DRAFT

BIOT 1

Maturation of stem cell-derived skeletal myocytes promoted by micropatterning and substrate stiffness

Wendy Crone, Nunnapas Jiwlawat, Brett Napiwocki, Eileen Lynch, Alana Stempien, Randy Ashton, Tim Kamp, Masatoshi Suzuki, [email protected]. University of Wisconsin-Madison, Madison, Wisconsin, United States

In this work we demonstrate the effectiveness of an engineered two- dimensional micropatterned cell culture platform for creating highly aligned myotubes from human induced pluripotent stem cell (iPSC)-derived myogenic progenitors. Myotube elongation was shown to be dependent on the micropattern feature width and spontaneous contractions were aligned with the long axis of the pattern. As a result of an optimized micropattern feature geometry on a substrate of physiologically relevant stiffness, the resulting myotubes were elongated, well-aligned, and similar to myofibers; they showed significant improvement in nuclear alignment, myotube fusion, and sarcomere formation. This platform holds great potential in further understanding the process of human muscle development, as well as opportunities for in vitro pharmacological studies with skeletal muscle diseases. As an example, we demonstrated that bundle-like myotubes can be produced using patient- derived iPSCs with a background of Pompe disease (glycogen storage disease type II). Culturing in this engineered platform enhanced the disease phenotype as demonstrated by the observation of abnormal lysosome accumulations.

BIOT 2

Cell free platform for rapid synthesis and testing of active oligosaccharyltransferases

Jasmine M. Hershewe, [email protected], Jennifer Schoborg, Michael C. Jewett. Chemical and Biological Engineering, Northwestern University, Evanston, Illinois, United States

Protein glycosylation, the covalent attachment of sugar moieties to proteins, is important for protein stability, activity, and immunogenicity. Recombinant glycoproteins are critical in biotechnology, comprising life-saving therapeutics and vaccines. However, despite its importance, biomanufacturing defined glycoproteins and understanding the structure/ function relationships of DRAFT glycosylation remains a significant challenge due to technological limitations. These limitations include a lack of available tools for high-throughput biochemical characterization of enzymes that carry out glycosylation. A particular challenge is the synthesis of oligosaccharyltransferases (OSTs), which catalyze the attachment of complex, pre-built glycans to specific amino acid residues in target proteins. The difficulty arises from the fact that canonical OSTs are large and contain many (>10) transmembrane helices. Here, we address this challenge by establishing a bacterial cell free protein synthesis platform that enables rapid production of a variety of OSTs in their active conformations. Specifically, by using lipid nanodiscs as cellular membrane mimics, we obtained soluble yields of up to 420 mg/L for the single subunit OST, PglB, from Campylobacter jejuni, as well as for three additional bacterial PglB homologs. Importantly, the cell free derived enzymes catalyzed glycosylation reactions in vitro with no purification or processing needed, and the ability to tightly control concentrations and ratios of glycosylation components in the in vitro system allowed us to quickly optimize for full glycosylation of target proteins. Since the publication of this work, we have extended the approach to various OST homologs, and have developed high- throughput mass spectrometry testbeds to analyze glycosylation. We anticipate that our technology will enable accelerated enzyme prototyping and expand the available enzyme toolbox for biomanufacturing defined glycoproteins.

BIOT 3

Encoding decision-making functions into cell metabolism: the marriage of synthetic biology, metabolic engineering and intelligent control

Peng Xu, [email protected]. Chemical, Biochemical and Environmental Engineering, University of Maryland Baltimore County, Baltimore, Maryland, United States

