. (1) Cornell University, Ithaca, New York, United States (2) Chemical and Biological Engineering, Northwestern University, Evanston, Illinois, United States (3) Chemical and Biological Engineering, Northwestern University, Evanston, Illinois, United States

In the last decade, the emerging discipline of bacterial glycoengineering has made it possible to produce designer glycans and glycoconjugates on demand, and to evolve enzymes, pathways, and host organisms that catalyze prescribed glycosylation reactions. Bacteria equipped with recombinant glycosylation pathways have the potential to both improve our fundamental understanding of the glycosylation process and generate structurally defined glycoproteins for use as vaccines and therapeutics. These developments not withstanding, cell-based production of homogeneous glycoproteins remains a significant challenge due to the complexity of this process in vivo, cell viability constraints, and the inability to control glycosylation components at precise ratios. To address these challenges, we describe a novel cell-free glycoprotein synthesis (CFGpS) technology that seamlessly integrates protein biosynthesis with asparagine-linked (N-linked) protein glycosylation. This technology leveraged a glyco- optimized chassis strain of Escherichia coli to source crude cell extracts that were selectively enriched with essential glycosylation components, including oligosaccharyltransferases (OSTs) and lipid-linked oligosaccharides (LLOs). The resulting extracts enabled a one-pot reaction scheme for efficient and site-specific glycosylation of target proteins in just hours, requiring only the addition of DNA to enable rapid synthesis of 100% glycosylated product. The platform is flexible and modular, allowing the use of multiple distinct OSTs and structurally diverse LLOs, including the eukaryotic trimannosyl core N-glycan (mannose3-N-acetylglucosamine2, Man3GlcNAc2) at higher conversions than those reported in vivo. Our CFGpS platform broadens the glycoengineering toolbox by providing a simplified and highly modular framework for investigating glycosylation outside the confines of a cell. As such, we anticipate that CFGpS will facilitate fundamental understanding in glycoscience and make possible new applications in on-demand biomanufacturing of glycoprotein products.

BIOT 469

Engineering synthetic consortia inspired by the rumen microbiome

Michelle A. O'Malley, [email protected]. Mail Code 5080, University of California, Santa Barbara, Santa Barbara, California, United States

Anaerobic microbes evolved to work together in complex communities that decompose and recycle carbon biomass throughout the Earth – from our guts to landfills and compost piles. Compared to microbes that thrive in the presence of oxygen, anaerobes are woefully understudied and are recalcitrant to culture; they represent a vast, untapped resource for novel enzymes that degrade woody biomass into sugars, as well as natural products that could find use as new drugs. Despite their importance, little information exists to parse the role of each microbial member within their dynamic community.

To address these knowledge gaps, we pioneered new techniques to isolate anaerobes from biomass-rich environments (e.g. guts and fecal materials of herbivores) and build synthetic consortia to uncover their interdependencies. Initially, we tracked the development of enrichment cultures from goat fecal pellets grown on four types of substrates: alfalfa stem, bagasse, reed canary grass, and xylan over several generations. We tested the hypothesis that the composition of these enriched consortia would stabilize to match the metabolic requirements needed to degrade each substrate. Metagenomic sequencing of the 16S rRNA (prokaryotes), 18S rRNA (eukaryotes), and ITS (fungi) population within the consortia revealed strong specialization of the microbes during selection, suggesting that the membership of each culture tuned to match the substrate. Using these natural systems as inspiration, synthetic consortia of fungi, bacteria, and methanogens were combined in culture and tested for stability and substrate hydrolysis. In nearly all cases, synthetic consortia demonstrated faster and more complete degradation of cellulosic substrates, as well as a wider range of utilized substrates compared to monocultures. We will further discuss the roles that interwoven metabolism and secondary metabolites play on microbial consortia dynamics, both in natural and synthetic systems. Overall, the stable microbial consortia we identified here directed the formation of synthetic, interdependent communities via a bottom up approach to compartmentalize biomass-degradation and bioproduct formation.

BIOT 470 Operon refactoring and construction assistant (ORCA): An integrated workflow to refactoring biosynthetic gene clusters

Ernst Oberortner, [email protected], Michael Wornow, Michalis Hadjithomas, Samuel Deutsch. DOE Joint Genome Institute, Walnut Creek, California, United States

Next-Generation Sequencing (NGS) and bioinformatics enable the sequencing of any organism and the identification of biosynthetic gene clusters (BGC) that produce chemical compounds that a relevant for a variety of applications in medicine, agriculture, or materials.

A synthetic biology approach to access the compounds of BGCs is to refactor the natural BGCs by organizing the genes into operons and to control their expression using regulatory elements, such as promoters, ribosome binding sites (RBS), and terminators. Easy and cost-efficient access to DNA synthesis and assembly, provided, for example, by commercial DNA synthesis providers and bio-foundries, enables to manufacture the refactored BGCs in a rapid manner.

Computer-aided Design (CAD) and Computer-aided Manufacturing (CAM) tools streamline the design and build process. Various CAD/CAM tools exist that automate the individual steps/tasks of refactoring. Here, we present the Operon Refactoring and Construction Assistant (ORCA), an integrated workflow for the design of pathways and the refactoring of biosynthetic gene clusters. ORCA unites the following software tools: (i) DNAplotlib is integrated for the visualization of the designed pathways and their operons (ii) The RBSCalculator calculates a (library of) RBS sequence(s) for the genes based on the 16S rRNA sequence of the target host organism. (iii) The Build-OptimizatiOn Software Tools (BOOST) automate the design of DNA sequences, ready for a minimal cost- and time-efficient manufacturing process with a maximum success rate.

Pathways of BGCs for dipeptides have been identified from, including bacteria, such as cyanobacteria, actinobacteria, algae, protists, and fungi. A variety of dipeptides have been described as useful compounds with activities as antibiotics, signaling molecules, quorom-sensing, UV protection, as well as coatings for materials. To evaluate ORCA’s applicability in the refactoring process of BGCs, we mined the Integrated Microbial Genomes (IMG) database for different types of dipeptide pathways, including cyclic dipeptides, MAAs, and RIPPs. We selected an initial set of genes for 20 bacterial pathways for design and construction. Once the list of genes have been identified, ORCA can rapidly generate the refactored biosynthetic pathways ready for manufacturing by a commercial DNA synthesis provider.

BIOT 471

Engineering species-like barriers to sexual reproduction for applications in public health, agriculture, and the environment Michael Smanski, [email protected]. Univerisity of Minnesota, Saint Paul, Minnesota, United States

Controlling the exchange of genetic information between sexually-reproducing populations has applications in agriculture, eradication of disease vectors, control of invasive species, and the safe study of emerging biotechnology applications. We introduce an approach to engineer a genetic barrier to sexual reproduction between otherwise compatible populations. Programmable transcription factors drive lethal gene expression in hybrid offspring following undesired mating events. As a proof of concept, we target the ACT1 promoter of the model organism Saccharomyces cerevisiae using a dCas9-based transcriptional activator. Lethal over-expression of actin results from mating this engineered strain with a strain containing the wild-type ACT1 promoter.

BIOT 472

Optimization, evolution, and control of pathways and metabolism in yeasts

Hal S. Alper, [email protected]. The University of Texas at Austin, Austin, Texas, United States

Blending pathway engineering with directed evolution, adaptive strain engineering, and synthetic biology enables a powerful approach for rapid metabolic engineering. This talk will highlight how the evolution of small parts, enzymes, pathways, and whole cells can rapidly accomplish metabolic engineering goals. This talk will describe two recent advances in speeding the design process for pathway optimization and evolution. First, this talk will discuss new approaches to multiplex gene expression within pathways and broader metabolism to identify key bottlenecks limiting flux. By utilizing a CRISPR- dCas9 system, we demonstrate a rapid method for assessing the importance of these metabolic targets in a manner that is scalable and combinatorial. Second, this talk will discuss new methods to speed and automate the processes of directed and adaptive evolution through this use of microdroplet technologies. Together, these two approaches enable rapid pathway design and optimization in yeast.

BIOT 473

Development of a method for purification of monoclonal antibodies utilizing precipitation

Alison Mutchler, [email protected]. GlaxoSmithKline, Jamison, Pennsylvania, United States

Purification of monoclonal antibodies (mAbs) after production in host cells typically consists of a series of laborious column chromatography steps to remove host cell protein (HCP) and DNA impurities. Protein A affinity capture and elution chromatography is the first step in this process and affords significant reduction of these impurities while also concentrating the protein of interest. Protein A resins, however, are quite costly and constitute a high portion of the overall mAb manufacturing cost. An alternative method to affinity chromatography was investigated to facilitate a more cost- effective and facile approach to crude purification. It was found that high molecular weight polyethelyne glycol (PEG) polymers reduce the solubility of mAbs in solution and facilitate their selective precipitation in the presence of other proteinaceous impurities. Furthermore, conditions were identified that allow for the preferential precipitation and removal of nucleic acids from mAbs in solution. A two-step precipitation process was then designed for mAb purification from clarified unprocessed bulk (CUB) (Figure 1). By lowering the pH and adding PEG to the clarified harvest, DNA and HCP can be preferentially precipitated and removed by filtration. After this, the mAb can be precipitated selectively in the presence of HCPs using a higher concentration of PEG. This process was shown to reduce the HCP and DNA load for three mAbs. A screening DOE was completed on one of these mAbs to identify optimized conditions for precipitation. The optimized conditions were used to produce purified material on a 400 mL scale. By utilizing this two-step process, 98.6% of HCP was removed from the mAb CUB while removing DNA to a level lower than LOQ. The mAb recovery was 80%. These levels are within range of Protein A’s purifying ability and the reconstituted pellet was concentrated for further processing without the need for a costly and time consuming Protein A chromatography step. Two step PEG precipitation may be a faster, less expensive method for initial crude mAb purification.

BIOT 474

Water on hydrophobic surfaces – modeling of precipitation

Steffen Grosshans, [email protected], Gang Wang, Juergen Hubbuch. Institute of Engineering in Life Sciences Section IV: Biomolecular Separation Engineering, Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany

In the last decades a lot of research was done to replace preparative chromatography steps with cost-efficient alternatives. The purification by precipitation is known as such a simple, cost efficient and easily scalable alternative process step. Specifically for the manufacturing of antibodies (mAb), precipitation using polyethylene glycol (PEG) as precipitant has shown promising results in terms of mAb yield and purity. Anyhow, the process development of a precipitation step is highly empirical until now. To speed up development times and to reach a deeper process understanding, in silico methods can be applied.

This work presents a new approach for modeling of protein precipitation. Inspired by recent results in hydrophobic interaction chromatography (HIC) modeling, a mechanistic protein precipitation model was derived. The modeling concept is based on similarities between precipitation and HIC and the volume exclusion theory. In the new model the formation of protein-protein interfaces is thought to be driven by hydrophobic effects. This involves a reorganization of the water structure on the hydrophobic surfaces of the protein-protein complex as well as on the surfaces of the precipitant. Therefore water was introduced as an additional component for the model building. In this work, the model was applied to describe the precipitation of lysozyme, bovine serum albumin (BSA), and a mAb with the precipitant PEG. In high-throughput experiments 192 conditions for each protein were investigated. The data were split into calibration and validation data sets. The validation showed prediction capability of the model with an overall normalized root mean square error of prediction (NRMSEP) smaller than 7 %. The model is capable to predict precipitant concentration as well as protein concentration, even beyond the calibration space. Furthermore it is capable to describe the precipitation behavior of proteins with a wide range of properties, such as size, hydrophobicity, and the isoelectric point.

BIOT 475

Single-step purification of small non-mAb biologics by peptide-ELP based affinity precipitation

Akshat Mullerpatan1, [email protected], André Nascimento1,2, Ronit Ghosh1, Erin Kane1, Pankaj Karande1, Steven M. Cramer1. (1) Rensselaer Polytechnic Institute, Troy, New York, United States (2) iBB, Instituto Superior Técnico, Lisbon, Portugal

The current workhorse of downstream bioprocessing is chromatography-based separations. However, there is an increasing requirement to develop economical and scalable capture strategies to handle the growing diversity and rising titers of therapeutic proteins. Affinity precipitation has shown potential as a non- chromatographic purification strategy that employs thermoresponsive biopolymers such as Elastin-like Polypeptides (ELPs) for monoclonal antibodies and can be extended to non-mAb protein biologics. New protein modalities have gained significant popularity as novel protein-based drugs. We carried out the identification and design of an affinity peptide ligand against a small non-mAb biologic. A 12-mer affinity peptide ligand (P10) was identified by a primary phage display screen followed by a secondary in-solution fluorescence polarization screen. Peptide P10 and the protein interacted with a Kd of 19.5 uM. A fusion of P10 with ELP was successful in selectively capturing the biologic from a highly crude mixture. However, no elution was observed at pH 3.0, while partial elution was observed at pH 10.0. A combination of alanine and histidine scanning mutagenesis of the P10 peptide was carried out to enhance protein elution at gentler conditions. While significant affinity maturation was achieved (leading to increased capacity), elution remained a challenge. Combining the effect of rational mutations to the P10-ELP fusion, with the effect of relevant fluid-phase modifiers such as NaCl, arginine and ethylene glycol ultimately helped reach milder elution conditions. SDS PAGE and reversed phase UPLC analyses indicated the successful single step purification of the biologic from an E. coli lysate resulting in >90% purity and >80% recoveries. These results demonstrate that phage display can be readily employed to identify a peptide ligand capable of performing the ELP-based affinity precipitation purification of a small non-antibody biomolecule. Moreover, the appropriate interplay of peptide mutagenesis and fluid phase modifiers was harnessed to both improve binding capacity and induce elution at relatively mild conditions. BIOT 476

One-step affinity capture and precipitation for enhanced purification of mAbs and Fc-fusion proteins using Z-ELP functionalized nanocages

Andrew Swartz2, [email protected], Xuankuo Xu1, Steven J. Traylor1, Zheng Jian Li1, Wilfred Chen2. (1) Biologics Process Development, Bristol-Myers Squibb, Devens, Massachusetts, United States (2) Chemical Engineering, University of Delaware, Newark, Delaware, United States

Protein A chromatography has been identified as a potential bottleneck in the monoclonal antibody (mAb) production platform, sparking interest in non- chromatographic capture technologies. Affinity precipitation using environmentally responsive, Z-domain-elastin-like polypeptide (Z-ELP) fusion proteins has been shown to be a promising alternative. However, elevated temperature and salt concentrations necessary for precipitation resulted in decreased antibody monomer content and reduced purification capacity. To improve upon the existing technology, we have demonstrated the benefit of conjugating Z-ELP to a 25 nm diameter, self-assembled E2 protein nanocage (Z-ELP-E2). The enlarged scale of aggregate formation and IgG- triggered crosslinking through multi-valent binding significantly outperformed traditional Z-ELP-based methods. A Z-ELP-E2 nanocage affinity precipitation process capable of purifying industrial mAbs at ambient temperature with minimal added salt was developed through optimization of key parameters. To challenge the capabilities of the process, affinity precipitation was investigated using four industrial mAbs (mAbs A-D) and one Fc fusion protein (Fc A) with diverse molecular properties. A 3:1 Z:mAb molar binding ratio was sufficient to precipitate >95% for all molecules at ambient temperature without added salt due to multi-valent crosslinking into large aggregates. The effect of solution pH on aggregation kinetics was studied to elucidate the crosslinking mechanism. Furthermore, all molecules were capable of washing at pH ≥ 5 and were eluted with >90% recovery at pH < 4. The proteins were purified from harvested cell culture using optimized process conditions and high yield (95%) and monomer content (>97%) were obtained. mAbs A- D purification resulted in 3 logs of host cell protein and 4-5.5 logs of DNA clearance from the cell culture fluids. Nanocage affinity precipitation yield and impurity clearance were equivalent to or exceeded expected protein A chromatography performance. Because of the operational flexibility afforded by this one-step affinity capture and precipitation process, the Z-ELP-E2 nanocage based approach has the potential to be a viable alternative to platform mAb purification.

BIOT 477

Optimization of downstream purification for Modified Vaccinia Ankara virus

Laura Thomas2, Alex Xenopoulos3, [email protected], Udo Reichl2,4, Michael W. Wolff1,2. (1) Life Science Engineering, University of Applied Sciences Mittelhessen, Giessen, Hesse, Germany (2) Bioprocess Engineering, Max Planck Institute for Dynamics of Complex Technical Systems, Magdeburg, Saxony- Anhalt, Germany (3) MilliporeSigma, Bedford, Massachusetts, United States (4) Chair of Bioprocess Engineering, Otto-von-Guericke University Magdeburg, Magdeburg, Saxony-Anhalt, Germany

There has been increased interest in production of vaccines based on the Modified Vaccinia Ankara (MVA) virus, as multiple indications are being addressed therapeutically. For this work we have focused on downstream purification of MVA produced in avian cells, with the goal of optimizing conditions and placement of Benzonase® digestion of nucleic acids. As a first step, clarification was optimized by testing multiple depth filter trains for both harvested cell culture fluid (HCCF), as well as Benzonase®-digested HCCF. In both cases it was possible to reduce turbidity, while maintaining high product recovery. Depth filter throughput was extremely high (>1000 L/m2 and >400 L/m2 for primary and secondary clarification, respectively) and enabled 0.45 μm microporous filtration with high throughput (>300 L/m2). Both achievements enable a single-use clarification process with low membrane area and low footprint. Nucleic acid digestion was optimized through use of DOE experimentation, with Benzonase® concentration, incubation time and temperature and salt concentration as variables. Furthermore, it was found that digestion can be most effective when performed after clarification. This would allow either a reduced processing time or a lower amount of required enzyme to reach the same final nucleic acid target. Finally concentration can be optimally performed using very open ultrafiltration membranes. Best results were obtained during screening experiments with 1000 kD regenerated cellulose and polyethersulfone, the latter offering better recovery and process flux. Concentration and diafiltration experiments are in progress.

BIOT 478

Understanding virus removal filtration in nanocellulose-based filter paper

Albert Mihranyan, [email protected]. Nanotechnology and Functional Materials, Uppsala University, Uppsala, Sweden

Virus removal size-exclusion filtration is a complex phenomenon that is still poorly understood. The introduction of new types of virus removal filters manufactured on other principles than phase inversion raises new questions on the mechanisms of virus removal. The virus removal filter paper developed at Uppsala University, Sweden, is a new type of size-exclusion filter capable of removing even the worst-case small size viruses, such as parvoviruses, with log10 reduction values above 5. The filter paper is manufactured using wet-laid cellulose nanofibers by adapting the traditional methods of papermaking to control the pore size distribution in the region most beneficial for virus removal. While virus removal capacity is consistently shown for a wide range of model viruses and bacteriophages, questions remain whether the processing parameters influence the filter's performance. This presentation discusses the effects of varying transmembrane pressure on the removal of small size model viruses and the theoretical background of the observed effects. BIOT 479

Challenges of implementing virus filtration into continuous manufacturing processes

Daniel Strauss, [email protected], Kazuya Kobayashi, Naokatsu Hirotomi. Asahi Kasei Bioprocess America, Glenview, Illinois, United States

Continuous downstream processing has the potential to transform the manufacture of biotherapeutic products through improvements to the efficiency, cost, and quality of these processes. Significant effort has been made in both academic and industry settings to adapt batch unit operations to their continuous counterparts, with the bulk of the work focused on the challenging task of developing continuous chromatography platforms. As a flow-through operation, virus filtration seems easily adaptable to the continuous process paradigm, and as such, it has received little attention from researchers thus far. However, implementing virus filtration into a fully integrated continuous downstream process has significant challenges which will need to be addressed. For instance, the output of a continuous chromatography step is a series of elution peaks with constantly changing concentrations of protein, salts, and impurities. The impact of such a dynamic system on a filtration process is not yet understood and may require engineering solutions to mitigate its effects. Another difficulty will be validation of these critical viral clearance steps. Typical batch spiking studies, which have been traditionally used for viral clearance validation, may not be suitable for these highly dynamic systems. Although in-line spiking and other validation strategies have been proposed, the feasibility and acceptance of these approaches is not fully understood. This presentation will discuss basic considerations in designing a virus filter unit operation for integration into a continuous downstream purification process, as well as some of the challenges which must still be overcome.

BIOT 480

The effects of buffer condition on viral clearance of model proteins

Namila Khereid2, Fan Rong3, Davar Sasongko4, S. Ranil Wickramasinghe5, Xianghong Qian1, [email protected]. (2) Biomedical Engineering, University of Arkansas, Fayetteville, Arkansas, United States (4) Chemical Engineering, University of Arkansas, Fayetteville, Arkansas, United States

Validation of adequate virus clearance is critical in order to obtain FDA approval for a manufacturing process. Clearance can be due to inactivation or removal from the process stream. During the purification, manufacturers of monoclonal antibodies must demonstrate reduction of 103 to 105 or more virus particles than is estimated in a single dose equivalent of the unprocessed bulk. Depending on the antibody titer, estimates of the number of virus particles in a single dose equivalent could be as high as 1010 to 1015 retrovirus-like particles per ml. In addition to the removal of adventitious virus, using model parvovirus is required. Today unit operations such as virus filtration are frequently used to validate virus clearance. Virus filtration is a size exclusion based virus clearance method and is conducted industrially in dead end mode. Membrane performance is shown to be affected by feed pH, ionic strength and feed buffer components. In this work, one of the FDA required model viruses, minute virus of mice (MVM) was spiked to evaluate the filtration performance of BSA protein using commercially available virus filters. The effects of feed conditions including buffer, pH and ionic strength on membrane performance and virus clearance were elucidated. Moreover, the interactions and possible associations among virus particles, proteins and membrane materials modulated by the feed conditions were investigated using confocal imaging studies.

BIOT 481

Continuous processing: Lessons learned from the past and strategies for success in the future

Sadettin Ozturk, [email protected]. MassBiologics, Mattapan, Massachusetts, United States

Although it is somehow branded as new, continuous bioprocessing has been utilized over 30 years that has resulted in more than 26 licensed biopharmaceuticals. Recent attempts of re-popularizing the continuous bioprocessing received mixed reactions form the industry: great enthusiasm, confusion, and skepticism. This talk will present a fair and balanced assessment of continuous bioprocessing against well- established fed- batch platform and provide a historical and futuristic perspective. The lessons learned from the past will be discussed and ideas for a successful implementation of continuous processing will be outlined.

BIOT 482

Development of a statistical model to support perfusion bioreactor scale-up

Kelly Schwarz1, [email protected], Jonathan Raley2. (1) GMS Technical, GlaxoSmithKline, Conshohocken, Pennsylvania, United States (2) Biopharm R&D, GlaxoSmithKline, King of Prussia, Pennsylvania, United States

Perfusion technologies have proven useful in the biopharmaceutical industry for improving the efficiency and productivity of cell culture processes. While there are many perfusion systems that have been evaluated by the industry, we chose to utilize the alternating tangential flow (ATF) system, a hollow-fiber membrane device, as the system has efficient waste removal and cell retention, but additionally sweeps flow in both directions to prevent membrane fouling. During development of the process we found that we could use the ATF device to support a high cell density at small scales, but observed large process variability at production scales. As identifying crucial process parameters with experiments can be difficult, expensive, and time-consuming, we concurrently developed a statistical model to gain a better understanding of ATF performance and to identify system properties or parameters that could account for the large discrepancies between scales. Using this modelling approach, we identified that a major difference between scales was the frequency of cells cycling between the ATF and the bioreactor, where cells ‘visited’ the ATF more frequency at small scales compared to larger scales. Our model was then adapted to help identify potential operational solutions to address this discrepancy between scales, which were subsequently evaluated experimentally. Overall, this work demonstrates the ability of modelling and experiments to be utilized in combination to solve issues encountered during scale-up of perfusion bioreactor systems.

BIOT 483

Understanding product sieving challenges in TFF perfusion cell culture and developing strategies for mitigation

Lisa Sawicki1, [email protected], Daisie Ogawa1, Hayden Tessman1, Felipe Strefling1, Scott Godfrey1, Jonathan L. Coffman2, Todd Luman1, Henry Lin1, Marcella Yu1. (1) Boehringer Ingelheim, Fremont, California, United States (2) Process Science, Boehringer Ingelheim, Fremont, California, United States

Continuous processing has gained significant interest for biologics manufacturing because of its potential to streamline production by integrating upstream and downstream processes. Combined with an intensified perfusion bioreactor, continuous processing can greatly reduce cost while improving production efficiency. Critical to perfusion operation are cell separation devices, such as hollow fiber membranes for tangential or alternating flow filtration (TFF, ATF), which allow product and spent media to be removed while cells are retained within the bioreactor. However, membrane fouling and issues with product sieving have a direct impact on total product yield from the process, and can cause hollow fiber reliability issues which in some cases can result in premature termination of a bioreactor run. It is hypothesized that membrane fouling from host cell proteins, cell debris, or additives such as antifoam can result in decreased product sieving as product transmission through the membrane decreases over time. Toward addressing issues with product sieving, we aim to identify the underlying causes of membrane fouling, associated with host cell proteins and antifoam. A proteomics approach has been used to identify host cell proteins in the bioreactor, hollow fiber, and permeate to identify target proteins that might play a role in membrane fouling. Additionally, we are studying the effect of antifoam on product transmission by implementing a WAVE bioreactor design for TFF perfusion to allow comparison of cultures treated with or without antifoam. In tangent, we are exploring different strategies to overcome sieving challenges, including filtration with hollow fibers of different pore sizes and other means to minimize microsparging to reduce antifoam addition and bubble-burst-associated cell death. The identification of factors within the bioreactor that lead to reduced product sieving and implementation of new strategies to mitigate them will improve the robustness of hollow fiber performance and the overall product yield. BIOT 484

Computational fluid dynamic modeling and design of a novel single-use airlift perfusion reactor prototype for human cells

Corinna Doris, [email protected], Cameron Bardliving, Parviz Shamlou. Bioprocessing, Keck Graduate Institute, Claremont, California, United States

Cell based therapeutics have made significant advancements recently, most notably with the approval of several CAR-T cell technologies. While the science of cell based therapeutics has advanced, manufacturing patient-specific treatments remains a challenge as it requires manufacturing separate batches per patient. This motivates the need for small-scale single patient bioreactors that are capable of expanding cells to therapeutic quantities in processes that can be scaled out. Reducing size and complexity of equipment required per patient enhances the process scale out ability. Perfusion bioreactors are routinely used to culture higher density cell cultures in bioreactors of reduced size; however, current standard perfusion processes use mechanical agitation and external cell separation techniques which are more challenging to scale out and can impact product quality. In order to improve upon the current standard perfusion systems to address the cell therapy manufacturing challenge, we designed a single-use airlift perfusion bioreactor with an internal cell separation system thereby minimizing the mechanical complexity and size.

Computational fluid dynamic (CFD) modeling in combination with rapid prototyping is employed to design, characterize, and optimize the novel bioreactor design. The predictive capability of the CFD modeling enables robust system understanding of the fluid conditions of interest including mixing, oxygen mass transfer (kLa), turbulence dissipation, and shear rate to provide a rationale for the geometric and operating parameter modifications for optimization.

BIOT 485

Optimizing the addition antifoam for process intensification: A case study using small scale models

Jonathan Wang2, [email protected], Jason Walther1. (1) Sanofi, Framingham, Massachusetts, United States (2) Upstream Development, Sanofi, Framingham, Massachusetts, United States

Process intensification provides many advantages for the production of therapeutic proteins. One of the strategies for process intensification is to maintain high cell densities to improve productivity. High cell density cell culture results in high demand in oxygen sparge which leads to foaming issues. This is even more prominent in intensified fed batch and intensified perfusion processes. The addition and accumulation of antifoam can cause a negative impact on cell culture performance. For better understanding, we have developed a small scale model to characterize a design space between the factors of shear, media components and additives, and antifoam addition. The goal is to minimize the effect that addition of antifoam has on cell culture growth and viability. We utilized orthogonal approaches to correlate shear stress between small scale and at scale models. We will present small scale and bioreactor data demonstrating the verification of our predicted design space.

BIOT 486

Process improvement of a high density fed-batch process using N-1 perfusion seed cultures

Joon Chong Yee, [email protected], Timothy Erlandson, Sarbajita Ray, Steven Sowa, Jun Tian, Qin He, Michael C. Borys, Zheng Jian Li. Biologics Process Development, Bristol Myers Squibb, Devens, Massachusetts, United States

The current bioprocessing landscape has relied on the development of nutrient-rich media and selection of high producing cell lines to drive titers higher and increase drug substance output in a manufacturing facility. In BMS, efforts in upstream process intensification have incorporated perfusion seed bioreactors, to reduce the duration of seed expansion, enable the implementation of a rolling seed and the inoculation of production bioreactors at high cell densities. This presentation demonstrate the utility of biomass capacitance probes to monitor online cell densities and control perfusion rate in the seed bioreactor. Based on a robust, linear correlation of biomass capacitance to viable cell densities across multiple platform cell lines, the online capacitance data was utilized to dynamically change the media exchange rate to maintain a constant cell-specific perfusion rate (CSPR). A dynamic adjustment of perfusion rate based on online VCD can reduce media utilization rates in large-scale manufacturing. CSPR rates was shown to impact cell growth in perfusion culture and the duration of the N-1 seed. Due to higher seeding densities in fed-batch process, the feed media components and feeding strategy was also adjusted to target a stoichiometric based feeding, where cultures were fed based on growth and nutrient consumption rates estimated from integrated viable cells. This presentation will highlight the impact on process titer yields and product quality profile using this intensified, high density fed batch process.

BIOT 487

A novel high density N-1 batch seed strategy development for CHO cell culture manufacturing

Andrew Yongky, [email protected], Jia Zhao, Jianlin Xu, Matthew Rehmann, Michael C. Borys, Zheng Jian Li. Bristol-Myers Squibb, Devens, Massachusetts, United States

Chinese hamster ovary (CHO) cells are the workhorse for the production of therapeutic recombinant proteins in the biopharmaceutical industry. Implementation of high seeding fed-batch production CHO cell cultures have been shown to increase productivity and shorten culture duration. With increasing demand for these therapeutics, high seeding cell culture is becoming necessary to increase the manufacturing capacity. However, generation of high quality N-1 seed with high viable cell density remains a significant challenge in those cell culture facilities such as large scale facilities with > 15,000-L bioreactors without an ATF perfusion equipment. In this study, we compared various N- 1 seed strategies to achieve high cell densities for inoculating production bioreactors. Finally we developed a batch N-1 high cell density strategy without the need of ATF perfusion equipment, thus removing additional complexity for large scale manufacturing. In addition, the impact of different high cell density N-1 seed strategies on CHO cell culture performance and quality attributes will be presented.

BIOT 488

Development and implementation of perfusion N-1 for mammalian cell culture to enhance process productivity

Ryan Sanford, [email protected]. Cell Culture Development, Biogen, RTP, North Carolina, United States

The use of perfusion in the seed train to increase volumetric productivity increases facility yield and flexibility. Here we use multiple case studies to show the optimization of the process to match the facility fit, adjust to program needs, and generate a platform process. Finally we show the transfer of the bench scale process to a validated manufacturing process with significant increase in volumetric productivity.

BIOT 489

Prediction of protein biophysical properties and chemical modifications by correlating in silico protein structural features

Lei Jia, [email protected], Dora Toledo Warshaviak, Yaxiong Sun. Amgen, Thousand Oaks, California, United States

Protein biophysical properties as well as chemical modification are major concerns in the development of protein therapeutics. They affect biologics therapeutics' efficacy, manufacturability, and safety. We are interested in developing and implementing prediction methods to forecast the potential liabilities in biologics, including thermostability, aggregation, viscosity, deamidation, isomerization etc.

Due to the limited size of data sets and complicated structural features of proteins, prediction accuracy of those models is challenging. More accurate prediction methods would allow elimination or reduction of problematic variants as early as possible in the drug discovery process. We try to develop accurate prediction models by mining the key features of proteins (feature engineering) and applying machine learning techniques. Features includes in silico biophysical properties, local and global 3D structural information, especially the surface properties. Machine learning algorithms are able to select key features which contribute most to the predictive models through feature selection. Accurate prediction models can be derived by learning from the key features. Those features can also guide protein engineers for better design of new variants.

BIOT 490

Insulin folding landscape and in-silico design of peptide excipients to inhibit aggregation

Gaurav Goel2, [email protected], Richa Singh2, Avinash Mishra2, Rohit Bansal2, Anurag S. Rathore1. (1) Department of Chemical Engineering, Indian Institute of Technology, New Delhi, India (2) Chemical Engineering, Indian Institute of Technology (IIT), Delhi, New Delhi, Delhi, India

Recent studies have highlighted the important role of protein unfolding thermodynamics and reconfiguration kinetics for protein aggregation. Further, it has been shown that identification of short-lived aggregation prone species can help in design of strategies to inhibit aggregation. These earliest events in the aggregation process are difficult to characterize in experiments. In this talk, I will present our work on using molecular simulations to probe short timescale, rare events associated with protein unfolding. This knowledge was subsequently used to design peptides that significantly inhibit insulin aggregation. We have used an enhanced sampling molecular dynamics technique, namely bias-exchange metadynamics (BEMD), coupled with a Markovian kinetic transition network model to construct a detailed thermodynamic and kinetic model of insulin folding. We find that metastable states identified in our study compare well with experimental data obtained in spectroscopic, crystallographic, and calorimetric measurements during early stages of insulin aggregation. We then identified metastable states with a high propensity for aggregation using a combination of protein-protein docking and scoring. Hot spot residues for insulin self-association were identified and used as a template to design a small peptide mimicking a significant number of interactions made by these binding interface residues. We incubated insulin at 60oC by adding designed peptides to formulation buffer and found up to 7 times reduction in monomer loss as quantified by size exclusion chromatography. ITC measurements were used to determine enthalpy of peptide binding to insulin at non-amyloidogenic conditions. Finally, binding of insulin to its receptor in presence of peptides was confirmed using SPR.

BIOT 491

Predicting pH and ionic strength effects on the colloidal stability of antibodies

Marco A. Blanco3,1, [email protected], Harold W. Hatch3, Maria Monica Castellanos3,1, Curtis W. Meuse3,1, Kevin W. Mattison2, Ioannis Karageorgos3,1, Joseph E. Curtis3, Vincent K. Shen3. (1) Institute for Bioscience and Biotechnology Research, Gaithersburg, Maryland, United States (2) Malvern Instruments Inc, Westborough, Massachusetts, United States (3) National Institute of Standards and Technology, Gaithersburg, Maryland, United States

Monoclonal antibodies (mAb) represent the fastest growing class of biotherapeutics, as their versatility makes them valuable for treating many chronic and fatal illnesses. However, the success (or failure) of a promising mAb candidate generally depends on stabilizing the protein against different degradation routes. Of particular interest are those issues related to colloidal stability or protein-protein interactions that result in protein self-association and other phenomena such as phase separation, high viscosity, or aggregation. Difficulties in controlling self-association propensity in proteins arise from their inherent marginal stability, where small changes in protein concentration or solution conditions may lead to attractive interactions and favor undesirable thermodynamic processes. Moreover, these challenges are augmented for mAbs when one considers the hinge flexibility, heterogeneous surface charge distribution, or highly non-globular morphology. Thus, understanding the relation between the physical mAb attributes, protein interactions, and self-association propensity is critical for mitigating colloidal instability problems during the development and manufacturing of mAb therapeutics. Using the NIST standard reference antibody (NISTmAb) and its separated Fc and Fab domains as model systems, the roles of pH, ionic strength, and protein concentration within the generalized colloidal stability framework for mAbs is assessed via a variety of biophysical techniques (e.g., SLS/DLS, SAXS, FTIR) for a wide range of solution conditions (pH 3–7, ionic strength 5–300mM, and mAb concentration < 250mg/mL). These experimental data are employed for the design of a minimalistic mAb computational model that coarse-grains antibodies at a sub-domain level (e.g., CH2, CH3, Fv) while preserving hinge flexibility and key physical aspects of protein interactions. Overall, the behavior of the NISTmAb correlates well with classical colloidal theory, where decreasing pH or ionic strength lead to stronger repulsions between proteins. However, these stronger repulsions do not relate to colloidal stability, as decreasing pH increases aggregation propensity and native self-association. Analysis of computational and experimental results shows that antibody stability is governed by the imbalance between domain-domain interactions; specifically, discrepancies in NISTmAb stability from classical colloidal stability arise from entropy-driven, native Fc-Fc interactions.

BIOT 492

Experimental and bioinformatics methods for identifying monovalent and bivalent antibodies with high colloidal stability

Magfur E. Alam1, [email protected], Steven B. Geng1, Christian Bender3, Seth D. Ludwig1, Lars Linden4, Rene Hoet3, Peter Tessier2,1. (1) Chemical & Biological Engineering, Rensselaer Polytechnic Institute, Troy, New York, United States (2) Departments of Chemical Engineering and Pharmaceutical Sciences, University of Michigan, Ann Arbor, Michigan, United States (3) Pharmaceuticals, Bayer AG, Nattermannallee, Cologne, Germany (4) Pharmaceuticals, Bayer AG, Aprather, Wuppertal, Germany

In vitro antibody discovery and/or affinity maturation are often performed using antibody fragments (Fabs), but most monovalent Fabs are reformatted as bivalent IgGs (mAbs) for therapeutic applications. One problem related to reformatting antibodies is that the bivalency of mAbs can lead to increased antibody self-association and poor biophysical properties (e.g., reduced antibody solubility and increased viscosity). Therefore, it is important to identify monovalent Fabs early in the discovery and/or optimization process that will display favorable biophysical properties when reformatted as bivalent mAbs. We have developed a facile approach (self-interaction nanoparticle spectroscopy, SINS) for evaluating Fab self-association in a multivalent assay format that is capable of identifying antibodies with low self-association and favorable colloidal properties when reformatted as bivalent mAbs. Importantly, we find that SINS measurements of Fab self- association are correlated with self-interaction measurements of bivalent mAbs and are useful for identifying antibodies with favorable biophysical properties. Surprisingly, we also find that the significant differences in the levels of self-association detected for Fabs and mAbs with similar frameworks can be largely explained by the physiochemical properties of the complementarity-determining regions. We expect that these findings will be useful for improving the development of therapeutic antibodies that are well suited for high concentration applications.

BIOT 493

Predicting the solubility of VLPs: Qualitative structure property relationship (QSPR) modeling applied to HBcAg VLPs

Philipp Vormittag1, [email protected], Cathrin Dürr1, Anja Wilming2, Thomas Hiller2, Thorsten Klamp2, Juergen Hubbuch1. (1) Institute for Engineering in Life Sciences, Section IV: Biomolecular Separation Engineering, Karlsruhe Institute of Technology, Karlsruhe, Germany (2) BioNTech Protein Therapeutics GmbH, Mainz, Germany

Engineered chimeric virus-like particles (VLPs) have shown great potential as cancer immunotherapeutic. Chimeric VLPs incorporate antigenic epitopes that can alter the assembly competence, solubility and stability of the virus proteins. To reduce those effects, the epitope inserts are fused to linkers. However, integration of a linker often does not lead to the desired effect of creating an assembly competent, soluble and stable VLP, suggesting there is a dependence of these properties on the VLP platform and the structure of the inserted epitope and linker. In this paper, we propose an in silico approach to predict VLP solubility in clarified harvest based on a chimeric Hepatitis B virus core antigen (HBcAg) VLP platform and reveal critical insert properties. Recent computational advances have rendered molecular dynamics (MD) simulations applicable to more complex protein structures. From X-Ray diffraction structures of the native HBcAg dimer we generated by homology modeling and MD simulations 3D models of 27 virus protein dimers associated with chimeric HBcAg VLPs, varying in the sequence of the insert. These 27 VLPs were characterized by their solubility in clarified E. coli harvest. We created in silico models to predict VLP solubility applying QSPR modeling. Based on the protein 3D structure and amino acid sequence, molecular descriptors related to geometry, charge and hydrophobicity were calculated. Besides well-established descriptors, we implemented descriptors that were specifically designed for this study, assuming that a rational design of the QSPR model potentially decreases the probability of including unnecessary or erratic descriptors. With principal component analysis (PCA) data dimension was reduced and solubility class membership was predicted by classification of the constructs into clusters. The model assigned more than 91% of the constructs correctly in cross-validation. Analysis of the model uncovered critical insert properties that may aid the design of novel inserts or be the basis for the modification of insoluble constructs. In conclusion, we developed a rational QSPR approach to predict VLP solubility. We established a tool to reduce screening efforts in early VLP drug development, cutting down the number of required experiments including DNA synthesis, plasmid transformation and expression of potential candidates by the ability of identifying insoluble VLP species thus potentially decreasing time-to-market and R&D costs.