Living organism is an intelligent system encoded by hierarchically organized information to perform precisely controlled biological functions. With a better understanding of cellular regulation, biomolecular engineers have been able to engineer both the chemistry modules (the mass flow) and the control modules (the information flow) inside the cell to design intelligent cells with desired functions. Instead of programing machine language in a chemical plant, synthetic biologists rewrite the genetic software and encode logic functions in living cells to improve cellular performance. In this lecture, I will present both computational and experimental approach to unravel the design principles underlying efficient biomanufacturing platforms – YIN and Yang DRAFT metabolic balance, autonomous metabolic switches, microbial social interactions for various biotechnological applications. I will present strategies to build genetic toolkits to streamline the genetic/genome modification for a promising industrial yeast Y. lipolytica, which allows us to harness the endogenous acetyl-CoA/malonyl-CoA/HMG-CoA metabolism to produce complex oleochemicals, terpenes, polyketides and aromatic commodity chemicals. By combining metabolic addiction with negative autoregulation, I will also present our recent effort to encode decision-making functions into cell metabolism to partition carbon flux and improve strain stability. Engineering feedback genetic circuits to encode decision-making functions into cell metabolism will present us exciting opportunities to solve the most pressing challenges in health, energy and environment in the 21stcentury.

BIOT 4

Quality and quantity?: Enhancing mammalian biomanufacturing performance through systems biotechnology approaches

Michael J. Betenbaugh, [email protected]. Department of Chemical and Biomolecular engineering, Johns Hopkins University, Baltimore, Maryland, United States

The traditional engineering paradigm states that you must sacrifice quality for quantity. This can be true for biomanufacturing processes in which quality can involve critical quality attributes such as the glycan structures attached to a produced glycoprotein therapeutic. As biotechnologists, we seek to break this axiom through the application of systems engineering, cell and process modeling, and synthetic biology and metabolic engineering. In our current work we are implementing these systems biotechnology tools to improve the performance of CHO cells and other production hosts. Engineering methodologies are being applied to both determine the glycosylation structures and, as needed, change these glycoforms. One approach being used is to alter the cell lines in order to both add and remove glycosylation capabilities in CHO cells. Determining which modifications are best to implement can be facilitated with the assistance of comprehensive models of glycosylation processes. Another alternative is to adjust the media and examine the role that supplementation with additives has on both the product titer and glycosylation character. It is equally important to be able to evaluate the effect of these engineering modifications and thus analytical tools are needed to determine the glycan patterns for glycoproteins produced in these hosts. Finally, it is also useful to dictate the final glycan structures and this is often best achieved through the use of control strategies based on our DRAFT understanding of cellular glycosylation and the impact of process modifications. Such a comprehensive systems biotechnology approach will enhance our ability to generate desired profiles in terms of glycosylation and other attributes while minimizing the effect on the performance of CHO cells in culture, leading to enhanced production of high quality target biopharmaceuticals in current and future mammalian biomanufacturing processes.

BIOT 5

Process development strategy for E. coli based cell-free protein synthesis reactions

Noelle Colant1, Jaime Teneb-Lobos1, Stephen Goldrick1, Stefanie Frank1, William Rosenberg2, Daniel G. Bracewell1, [email protected]. (1) Biochemical Engineering, University College London, London, United Kingdom (2) Institute for Liver and Digestive Health, UCL Division of Medicine, London, United Kingdom

Over the last decade, cell-free protein synthesis (CFPS) has been utilized as a production platform for antibodies, therapeutic proteins, and vaccine candidates. CFPS is advantageous as a production platform because reactions can achieve relatively high titers in a few hours and reactions have been demonstrated to scale linearly up to 100 L. CFPS reactions also tolerate conditions that are not typically attainable in traditional in vivo cultivations. The open nature of these in vitro systems allows for non-physiological conditions as well as the addition of components not naturally found in E. coli, like non- standard amino acids which can then be incorporated into proteins or chaperones and other agents for post-translational modifications.

CFPS process development strategies avoid the need for cell line development but must account for these differences. As CFPS reactions can be completed rapidly we were able to design a rational process development strategy that allows us to optimize titers in ~48 hours. The approach begins by examining the impact of the E. coli strain chosen for the extract, the formulation of the reaction mixture, and the optimization of the plasmid sequence on product titer. Using the system selected by this study we then investigated the key process parameters; temperature, pH of the reaction mixture, concentration of plasmid, percentage of extract in the reaction, and length of the reaction. This experimental design was then analyzed using multivariate data analysis (MVDA) to determine the reaction conditions that would maximize product titer. DRAFT

The methodology was applied to two different products; hepatitis B core antigen (HBcAg) and superfolder green fluorescent protein (sfGFP). We found that the impact of the process parameters on titer is highly dependent on the product. The strain of the cell extract – in particular, whether or not the cells used for the extract had been induced during cultivation – was the most influential on product titer. Titers for sfGFP were improved in reactions with a non-induced BL21 StarTM (DE3) extract while an induced BL21 StarTM (DE3) extract enhanced HBcAg production. Following that, adjustments in temperature or pH of the reaction mixture were most influential and resulted in over a 40% increase in titer in some cases but again optima varied widely between the two products.

BIOT 6

3D-printed villi-on-a-chip with bioelectronics for drug development and manufacture

Chen-Yu Chen2, [email protected], Gregory F. Payne1, Ryan D. Sochol3, William E. Bentley2, Andrew Lamont2, Kimberly Lo2. (1) Univ of Maryland Biotech Inst, College Park, Maryland, United States (2) Fischell Dept of Bioengineering, University of Maryland, College Park, Maryland, United States (3) Mechanical Engineering, University of Maryland, College Park, Maryland, United States

Gut dysbiosis (microbial imbalance) is known to influence the physiological function of the gut epithelium and is involved in several intestinal diseases and even mental disorders. Owing their ability to host multiple human cell types and gut microbiota, synthetic in vitro models commonly referred to as “Organs-on-a-chip” (OOC) offer the capability to interrogate and modulate biological signaling at cellular length scales and in real time. Importantly, few systems recapitulate the spatial and chemical environments of the GI tract that are known to have a dramatic impact on the molecular signaling occurring in vivo. For example, oxygen tension and redox environments that extend from the crypts to the villous tips vary tremendously. To gain insight, we have developed a 3D-printed human villi-on-a-chip model that incorporates an internal capillary system that recreates lymphatic and circulatory conduits for molecular exchange. This capillary system provides for real-time monitoring of cellular permeability and redox signaling, and further recapitulates apical and basal transport in a structurally relevant model. The villus scaffold is assembled with ultra-high spatial resolution owing to a new two-photon direct laser writing (DLW) 3D printing system. The 3D fabrication DRAFT strategy offers finely tuned parameters including pore-size, curvature, surface roughness, and channel size. Results are consistent with earlier efforts showing the upward migration of cells towards the villus tips. By coupling novel bioelectronic sensing modalities on adjacent “chips”, we can rapidly evaluate a chemical/biological landscape that is needed for the development of precision medicines. Conversely, owing to the integrated electronics, the device can also modulate the redox nature of the microenvironment, providing new insight in biological responses. That is, by advancing 3D printing configurations that faithfully recapitulate the geometric landscape and at the same time enable bidirectional regulation of the signaling events, we are in a better position to discover new insights on the complexities of the GI tract. These will serve to more rapidly understand disease progression and at the same time, facilitate the development of new therapies.

BIOT 7

Enabling chemical and structural biology with synthetic binding proteins

Shohei Koide1,2, [email protected]. (1) Biochemistry and Molecular Pharmacology, New York University School of Medicine, New York, New York, United States (2) Perlmutter Cancer Center, NYU Langone Health, New York, New York, United States

The protein engineering community now has the capacity to rapidly create synthetic binding proteins with high potency and exquisite selectivity. Although its major focus remains on the development of biologic therapeutics, synthetic binding proteins have proven also to be transformative tools in mechanistic biomedical investigation. I will present applications of the Monobody technology, a synthetic binding protein system based on the fibronectin type III scaffold, to challenging questions in chemical biology and structural biology in which tailor-made Monobodies enable novel investigations and help gain new mechanistic insights.