BIOT 494

Characterizing long-term protein phase behavior via multidimensional empirical protein phase diagrams containing image-based features

Marieke Klijn, [email protected], Juergen Hubbuch. Karlsruhe Institute of Technology, Karlsruhe, Germany

Stable storage conditions are identified via long-term experiments, where formulations can be stored 2 to 5 years under multiple conditions. Solution conditions frequently varied include protein concentration, additives, additive concentration, pH and temperature. These experiments generate information on protein phase behavior over time and its results allow for characterization of underlying (de)stabilizing protein-protein and protein-solvent interactions. The underlying interactions do not only determine formulation instability but also the aggregation type that occurs after storage. Typically, formulations are optically scored where distinctions between different morphologies are made. Crystals, precipitates, phase separation, skin formation or gel formation are the main morphology categories used for separating aggregated solutions.

The final morphology indicates underlying protein-protein and protein-solvent interactions, but the effect of these interactions on kinetics and subtypes within the morphology categories is also essential for understanding protein phase behavior. For example, underlying interactions influence crystal dimensions and growth properties such as formation time. An automated imaging setup allows for monitoring formulations during storage, but this information is rarely included when characterizing long-term protein phase behavior. Conventional protein phase diagrams visualize results with unique symbols for each morphological category. Incorporation of more symbols or altering symbol dimensions to represent additional information results in incomprehensive diagrams. In addition to visualization issues, systematic extraction of features from image time series has not been included yet. This study has extracted various image features from a 40 day storage experiment to enhance understanding of long-term protein phase behavior as a function of varying solution conditions. The image-based data is compiled into an empirical protein phase diagram. The impact of additional information on the interpretation of protein phase behavior is shown via comparison of the same data represented in a conventional protein phase diagram.

BIOT 495

Protein-protein interactions and viscosity of highly concentrated monoclonal antibody solutions

Barton Dear2, [email protected], Jessica Hung2, Jon Bollinger2,3, Logan Wilks2, Ayush Sharma2, Carl Karouta2, Maria Nieto2, Thomas Truskett2, Keith P. Johnston1. (1) Univ of Texas, Austin, Texas, United States (2) Department of Chemical Engineering, University of Texas at Austin, Austin, Texas, United States (3) Sandia National Laboratoreis, Albuquerque, New Mexico, United States

In solution, monoclonal antibodies (mAb) interact via both electrostatic and hydrophobic forces (among others), resulting in complex interaction potentials. In the simplest forms these potentials are believed to include a short-ranged attraction, caused by oppositely charged patches, multi-poles and hydrophobics, and a long-ranged repulsion due to their similar net charge. At high concentration (>200 mg/ml) these short-ranged interactions become more significant and cause the formation of protein networks, often resulting in aggregation and viscosities too high for subcutaneous delivery. In the present study, we use small molecule co-solutes to screen these interactions, breaking networks and reducing the viscosity. Previous attempts to relate the effects of co- solutes on interactions to their effects on viscosity have not been definitive. We believe these inconsistencies have occurred, because many co-solutes screen both short- ranged attractions and long-ranged repulsions leading to conflicting effects on net measurements. Since most measurements of protein interactions are performed in the dilute state, the short-range interactions are less significant. Therefore, we delineate between these interactions, by characterizing mAb solutions at low and high concentration, using a series of techniques that probe various length/time scales, including Small Angle X-ray Scattering (SAXS), light scattering (LS) and rheology. Additionally, we compare molecular dynamics simulations to our SAXS results to show further evidence that these viscosity reducing co-solutes screen short-ranged attractive forces.

BIOT 496

Microrheology of antibody solutions Eric M. Furst1, [email protected], Mahlet Woldeyes1, Christopher J. Roberts2. (1) Univ of Delaware, Newark, Delaware, United States (2) University of Delaware, Newark, Delaware, United States

A key challenge encountered in the current development of therapeutic protein solutions is the need to measure their viscosity to identify stable, syringeable formulations in a large composition space. Commercially available techniques such as capillary viscometry and rotational rheometers are frequently used, but require relatively large sample volumes. This sample size requirement restricts the number of rheological measurements in the early development stage when only small amounts of protein are typically available. This talk will focus on characterizing the viscosity of protein therapeutics over a wide range of compositions and temperatures with a minimal amount of material. Microrheology techniques are powerful methods to study scarce biomaterials. We use multiple particle tracking (MPT) to provide a high throughput sample processing platform for protein therapeutics. This work examines the microviscosity of three IgG monoclonal antibodies and two bispecific monoclonal antibodies assembled from IgG half-antibodies over a large range of concentrations (1 - 180 mg/mL). The protein solutions exhibit Newtonian fluid behavior over a frequency range of 0.05 to 30 s-1. An analysis of covariance demonstrates the high accuracy of small volume microrheology measurements. Based on the relative error between measured and tabulated viscosities, the uncertainty of viscosities derived from particle tracking is less than 2% of the true value.

BIOT 497

Revoke immunosuppression of tumor microenvironment using engineered molecular traps

Rihe Liu, [email protected], Lei Miao, Jingjing Li, Leaf Huang. Eshelman School of Pharmacy, UNC Chapel Hill, Chapel Hill, North Carolina, United States

Activation of the immune system against cancer has major advantages over conventional therapies, including greater specificity, stronger potency, and longer term efficacy. Progress made recent years focuses on overcoming T-cell immunological checkpoints with blocking monoclonal antibodies (mAbs) to CTLA-4 and PD-1/PD-L1. Despite targeting PD-1/PD-L1 in cancer has shown beneficial responses in some cancer types, it is not effective in colorectal cancer liver metastasis and pancreatic cancer, two most deadly malignant human diseases. We address the challenges by developing affinity molecules based on protein domains (called traps) that potently and specifically bind and disrupt the biological functions of signaling molecules that play pivotal roles in the immunosuppression of tumor microenvironment (TME). To deliver traps to the diseased tissues, we have used targeted nanoparticles encapsulating a plasmid encoding the trap protein and delivered it to TME. Local and transient expression of the traps in TME profoundly remodeled and revoked its immunosuppression, resulting in activation of CD8 T-cells to inhibit tumor growth and metastasis. Using an orthotopic, syngeneic CT26 mouse model, we demonstrated that the liver metastasis of colorectal cancer was drastically blocked by delivering a CXCL12 trap (Science Translational Medicine, 2016 Nov 9; 8(364):364ra153). Using an orthotopic, syngeneic KPC mouse model, we demonstrated that the local and transient expression of a combination of a CXCL12 trap and an innovative trimeric PD-L1 trap potently shrunk pancreatic cancer and significantly prolonged the host survival (ACS Nano, 2017 Sep 26; 11(9):8690-8706). Combo trap therapy also significantly reduced metastasis of pancreatic cancer to other organs. Significantly, our mechanistic studies showed that the CXCL12 trap allowed the penetration of T-cells into the tumor, while the PD-L1 trap allowed the activation of infiltrated T-cells to kill cancer cells. This novel cancer immunotherapy has distinctive advantages over conventional mAbs in that the gene therapy approach allows local and highly effective reversal of the immunosuppression of TME, as well as tunable combination of more than two traps targeting signaling pathways with synergistic therapeutic effect, making it possible to avoid the immunodeficiency and other side toxicities resulted from the systemic use of immunotherapeutics with long half-lives.

BIOT 498

Driving targeted, reversible cell-cell interactions with a universal membrane engineering method

Clifford M. Csizmar1, [email protected], Jacob R. Petersburg1, Lawrence A. Stern2, Benjamin J. Hackel2, Carston R. Wagner1. (1) Medicinal Chemistry, University of Minnesota , Minneapolis, Minnesota, United States (2) Chemical Engineering and Material Science, University of Minnesota, Minneapolis, Minnesota, United States

Membrane-engineered cells displaying antigen-targeting ligands have been developed for a variety of therapeutic applications, including immunotherapy, tissue engineering, and regenerative medicine. While genetically-encoded artificial receptors have proven efficacious in certain settings, their scope is currently limited by the genetic-engineering that underlies the approach. To circumvent some of these limitations and expand the use of membrane-engineered cells, our group has developed a non-genetic method to stably and reversibly modify essentially any cell membrane with a targeted protein scaffold.

First, we engineered a protein ligand based upon the human tenth type III fibronectin domain (Fn3) that binds to epithelial cell adhesion molecule (EpCAM), an overexpressed tumor antigen and cancer stem cell marker. Using yeast surface display, mammalian cell panning, and a novel titratable avidity-reduction selection technique, we successfully evolved Fn3 clones exhibiting high affinity (11 nM) and robust selectivity for cellular EpCAM.

We then incorporated these EpCAM-targeting Fn3’s into a multivalent chemically self- assembled nanoring (CSAN) for use as a cell-directing protein scaffold. To install the CSAN on a cell surface, phospholipids conjugated to either biotin or dibenzocyclooctyne (DBCO) are hydrophobically inserted into the membrane. Heterobifunctional CSANs displaying the Fn3’s and either monovalent streptavidin domains or free azide groups are then stably bound to the biotin or DBCO moieties, respectively, granting EpCAM- targeting capabilities to the cell. Importantly, the CSAN scaffold can be rapidly disassembled via exposure to the FDA-approved antibiotic trimethoprim, providing a clinically-relevant mechanism for reversing the artificial cell-cell interactions. Using this system, we have successfully directed and reversed targeted cell-cell interactions in vitro.

In conclusion, we have engineered novel, EpCAM-binding Fn3 domains and incorporated them into a modular scaffold that can be used to direct pharmacologically- reversible cell-cell interactions. Thus, our CSAN platform is directly applicable to several therapeutic arenas and may offer distinct advantages over established cell-directing methods.

BIOT 499

Development of late-stage ready CMC approaches to enable recombinant AAV biomanufacturing

K. Reed Clark, [email protected]. Pharmaceutical Development, Dimension Therapeutics, Cambridge, Massachusetts, United States

Use of recombinant AAV (rAAV) vectors continues to expand beyond hPOC studies in rare diseases and has begun to transition to late-stage clinical and process development. Accordingly, the establishment of robust manufacturing platforms and strategies to support development represents a clear opportunity within the space and will be highlighted in this presentation. Current approaches to upstream, downstream and analytical operations to meet these demands and challenges with scale, access, supply chain and COGs will also be discussed. An in depth understanding of product quality and comparability metrics will continue to drive process maturation with robust analytical assays essential to informing these assessments as this novel biologic moves into an ever increasing number of pivotal clinical trials and eventual commercialization.

BIOT 500

Development of a scalable production process for the manufacturing of exosome- based biotherapeutics

Aaron Noyes, [email protected], Kim Ellis, Michael Doherty, Raymond Bourdeau, Michael Mercaldi, Konstantin Konstantinov. Chemistry Manufacturing and Controls, Codiak BioSciences, Cambridge, Massachusetts, United States

Exosomes have emerged as a therapeutic platform that could allow for the development of highly potent and directed therapeutics. As efforts accelerate to translate exosome biology into new medicines, clear technology gaps have emerged between the current state of the art for producing extracellular vesicles, including exosomes, and the capabilities necessary to support large scale clinical and commercial manufacturing. Codiak BioSciences has created a scalable exosome production platform based on chemically-defined, suspension cell culture with purification using conventional downstream unit operations. In this presentation, the potential of exosome-based therapeutics and the bioprocessing technologies for their future commercialization will be discussed in the context of the capabilities and limitations of the existing biomanufacturing platforms. The development strategy employed by Codiak BioSciences will be described and supported by case studies from process development. This will include examples of challenges specific to working with large macromolecular complexes.

BIOT 501

Hepatitis B viral-like particles as protein delivery vehicles for targeted therapy and genome editing

Emily Hartzell2, [email protected], Heejae Kim1, Wilfred Chen1. (1) Chemical Engineering, University of Delaware, Newark, Delaware, United States (2) Chemical and Biomolecular Engineering, University of Delaware, Newark, Delaware, United States

In an effort to develop smarter, targeted therapies for disease, nanodevices, which can interact with specific cells via molecular level signals and deliver high therapeutic payloads, are of great interest. We have developed a modular Hepatitis B Viral-Like Particle (HBV VLP) nanocarrier platform for the tunable display of moieties which can target over-expressed surface receptors of cancer cells and escape endosomal compartments. What remains a significant challenge in engineering these nanodevices for the delivery of therapeutic protein cargo is providing the ability to efficiently release these proteins from the carrier, so that they can execute their therapeutic functions. The reducing environment in the cytosol has often been exploited as a cargo release mechanism for proteins tethered through disulfide bonds; however, the ability to form these reversible bonds in vitro is complicated by nondiscriminatory free cysteines and unwanted homodimeric byproducts. To combat this, we have developed a genetically encoded protein module, which uses affinity domains to drive the formation of disulfide bonds to the respective cargo. We have demonstrated the redox dependent tethering of a GFP reporter to the HBV VLP nanocarrier and have shown the released protein is able to localize to the nucleus with the addition of a nuclear localization sequence. We further intend to apply our device towards targeted genome editing by similarly delivering Cas9 and demonstrating its activity through the knockdown of a fluorescent reporter.

BIOT 502

Effective cancer therapy via therapeutic-loaded, cell-derived nanovesicles induced by sulfhydryl-blocking (NIbS)

Dominique Ingato2, [email protected], Julius Edson3, Michael Zakharian4, Young J. Kwon1. (1) Chems, Uci, Irvine, California, United States (2) Chemical Engineering and Materials Science, University of California, Irvine, Irvine, California, United States (3) Kwon Lab, University of California, Irvine, Irvine, California, United States (4) University of California, Irvine, Irvine, California, United States

Extracellular vesicles (EVs) are a promising carrier for targeted therapeutic delivery, but the development of EV technology has been hindered by poor scalability. We have developed a method to scale-up cellular production of nano-sized EVs by an order of magnitude using a sulfhydryl-blocking reagent. The Nanovesicles Induced by Sulfhydryl-blocking (NIbS) were loaded with chemotherapeutic doxorubicin. Tumor- bearing mice treated with drug-loaded NIbS showed improved survival outcomes and diminished tumor growth in comparison to those treated with free drug or the commercial liposomal formulation. Additionally, NIbS showed lower non-specific accumulation in the lungs and the liver. A versatile therapeutic carrier, NIbS overcome the traditional scalability problems associated with EV therapeutics.

BIOT 503

Incorporation of a weak poly(acid) into multilayer polymer coatings enables rapid contact-transfer of DNA to soft surfaces

Visham Appadoo2, [email protected], Yan Yu1, Jun Ren3, Bo Liu3, Timothy Hacker4, David M. Lynn1. (1) Chemical and Biological Engineering, University of Wisconsin-Madison, Madison, Wisconsin, United States (2) Chemistry, University of Wisconsin-Madison, Madison, Wisconsin, United States (3) Department of Surgery, UW-Madison, Madison, Wisconsin, United States (4) Cardiovascular Physiology Core Facility, University of Wisconsin-Madison, Madison, Wisconsin, United States Thin films and coatings that promote the contact-transfer of nucleic acids to cells and tissues have the potential to be useful in a broad range of biomedical and biotechnological contexts. Contact-mediated delivery often requires quick release of DNA or the rapid separation of a coating from its underlying substrate (e.g., over periods of seconds or minutes), but many polymer films fall apart or release DNA on much slower time scales (e.g., over many hours, days, or weeks) and are thus not useful for such applications. Here, we report a strategy for the design of thin polyelectrolyte-based multilayer coatings that enables the ‘quick release’ and rapid contact-transfer of plasmid DNA to other secondary surfaces in as little as 30 seconds when pressed into contact with tissue or other soft objects. This approach is based on the incorporation of layers of weak polyelectrolytes such as poly(acrylic acid) (PAA) into polyelectrolyte multilayers fabricated using layer-by-layer assembly. We demonstrate that this approach can be used to promote the rapid disassembly and contact transfer of DNA-containing films fabricated using a range of different degradable and non- degradable polyamines, through a mechanism that involves the ionization of PAA at physiological pH. This approach thus enables contact-transfer of DNA to objects surrounded by or bathed in physiological media. We demonstrate that this approach can be used to coat the surfaces of inflatable catheter balloons and used to promote the inflation-induced contact-transfer of plasmid DNA to the peripheral arteries of pigs as well as other soft, deformable objects, including hydrogels with planar or curved geometries. Aspects of film fabrication, polymer structure, environmental conditions, and physical forces and parameters that influence rates and extents of film transfer will be discussed. These new quick-release materials enable rapid, contact-transfer of DNA and potentially other biomolecular agents, and thus open the door to the design of new therapeutic strategies, including microneedle and angioplasty balloon-based therapies, that would be difficult to achieve using conventional polymer-based materials. Advances toward potential applications and new approaches to promoting the rapid release of DNA-containing coatings based on the design of rapidly dissolving hydrogen-bonded sacrificial layers will also be presented.

BIOT 504

Adventures in RNA synthetic biology

Chase Beisel, [email protected]. North Carolina State University, Raleigh, North Carolina, United States

RNA was once considered a passive information carrier in biology, with limited importance beyond common processes in gene expression. However, accumulating examples have illustrated the versatility of RNA and its important regulatory roles in all forms of life. These insights in turn have motivated the harnessing of RNA for synthetic biology. In this talk, I will describe the emergence of RNA synthetic biology as a field through the perspective of my graduate, postdoctoral, and independent research. I will also describe my group's efforts to harness RNA-guided immune systems called CRISPR-Cas systems and how the resulting CRISPR technologies are revolutionizing biotechnology, medicine, agriculture, and basic research as we know it. BIOT 505

Efficient technology transfers to increase agility, flexibility, and productivity

Aaron Goerke2, [email protected], Kara A. Calhoun1. (1) Genentech, S San Fran, California, United States (2) Global Biologics MSAT, Hoffmann-La Roche, Salem, Oregon, United States

The current pharmaceutical business environment requires increasing agility, flexibility, and efficiency in bioprocess manufacturing to ensure competitiveness. Excellence in technology transfer execution is one way to become more agile and to develop a competitive advantage. Roche’s large manufacturing network, using both internal and external (CMO) manufacturing capabilities, provides many examples of various bioprocess technology transfers. A consistent framework for technology transfer is used that incorporates elements of quality, speed, and risk management. The amount of complexity is dictated by the level of changes in process, scale, or analytical methods as well as by the experience level of donor and receiving parties. When external manufacturing (CMO) is included, additional complexities can arise due to differences in transfer methodologies, cultural norms, and communication styles. Despite the complexities, a unified framework for transfers has allowed for significant reductions in transfer time. A series of recent transfers for Roche commercial legacy processes will be described. Facility fit considerations for unit operations, scale-up/scale-down approaches, and obstacles encountered during manufacturing campaigns illustrate challenges encountered and overcome.

BIOT 506

Effects of bed compression on protein separation on gel filtration chromatography at bench and pilot scale

Darryl Kong1, [email protected], Spyridon Gerontas1, Ross McCluckie2, Martin Mewies2, David Gruber2, Nigel Titchener-Hooker1. (1) Biochemical Engineering, University College London , London, United Kingdom (2) Ipsen Biopharm Limited, Wrexham, United Kingdom

Poorly packed chromatography columns are known to reduce drastically the column efficiency and produce broader peaks. Controlled bed compression has been suggested to be a useful approach for solving this problem. Here the relationship between column efficiency and resolution of protein separation are examined when preparative chromatography media were compressed using mechanical and hydrodynamic methods. Sepharose CL-6B, an agarose based size exclusion media was examined at bench and pilot scale. The asymmetry and height equivalent of a theoretical plate (HETP) was determined by using 2% v/v acetone, whereas the void volume and intraparticle porosity (εp) were estimated by using blue dextran. A protein mixture of ovalbumin (chicken), bovine serum albumin (BSA) and gamma- globulin (bovine) with molecular weights of 44, 67 and 158 kDa, respectively, were used as a “model” separation challenge. Mechanical compression achieved a reduction in plate height for the column with a concomitant improvement in asymmetry. Furthermore, the theoretical plate height decreased significantly with mechanical compression resulting in a 40% improvement in purity compared to uncompressed columns at the most extreme conditions of compression used. The results suggest that the mechanical bed compression of Sepharose CL-6B can be used to improve the resolution of protein separation.

Comparison of reduced plate number and asymmetry for compressed beds achieved by hydrodynamic and mechanical methods. Columns packed with Sepharose CL-6B 0.016 m I.D. 20 cm bed height. (a) hydrodynamic compression achieved by multiple incremental steps (b) hydrodynamic one step compression. (c) mechanical compression achieved by multiple incremental steps (d) mechanical one step compression. (■) reduced plate height; (Δ) asymmetry.