BIOT 8

Innovative automation and mass spectrometry infrastructure: Bridging processing parameters with CQAs of biotechnology products

Patrick J. Faustino, Jinhui Zhang, [email protected]. FDA, Sliver Spring, Maryland, United States DRAFT

Regulators worldwide (FDA, EMA, ICH) have drawn attention to critical quality attributes (CQAs) that should be identified and monitored to evaluate consistent batch production and thus help to ensure optimum control of the manufacturing process of biotechnology products for regulatory acceptance. The purpose of this presentation is to provide a holistic understanding how innovation in regulatory science can bridge processing parameters with CQAs of biotechnology products. The presentation will first highlight FDA lab office CDER/OPQ/OTR’s vision on how innovative platforms may support CDER/OPQ’s review and research on biotechnology products; subsequently, the presentation will describe a landscape of evolving analytical platforms for biotechnology products CQAs’ identification and evaluation; in the end, several case studies will be presented to showcase the scientific efforts that OTR has taken to bridge process parameters with CQAs of biotechnology products. In summary, the presentation will highlight the innovative promise of these evolving platforms and emerging technologies to support process understanding and advance the state of product quality for biotechnology products.

BIOT 9

Evaluation and mechanistic elucidation of recombinant hydrophobin as a crystallization inhibitor of flufenamic acid

Nathanael Sallada1, [email protected], Yongjun Li2, Matthew Lamm2, Bryan Berger3. (1) Biomedical Engineering, University of Virginia, Charlottesville, Virginia, United States (2) Formulation Sciences, Merck & Co., Rahway, New Jersey, United States (3) Chemical Engineering, University of Virginia, Charlottesville, Virginia, United States

Formulation of poorly water-soluble small molecule drugs continues to be a major obstacle in bringing new drugs to market due to issues with low bioavailability. Maintaining a high degree of supersaturation in the gut is one method to increase bioavailability of poorly water-soluble drugs through addition of antinucleating agents in amorphous drug delivery systems. Hydrophobins are small, highly surface-active proteins from filamentous fungi that display resistance to degradation and potential immunological inertness, making them particularly amenable to drug formulation and delivery applications. In this work, we evaluate an engineered, recombinant hydrophobin (HFB1, class II) as a novel inhibitor of crystallization to maintain drug supersaturation of a model hydrophobic drug, flufenamic acid (FA). Model drug precipitation kinetics were tracked using a UV fiber optical method in a supersaturationprecipitation system. HFB1 was compared to several DRAFT widely used polymeric crystallization inhibitors: methocel™ (A4C grade), methocel™ (K15M grade), Kollidon VA64 and HPMCAS (MF grade). Supersaturation-precipitation experiments show that HFB1 outperformed all polymers tested and has potential for use as a crystallization inhibitor in amorphous drug delivery systems. Ongoing work to be discussed in this presentation will focus on elucidation of the crystallization inhibition mechanism and design of hydrophobin variants with improved drug formulation properties.

BIOT 10

Overcoming mAb manufacturing process challenges for high concentration drug substance to facilitate subcutaneous administration

Swarnim Ranjan, [email protected], Jessica Hung, Melissa Holstein, Sanchayita Ghose. Biologics Process Development, Bristol-Myers Squibb, Devens, Massachusetts, United States

Ultrafiltration/diafiltration is a key step for achieving the desired final protein formulation in nearly all downstream processes. With the current industry focus on subcutaneous delivery due to ease of administration and better patient experience, there is an increasing need for higher concentration (≥100 g/L) and low volume (≤1.5 mL) formulations. However, obtaining high protein concentration using ultrafiltration poses numerous challenges, both from manufacturing process and protein stability perspectives. The objective of this work is to identify these challenges and evaluate strategies to mitigate them. The exponential increase in viscosity with increasing protein concentrations, particularly above 150 g/L causes high pressure leading to operational limitations. High viscosity also leads to low flux, resulting in long processing times during ultrafiltration and sterile filtration operations. The higher volumetric ratio of system holdup to final product pool results in undesirable yield loss and/or product dilution during product recovery. Additionally, during diafiltration and ultrafiltration, significant differences are observed in the pH and excipient concentrations between the diafiltration buffer and the formulated protein solution due to volume exclusion, charge interaction and preferential hydration effects. These effects become more pronounced with increasing protein concentration, as molecular ‘crowding’ affects protein charge due to protein-protein interactions and leads to increased volume exclusion, and can also cause protein aggregation. This work includes understanding the pH and excipient behavior during ultrafiltration/diafiltration operation and the impact of molecular properties to achieve the target DRAFT excipient concentration; implementation of a plug flow strategy at low flow rate to maximize the recovery of the drug substance with minimal dilution; better understanding of the impact of viscosity on pressure drop during concentration, and evaluation of filter membranes with improved flow paths to address high pressure and low flux concerns during sterile filtration, both at lab and commercial manufacturing scales, thereby, overcoming challenges associated with manufacture of high concentration mAb product.

BIOT 11

Next generation protein A capture – harnessing the power of nanofiber technology to intensify biomanufacturing and facilitate rapid process development

Richard Kucia-Tran, [email protected]. GSK, Stevenage, United Kingdom

Traditional packed bed chromatography has been a workhorse for the biomanufacturing industry for the past five decades, deployed in the production of everything from plasma proteins, through monoclonal antibodies to modern applications such as the manufacture of cell based vaccines. A limitation of this technology has been the inherent mass transfer constraints, inversely linking how quickly processes can be operated, and the mass throughput of the operation. As a result, it has been the long held wisdom that packed bed chromatography is a capacity and productivity bottleneck for most modern biomanufacturing facilities.

Nano-fibre based chromatographic media developed by Puridify, subsequently acquired by GE Healthcare, decouples mobile phase flowrate and protein mass transfer, enabling chromatography cycles to be completed in single digit minutes, rather than hours. The work presented will highlight how this shift in processing paradigm has been utilised at GSK. Case studies will illustrate the process development work that has been performed using the PrismA Fibro for antibody capture, and to evaluate its performance, particularly at pilot scale. The data gained from these studies has been used to map out potential implementation strategies, providing either an increase in facility flexibility and therefore utilisation and/or reducing the CoGs, particularly in the context of clinical manufacturing.

Furthermore, whilst the operational and financial benefits of adopting Fibro technology can be easily conceptualised, perhaps less well documented is the potential for using this technology to facilitate more rapid process DRAFT development. The work presented will show how PrismA Fibro has been implemented at GSK, outside of the DSP space to which its application is most immediately apparent, in order to facilitate more rapid cell line screening and selection.

BIOT 12

Shikimate kinase: Potential target for new antibacterial agents

Marcin Ogrodniczuk, [email protected], Rene Fuanta, Anand A. Motilall. Biochemistry, East Stroudsburg University, Stroudsburg , Pennsylvania, United States

Tuberculosis is one of the world’s leading cause of mortality from a single bacterial pathogen, with over 10 million reported cases each year. There is an alarming increase in the prevalence of drug-resistant strains, thus the need for the discovery of novel anti-tubercular agents. Mycobacterium tuberculosis shikimate kinase (MtSK) catalyzes the 5th step of the shikimate pathway, converting shikimate to shikimate-3-phosphate. The overall goal of this project is to express and characterize MtSK and screen for potential anti- tubercular agents. Transformation of XL-1 blue competent cells was performed using a pET 21b plasmid with aroK gene inserted at the multiple cloning site. Plasmids were cloned and purified and used to transform BL 21 DE3 competent cells for subsequent protein expression. With further small and large scale expression, purification and characterization, MtSK kinetic parameters would be determined prior to enzyme inhibition studies using inhibitors like avarone, a marine sponge sesquiterpene quinone and derivatives thereof.