BIOT 507

Development of in-line PAT tool for real-time concentration measurement of biologics during ultrafiltration/diafiltration

Sushmitha Krishnan, [email protected], Sobhana Sripada, Sayantan Bose, Dharmesh Bhanushali. Advanced Manufacturing Technology, GlaxoSmithKline, King of Prussia, Pennsylvania, United States This work aims to describe the use of an in-line UV sensor to monitor product concentration in real-time during the Ultrafiltration/Diafiltration (UF/DF) unit operation of the biologics purification process. A typical UF/DF involves concentrating purified protein from approximately 10g/L to over 200g/L. Each phase of the operation ends based on a target protein concentration, which is currently estimated through manual mass balance calculations and verified with benchtop UV spectrophotometer measurements. In a manufacturing environment, the mass balance estimates are prone to error, particularly at low operating volumes, and the offline measurements require process pauses and quality checks, which add significant time to operation. The large concentration range causes challenges in quantitation through the traditional method of measuring absorbance at 280nm as the signal of concentrated proteins saturates the detector. While sample dilutions and pathlength alterations can be performed offline to achieve a signal within the instrument’s linear range, this is not feasible for an in-line measurement. Instead, the proposed sensor will measure absorbance of a flowing sample at an off-peak wavelength of 313nm, which allows for a decreased, unsaturated signal even at high concentrations, but also detectable signal at low concentrations. Proof-of-concept work with this UV sensor demonstrated capability to measure concentrations in real-time with multiple monoclonal antibodies at ranges from 5 – 230g/L in both lab and pilot scale UF/DF and established excellent comparability in accuracy to traditional offline measurements. The use of this technology in a GMP environment will enable improved process control, decreased risk of batch failure due to overconcentration, and reduction in personnel needs and processing time, leading to simplification of operation and reduction of cost.

BIOT 508

A case study of tech transfer and growing pains: Shoehorning a process to fit a facility

Alexis Henry1, [email protected], Ling Zhang1, Yik Lam1, Josh Walker1, Devin McCann1, Matthew Westoby1, Edward K. Koepf2. (1) Process Biochemistry, Biogen, RTP, North Carolina, United States (2) Biogen, Holly Springs, North Carolina, United States

Transfer and scale-up of a biological therapeutic manufacturing process is often required when clinical development programs transition from early to late stage clinical trials. An initial assessment of a therapeutic protein manufacturing process as run at a CMO identified several limitations in processing capabilities if it were to be run in our 15K RTP manufacturing facility. The downstream facility fit challenges/limitations that were identified centered largely around the harvest, the capture column load UF/DF, and two downstream columns. The harvest was challenged with an extremely large depth filtration area, along with a high buffer volume for the capture column load UF/DF unit operation, while the capture column was development with a unique pH slope change for end of eluate collection, and the first polishing column was run using a linear gradient with the eluate collection stop calculated upon mass load. To enable running this process in our 15K facility, several processing options were explored with the requirement of maintaining product quality comparability. Process modifications were initially tested at bench scale to assess the impact to product quality and ease of processing, and then verified at pilot scale. Modifications made to the original CMO process included alteration of the harvest flocculation and clarification operations, replacement of the pH dependent eluate collection with a fixed volume collection, and implementation of an isocratic and fixed volume elution in place of a linear gradient with the calculated end of eluate collection criteria.

BIOT 509

Scale up design and optimization for an intensified downstream process utilizing multi-column operations

James M. Angelo1, [email protected], Srinivas Chollangi1, John Pagano2, Daniel Baur3, Kathleen Mihlbachler4, Thomas Mueller-Spaeth5, Xuankuo Xu1, Massimo Morbidelli3, Sanchayita Ghose1. (1) Biologics Process Development, Bristol-Myers Squibb, Devens, Massachusetts, United States (3) ETH Zurich, Zurich, Switzerland (4) LEWA Process Technologies, Inc., Devens, Massachusetts, United States (5) Chromacon AG, Zurich, Switzerland

Biologics manufacturing still remains dominated by batch operations. As upstream titers continue to increase due to developments in higher producing cell lines, the downstream purification process must adapt to meet the demands of productivity and processing cadence as to avoid becoming the bottleneck in manufacturing operations. In this work, several next generation manufacturing technologies were evaluated and implemented at both benchtop and pilot scale for unit operations used in monoclonal antibody purification. Multi-column continuous capture chromatography was utilized in order to increase both specific productivity and resin capacity utilization by up to 2-3 fold and 30-40%, respectively. A variety of capture column scenarios were investigated in a DoE format to determine the optimal performance range depending on several different factors, including column size, operating flowrates and resin type. Model assisted design of multi-column capture was performed in order to allow the prediction of performance attributes based on process parameters and was validated by experimentation. Utilizing optimized process parameters from this modeling exercise, a 2-column continuous capture operation was carried out at pilot scale and showed good agreement to scale down models, achieving a productivity of 25 grams processed per liter of resin per hour and a 93% resin capacity utilization. The interconnection of two separate polishing operations was also carried out at benchtop and pilot scale, utilizing UV based criteria to allow for automated valve switching, effectively removing the need for in-process holding vessels. The streamlining of the polishing operation showed no sacrifice in yield or purity while decreasing processing and switchover time by up to 2 fold. The multi-column capture and integrated polishing operations were linked with an automated viral inactivation vessel to allow for automated pH control of intermediate pools, removing the necessity of in-process pool handling from the harvest step to viral filtration or concentration steps. BIOT 510

Multi-column continuous chromatography as a strategy for reducing cost of drug development

Michelle Najera, [email protected]. CMC Biologics, Bothell, Washington, United States

Generation of a batch pre-clinical material can be a cost prohibitive step for drug innovators due to the high cost of goods associated with the manufacture of biologics. For monoclonal antibodies, the cost associated with chromatography resin procurement can make up a significant portion of the total manufacturing cost of the batch. Multi- column continuous chromatography offers a strategic solution to this challenge by dramatically reducing the required resin volume to perform a given chromatographic purification. However, since manufacture using continuous chromatography involves new challenges related to the added process complexity, few contract development and manufacturing organizations (CDMOs) offer access to this technology despite its cost saving potential. Our work is focused on simple and straightforward methods for CDMOs to incorporate continuous chromatography strategies during pre-clinical and clinical drug production. As part of this work, we have converted Protein A capture steps for 2 monoclonal antibodies from batch to multi-column continuous mode using the Cadence BioSMBTM, producing eluate streams with comparable product quality and impurity levels. The continuous process was then executed at the pilot scale as part of a project deliverable to produce high purity product to supply pre-clinical studies. This proof-of-concept study shows the important potential for continuous technology to drive down cost of goods, particularly during the early phases of a drug development program. This study also demonstrates the role that contract manufacturers can play by offering continuous manufacturing as an enabling technology for future cost savings and bringing new drugs to the market at reduced cost.

BIOT 511

Technical and practical solutions to the challenges of low pressures and process interruptions in optimal continuous virus filtration

Nigel Jackson2,1, Nigel_Jackson@europe.pall.com, Kyle Jones2,1. (1) PALL Europe Ltd., Portsmouth, United Kingdom (2) Biotechnology, Pall Corp, Portsmouth, Hampshire, United Kingdom

Continuous bioprocessing is rapidly gaining momentum, and will be implemented in more manufacturing processes in the next few years. The process intensification produced creates benefits of operational flexibility and efficiency, reduced facility footprint and capital expenditure, and significant cost of goods savings. A fully continuous bioprocess from end to end enables greater utilization of all unit operations, which share all these benefits. Breaking that chain during virus filtration to allow semi- batch operation risks losing some of the advantages of continuous operation throughout the process. This can also lead to a complicated filtration process with either an excessive number of small virus filters or large holding tanks surrounding the virus filter.

Continuous virus filtration also generates risks by driving operation to low flows and pressures and increasing the risk of process interruptions. We look into the detailed technical drivers and governing equations behind the risk of low pressure operation and process interruptions, looking at the balance between convective forces and hindered diffusion of viruses. The solution lies in the newest generation of virus filters which are ready for continuous processing due to unique pore structures and high permeabilities. We demonstrate the ability of PegasusTM Prime to deliver strong retention at operating pressures as low as 70 mbar (1 psi) for several days of operation.

We also present methods which resolve practical problems relating to process scale continuous virus filtration and the challenges of virus validation. The addition of practical solutions to the technical advances in virus filter design help to facilitate robust continuous virus filtration. This then contributes towards the future of continuous processing by maintaining the benefits provided by the enabling technologies that have advanced continuous bioprocessing from potential theory to economical practice.

BIOT 512

Investigation of biomimetic resin structure using CFD modeling of nano/meso fluid behavior for chromatographic gene therapy vector purification

Kevin Vehar, [email protected], Cameron Bardliving, Parviz Shamlou. Bioprocessing, Keck Graduate Institute, Fullerton, California, United States

Demand for improve the biomanufacturing process of gene therapy products has grown with the recent approval of several gene therapy based therapeutics. While the number of clinical trials investigating the therapeutic use of viral vectors continues to grow, the absence of an ideal scalable process limits both clinical advancement for many investigational products, as well as, successful commercialization of approved products. A major focus in development of gene therapy products is improvement of purification strategies for viral vectors. Currently, most viral vector downstream processes are based off methods developed for recombinant protein purification. While these methods are attractive for their scalability and routine usage in industry, challenges arise when applying these methods to much larger molecules, such as DNA and viruses. Column chromatography using porous particle-based media is one such protein purification technique, whose yield is low when adapted to gene therapy vector purification due to the heavy reliance on diffusive mass transport. This diffusive limitation, coupled with the fact that viral diffusion can be orders of magnitudes slower than proteins, prompts the need to investigate how porous particle morphology can be modified for improved viral absorption. Recent advances in computational modeling have allowed us to study and better understand the behavior of biological species in chromatographic separation all the way down to the bead/pore level, thereby empowering the simulation of different resin topology for improved viral absorption. This work examines the effect of chromatographic resin morphology on viral vector purification using computational fluid dynamics (CFD), scanning electron microscopy (SEM) and confocal laser microscopy. The structure characteristics for resins with different morphologies are determined using SEM. Resin morphology was varied though the use of chemically modified diatom silica frustules and compared with commercially available POROS 50 resin. In this presentation, CFD simulations will be provided for comparisons of flow patterns, peclet numbers, and tracer absorption for different resins. Comparison of experimental flow cell data for absorption kinetics provided a rational basis for CFD validation.

BIOT 513 Towards automated model-based process development

Tobias Hahn1, [email protected], Alexandros Papadopoulos1, Alexander Gutzler1, Thiemo Huuk1, Teresa C. Beck1, Juergen Hubbuch2. (1) GoSilico GmbH, Karlsruhe, XX, Germany (2) KIT, Karlsruhe, Germany

While high-resolution design space screening based on Design-of-Experiments (DoE) works for most of the established molecular scaffolds, it requires a high sample amount, elaborate offline analytics, and is restricted to the calibrated experimental range. Mechanistic modeling of biopharmaceutical processes has been in the focus of academic research for a long time. Nevertheless, industry failed to adapt these technologies, mainly due to the complexity of the underlying technology. At current time, mechanistic modeling building is mostly performed by highly skilled engineers with a scientific background in both biotechnology and mathematical modeling.

This contribution presents an industrial chromatography development challenge for which the traditional DoE-based process development carried out by an industrial partner failed to identify an economically feasible process. In order to prevent a delay of the biologically promising antibody, a mechanistic model was built for a cation-exchange step based on only three experimental chromatograms.

To quickly arrive at a calibrated model for in silico optimization, the process of model building was automated to a large extend. The log book of the lab scale chromatography system was parsed to extract all relevant events. Conductivity, pH, and temperature traces were used to deduce molar salt concentration from the given dominant salt species. Thereafter, all UV signals were treated with a Savitsky–Golay smoother to remove noise without losing information. Automatic peak detection and deconvolution was performed based on polynomial-exponential modified Gaussian peaks. Using well-established correlations between elution behavior and isotherm parameters, the search space for molecule-specific parameters could be reduced to a minimum. Effectively the process of model calibration succeeded with minimal interaction.

Afterwards, a much larger range of different process options compared to the initial DoE was evaluated by means of simulations. This in silico process development identified a feasible chromatographic process by varying the column aspect ratio and the load density in particular.

BIOT 514

Optimization of biologics process development enabled via data mining and quantitative structure analysis

Francis K. Insaidoo2, [email protected], Gianna Stella4, Hong Li3, Nihal Tugcu1, David J. Roush3. (1) K15-2-H206, Merck Co Inc, Kenilworth, New Jersey, United States (2) Process Development and Engineering, Merck & Co, Rahway, New Jersey, United States (3) Process Development and Engineering, Merck, Sharp and Dohme, Colts Neck, New Jersey, United States (4) Stevens Institute of Technology, Hoboken, New Jersey, United States

The prediction of robust process development (PD) modalities for biologics purification is of paramount significance in the drug development process. Now that advanced computational tools have been developed, computational modeling can inform and expedite the development of biologics. The selection of appropriate purification (capture and polishing) modalities is governed by the biophysical properties of the associated variants and impurities (e.g., aggregates, host cell protein etc) to be purified. In this study, in-silico biophysical properties of a diverse set of monoclonal antibodies (~400 unique molecules) were calculated from the secondary structure (determined either experimentally or via homology modelling). For protein biophysical properties that vary as a function of protein conformation, multiple structural conformations were generated to introduce heterogeneity in the calculated properties. This allowed for ensemble averaging of the calculated properties. Further, protein surface properties (hydrophobic, positive, and negative patches) were calculated for the CDR loops, Fab, and entire antibody. The calculated in-silico properties were compared to experimental data for each antibody molecule to determine correlated parameters. Using this relational database and correlation matrix, a Quantitative Structure Property Relation (QSPR) model was developed to derive a relationship between in-silico biophysical properties and PD outcome. A directed PD approach based on a more detailed biophysical characterization of the effect of ligands on biologics biophysical properties is required to ensure successful advancement of the currently developed model. This model and principles derived from this study can be deployed to identify the platform fit and expedite the discovery of lead candidates. Further, appropriate modalities for biologics purification can be selected or prioritized based on in-silico biophysical properties of lead candidates. Principles from this study are readily applicable to chromatographic, formulation, and developability assessment of lead candidates to speed the development of molecules from DNA to first-in-human (FIH). Hence, implementation of this computational modeling tool at the onset will significantly enhance the success of biologics development from discovery, through purification to formulation activities.

BIOT 515

Modeling light transmission and chemistry in production-scale ultraviolet thin- film reactors

Roger Hart, [email protected]. Amgen, Cambridge, Massachusetts, United States

Viral infection of cell-culture processes represents a significant risk in biopharmaceutical processing using mammalian cells. Treatment of media with ultraviolet light is an effective means for inactivation of a broad set of adventitious virus. Owing to the high optical absorbance of media, delivery of a uniform dose of light is a significant challenge with this treatment. Use of thin-film optical reactors is one approach for over-coming this challenge but the heterogeneity of the resulting dose can hamper process design, scale-up, technology transfer and process qualification.

Mechanistic modeling can be used to quantify the dose received by particles flowing through different regions of the photo-reactor and provide the means to link laboratory results utilizing virus with production-scale conditions. Further, use of photo-sensitive beads as photon detectors provides means for ensuring models are fit-for-purpose in process design and qualification. This paper describes the development and use of mechanistic stochastic models which enable quantification of the dose distribution received by particles transgressing thin-film photoreactors. Models were qualified using NIST-traceable standards and applied to different reactor types and media composition. Viral inactivation was assured using viral propagation and PCR methods.

BIOT 516

Dependence of elution peak shape on binding models and resin structure in ion- exchange chromatography of proteins

Vijesh Kumar3, [email protected], Karin Westerberg1, Christian Kunert1, Fabrice Schlegel1, Abraham M. Lenhoff2. (1) Amgen, Newbury Park, California, United States (2) Univ of Delaware, Newark, Delaware, United States (3) Chemical and Biomolecular engineering, University of Delaware, Newark, Delaware, United States

Correct prediction of protein elution behavior on an ion-exchange resin depends largely on an accurate description of the binding models, in terms of both the model form and the model parameters. At low coverages, several widely-used isotherms approach linear behavior, but at higher surface coverages, encountered during saturation loading, the deviation from linearity and approach to a plateau level deviate among different models. This behavior leads to different peak shapes for different models when saturation levels of loading are reached. In the current work, several models, including steric mass action (SMA), modified Langmuir and a colloidal model, were examined for prediction of elution peak shape from low loading to saturation of protein load onto the column. The elution peak shape is found to be dependent not only upon the protein size and its physiological properties but also on details of the resin morphology. This is demonstrated using lysozyme, cytochrome c and a monoclonal antibody on different - resins (SP Sepharose FF, Fractogel EMD SO3 , Capto S, Capto SP ImpRes, and Poros HS ). To address another important aspect of mechanistic modelling, namely accurate and easy estimation of isotherm parameters, the binding models have been also evaluated based on the capacity and ease of estimation of parameters by independent measurements. Among the models evaluated, the colloidal model parameters were easily obtained from pulse and breakthrough curves. The colloidal isotherm descriptions depend mainly on two independently-measured parameters: (a) the adsorption equilibrium, Keq, which captures protein–surface interactions and (b) the Yukawa parameter, Bpp, which captures lateral protein-protein interactions, particularly electrostatic ones, with the range determined by the Debye parameter κ. Keq is independent of Bpp and was easily obtained from retention factors for pulse injections at different ionic strengths. Bpp plays a more dominant role at higher surface coverage and was determined from breakthrough curves at different ionic strengths. Some of the systems such as lysozyme on Fractogel were not adequately described by the existing binding models. For such systems isotherms were approximated by frontal analysis using the characteristic point method and insights were examined for possible mechanisms.

BIOT 517

Effects of protein and ligand structure on protein-multimodal ligand interactions in antibody systems

Camille Bilodeau4, [email protected], Edmond Y. Lau5, David J. Roush1, Shekhar Garde2, Steven M. Cramer3. (1) Process Development and Engineering, Merck, Sharp and Dohme, Colts Neck, New Jersey, United States (2) Dept of Chemical and Biological Engr, Rensselaer Polytechnic Inst, Troy, New York, United States (3) Ricketts Bldg, Rensselaer Polytechnic Inst, Troy, New York, United States (4) Chemical Engineering, Rensselaer Polytechnic Institute, Troy, New York, United States (5) Physical Life Sciences, Lawrence Livermore National Laboratory, Livermore, California, United States

Multimodal chromatography uses multiple modes of interaction ranging from charge and aromatic to hydrophobic and hydrogen bonding to achieve challenging separations and unique selectivities. These modes of interactions affect each other in non-intuitive ways making overall chromatographic behavior difficult to understand and predict. The lack of fundamental understanding of these systems poses a significant barrier to multimodal process development and multimodal ligand design. In this work, we employ molecular modeling and simulations to obtain fundamental understanding of how multimodal ligands interact with proteins. Specifically, we examine interactions between a set of multimodal cation exchange ligands and a set of chemically diverse antibody fragments. First, we focused on how ligand chemistry and architecture affect hydration and dynamics using molecular dynamics simulations of ligands in aqueous solution. Second, we investigated ligand-protein binding phenomena using molecular simulations of proteins in aqueous solutions of free ligands. We also performed simulations of ligands immobilized on a SAM surface to study the effect of immobilization and ligand density on ligand solvation and conformation. Finally, specific faces of the protein surface were chosen based on the free ligand simulations results and umbrella sampling simulations were performed to obtain free energies of binding of these faces to multimodal ligands immobilized on a SAM surface. Collectively, our work provides a deeper understanding for how protein and multimodal ligand attributes result in different chromatographic selectivities. This will serve to guide design of new multimodal ligands as well as resin selection for multimodal polishing steps.

BIOT 518 Development of a mechanistic model for the peristaltic pump filling unit operation to enable drug product process design

Joseph Bernacki2, [email protected], Daniel Marsiglio2, Nitin Rathore1, Harrison Rose2. (1) MS 30W-3-A, Amgen Inc, Thousand Oaks, California, United States (2) Amgen, Inc., Thousand Oaks, California, United States

Peristaltic pump filling technology is becoming increasingly common for drug product fill- finish facilities due to its amenability to the single use system concept, which eliminates the need for cleaning and sterilizing stainless steel filling equipment and the associated infrastructure requirements. This feature of peristaltic pump filling also minimizes product contact surfaces, since the material being filled does not directly contact the filling equipment. A mechanistic model for peristaltic filling can expedite new product introductions, reducing the characterization activities needed to determine the proper tubings and process parameters for accurate dosing and a clean fill into the primary container. This presentation will describe such a model for simulating the pressure drop and flow dynamics during peristaltic pump filling. The model predicts the fluid flow during a peristaltic pump filling stroke based on the pump head geometry, tubing dimensions, fluid physical properties, and rotor rotation dynamics. The model enables in silico estimation of important attributes such as fill volume and filling time, providing operators a good starting point for confirmatory process development experiments. The model can also facilitate technology transfer across filling systems with different pump head geometries or parameter control logic, thereby minimizing experimental work needed to optimize process parameters.

BIOT 519

Online monitoring of quantity, purity and potency of an antibody capture process

Nicole Walch2, [email protected], Dominik Sauer2, Edit Felföldi2, Theresa Scharl- Hirsch2, Michael Melcher1, Friedrich Leisch1, Alois Jungbauer3, Astrid Dürauer3. (1) Institute of Applied Statistics and Computing, BOKU, Vienna, Austria (2) Austrian Centre of Industrial Biotechnology, Vienna, Austria (3) Institute of Biotechnology, BOKU, Vienna, Austria

Quantity, purity and potency are the key attributes that need to be observed during downstream processing of biopharmaceuticals. State of the art in protein chromatography is to use sensors like UV monitor or conductivity probe to control the process itself. But these sensors are usually not specific enough to accurately determine product concentration, purity and bioactivity. Therefore fractions have to be collected and analyzed in the laboratory which is both time and labor intensive. We have developed a chromatographic system with additional integrated online sensors to gain supplementary process information. We used a combination of signals obtained from an ATR-FTIR spectrometer, a static light scattering and a refractive index detector as well as a fluorescence spectrometer. The monitoring system was applied to an antibody capture process with focus on the elution step. A set of online data was collected as well as corresponding offline values, which were used to generate a statistical model. This model correlates the data sets and enables prediction of concentration, purity and potency of the target protein during elution. Coupled to a software framework, which provides sensor communication, data storage and treatment as well as the application of the model, this system makes additional process understanding available and enables peak cutting in real time. This knowledge also facilitates process control and thus enables constant product quality representing a promising future tool in biopharmaceutical manufacturing.

BIOT 520

Application of biophysical and modeling tools for formulation and process development of biologics

Valentyn Antochshuk, [email protected]. Sterile Formulations, Merck & Co, Kenilworth, New Jersey, United States

The advances in understanding of human disease bring diverse targets and biological molecules into development pipeline. Development of the novel biotherapeutic modalities is often complex, lengthy and expensive process. The robust screening, optimization and final candidate selection require disciplined assessment of biological, computational and experimental criteria against set of key metrics while balancing time and resources. Thorough candidate ranking is aiding with the liability identification and development process (e.g. protein engineering, ligand selection for purification, formulation selection and speed to clinic). Application of biophysical and modeling tools for molecular profiling, understanding liabilities and their impact, is discussed using practical examples of challenging candidate development.

BIOT 521

Influence of different freeze and thaw ramps on the phase behavior of lysozyme from chicken egg white

Anna K. Wöll, [email protected], Monika Desombre, Lena Enghauser, Juergen Hubbuch. IBLT Section IV: MAB, Karlsruhe Intitute of Technology, Karlsruhe, Germany

During the downstream process and formulation of proteins, certain lag and storage phases are unavoidable. Thus, proteins are frozen in order to achieve higher long term stability. The stress put on the protein during this process can result in decreased stability and a loss of conformation. In order to increase stability during freezing and thawing, understanding the mechanisms of destabilization during that process is of paramount importance. Well-known influencing parameters on stability are cryoprotectants, buffer components and pH value. Additionally, the process of freezing and thawing itself can be optimized by using ramps with different temperature change rates and lag periods. The stability of protein solutions can be assessed using phase diagrams. The final phase behavior and the morphology, observed after 40 days at 20 °C, indicate different protein-protein and/or protein-solvent interactions, which are influenced by the previously mentioned parameters. By varying these parameters in high-throughput screenings, conclusions about their effect on interactions and protein conformation can be drawn. This method is useful for the better understanding of freezing and thawing process and how to influence it. For buffer substances, the pH and additives like salt, the influences are well understood and to some extend predictable. The influence of different freezing and thawing ramps is not understood and has to be investigated. This project aimed at analyzing the effect of different freezing and thawing ramps on the phase behavior of lysozyme from chicken egg white. Phase diagrams were prepared in a small scale using either different freezing or thawing ramps in order to observe the effects of freezing and thawing separately. It was shown that freezing and thawing influences phase behavior, solubility and morphology of lysozyme. A number of one to three cycles were performed to simulate several freezing and thawing steps. Both tested parameters could be shown to have a significant impact on the solubility and morphology.

BIOT 522

Connecting protein conformation and dynamics with ligand-receptor binding using 3-color Förster resonance energy transfer tracking

David Marruecos3, [email protected], Mark Kastantin3, Navdeep Grover3, Sean Y. McLoughlin3, Daniel K. Schwartz1, Joel Kaar2. (1) Univ of Colorado, Boulder, Colorado, United States (2) Chemical Biological Eng, University of Colorado, Boulder, Colorado, United States (3) Chemical and Biological Engineering, University of Colorado Boulder, Boulder, Colorado, United States

Specific binding between biomolecules, i.e., molecular recognition, controls virtually all biological processes including the interactions between cells and biointerfaces, both natural and synthetic. Such binding often relies on the conformation of biomacromolecules, which can be highly heterogeneous and sensitive to environmental perturbations, and therefore difficult to characterize and control. An approach is demonstrated here that directly connects the binding kinetics and stability of the protein receptor integrin αvβ3 to the conformation of the ligand fibronectin (FN), which are believed to control cellular mechanosensing. Specifically, we investigated the influence of surface-adsorbed FN structure and dynamics on αvβ3 binding using high-throughput single-molecule three-color Förster resonance energy transfer (FRET) tracking methods. By controlling FN structure and dynamics through tuning surface chemistry, we found that as the conformational and translational dynamics of FN increased, the rate of binding, particularly to folded FN, and stability of the bound FN−αvβ3 complex decreased significantly. These findings highlight the importance of the conformational plasticity and accessibility of the arginine-glycine-aspartic acid (RGD) binding site in FN, which, in turn, mediate cell signaling in physiological and synthetic environments. BIOT 523

Effect of polysorbate 20 and 80 on higher-order structure of a monoclonal antibody and its Fab and Fc fragments probed using 2D-NMR

Surinder Singh1, Swati Bandi1, [email protected], David Jones2, Krishna Mallela1. (1) Department of Pharmaceutical Sciences, University of Colorado Anschutz Medical Campus, Aurora, Colorado, United States (2) Department of Pharmacology, University of Colorado Anschutz Medical Campus, Aurora, Colorado, United States

Polysorbates, such as polysorbate 20 (PS20; Tween 20) and polysorbate 80 (PS80; Tween 80), are commonly used in antibody therapeutic formulations to prevent agitation and interface-induced protein aggregation. However, the effects of polysorbates on the antibody structure and stability is less understood. We examined how PS20 and PS80 affect the higher-order structure of a monoclonal antibody (mAb) and its antigen-binding (Fab) and crystallizable (Fc) fragments, using near-UV circular dichroism and 2D nuclear magnetic resonance (NMR). Both polysorbates bind to the mAb with submillimolar affinity. Binding causes significant changes in the tertiary structure of mAb with no changes in its secondary structure. 2D 13C-1H methyl NMR indicates that with increasing concentration of polysorbates, the Fab region showed a decrease in crosspeak volumes. In addition to volume changes, PS20 caused significant changes in the chemical shifts compared to no changes in the case of PS80. No such changes in crosspeak volumes or chemical shifts were observed in the case of Fc region, indicating that polysorbates predominantly affect the Fab region compared to the Fc region. This differential effect of polysorbates on the Fab and Fc regions was because of the lesser thermodynamic stability of the Fab compared to the Fc. These results further indicate that PS80 is the preferred polysorbate for this mAb formulation, because it offers higher protection against aggregation, causes lesser structural perturbation, and has weaker binding affinity with fewer binding sites compared to PS20. Published accounts of the application of 2D NMR techniques to study antibodies are only two years old, and this study for the first time shows the power of simple 2D NMR techniques in probing how excipients in therapeutic formulations interact with mAbs.

BIOT 524

Establishing a pathway for robust implementation of higher order structure assessment of monoclonal antibodies therapeutics by 2D-NMR

Robert G. Brinson, [email protected], Luke W. Arbogast, Frank Delaglio, John P. Marino. Institute for Bioscience and Biotechnology Research, National Institute of Standards and Technology and the University of Maryland, Rockville, Maryland, United States

Techniques for the facile characterization of the critical quality attribute (CQA) of higher order structure (HOS) of biologic drugs are critical for establishing consistency in drug manufacturing, detecting process-related drug product variations and comparing a biosimilar to an innovator reference product. Recently, two-dimensional nuclear magnetic resonance (2D-NMR) methods have been introduced for HOS assessment of biopharmaceutical samples, including monoclonal antibody (mAb) drugs, at atomic resolution. Here, an inter-laboratory study was performed using samples of the Fab domain derived from the NISTmAb standard. A total of 451 spectral maps were acquired on thirty-nine different NMR spectrometers ranging from 500 MHz to 900 MHz in twenty-six different laboratories. These spectral maps from all instruments could be correlated directly to HOS through precision metrics of combined chemical shift deviation and similarity metrics through principal component analysis. Results from the study demonstrate the chemometric-structural relationships and provide a pathway for harmonized implementation of 2D-NMR methods within the biopharmaceutical laboratory for drug discovery, process development, and quality control.

BIOT 525

Preferential interaction of osmolytes with IgG1 monoclonal antibodies

Chaitanya Sudrik1, [email protected], Theresa Cloutier2, Hasige A. Sathish3, Bernhardt L. Trout4. (1) Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts, United States (3) Formulation and Drug Product Process Development, MedImmune, Gaithersburg, Maryland, United States

In recent years, there has been an increasing push towards the development of high- concentration mAb formulations for sub-cutaneous administration. With increasing concentrations, stability issues like aggregation and viscosity are aggravated. These issues can be overcome through formulation development and typically involve the use of empirical approaches to identify excipients that will mitigate these instabilities to an acceptable level, for a given antibody molecule. Such approaches are limited by the relatively high cost and time required for development and the inability to readily translate the findings to other novel therapeutic protein modalities. A better understanding of the underlying mechanisms of excipient-mediated antibody stabilization would enable a rational design of protein formulations.

Osmolytes are widely used in protein formulations to reduce the aggregation rates of antibody molecules in liquid solutions. To compare the stabilization afforded by excipients within this class, we have studied the preferential interactions of trehalose, sucrose, L-proline, mannitol, and sorbitol with protein surfaces. Specifically, we have characterized preferential interactions of these osmolytes with three therapeutically relevant IgG1 antibody molecules using the method of vapor pressure osmometry. The three IgG1 antibody molecules have similar size and shape but have dissimilar net charge and surface hydrophobicity. We find that for a given antibody molecule, osmolytes differ in their magnitude of interaction in the concentration range of 0-0.4 molar. We also find that the degree of exclusion for a given excipient varies with the antibody molecule and thus shows a dependence on surface features. These results provide greater insight into stabilization of antibody molecules with various osmolytes in solution. BIOT 526

Single particle virus isoelectric point determination with chemical force microscopy

Xue Mi, Caryn Heldt, [email protected]. Michigan Technological University, Houghton, Michigan, United States

Virus surface characteristics plays a key role in virus sorption processes, but there is very limited information in the literature on viral surface properties. One important surface characteristics is the isoelectric point (pI), which corresponds to the pH where the net charge on the virus particle is zero. The surface charge of virus particles is pH dependent, which dictates the mobility and controls the colloidal behavior in virus sorption processes. While traditional characterization methods, including zeta potential and viral adsorption to a charged surface chemistry, have been used to study viral surface charges and determine the pI of virus, they are bulk measurements that are limited by virus purification methods. The single particle method of chemical force microscopy (CFM) allows not only the quantitation of the virus surface charges, but can probe the effect of purification on the charge. CFM can measure the isoelectric point of complicated virus particles by using an AFM probe terminated with ionizable groups. The model non-enveloped porcine parvovirus (PPV) and enveloped bovine viral diarrhea virus (BVDV) were used to demonstrate the use of CFM for viral particles. The change of adhesion forces measured as a function of pH is due to a change of virus surface electrostatic properties. The virus immobilization method used helps to maintain a more natural form of the virus without deformation, disassembly, or dehydration. With a thorough understanding of virus surface characteristics, virus purification and future virus removal process could be significantly improved to specifically targeting viral particles. This will lead to improvements in virus removal, purification and detection.

BIOT 527

Bioengineered hydrogel models of Alzheimer's disease: Studying β-amyloid protein aggregation within 3D scaffolds

Laura W. Simpson, [email protected], Jennie B. Leach. Chemical, Biochemical and Environmental Engineering, University of Maryland Baltimore County, Columbia, Maryland, United States

Background: Amyloids are native proteins that are misfolded such that aggregation is initiated, yielding insoluble fibrous deposits found in various diseases, including Alzheimer’s (AD), Parkinson’s, and Huntington’s. In AD, unstable β amyloid (Aβ) oligomers have a higher diffusibility, hydrophobicity and ability to interact with lipid membranes, as compared to Aβ fibrils. When confined in a 3D environment such as tissue or a tissue-like hydrogel, Aβ transport is altered vs in solution due to covalent and noncovalent binding and hindered diffusion. We hypothesize that 3D environment factors may stabilize unfolded oligomers and alter aggregation kinetics resulting in fewer toxic Aβ species formed. Methods: Pretreated Aβ42 films were dissolved in NaOH then diluted in PBS before being incorporated in gels including collagen I, SeaPlaque agarose, 4-arm PEG maleimide crosslinked with PEG dithiol and maleimide modified hyaluronic acid crosslinked with PEG dithiol. Neuroblastoma derived SY5Y cells were cultured in phenol red free DMEM/F12 w/ 1% B27. Multiple protein detection methods were utilized to classify the aggregate species over time including: Thioflavin T (ThT), Fluorescence Correlation Spectroscopy (FCS), TEM, and Dot blot. Cell viability assays and calcium imaging were used to determine the toxicity of the aggregate species on SY5Y cells. Results: ThT assays using 20µM Aβ showed a lag phase (~9 hr) in medium before β- sheet structures initiated the rapid increase in fluorescence vs minimal lag phase seen in the 3D gels representing rapid fibrilization. FCS found Aβ species in medium remained as small aggregates with only 5x slower diffusivities than monomer Aβ during the lag phase vs immediate larger aggregate species in all gels. SY5Y cells cultured on collagen substrates consistently have a cell viability >90% but in the presence of 20µM Aβ, the cell viability decreases <60% after 3 days compared to in 3D gels, the cell viability is no different with or without Aβ. Discussion: Results identify a lack of a lag phase during aggregation in 3D gels suggesting stabilization of β-sheet structures vs Aβ aggregation kinetics in medium. FCS confirms low molecular weight Aβ during the lag phase in medium and high molecular weight Aβ in gels. SY5Y cells cultured in 2D are susceptible to Aβ toxicity over 3 days while in 3D gels there is minimal difference in cell viability with or without Aβ, suggesting the 3D environment promotes fewer toxic Aβ species.

BIOT 528

Integrating experimental and computational approaches to elucidate mechanisms of binding in multivalent proteins

Wesley J. Errington1, Bence Bruncsics2, Casim A. Sarkar1, [email protected]. (1) Department of Biomedical Engineering, University of Minnesota, Minneapolis, Minnesota, United States (2) Budapest University of Technology and Economics, Budapest, Hungary

Multivalent proteins are ubiquitous in nature and can provide unique, exploitable properties in therapeutic applications such as increased affinity or multi-target specificity. Despite the importance of these proteins in fundamental and applied biomedical research, mechanistic and quantitative descriptions of their binding kinetics are limited. We have considered such multivalent protein-protein interactions to be driven by three key variables: the binding affinity of individual monomer units, the linker length/structure between the monomers, and the overall valency of each multivalent protein. Using model synthetic proteins in which all three of these variables can be independently tuned, we performed surface plasmon resonance experiments to quantify the kinetics of association and dissociation as a function of affinity, linker, and valency. In parallel, we developed a mechanistic model based on mass-action kinetics that explictly enumerates all possible microstates that participate in the binding reaction. Integration of these quantitative experimental and computational approaches has elucidated a number of interesting findings, including the role of valency in generating non-canonical dissociation kinetics, that will be discussed. Our approach should enable better understanding of dynamic behaviors in natural multivalent proteins and lead to more rational optimization of multivalent therapeutics.

BIOT 529

Expression of snake-antivenom peptide in Pichia pastoris

Israel Juarez Contreras, Sabrina Tang, Corey Laperyi, Khang Le, Claire F. Komives, [email protected]. Dept of Chemical Engineering, San Jose State University, San Jose, California, United States

Snakebite claims the lives of over 100,000 people in India every year in spite of the availability of horse serum-based antivenoms. A short peptide from the N-terminus of a serum protein from the North American Opossum has been shown to completely neutralize the venom of several snakes from the viper family, including the Indian Saw Scaled Viper. We have expressed this peptide as a concatenated chain of peptides and expressed it in both E. coli and P. pastoris. P. pastoris strains GS115 and a protease knockout strain SMD1168 were both tested and found to express high levels of the product and the 6XHis tag remained intact on the secreted peptide chain. We have tested the peptide chain and demonstrated activity against the Saw Scaled Viper. The presentation will discuss the transformation strategy and the assessment of the activity of the produced peptide.

BIOT 530

Expression and in vitro reconstitution of novel lasso peptide gene clusters

Joseph D. Koos2, [email protected], A James Link1. (1) Princeton University, Princeton, New Jersey, United States (2) Molecular Biology, Princeton University, Princeton, New Jersey, United States

Lasso peptides are members of a class of small molecules called ribosomally synthesized post-translationally modified peptides (or RiPPs) characterized by a unique three-dimensional structure. Lasso peptides are defined by a single isopeptide bond formed between the N-terminus and an acidic side chain, forming a ring. The remaining C-terminus of the peptide is threaded through this ring, forming the characteristic lariat shape. Recent research has shown that potential lasso peptide clusters are widespread in a range of genomes, but the set of isolated lasso peptides is more limited. A set of gene clusters were found through genome mining with peptides predicted as more hydrophobic than known examples and with previously unobserved N-terminal residues. These clusters were heterologously expressed and the properties of the products were investigated. In addition, the associated maturation proteins for each cluster were expressed and purified. Using purified components, in vitro experiments were performed and mature peptides were identified by LC-MS, confirming the activity of the proteins. The in vitro reconstituted system also allowed quantification of the activity of the enzymes in a manner that was not possible through heterologous expression of clusters and enables future studies to further understand the details of lasso peptide production.

BIOT 531

New fungal RiPP family changes the rules for modified peptide biosynthesis

Michael F. Freeman, [email protected]. Biochemistry, Molecular Biology and Biophysics, University of Minnesota, Minneapolis, Minnesota, United States

Microorganisms communicate using the molecular language of natural products. Amazingly, the vast majority of microorganisms on Earth are as-of-yet uncultivated or too difficult to genetically manipulate and directly study. Thus, the discovery of new natural products or description of new biocatalytic transformations occurring within NP pathways from these niche microbial sources often require heterologous expression in tractable microbes. This work describes the unexpected biosynthetic origins of a ribosomally synthesized and post-translationally modified peptide (RiPP) produced by a basidiomycete fungus. We provide evidence, through heterologous expression, for iteratively incorporated alpha-N-methylations into a ribosomally encoded precursor protein. This never-before-seen post-translational modification was thought for over forty years to be only accessible through nonribosomal peptide synthesis. Along with a very unusual leader peptide composition, this pathway hallmarks the designation of a new RiPP natural product family we termed borosins. We go on to show this borosin RiPP family can be easily engineered to create new alpha-N-methylated peptides in vivo through genetic recoding, a first for this type of tailoring reaction. This research adds to the handful of RiPPs so far discovered or described in fungi and highlights the vast metabolic potential of uncharacterized microorganisms.