BIOT 13

Inductive niche module: A novel angiogenesis in vitro assay system

Abdulkader Rahmo, [email protected]. SMSbiotech, San Marcos, California, United States

A unique Angiogenesis assay system (ANGIOstream) that uses a proprietary human extracellular matrix, derived from small mobile stem cells was used to capture tissue level organization of endothelial cells reminiscent to the polygonal capillary network. The mere application of an Inductive Niche Module (INM) to endothelial cells within a thin layer of Extracellular matrix initiated a process involving cell DRAFT communication, differentiation, migration and collective organization, leading to a polygonal superstructure. This process initiated within 2-3 hours of INM application, was captured in detail using time laps photography in vitro. A complex multi-stage, multi-step process of organized micro vessel spatial structuring was accessible to observation without interruption. This in vitro capability would provide ample opportunities for interrogating biochemically some fundamental aspects of tissue self-organization, cell networking and spatial cell structuring.

BIOT 14

Self-assembling peptides to mitigate familial hypercholesterolemia

Sreya Sanyal1, [email protected], Victoria Harbour2, Zain Siddiqui2, Biplab Sarkar2, Ka Kyung Kim2, Vivek Kumar3. (1) New Jersey Institute of Technology, Newark, New Jersey, United States (2) Biomedical Engineering, New Jersey Institute of Technology, East Brunswick, New Jersey, United States (3) Biomedical Engineering, New Jersey Institute of Technology, Newark, New Jersey, United States

Elevated levels of low-density lipoprotein cholesterol (LDL-C) is one of the major contributors to ischemic cardiovascular disease (CVD), the leading cause of death in the United States and globally. PCSK9, a secretory mammalian serine protease, interacts with low-density lipoprotein receptors (LDLR) on the surface of hepatocytes, leading to the receptor’s internalization and degradation. As LDLR is primarily responsible for removal of LDL-C from circulation, PCSK9 activity decreases expression of LDLR and increases serum levels of LDL-C. Accordingly, gain-of-function PCSK9 mutations lead to autosomal dominant familial hypercholesterolemia. FDA-approved anti- PCSK9 monoclonal antibodies (mAbs) have shown clinical efficacy in LDL-C reduction, lowering plasma LDL-C levels by approximately 60% without adverse events, even in patients receiving maximum dose statins. A significant drawback to anti-PCSK9 mAbs is high list price, limiting affordable access. We have engineered a selfassembling therapeutic peptide (Sle- Pep28) targeting PCSK9 to treat familial hypercholesterolemia, as an affordable alternative to mAb therapies. Sle-Pep28 is formulated as a polymeric hydrogel that gradually releases therapeutic peptides at the subcutaneous implant site. Subcutaneous delivery promotes improved pharmacodynamics over oral and intravenous administration. Promisingly, Sle-Pep28 shows specific, irreversible binding to PCSK9 in surface plasmon resonance (SPR) studies as well as significantly increased LDLR expression in hepatocytes. DRAFT

BIOT 15

High-sensitivity detection and quantification of dengue virus marker NS1 by nanophosphor lateral flow immunoassay

Heather Poeck-Goux2,3, Victoria Hlavinka1, [email protected], Kristen Brosamer4, Ujwal Patil3, Atul Goyal1, Suman Nandy1, Binh V. Vu1, Tesfaye Gelanew6, Elizabeth A. Hunsperger7, Katerina Kourentzi1, Richard C. Willson1,5. (1) Department of Chemical and Biomolecular Engineering, University of Houston, Houston, Texas, United States (2) Maryland Department of Health Laboratories Administration, Baltimore, Maryland, United States (3) Department of Biology and Biochemistry, University of Houston, Houston, Texas, United States (4) Department of Biomedical Engineering, University of Houston, Houston, Texas, United States (5) Escuela de Medicina y Ciencias de la Salud, Tecnológico de Monterrey, Monterrey, Nuevo León, Mexico (6) Armauer Hansen Research Institute, Addis Ababa, Ethiopia (7) Centers for Disease Control and Prevention, Nairobi, Kenya