BIOT 532

Enabling natural products discovery from anaerobic fungi via integrated ‘omics’ approaches

Candice Swift1, [email protected], Katherine Louie2, Benjamin Bowen2, Kerem Bingol3, Heino Heyman3, Trent Northen2,4, Michelle A. O'Malley1. (1) Chemical engineering, UC Santa Barbara, Santa Barbara, California, United States (2) Department of Energy, Joint Genome Institute, Walnut Creek, California, United States (3) Department of Energy, Environmental Molecular Sciences Laboratory, Richland, Washington, United States (4) Department of Energy, Joint BioEnergy Institute, Emeryville, California, United States

Fungal secondary metabolites have provided humans with abundant natural products with applications in medicine, agriculture, and energy. Despite this successful history, significant challenges exist in the field of natural products discovery. These include avoiding rediscovery, awakening silent biosynthetic gene clusters (BGCs), linking orphan BGCs to their products, and prioritizing BGCs for pathway characterization. Here, we have combined complementary approaches in genomics, transcriptomics, mass spectrometry, and NMR to characterize the secondary metabolites of anaerobic fungi. Our integrated approach allows us to study secondary metabolism at all levels, from DNA to RNA to metabolites, thus maximizing discovery of novel metabolites and unmasking their native functions. Anaerobic fungi thrive in competitive microbial environments such as the digestive tracts of many large herbivores despite being vastly outnumbered by other microorganisms. We hypothesize that these fungal secondary metabolites may thus function as defense compounds, and could be harnessed to develop valuable antimicrobials. First, we uncovered a plethora of gene clusters encoding BGCs from diverse chemical classes by mining the genomes and transcriptomes of anaerobic fungi. Key secondary metabolite clusters include canonical polyketide synthases and nonribosomal peptide synthetases. Several of the clusters are expressed under standard laboratory growth conditions, whereas others are silent. Via RNA-Seq, we found that growth in minimal media enhances transcription of some BGCs. Other environmental factors under investigation include oxidative stress and co-cultivation with anaerobic bacteria. We detected by LC-MS/MS ~100 likely secondary metabolites and putatively identified two known polyketides: the antioxidant baumin, and 3-methylorcinaldehyde, which both have pathways that are sequence similar to BGCs in anaerobic fungi. We are performing targeted fractionation to isolate unknown secondary metabolites and structurally characterize them using NMR. Finally, we are striving to heterologously express selected BGCs in Saccharomyces cerevisiae in order to definitively link BGCs to products.

BIOT 533

Redefining polyketide synthase modules

Adrian Keatinge-Clay, [email protected], Drew Vander Wood. UT Austin, Austin, Texas, United States

The biosynthetic community is working to decipher the logic of two large classes of enzymatic assembly lines that contain polyketide synthase (PKS) machinery - cis- acyltransferase (cis-AT) assembly lines that primarily rely on embedded ATs and trans- acyltransferase (trans-AT) assembly lines that primarily rely on separately-encoded ATs. From the sequencing of the erythromycin synthase in 1990 until 2017, the boundaries of the modules of cis-AT assembly lines were incorrectly defined - comparisons with the domain organization of the mammalian FAS had led to the module being defined with KS at its upstream boundary and ACP at its downstream boundary. The ACP domains of several cis-AT assembly lines are now known evolutionarily co-migrate with the KS downstream of them, leading to the redefinition of cis-AT modules as possessing a downstream KS. When trans-AT assembly lines started to be discovered fifteen years ago, often from difficult-to-culture, symbiotic bacteria, the use of cis-AT nomenclature resulted in their modules being incorrectly defined as well. Ironically, only a few years later KSs that accept similar substrates in trans-AT assembly lines were oberved to clade together, suggesting that they collaborate most closely with the domains preceding them, not following them.

In this seminar, bioinformatic analysis of 526 ACPs from 33 well-characterized trans-AT assembly lines will be presented. ACPs from the same module type are related to one another, reflective of their interactions with cognate processing enzymes that have been relatively conserved through evolution. KSs downstream of ACPs from the same module type also clade together, whereas KSs upstream do not, contrary to the traditional definition of a module. Beyond nomenclature, how the updated definition of a module impacts our understanding of modules, the evolution of assembly lines, pathway prediction, and assembly line engineering will be discussed.

BIOT 534

Nature as the ultimate combinatorial biosynthetic chemist: Inspiration for engineering natural products biosynthesis

Ben Shen, [email protected]. Departments of Chemistry and Molecular Medicine, The Scripps Research Institute, Jupiter, Florida, United States

Natural products are among the best sources of drugs and drug leads and serve as outstanding small molecule probes for dissecting fundamental biological processes. Natural product biosynthesis continues to push the frontier of modern chemistry, biochemistry, and molecular biology by revealing novel chemical reactions, complex enzyme systems, and intricate regulatory mechanisms. The progress made in the last two decades in connecting natural products to the genes that encode their biosynthesis has fundamentally changed the landscape of natural products research and sparked the emergence of a suite of contemporary approaches to natural products discovery. While Nature’s ability to generate diverse natural products from simple building blocks has inspired combinatorial biosynthesis and made it possible to produce designer natural products and their analogues by rational metabolic pathway engineering, technological advances in DNA sequencing, genome mining, and bioinformatics has also allowed targeted discovery natural product scaffolds by exploring the vast combinatorial biosynthesis repertoire found in Nature. Selected examples from our current studies will be presented to highlight Nature as the ultimate combinatorial biosynthetic chemist, inspiring natural product biosynthesis and engineering.

BIOT 535

A synthetic biology approach to precursor-directed biosynthesis of complex polyketides and terpenes

Gavin J. Williams, [email protected]. Chemistry Dept Dabney Hall, North Carolina State University, Raleigh, North Carolina, United States

Many natural products are biosynthesized in a modular fashion by the selection and condensation of small molecule building blocks. Chimeric biosynthetic pathways can be constructed in an attempt to produce analogues for drug discovery. Yet, the scope and utility of this combinatorial approach is limited by the inherent substrate specificity and poor functional modularity of most biosynthetic machinery. Here, we show that biosynthetic pathways are tolerant towards non-natural building blocks and that this promiscuity forms a platform for accessing new natural products modified via precursor- directed biosynthesis. Accordingly, we describe a comprehensive program of enzyme engineering, directed evolution, and synthetic biology aimed at constructing artificial bacterial strains capable of producing complex natural products that are modified with non-natural chemical functionality. Key to our synthetic biology approach is the development of genetically encoded biosensors for non-natural small molecules that enable ultra-high-throughput methods to engineering biosynthetic pathways. Our approach expands the synthetic capabilities of natural product diversification strategies, and provides an improved understanding of the molecular basis for specificity in complex molecular assemblies.

BIOT 536

Engineering biosynthesis of the anticancer alkaloid noscapine in yeast

Yanran Li, [email protected]. Department of Chemical and Biomolecular Engineering, University of California Los Angeles, Los Angeles, California, United States Noscapine is a potential anticancer drug isolated from the opium poppy Papaver somniferum, and genes encoding enzymes responsible for the synthesis of noscapine have been recently discovered to be clustered on the genome of P. somniferum. Here, we reconstitute the noscapine gene cluster in Saccharomyces cerevisiae to achieve the microbial production of noscapine and related pathway intermediates, complementing and extending previous in planta and in vitro investigations. Our work provides structural validation of the secoberberine intermediates and the description of the narcotoline-4′- O-methyltransferase, suggesting this activity is catalysed by a unique heterodimer. We also reconstitute a 14-step biosynthetic pathway of noscapine from the simple alkaloid norlaudanosoline by engineering a yeast strain expressing 16 heterologous plant enzymes, achieving reconstitution of a complex plant pathway in a microbial host. Other engineered yeasts produce previously inaccessible pathway intermediates and a novel derivative, thereby advancing protoberberine and noscapine related drug discovery.

BIOT 537

Assembly and purification of a novel format Fab’2 bispecific

Glen Giese, [email protected], Ambrose Williams, James Dvornicky, Ben Tran, Mark Fedesco. Purification Development, Genentech, Inc., South San Francisco, California, United States

The formation of a Fab’2 bispecific by the assembly of two distinct Fab’s in order to form a Fab-Fab heterodimer poses several unique challenges. The assembly mixture from combining the two Fab’s must be purified in order to remove unique assembly byproducts such as homodimers, free light chain or heavy chain, adducts, unreacted Fab’ and others, in addition to typical process related impurities including host cell protein, DNA and high molecular weight species. Stability of the process intermediates must also be investigated. Multiple novel unit operations have been developed in order to achieve effective assembly and separation. High throughput screening was leveraged to rapidly develop a significantly off platform purification process. The manufacturability of this new molecule format and facility fit were assessed. Non-standard analytical methods were also implemented in order to support the purification process development.

BIOT 538

Affinity capture options for monovalent bispecific antibodies

Matthew T. Aspelund, [email protected], Dhanesh Gadre, Timothy Pabst, Alan K. Hunter. Purification Process Sciences, MedImmune, Gaithersburg, Maryland, United States

Monovalent bispecific antibodies are finding increased use in a broad range of clinical applications. Production of this types of antibodies require co-expression of multiple heavy and light chains leading to the formation of product related impurities and light chain mis-pairs that pose significant purification challenges. Capturing using a Protein A resin results in the co-purification of mis-pairs and antibody fragments. This study provides a detailed examination of the affinity capture options that can be used to purify monovalent bispecific antibodies. The effectiveness of the affinity capture resin is strongly dependent on the construction of the monovalent bispecific target molecule; however, certain capture options provide advantages in elution conditions and resin lifetime. The use of novel light chain affinity resins for capture provides significant reduction of the levels of product related impurities and the potential to utilize simplified polishing options in the purification process.

BIOT 539

Integrated purification process development for the single-domain variable fragment of camelid antibodies

Chaz Goodwine1, [email protected], Nicholas Vecchiarello1, Kerry Love2, John C. Love2, Steven M. Cramer1. (1) Rensselaer Polytechnic Institute, Troy, New York, United States (2) Massachusetts Institute of Technology, Cambridge, Massachusetts, United States

Single-domain variable fragments of camelid antibodies, also known as VHHs or nanobodies, offer several beneficial features that make them desirable as a new platform modality for biotherapeutics, such as high solubility, proteolytic resistance, and chemical and thermal stability. A standard affinity capture process, such as Protein A for mAbs containing an Fc region, has not yet been thoroughly established for the purification of these molecules, making downstream processing of a non-tagged VHH a challenge to their broad utilization. Our lab has recently developed an approach for generating integrated purification trains for three non-mAb biologics produced recombinantly in the yeast Pichia pastoris. Here we report application of this in silico approach to the development of a purification process for a Pichia-expressed nanobody. Our method relies upon the generation of both product-related and process-related impurity databases containing retention data acquired from linear gradient screens on a variety of chromatographic resins at different salt and pH conditions. Given the consistency routinely observed in the process-related impurity profile when expressing various proteins in P. pastoris, we combined our previously reported process-related database with a condensed set of product-related chromatographic screens to further streamline assembly of the complete database required for in silico process identification. Our in silico tool was able to efficiently search and rank potential integrated purification processes based on end product quality goals. Top candidates were then assessed and refined for a finalized downstream purification process. The demonstration of this rapid approach to the downstream process development for a nanobody produced in P. pastoris highlights the utility of this method and its capability for identifying purification processes for a range of biologic molecules produced in a well-characterized host system.

BIOT 540 Resolution and enrichment of antibody fragments through pH modulation on an affinity chromatography step

James Woo, [email protected], Nooshafarin Sanaie, Rajesh Krishnan. Biological Process Development, Gilead Sciences, Oceanside, California, United States

During the production and harvest of a recombinant antibody, antibody fragments may be produced which can be difficult to resolve from the intact antibody using conventional IEX and HIC chromatography steps. This case study demonstrates that antibody fragments can be resolved from intact antibody using affinity chromatography by exploiting differences in affinity and avidity between the affinity ligand and its binding site in different pH conditions. Selection of an optimal bead size and ligand density for the affinity resin further contributes to a 4-fold improvement in the product throughput of this step while maintaining the resolution of antibody fragments. This mode of antibody fragment resolution was found to be effective on four commercially-available resins, some with different binding sites on the target antibody. Using multiple rounds of processing, this optimized chromatography step was able to generate enriched material with > 50% antibody fragment species for future characterization studies.

BIOT 541

Strategies for achieving clearance of challenging host cell impurities in development of purification processes for biotherapeutics

Joshua H. Lee, [email protected], Asif Ladiwala, William O'Dwyer, Kevin Shomglin, Michelle Butler. Purification Development, Genentech, Inc., South San Francisco, California, United States

One of the key objectives during development of downstream purification processes for biotherapeutic molecules is achieving sufficient and robust removal of host cell impurities, which can have implications for patient safety and may confound clinical trial results. Occasionally, specific host cell proteins can be challenging to remove, often times requiring fine-tuning of processing conditions or even dedicated unit operations to attain adequate clearance. One familiar example is the CHO host cell protein phospholipase B-like 2 (PLBL2), which has presented a purification challenge for several monoclonal antibody therapeutics and led to the identification of several process options for its successful removal. Besides PLBL2, other instances of challenging host cell impurities have recently been encountered during downstream process development at Genentech, encompassing different therapeutic molecule formats and cell culture processes.

In the first case study of an E. coli-derived bispecific antibody, overexpression of the molecular chaperone, FkpA, and oxidoreductases, DsbA and DsbC, was leveraged to boost cell culture titers. However, the standard downstream purification process was unable to clear these impurities to acceptable levels. Extensive high-throughput screening was performed exploring flocculation conditions, Protein A washes, and various modes of chromatography, resulting in the identification of various processing alternatives for achieving acceptable removal of these impurities. In a second case study related to a CHO-derived molecule, the late discovery of the host cell protein glutathione synthetase (GSS) spurred additional process development, leading to the implementation of a modified wash on the capture column step as well as a dedicated unit operation utilizing selective precipitation of GSS.

This paper will review the strategies and tools employed for identifying and optimizing process options to address the impurity clearance challenges faced during process development for these molecules. In addition, head-to-head comparisons of overall process performance will be presented for the various process options identified through this work.

BIOT 542

Downstream process optimization to prevent fragmentation caused by cathepsin L

Nicholas E. Levy, [email protected], Lakshmi Madhavan, Jessica R. Molek, Kent E. Goklen. Downstream Process Development, GlaxoSmithKline, Philadelphia, Pennsylvania, United States

High levels of fragment were observed during early phase process development and clinical manufacturing of an IgG1 monoclonal antibody (mAb). Fragment concentration increased across the downstream process, and stability testing demonstrated rapid fragmentation of the mAb in bulk drug substance at both 5 and 25 degrees Celius.

Mass spectrometry and protease inhibitor screening confirmed that the fragmentation was caused by a protease, cathepsin L. While fragment concentration increased across the downstream process, an elevated fragmentation rate occurred following the bind- and-elute cation exchange chromatography step. Investigations confirmed that cathepsin L – expressed by CHO cells in the inactive pro-form – becomes activated when adsorbed on negatively charged surfaces, such as the cation exchange chromatographic resin.

Late phase process development focused on improving cathepsin L clearance throughout the downstream process and avoiding protease activation. Several process changes were explored to improve protease clearance. Optimizing the Protein A wash buffer for cathepsin L clearance was the most successful mitigation strategy. Several different Protein A wash buffers provided sufficient cathepsin L removal. The improved Protein A process also greatly increased host cell protein impurity clearance, enabling the removal of the final chromatographic polishing step. Bulk drug substance stability testing confirmed that protease-related fragmentation was eliminated.

BIOT 543 Engineering TNB (5-Thio-2-NitroBenzoic acid) on antibody to enable selective reduction

Amarnauth Prashad, [email protected]. Bioprocess R&D, Pfizer, Inc, New City, New York, United States

The cysteine mutant technology is a novel site specific antibody drug conjugate (ADC) technology. In this technology the mutant cysteine residues of the antibody are typically protected with cysteine or glutathione protecting groups. The conjugation protocol (total reduction & re-oxidation) for these constructs may impact the quality attributes of the resulting ADCs and in addition, involves multiple steps. A labile protecting group, 5-thio-2-nitrobenzoic acid (TNB), that can be selectively removed using tris(3-sulfophenyl) phosphine (TSPP) was discovered at the Bioprocess R&D group within Pfizer. Proof of concept using partially uncapped cysteine mutants was done to demonstrate the concept is viable. This labile TNB capping group has enabled the selective reduction for direct conjugation and therefore eliminates the harsh conditions of conventional reduction-reoxidation and buffer exchange of totally reduced antibody. As a result, the three dimensional structure of the original antibody is preserved and in addition, this protocol only involves 2 steps – selective reduction & conjugation. The BioMedicines Design department within Pfizer discovered that the Cys-capping modifications likely take place outside mammalian cells and the capping status of unpaired Cys residues can be manipulated through medium optimization. They have demonstrated that novel TNB-capping can be engineered by adding Ellman’s reagent [DTNB – 5,5’-Dithiobis-(2-nitrobenzoic acid)] into cysteine/cystine-depleted medium. Currently the Bioprocess R&D group is developing & optimizing culture conditions to maximize efficiency & homogeneity of TNB-capping and yield of protein. The progress and advantages of this alternative TNB technology for cysteine mutants in the preparation of ADCs will be presented.

BIOT 544

Effective removal of Bacillus cereus exotoxin by a two-step mAb chromatography process

Anna Graanberg1, [email protected], Sara Grönlund1, Tomas Bjorkman1, Magnus Wetterhall1, Sravani Musunuri1, Katherine Chaloupka2, Patrick Gammell2. (1) R&D, GE Healthcare, Uppsala, Sweden (2) Process Development, Amgen Inc., Cambridge, Massachusetts, United States

Bioburden may occur both in the upstream and the downstream process and can have a detrimental effect on biopharmaceutical products by affecting the safety and efficacy of the final drug. Even if bacteria are easily removed by filtration steps during the purification process, residual microbial byproducts can remain. Bacillus cereus (B. cereus) is a common bacterial contaminant which produces both protein and peptide based exotoxins. The lack of available GMP assays for exotoxins makes it difficult to monitor and prove clearance of these byproducts in bioprocesses. Typically, quality impact assessments are performed based on worst case calculations.

Here, we have developed an in-house liquid chromatography-mass spectrometry based analytical assay with high resolution and sensitivity for quantification of the B. cereus peptide exotoxin cereulide. The highly toxic cereulide is challenging to analyze due to the hydrophobic nature of the cyclic peptide structure. The clearance of cereulide was evaluated over a mAb capture step using protein A. The major part of the cereulide was detected in the protein A flow through fraction, whereas only residual amounts were found in wash, acidic strip, elution pool and extracted from the protein A resin. Carryover between cycles was assessed by analysis of mock eluates from blank cycles and cleaning optimization carried out by investigation of NaOH concentrations and contact times during cleaning-in-place (CIP). The generality of cereulide clearance over the protein A step was tested using three different mAbs. Finally, polishing with a multi- modal anion exchange resin in flow-through mode was evaluated for clearance of the residual amount of cereulide from the protein A product pool. A Log purification factor of >4 for the cereulide was obtained over the two-step chromatography process.

BIOT 545

Microdroplet-assisted evolution of Yarrowia lipolytica for the production of secreted riboflavin (vitamin B2)

James M. Wagner1, [email protected], Leqian Liu2, Shuo-Fu Yuan3, Eden Williams1, Maya Venkataraman1, Adam R. Abate2, Hal S. Alper1,3. (1) Chemical Engineering, University of Texas at Austin, Austin, Texas, United States (2) University of California San Francisco, San Francisco, California, United States (3) Institute for Cellular and Molecular Biology, The University of Texas at Austin, Austin, Texas, United States

Microdroplet and microfluidic technologies are gaining traction as a means to accelerate the design-build-test cycle for strain engineering. One of the advantages of microdroplets is the capacity to physically link genotype and extracellular phenotype, which is especially critical for secreted bioproducts. In this talk, we present recent work on creating a microbial cell factory for secreted riboflavin production using an approach that combines rational metabolic engineering and droplet-sorting based adaptive evolution. Specifically, we chose the non-conventional yeast Yarrowia lipolytica as the host for riboflavin production due to its ‘Generally Regarded as Safe’ (GRAS) status, strong pentose phosphate pathway, high tolerance to inhibitors, fast growth on a wide variety of low-value carbon sources, and proven amenability to adaptive laboratory evolution (ALE) schemes. The strong, innate yellow color and fluorescence of riboflavin makes screening feasible using a variety of approaches, including fluorescence activated cell sorting (FACS) and microfluidic droplet sorting. Initially we demonstrate that a rational metabolic engineering approach can improve initial riboflavin titers by 12.6-fold. Next, we use a variety of screening approaches coupled with adaptive evolution to further increase titer. Specifically, we show quantitative evidence that the choice of screening method can greatly impact the resulting production capacity and balance between intracellular and extracellular titers. Droplet sorting yielded an additional 3.6-fold improvement over the engineered strain, and a final shift from 39% to 90% secretion of the product. In contrast, matched throughput screens that relied upon only intracellular content as the surrogate for high production (i.e. FACS) failed to yield strains with similar levels of improvement in titer or secretion rate. Overall, engineering and evolution of Yarrowia lipolytica coupled to high throughput droplet screening resulted in a collection of strains with a nearly 41-fold mean increase over basal vitamin production, opening up a new possibility for the biorenewable production of this important food additive. We conclude this talk by discussing important implications for the screening and selection of large strain libraries for potentially secreted products.