An estimated 3.9 billion people live in tropical and subtropical areas affected by the mosquito-borne dengue virus (DENV), with nearly 400 million people infected each year. Diagnosis of dengue is primarily by reported symptoms and physical examination, resulting in frequent under- and misdiagnosis. We have developed an inexpensive, rapid lateral flow assay (LFA) to quantify DENV nonstructural protein 1 (NS1), a known marker of early dengue infection, in human serum. Using strontium aluminate persistent luminescent nanoparticles (nanophosphors) as reporters, we achieved a limit of detection of 1 ng/mL DENV serotype 1 NS1 antigen in buffer. Our assay is comparable to an analogous NS1 ELISA that has a reported limit of detection of 2 ng/mL. This low limit of detection suggests that, with further testing and validation, nanophosphor NS1 LFA may be beneficial for early diagnosis of dengue.

BIOT 16

Quorum quenching enzymes site-specifically functionalized via protein farnesylation for visualization and surface immobilization

Keun-Young Park1, [email protected], Nathan Hoekstra2, Mikael Elias2, Mark D. Distefano2. (1) Chemistry, University of Minnesota, Minneapolis, DRAFT

Minnesota, United States (2) Univ of Minnesota, Minneapolis, Minnesota, United States

Biofilm formation is a microbial assembly which is detrimental in human health, agriculture and aquaculture, which generates losses worth of billions of dollars per year. Bacterial communication known as quorum sensing is essential for this detrimental phenomenon, which involves secretion of signaling molecules such as acyl homoserine lactones to the local environment. Quorum quenching enzymes (QQE) prevent biofilm formation by continuously degrading these signaling molecules, and hence have been gaining attention. For example, the lactonases developed by Elias and coworkers were shown to be easily applicable for antifouling purposes due to its high thermal stability (Tm = 88 °C) and robustness. However, there were improvements that needed to be addressed. In this work, enzymatic protein labeling via protein farnesylation, click chemistry, and in situ polymerization are exploited to enhance the utility of QQEs. To enable visualization of enzymes applied to surfaces, a fluorophore was site-specifically linked to QQEs via introduction of an azide-moiety via enzymatic modification followed by click reaction with DBCO-TAMRA. To enhance the retention of surface applied QQEs, an ATRP (Atom Transfer Radical Polymer) initiating moiety was enzymatically transferred to a QQE, followed by in situ polymerization directly from the QQE-macroinitiator with “sticky” monomers to yield QQE-polymer conjugates. We anticipate that the combination of covalent crosslinks and hydrophobic interaction between the polymer and various coating materials will allow efficient protein- immobilization. Additionally, the use of an enzymatically installed diazirine moiety to facilitate photochemical crosslinking to surfaces for immobilization is being explored.

BIOT 17

Colorimetric response of borohydride stabilized silver nanoparticle on interaction with organophosphates shalini shikha, [email protected], Sudip Pattanayek. CHEMICAL ENGINEERING, IIT DELHI, New Delhi, NEW DELHI, India

Detection of residual organophosphates in food is of paramount importance in the current scenario. Biosensors for pesticide detection may be developed by monitoring the interaction of nanoparticles with pesticides. In this report, we have studied the colorimetric response of pesticide phorate, chlorpyrifos and malathion with borohydride stabilized silver nanoparticles. The colorimetric DRAFT changes are corroborated with UVVisible spectra along with the change in particle size and zeta potential. Nanoparticle on interaction with phorate shows the visible change in color to dark green while the other two pesticides haven’t shown any visible color change. The colorimetric response of nanoparticle pesticide mixture was further studied for 12 hours of incubation and it was observed that nanoparticle-phorate mixture started decolorizing after an hour and becomes colorless after 12 hours of incubation, while other two pesticide mixtures remain same. This decolorization of phorate was correlated with the degradation of pesticides in the presence of bare silver nanoparticles. The reaction mechanism was explained by the formation of silver and sulfur atom (Ag→S) complex. The uncapped nanoparticles get stabilized by pesticides over time for chlorpyrifos and malathion while phorate undergoes degradation due to the scission of S-CH2 present in the molecule. This color change may be utilized as a colorimetric method for the detection of phorate pesticide.