BIOT 546

Discovery of metabolic pathways for lipid-production from lignin-derived phenolics in a non-model oleaginous yeast

Allison Yaguchi2, [email protected], Michael Spagnuolo2, Alana Robinson1, Erin Mihealsick1, Mark A. Blenner2. (1) Clemson University, Clemson, South Carolina, United States (2) Chemical and Biomolecular Engineering, Clemson University, Clemson, South Carolina, United States

Oleaginous yeasts have long been a target for developing industrial-scale biochemical processes due to their ability to accumulate high amounts of lipids, synthesize complex chemicals and proteins, and robustly metabolize diverse feedstocks. In parallel, interest in lignocellulosic biomass as a feedstock has grown. Cellulosic biomass is rich in C6 and C5 sugars, such as glucose and xylose, respectively, while the lignin is rich in aromatics. Cutaneotrichosporon oleaginosus, previously known as Cryptococcus curvatus, is a non-model oleaginous yeast that is known for its ability to metabolize many alternative sugars, including xylose, and tolerate toxic lignocellulosic hydrolysate inhibitors such as 5-hydroxymethylfurfural (5-HMF) and furfural. We discovered C. oleaginosus also tolerates and metabolizes lignin-derived phenolics, highlighting this organisms’ potential to utilize all components of lignocellulosic biomass. C. oleaginosus is able to fully metabolize phenol, 4-hydroxybenzoic acid (pHBA), and resorcinol as sole carbon sources, as well as in co-utilization with glucose and xylose. At low concentrations, cells were able to grow as robustly in pHBA and resorcinol as in glucose. Phenol caused an extended lag phase due to toxicity, but cells were able to achieve comparable biomass as cells in other carbon sources. We explored different feeding strategies to overcome aromatic toxicity and increased lipid accumulation to over 69% of biomass by weight. Aromatic metabolism is well-characterized across many organisms, however, there are many different mechanisms utilized between species. An initial BLAST analysis revealed ortho-cleavage may be the mechanism this yeast uses, but qPCR data was inconclusive. RNAseq data enabled analysis of all putative aromatic degradation genes, and revealed other participating genes that were missed by BLAST. Using transcriptomic analysis, we have elucidated pathways for aromatic metabolism in C. oleaginosus, and are in the process of confirming it with metabolomics data. Our work demonstrates the potential for this yeast to convert all components of lignocellulosic biomass into value-added products.

BIOT 547

Global regulation of lignocellulolytic proteins from anaerobic fungi in response to substrate lignin content and composition

Casey Hooker1,3, Maklayla Schacht2, Abigail Hunnicutt1, Jonathan Overton1,3, Kevin Solomon1,3, [email protected]. (1) Agricultural & Biological Engineering, Purdue University, West Lafayette, Indiana, United States (2) Biological Sciences, Purdue University, West Lafayette, Indiana, United States (3) Laboratory of Renewable Resources Engineering (LORRE), Purdue University, West Lafayette, Indiana, United States

Plant cell walls form an abundant source of fermentable sugars for biofuels and biochemicals yet resist microbial decomposition due to their variable lignin content and composition. Model and more established platforms such as T. reesei and brown rot display reduced hydrolytic activity and inhibited growth with syringyl-rich lignin. Early- diverging anaerobic fungi resident in the digestive tracts of large herbivores, however, are among the most active plant-degrading microbes secreting the widest known array of biomass-degrading enzymes seemingly unbiased in substrate preference. More importantly, anaerobic fungi degrade diverse substrates with comparable efficiency through remodeling of their secretome in a substrate dependent manner. However, the regulatory inputs that are recognized remain unclear. Insight into this process would inform future genomic studies that dissect this behavior to study efficient fungal hydrolytic strategies for sustainable chemical production. Here, we evaluate the ability of these fungi to adapt to substrates of variable lignin content and composition. We have assembled a panel of novel fungal isolates from both zoo and farm animals representing several types of fungal genera. We demonstrate that these fungi express and secrete tightly regulated carbohydrate active enzymes (e.g. cellulases) that are active on a broad array of untreated agricultural residues, food wastes, and woody forestry products. More importantly, these enzymes are robust against increasing levels of syringyl lignin, which is a natural plant defense against microbial degradation. Compositional analyses of these substrates coupled with surveys of the fungal proteome reveal that anaerobic fungi recognize both polymeric content (e.g. lignin) and its monomeric composition. Increasing levels of S-lignin, while keeping lignin:cellulose:hemicellulose constant, led to the regulation of several biomass degrading enzymes. This behavior is conserved across several genera of anaerobic fungi. We are beginning to characterize this response in detail to identify the specific activities of the enzymes involved. Our work probes the rich regulatory response of plant biomass hydrolysis in anaerobic fungi, and highlights strategies honed and evolved for millennia that may be key to the efficient hydrolysis of untreated lignocellulose.

BIOT 548 Elucidating core design principles to engineer nonconventional yeasts as novel microbial factories

Zengyi Shao, [email protected], Mingfeng Cao, Meirong Gao, Le Zhao, Wan Sun. Iowa State University, Ames, Iowa, United States

Saccharomyces cerevisiae is far from being the only yeast of potential scientific and economic importance. Many of the 1800 other known yeast species have highly unusual metabolic, biosynthetic, physiological, and fermentative capacities. As outcomes of long-term natural evolution in particular environments, these high-performance characteristics are conferred by a network of genes via a hierarchy of regulations that are intrinsically complex, making the horizontal transfer of these functions into model hosts very challenging.

Three nonconventional yeast platforms will be discussed in this talk. First, we recently demonstrated Scheffersomyces stipitis as a better-suited platform to produce shikimate pathway derivatives, many of which have desired health-promoting activities. S. stipitis offers a higher availability of the rate-limiting precursor of the shikimate pathway due to its much more active pentose phosphate pathway. In the second platform, we are exploring Yarrowia lipolytica in producing long-chain wax esters as one of the major additives in the skin care products. Y. lipolytica, which is well known for its superior ability to accumulate triacylglycerol (TAG) under appropriate culturing condition and genetics combinations. Due to the biosynthesis of wax esters shares the same precursor with TAG biosynthesis, Y. lipolytica produced wax esters at a titer much higher than other reported hosts. Lastly, we will talk about how we engineer an acid tolerant strain Issatchenkia orientalis to produce itaconic acid, an important biopolymer precursor.

With these three exemplary projects, we will demonstrate how to overcome technology hurdles lay in front of engineering nonconventional microorganisms. How to design systematic rules and enable generalizable platform technologies that can be effectively applied from one species to another will be the focus of this presentation.

BIOT 549

Developing the thermotolerant yeast Kluyveromyces marxianus as a microbial host for volatile ester biosynthesis

Ann-Kathrin Loebs2, [email protected], Cory M. Schwartz2, Ian R. Wheeldon1. (1) A242 Bourns Hall, UC, Riverside, Riverside, California, United States (2) University of California Riverside, Riverside, California, United States

The yeast Kluyveromyces marxianus is a promising candidate for chemicals biosynthesis. Its natural capacity to produce short and medium chain volatile esters and ethanol at high rates, along with rapid growth kinetics at temperatures upwards of 45 °C make it especially interesting as platform for bioprocessing and biochemical production. The lack of synthetic biology tools along with limited knowledge about the metabolism, especially ester biosynthesis, have thus far hampered the development of K. marxianus as a platform for ester production. We have developed an efficient CRISPR-Cas9 genome-editing tool that allows for genomic disruptions and heterologous gene integrations in K. marxianus. This system was applied to elucidate the function of alcohol acetyl/acyltransferases and alcohol dehydrogenaseses (Adh) in acetate ester production. In contrast to prior claims, the alcohol acetyltransferase Atf was found to marginally contribute to bulk ethyl and isoamyl acetate formation. Our results show that the newly discovered Eat1 (ethanol acetyltransferase) is responsible for the formation of ethyl acetate as well as isoamyl acetate in K. marxianus. Furthermore, Adh2 activity is essential for ethanol and subsequent ethyl acetate production. In contrast to many other AATases in S. cerevisiae and other yeasts, Eat1 is localized to the mitochondria, allowing for high ethyl acetate production through utilization of the mitochondrial acetyl- CoA pool that is fostered by K. marxianus’ rapid growth kinetics. To further increase acetate ester production in K. marxianus we developed a CRISPR interference (CRISPRi) system to manipulate the expression of TCA cycle and electron transport chain enzymes with the goal of increasing the acetyl-CoA and alcohol pools available to ester production. Application of this system lead to an increase in ester formation up to 50% compared to a scrambled DNA control. Along with classical metabolic approaches, this yield can further be boosted to maximize ethyl acetate production in K. marxianus.

BIOT 550

Global metabolic rewiring for 2,3-butanediol production in cyanobacteria

Austin Carroll2, [email protected], Masahiro Kanno1, Nicole Nozzi1, Anna Case1, Shota Atsumi1. (1) Chemistry, UC Davis, Davis, California, United States (2) Chemistry, University of California-Davis, Davis, California, United States

Cyanobacteria have attracted much attention as hosts to recycle CO2 into valuable chemicals. Our work focuses on the production of 2,3-butanediol, a commodity chemical, in both marine and freshwater cyanobacteria. Recent efforts to establish a novel production system in the marine cyanobacterium, Synechococcus sp. PCC 7002, for 2,3-butanediol have been successful through a combination of synthetic biology tools, including gene integration, inducible systems, and production media composition. Although the freshwater cyanobacterium, Synechococcus elongatus PCC 7942, has been previously engineered to produce 2,3-butanediol, increasing production efficiencies remains a major challenge for commercialization. We have engineered a system in S. elongatus PCC 7942 that allows for simultaneous utilization of glucose and CO2 for production of 2,3-butanediol. Recently, we have made modifications to glycolytic pathways and the Calvin-Benson cycle in this system to improve glucose utilization, enhance CO2 fixation, and increase chemical production. These modifications are designed to increase carbon flux and redirect it towards carbon fixation. The engineered strain efficiently uses both CO2 and glucose, and produces 12.6 g/L of 2,3-butanediol with a rate of 1.1 g/L/d under continuous light conditions. This presentation will cover ongoing work with this system, which focuses on improving 2,3- butanediol titers in a variety of lighting and media conditions through further exploration of modifications to sugar metabolism and CO2 fixation.

BIOT 551

Synthetic biology tools for the purple, non-sulfur bacterium Rhodopseudomonas palustris CGA009

Cheryl Immethun, [email protected], Xinyuan He, Rajib Saha. Chemical & Biomolecular Engineering, University of Nebraska, Lincoln, Nebraska, United States

Non-model organisms’ unique metabolic processes and resilience in stress conditions make them appealing biotechnology chassis; yet, the engineering tools for these promising hosts can be limited. The purple, non-sulfur bacterium Rhodopseudomonas palustris CGA009 adapts to a huge range of environmental conditions through four different metabolic modes; photoautotrophic, photoheterotrophic, chemoautotrophic, and chemoheterotrophic. It can fix carbon dioxide and nitrogen or break down organic compounds for its carbon and nitrogen requirements while using light, inorganic, or organic compounds for its source of energy. R. palustris can also remain metabolically active in a non-growing state for months. This versatile bacterium has both aerobic and anaerobic pathways for degrading aromatic compounds, including many monolignols. Few microorganisms can consume monolignols and aerobic pathways for degradation can be limited by insufficient oxygen. A host that breaks down monolignols anaerobically could be better suited for industrial applications. To take advantage of R. palustris’ biochemical processing potential to convert waste products, such as lignin and carbon dioxide, into value-added products, tools for the synthetic gene regulation need to be constructed. Similar to many non-model microorganisms, a synthetic biology toolbox for R. palustris does not currently exist, which limits its use as a biotechnology platform. To begin addressing this limitation, a library of transcriptional regulators, starting with orthogonal, IPTG-inducible promoters, are being developed to precisely control gene expression. We will present progress towards creating these synthetic biology tools for the purple non-sulfur bacterium.

BIOT 552

Designing a continuous perfusion process for manufacturing antimicrobial peptides in Drosophila melanogaster S2 cells

Tobias Weidner1, [email protected], Jan Zitzmann1, Denise Salzig1, Peter Czermak1,2. (1) Life Science Engineering, University of Applied Sciences Mittelhessen, Giessen, Hesse, Germany (2) Bioresources, Fraunhofer Institute for Molecular Biology and Applied Ecology, Giessen, Germany

In the fight against multi-restistant pathogens, antimicrobial peptides (AMPs) can serve as leads for the development of novel antibiotics. In order to access this potential for pharmaceutical industry, effective and robust recombinant expression processes are mandatory. This includes the selection of a suitable expression host as well as process optimization during scale up. Here we describe a hollistic approach for the optimization of the production of two model AMPs, derived from the greater wax moth Galleria mellonella or the ladybug Harmonia axyridis. The proteins of interest were stably transformed Drosophila melanogaster S2 cells. Based on the polyclonal population, that was obtained after transfection of respective expression plasmids, we isolated highly productive single cell clones by limiting dilution and achieved at least a 100% increase in productivity. Further optimization on the cellular level included a statistical planned screening to determine optimal conditions for induction of the employed Metallothionein promoter. The online measurement of the cell suspensions dielectric properties and turbidity enabled an efficient process control and monitoring of the cells physiological status. Based on this information, 25 mg of the AMP was expressed at the 1-L bioreactor scale ensuring efficient timing for induction and harvest. Current focus of this work is the employment of a tangential flow filtration to enable efficient perfusion processes. A first continuous cultivation resulted in the production of 65 mg of target protein per day, using the same equipment as for batch culture. Finally, the functional AMPs were recovered by affinity chromatography and the products antimicrobial properties against model strains were tested, indicating the successful isolation of active peptides.

BIOT 553

Continuous manufacture of biopharmaceuticals: Engineering solutions to enable improved productivity and facility capacity management

Joanna Pezzini, [email protected]. MedImmune, Gaithersburg, Maryland, United States

Continuous processing is a paradigm shift for the manufacture of biopharmaceuticals. The technology provides numerous efficiency improvements with equivalent or better product quality. Simultaneous operation of each purification step combined with improved productivity results in dramatically reduced run rates. Additionally, the purification train can be synchronized to the flow rate of a perfusion bioreactor. Significant consumable cost savings are achieved due to higher binding capacities and increased cycling of resins. Reduced capital funding is required with the elimination of intermediate product hold tanks, and lower flow rates enable the use of single use tubing and vessels, even at the 20,000 L bioreactor scale.

This presentation outlines the new technologies, engineering solutions and practical considerations of a fully integrated purification process from cell culture through final formulation, demonstrated at pilot scale. The overall purification process design and the individual unit operation technologies are described, including novel viral inactivation and diafiltration steps. The implementation of a closed purification process, to enable a run over extended durations, is discussed. Lastly, automation control strategies and process analytical technologies are presented. BIOT 554

Quantitative assessment of environmental impact of biologics manufacturing using process mass intensity (PMI) analysis

Sri Madabhushi1, John Gavin1, [email protected], Sen Xu1, Collette Cutler1, Rebecca Chmielowski1, William Rayfield1, Nihal Tugcu2, Hao Chen1. (1) Merck, Kenilworth, New Jersey, United States (2) K15-2-H206, Merck Co Inc, Kenilworth, New Jersey, United States

Process mass intensity (PMI) is the benchmarking metric to evaluate the efficiency of a manufacturing process, which also provides an indication of the environmental impact of the process. The PMI for different types of manufacturing processes for a monoclonal antibody was calculated in order to better understand any efficiency differences between them. For the upstream manufacturing process, three operational modes were compared: fed-batch, fed-batch with N-1 perfusion, and perfusion in the N-stage bioreactor. Three downstream operational modes were also evaluated: a standard column chromatography process utilizing protein A resin and anion exchange (AEX) resin, a process using a protein A column and a single-use AEX membrane instead of the AEX column, and a third option using three-column periodic counter-current (3C- PCC) chromatography process for protein A followed by an AEX membrane. Sensitivity analysis was performed around selected unit operations to potentially identify opportunities to improve the efficiency of the manufacturing process further. The data indicate that a semi-continuous manufacturing process that includes perfusion, 3C-PCC and AEX membrane is the most efficient process, and therefore likely has the lowest environmental impact, equivalent to a 23% reduction of the overall PMI when compared with the fed batch and two-column chromatography process. Of all the process inputs, water contributes about 92-94% of the overall PMI. Additionally, the upstream processes account for 26-38% of the overall PMI, with the chromatography steps combining for 28-47%, and the ultrafiltration/diafiltration (UF/DF) step contributing 12- 17%. Selected sensitivity analysis revealed potential opportunities to further reduce the overall PMI of manufacturing process. These data indicate that PMI can be used to guide process development towards a more efficient, and in turn, a more environmentally sustainable monoclonal antibody manufacturing process. The data from this study also complements the monoclonal antibody process benchmarking data collected by the ACS Green Chemistry Institute Pharmaceutical Roundtable BioPharma Focus Group.

BIOT 555

Pall Cadence Acoustic Separator for continuous perfusion cell culture: Cell retention device and downstream filtration

Kevin Galipeau2,1, [email protected]. (1) Pall Corporation, Westford, Massachusetts, United States (2) BioPharm Applications R&D, Pall Life Sciences, Westborough, Massachusetts, United States As the biopharmaceutical industry evolves toward fully continuous purification processes, perfusion bioreactor systems become more and more attractive for therapeutic manufacturing. Conventional perfusion cell culture processes are designed to reach a “steady state” where viable cell density is kept relatively constant enabling consistent product expression for the duration of the process. Thus, ideal perfusion processes output predictable product concentrations at constant rates ready for downstream. Leveraging this principle naturally links perfusion bioreactors with continuous downstream bioprocessing.

Most perfusion cell culture processes rely on membrane based cell retention devices to keep cells in the bioreactor while allowing the molecule of interest to permeate through. These cell retention devices often foul over time, restricting the functional range of the device, and create a barrier to product transmission.

The Cadence™ Acoustic Separator is a filterless technology that delivers 100% product transmission. This disruptive single-use device facilitates continuous processing across a wide range of operating conditions. The process development scale system has been designed to operate continuously, supporting perfusion rates up to 20 L/day and is capable of processing bioreactor cell densities up to 100 x 106 cells/mL.

Following a series of performance mapping and optimization studies, we present a detailed operating window for the acoustic separator and logical filter selection guidelines based on current offerings. This information will enable users to harness their perfusion cell culture platform and produce a continuous harvest stream ready to be integrated with Cadence BioSMB PD chromatography systems.

BIOT 556

Demonstrating viral clearance for a novel approach to continuous viral filtration

Tarl Vetter1, [email protected], Michael J. Coolbaugh1, Veena Warikoo2, Kevin Brower1. (1) Bioprocess Development, Sanofi Genzyme, Framingham, Massachusetts, United States (2) Axcella, Cambridge, Massachusetts, United States

Nanofiltration is a robust technique for clearing viral contaminants and is generally positioned as one of the final steps in the purification of biopharmaceutical products. The nanofiltration process is almost universally operated as a batch process with specially designed dead-end viral filters. While the industry has moved toward implementing continuous processes for other parts of the production and purification of biopharmaceutics, continuous nanofiltration has received less consideration. Presented here is a new approach to viral filtration based on standard equipment and materials. The simple setup allows for periodic inputs while producing a continuous product stream. Initial studies have shown that this novel operating mode allows for a single viral filter to be operated continuously for multiple days with no indications of flux decay. Viral clearance studies, run continuously over multiple days, have demonstrated clearance of virus to levels below detection. These data indicate that this new technique can increase the capacity and lifetime of the viral filter while making the operation more amenable to end-to-end continuous production of biopharmaceuticals.

BIOT 557

Can integrated continuous processing occur at the 12kL scale?

Jonathan L. Coffman3, [email protected], Henry Lin2, Jeff Salm4, Gregory Hiller4, Scott Godfrey2, Samantha Wang1, Robert Fahrner4, Robert Kottmeier4, Raquel Orozco1, Samet S. Yildirim1, David Sullivan4. (1) Boehringer Ingelheim, Fremont, California, United States (2) Process Science, Boehringer Ingelheim, Fremont, California, United States (3) Biopharma Global Innovation and Technology, Boehringer Ingelheim, Fremont, California, United States (4) Bioprocess Development, Pfizer, Andover, Massachusetts, United States

We are building a new disposable manufacturing system to support the development and manufacturing of mAb and mAb-related products. We have made choices that are different than many others in the field of continuous and integrated processing. These choices avoid many misperceptions about continuous processing, are consistent with a staged approach to implementation, and facilitate manufacturing in either large-scale disposable or stainless manufacturing facilities. These choices leave open the possibility of operating continuous bioreactor at the 12,000L scale without significant equipment modifications. The design of the single use manufacturing system will be explained. The conversion of the process to run in a 12,000L stainless steel facility will be discussed.

BIOT 558

Viral clearance validation across continuous capture chromatography

Srinivas Chollangi3, [email protected], James M. Angelo1, Xuankuo Xu1, Sanchayita Ghose1, Massimo Morbidelli2. (1) Biologics Process Development, Bristol- Myers Squibb, Arlington, Massachusetts, United States (2) ETH Zurich, Zurich, Switzerland (3) Biologics Development, Bristol-Myers Squibb, Devens, Massachusetts, United States

Use of multi-column capture chromatography has been demonstrated to offer significant advantages over traditional batch chromatography in regards to improvement in productivity and capacity utilization of the resin. In our recent study, we have compared several multi-column formats (2-column, 4-column and 6-column) and the impact of process parameters such as resin type, feed titer, feed flow rates, % breakthrough and process duration on optimal performance during continuous capture. Upon identifying the optimal conditions, we have successfully demonstrated scalability of the process to a pilot-scale and purified 500L of cell culture harvest using 1.6 L of Protein-A resin. However, for this process to be successfully implemented in a clinical/commercial manufacturing setting, successful viral clearance validation is essential. In this study, we have conducted viral validation studies under cGLP guideline setting to assess viral clearance across continuous capture chromatography. Specifically, we have assessed the clearance of both enveloped and non-enveloped viruses across continuous capture chromatography in each of the three phases of continuous chromatography i.e. a) Start- up b) Steady State and c) Shut-Down. Further, we have evaluated a scale-down model using standard chromatographic operation based on valve switching to mimic the steady state operations using a continuous chromatography skid. Results from these studies are compared against viral clearance achieved by standard batch capture chromatography operated at 1% breakthrough. Finally, we have also assessed the resin cleaning strategy during continuous chromatography and assessed its impact on impurity carry over including HCPs, DNA, rPA and Viruses. As executed for continuous chromatography, a scale-down model has been developed to assess the virus carry- over as well. Together, these results show that cGLP viral validation studies can be successfully executed using continuous chromatography and a scale-down model can be successfully generated using standard chromatographic skids. Further insights will be shared on viral clearance across start-up, steady state and shut-down phases as well as effective cleaning strategies to prevent virus carry over between cyclic operations.

BIOT 559

Eliminating architectural segregation of pre and post viral manufacturing suites

Phil Clark, [email protected]. Development Supply Chain , Amgen, Thousand Oaks, California, United States

Regulatory agencies expect biotherapeutic plants to maintain appropriate segregation between product that has been through a viral reduction or removal step and product that has not been through such a step. This pre and post viral segregation is typically achieved with architectural segregation that includes: separate manufacturing suites with independent gowning vestibules, segregated air handlers, dedicated post viral clearance cleaning systems and lab instruments. Architecturally segregated manufacturing suites add cost to the construction and operation of the plant and reduce efficiency due to the additional gowning and movement of staff. We have created a second generation manufacturing process for a commercial product that is operated in a fully closed mode where the product stream is never exposed to the operating environment. Closed manufacturing provides an opportunity to eliminate architectural segregation of pre and post virus removal operations. In a fully closed operating mode appropriate segregation is provided by the closed nature of the systems used for manufacturing. All manufacturing operations; bioreactor, harvest, purification, viral filtration and drug substance final filtration, are competed in a single manufacturing suite. Eliminating architectural segregation with closed processing significantly reduces the cost of constructing and operating the plant.

BIOT 560 General platform for development of integrated downstream processes

Bernt Nilsson1, [email protected], Anton Löfgren1, Joaquin F. Gomis1, Niklas Andersson1, Lotta Berghard2, Peter Tiainen3, Arne Staby3. (1) Chemical Engineering, Lund University, Lund, Sweden (2) Sobi AB, Stockholm, Sweden (3) Novo Nordisk A/S, Bagsvaerd, Denmark

Smart downstream processing can be performed with a sequence of integrated purification steps, which minimize the number of storage tanks and reduce hold-up time. The result is an integrated multiple unit operation sequence that performs straight through processing of the target protein, with minimal time from expression to formulation. This downstream processing technique is well suited to be connected to a continuous upstream process based on perfusion. To develop these kinds of processes it is important to do studies in small-scale in a convenient way. This paper presents a methodology for process development and supervisory control of integrated downstream processes in lab-scale. A general platform is developed for sequential processing of lab-scale integrated downstream processes. The platform is implemented using ÄKTA/UNICORN-systems for demonstration but is not limited to this setup. The modification of the physical setup to handle multiple processing steps in sequence on one single machine makes it possible to study advanced and complex process configurations without a lot of resources. To make it easy to program and run the complicated setup a new supervisory controller is developed on top of UNICORN. The new controller, called orbit, is extendable and flexible to handle very different configurations and processes. To facilitate the usage even further the actual controller code is automatically generated from a high level presentation of the separation problem. Tools for design, control and verification makes it possible to virtual test the concept before making the actual experiment. The power of this concept is illustrated by some case studies. The general platform is illustrated by the chromatogram below from a process with four steps. The sensors measure UV and conductivity after each separations step in a cycle of four chromatography columns with in-line mobile phase conditioning. The product elutes after the fourth step in the small broad peek at 46 min.

The chromatogram of an integrated process with four separation steps in a sequence. The product is the small broad peek at 46 min.

BIOT 561

Differences in antibody transport at the human blood-brain barrier

John Ruano-Salguero1,2, [email protected], Kelvin Lee1,2. (1) Chemical & Biomolecular Engineering, University of Delaware, Newark, Delaware, United States (2) Delaware Biotechnology Institute, University of Delaware, Newark, Delaware, United States

The blood-brain barrier (BBB) is the generally impermeable network of cerebral capillaries that tightly regulates molecular trafficking between the blood and the brain. Notably, antibody therapeutics targeting neurodegenerative disease-related antigens demonstrate only 0.1% brain uptake. To help elucidate the transport of such medicines across the BBB, we use a hydrogel-based in vitro model of the BBB comprised of human induced pluripotent stem cell-derived brain microvascular endothelial cells. Time-lapse live-cell microscopy of fluorescent antibody conjugates is used to quantify transport and track intracellular processing. Differences in transport and intracellular processing between various antibody types were observed. Based on the different properties of the antibodies and molecules studied, a hypothesis on antibody transport at the BBB is integrated with current knowledge. In conclusion, these studies demonstrate the utility of a hydrogel-based stem cell-derived BBB model to characterize important attributes of therapeutics interfacing with the BBB.

BIOT 562 EXO-TIP: A novel immuno-photolytic method for the isolation of highly purified neuronal exosomes

Mohammad Parvez Alam, [email protected], Tina Bilousova, Jesus Campagna, Varghese John. Neurology, University of California Los Angeles, Los Angeles, California, United States

Currently, there is no approved therapeutic for Alzheimer’s disease (AD), which provides more than temporary symptomatic relief. Due to the aging population, the AD incidence is expected to reach 15 million by 2030 in the US alone. Many potential therapeutics have been tested in the clinic but have failed. Two main reasons why clinical trials of potential new AD therapeutics have not been successful include the late stage of intervention and the absence of sensitive, noninvasive methods to follow the effect of treatment on pathology-related biomarkers to enable optimization of the drug dose. We are attempting to address lack of non-invasive methods for diagnosis and prognosis by using an extracellular vesicle-based approach. Extracellular vesicle (EVs) carry proteins, lipids, and RNAs derived from their parent cells, and have potential as biomarkers specific to cell types and even disease/cellular states. The ability to detect brain-derived EVs in blood opens a window into the brain and creates the possibility of development of brain-specific liquid biopsy for AD and other neurological conditions. Conventional methods including ultracentrifugation or polymeric precipitation for isolation of EVs suffer from poor purity level and feasibility. Trending method such as affinity purification by using antibodies against exosome-specific surface proteins such as CD9, CD63, and CD81 cannot be used for analyzing the functions of intact EVs, as it is hard to detach the bound EVs from the antibodies under mild elution conditions. Because of these problems, developing a feasible method to isolate highly purified and intact EVs has been long warranted. Here, we report our novel method which comprises of immuno-photolytic isolation of brain-derived EVs from plasma with consecutive analysis of EV surface markers using a recently developed “ExoScreen” method based on Perkin-Elmer’s AlphaLISA technology, and downstream analysis of the disease stage and treatment-specific biochemical changes in EV compositions. The method is aimed to overcome heterogeneity in brain-derived EVs within plasma samples and detect small scale changes in the disease-specific and treatment-related biomarkers. When fully developed, our method may be utilized for non-invasive early diagnosis of AD, monitoring of drug treatment effects, and identification of markers and targets for new therapeutic approaches in AD.

BIOT 563

Inhibition of bacterial toxin activity using receptor-based peptides

Eric Krueger, Evan Koufos, Angela C. Brown, [email protected]. Department of Chemical and Biomolecular Engineering, Lehigh University, Bethlehem, Pennsylvania, United States Pathogenic bacteria produce numerous virulence factors, including protein toxins, to enhance their growth and survival within the host. Our lab seeks to understand the mechanisms of bacterial toxin delivery to identify and exploit therapeutic targets. In this project, we have focused on a member of the repeats-in-toxin (RTX) family of proteins, which are secreted by many Gram negative bacteria, including Bordetella pertussis, Escherichia coli, and Aggregatibacter actinomycetemcomitans. The leukotoxin (LtxA) secreted by A. actinomycetemcomitans specifically kills human white blood cells to disrupt the host immune response and therefore plays a key role in bacterial colonization of the host. The cell type specificity of LtxA arises from its reported targeting of the lymphocyte function-associated antigen-1 (LFA-1) integrin, which is only expressed by human white blood cells. In addition, we recently showed that LtxA binds with a strong affinity to cholesterol, and inhibition of this binding to cholesterol prevents LtxA-mediated cytotoxicity. Our goal in this project was to use information about these two targets of LtxA to design and test a panel of anti-toxin peptides, as an anti-virulence strategy to inhibit the pathogenesis of A. actinomycetemcomitans. (A) Cholesterol. We have identified a cholesterol recognition amino acid consensus (CRAC) motif in the toxin that regulates the interaction of the toxin with cholesterol. Using this CRAC sequence, we designed a cholesterol-binding peptide that inhibits LtxA-mediated cytotoxicity of human immune cells by blocking the binding of the toxin to cholesterol. (B) LFA-1. Using an analogous approach, we synthesized peptides corresponding to the hypothesized binding domains on the β-propeller of the αL subunit of LFA-1 and characterized their capability to inhibit LtxA binding and subsequent cytotoxicity. We found that four of the five peptides, specifically those corresponding to sequential β- strands in the β-propeller domain, inhibited LtxA activity. We are currently applying biochemical methods to investigate the mechanism by which these peptides inhibit LtxA binding to its targets to fully characterize the inhibitory mechanisms of these peptides. Our results demonstrate the possibility of using target-based peptides to inhibit LtxA activity, and we expect that a similar approach could be used to hinder the activity of other RTX toxins.

BIOT 564

New myeloperoxidase detection system based on enzyme-catalysed dye oxidative polymerization for paper-based diagnostic device

Arnau Bassegoda, [email protected], Guillem Ferreres, Sílvia Perez Rafael, Tzanko Tzanov. Chemical engineering, Universitat Politècnica de Catalunya, Terrassa, Spain

Chronic wounds represent a challenge to wound care professionals consuming a great deal of healthcare resources and, at the same time, reducing patient life quality with increased hospitalization times, heavy pain and eventually sepsis and death. Heavy bacterial colonisation is the main reason for non-healing chronic wounds, consequently wounds are often treated with antibiotics prophylactically, thus leading to unnecessary selection for bacterial resistance. Hence, there is a need for point of care testing (PoCT) devices for the evaluation of infection biomarkers allowing an early and appropriate treatment to reduce the severity of the disease and avoid the chronicity. In the last decade, paper based PoCT devices has showed great potential with the development of cheap and versatile microfluidic and lateral flow devices. These devices incorporate sensing molecules (e.g. enzyme substrates) immobilized in specific spots within the paper platform where they will react with determined biomarkers when the liquid sample flows through the device. Myeloperoxidase (MPO) an enzyme secreted by neutrophils and detected in fluids of infected wounds has been postulated as a suitable biomarker for wound diagnostics. MPO catalyzes the oxidation of chloride ions to hypochlorous acid (HClO), a powerful bactericidal oxidant, using hydrogen peroxide as co-substrate. At the same time, MPO can oxidize a variety of molecules including phenols, quinones, hydrazines and also proteins. Taking advantage of MPO substrate promiscuity, here we present an unexplored system for MPO detection based on enzyme-catalysed oxidative dye polymerization which can be incorporated into paper-based PoCT devices. Visual MPO detection has been achieved through the use of phenylenediamines, a common dye component, which its oxidation by MPO yielded bright coloured products distinguishable from the colour of the wound environment. Using paper strips as model of paper-based lateral flow device, immobilisation of the dye substrate was achieved through in situ interaction of the oxidised coloured product with a polycationic polymer. The colour reaction of the immobilised substrates, detectable by naked eye, responds to the MPO levels present in infected wound fluids. Thus revealing an easy system for incorporation of MPO detection in paper based diagnostic devices.

BIOT 565

Fluid shear stress as a circulating tumor cell model for testing chemotherapy drug resistance in breast cancer cell lines

Ursula L. Triantafillu1, [email protected], Yonghyun Kim2. (1) The University of Alabama, Tuscaloosa, Alabama, United States (2) Dept of Chemical and Biological Engineering, The University of Alabama, Tuscaloosa, Alabama, United States

Cancer is the second leading cause of death in the United States. Over 90% of these cancer deaths are due to the spread of cancer from the initial tumor to distant locations in the body during metastasis. The metastasizing cancer cells in the blood circulation are known as circulating tumor cells (CTCs). CTCs are responsible for secondary tumor formation. While solid tumors have been extensively studied, the dynamic nature of CTCs has been a challenge as they are under constant fluid shear stress. Another challenge in cancer is that of ineffective cancer treatment. Conventional cancer treatments mainly consist of chemotherapy regimen. However, chemotherapy drugs are largely tested under static 2D conditions, which does not resemble the CTC environment. Therefore, we created an in-vitro model for CTCs. Our CTC model allowed for physiological fluid shear stress to be applied on breast cancer cell lines. Breast cancer cells were treated with chemotherapy drug agents to test for drug resistance under our CTC model. By testing fluid shear stress on breast cancer cells, this will allow for a better model for CTCs as well as a better drug testing platform. BIOT 566

Investigating H2O2-induced cell death using a tunable, localized generator of peroxide in mammalian cell mitochondria

Kassi Stein, [email protected], Hadley Sikes. Chemical engineering, MIT, Cambridge, Massachusetts, United States

Among reactive oxygen species (ROS), H2O2 alone acts as a signaling molecule that promotes diverse phenotypes depending on the intracellular concentration. Mitochondria are a putative source of H2O2 as well as a site of redox signaling, and mitochondrial dysfunction has been implicated in diseases such as cancer. A genetically-encoded H2O2 generator, D-amino acid oxidase (DAAO), was targeted to the mitochondria of human cells; this was paired with a genetically encoded sensor of H2O2, HyPer-mito, and was used to investigate a range of H2O2 doses over time. Mitochondrially-targeted perturbations induced significant cell death in a concentration- and time-dependent manner. This work presents the first oxidant-specific perturbation of mitochondria in order to investigate cell death, and it reveals a marked sensitivity of this organelle to increases in H2O2 in comparison with prior studies that targeted the cytosol. These data have implications for the targeting of redox-based chemotherapeutics.

BIOT 567

Investigating how physicochemical properties of extracellular matrix influence apparent diffusion coefficient in MRI

Hannah Macdonald1,2, [email protected], Jeffrey Bamber1, Maxim Ryadnov2, David Collins1, Mihaela Rata1, Nandita deSouza1,3. (1) Institute of Cancer Research, London, United Kingdom (2) National Physical Laboratory , London, United Kingdom (3) Royal Marsden Hospital, London, United Kingdom

Quantitative Magnetic Resonance Imaging (MRI) parameters, such as the apparent diffusion coefficient (ADC) derived from diffusion weighted MRI (DW-MRI), have been proven to be valuable biomarkers of tumor grade and aggressiveness. Although the ADC potentially informs on the microstructural properties of the tissue, the contribution of various components of the extracellular matrix (ECM) to the hindered water diffusion observed in tissues is not fully understood. As tumor ECM differs substantially in its microstructure from normal tissues, it is desirable to investigate which features of the ECM influence the ADC, in order to understand the changes in the ECM that relate to tumor behaviour. We therefore sought to establish the relationship between ECM microstructural properties and ADC.

The structural components of the ECM may be divided into fibrillar proteins, (collagens being the most abundant), and glycosaminoglycans (GAGs). We used collagen I based materials to model the ECM and investigate how changes in matrix microstructure relate to ADC, longitudinal (T1) and transverse (T2) relaxation times. This analysis was combined with torsional rheometry measurements of viscoelastic moduli, an established way to characterise properties of the ECM that is frequently used in cell culture studies.

Preliminary data demonstrate that whilst viscous and elastic moduli of collagen I gels increase with increasing collagen I concentration, there is little change in ADC. Changes in T1 and T2 were, however, observed, with a 70% decrease in T2 when the collagen concentration was increased from 1.6 to 16 mg/ml. This indicates that while collagen I fibres at these concentrations may not be a major contributor to the reduction of ADC seen in tumors, they do influence the relaxation properties of water. Adding a GAG component to the gel increases the complexity of the ECM model and allows us to further understand the differential effect of ECM components on the MRI parameters of water.

T2 map (above) and ADC map (below) of collagen I gels (1.6-16 mg/ml) at 1.5T. As collagen density increases, the decrease in T2 is visible on the T2 map, but changes in ADC cannot be distinguished on the ADC map

BIOT 568

Modeling diseased blood-brain barrier utilizing patient-derived iPSCs and potential for rescue via amphiphilic block copolymer treatment

Hannah Seo1, [email protected], Catherine Lee2, Frank Bates1, Jakub Tolar3, Samira Azarin1. (1) Department of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, Minnesota, United States (2) Department of Genetics, Cell Biology and Development, University of Minnesota, Minneapolis, Minnesota, United States (3) Department of Pediatrics, Division of Blood and Marrow Transplantation, University of Minnesota, Minneapolis, Minnesota, United States As the interface between the microvascular system and the brain, the blood-brain barrier (BBB) is responsible for maintaining homeostasis in the central nervous system. As such, disruption to the BBB is implicated in many neurological diseases, but whether the loss in BBB integrity is an upstream or downstream effect of the disease is difficult to ascertain. We investigated this question for the disease childhood cerebral adrenoleukodystrophy (ccALD), a metabolic storage disorder caused by ABCD1 mutations ultimately leading to rapid neurological deterioration. We hypothesized that the BBB integrity of ccALD patients is inherently decreased compared to wild-type (WT) individuals. To model the BBB, we differentiated induced pluripotent stem cells (iPSCs) originating from WT individuals and ccALD patients into brain microvascular endothelial cells (BMECs) to compare their molecular and functional characteristics. PCR analysis and immunocytochemistry confirmed that the differentiated BMECs possess requisite BBB markers. Comparison of functional characteristics by the transendothelial electrical resistance (TEER) and sodium fluorescein permeability signified that the barrier function of ccALD-BMECs is inherently worse than that of WT-BMECs, with ccALD-BMECs displaying significantly reduced TEER (2600 ± 190 Ω cm2) compared to WT-BMECs (5000 ± 300 Ω cm2). Differences in ultrastructural properties were evaluated using transmission electron microscopy, which revealed greater accumulation of lipid droplets in ccALD-BMECs compared to WT-BMECs. These results were confirmed with image analysis of Oil-Red-O staining, which showed a two-fold increase in the amount of lipid droplets in ccALD-BMECs compared to the WT. With experimental evidence supporting our hypothesis, we aimed to improve the barrier function of ccALD-BMECs utilizing block copolymers. Amphiphilic block copolymers of poly(ethylene oxide) (PEO) and poly(propylene oxide) (PPO) have been previously demonstrated to be able to stabilize damaged cell membranes. Treatment of ccALD-BMECs with a diblock copolymer of PEO and PPO resulted in an approximate 60% increase in TEER and 25% decrease in lipid droplet accumulation. In summary, we have provided evidence for functional differences between BBB cells derived from iPSCs that originated from ccALD and WT individuals. Furthermore, we have demonstrated that the deficient functionalities of the ccALD-BMECs can be improved by treatment with block copolymers.