BIOT 18

Hitchhiking probiotic vectors to deliver ultra-small Hafnia nanoparticles for ‘color’ gastrointestinal tract photon counting X-ray imaging

Fatemeh Ostadhossein1, [email protected], Michael Nelappana2, Chiara Lowe3, Mahdieh Moghiseh3, Anthony Butler3, Indu Tripathi2, [email protected], Dipanjan Pan2, [email protected]. (1) Stanford University, Palo Alto, California, United States (2) University of Illinois Urbana, Urbana, Illinois, United States (3) University of Otago Christchurch, Christchurch, New Zealand

Gastrointestinal (GI) tract is one of the hard-to-reach target tissues for the delivery of the contrast agents and drugs mediated by the nanoparticles due to its harsh environment. Herein, we overcame this barrier by the design of the orally ingestible probiotic vectors for ‘hitchhiking’ the ultrasmall hafnia (HfO2) (~1-2 nm) nanoparticles. The minute-made synthesis of these nanoparticles is accomplished during a simple reduction reaction. These nanoparticles were incubated with probiotic bacteria with potential health benefits and non-specifically were taken up due to their small size. Subsequently, the bacteria were lyophilized and packed into capsule to be administered orally as the radiopaque contrast agents for delineating the GI features. These nano-bio-hybrid entities could successfully be utilized as contrast agents in vivo in the conventional and multispectral computed tomography (CT). We demonstrated in ‘color’ the accumulated nanoparticles using advanced detectors of the photon counting CT. The enhanced nano-bio- DRAFT interfacing capability achieved here can circumvent traditional nanoparticle solubility and delivery problems while can offer a patient friendly oral delivery approach for the GI imaging to replace the currently practiced barium meal.

BIOT 19

Controlled Drug Release by Varying Polymer Structure-Property Relationships in Microparticles

Catherine Miles, [email protected], Jarrod Cohen, Fatima Buevich, Joachim B. Kohn. Chemistry and Chemical Biology, Rutgers University, Piscataway, New Jersey, United States

Limited control of drug release of commonly used poly(lactide-co-glycolide) (PLGA) and poly(lactic acid) (PLA) microparticles has led to the investigation of a tyrosine-derived polymer library with varying structure-property relationships that has the capability to control drug release rates from microparticles. A microparticle preparation method has been optimized using a continuous flow method to prepare particles ranging from 100-130 µm in size loaded with both hydrophilic and hydrophobic drugs. This delivery system allows for localized delivery of a wide range of therapeutic drugs. A thorough investigation into how polymer structure-property relationships effect drug encapsulation and release has been performed evaluating the tyrosol-derived polymer library against PLGA and PLA. The tyrosol-derived polymer library is an alternating AB polymer system with A either tyrosol or desaminotyrosine and B a diacid of varying carbon length. High encapsulation of a hydrophobic drug, dexamethasone, was achieved for all formulations prepared and all formed a miscible drug-polymer interface at 20% w/w loading which is key to achieving a controlled release system. PLGA and PLA exhibited 43% and 24% cumulative release respectively after 41 days, while tyrosol-derived polymers exhibited between 37-65% cumulative release depending on the polymer. By selecting a polymer with certain structure-property relationships, such as polymer hydrophobicity and mixed Tg (polymer-drug) it is possible to directly control the release profile. Incorporation of drug at multiple loadings predicts the mixed Tg and determines when the polymer-drug is no longer miscible, which has shown to correlate with release rates. Ongoing studies are using polymers containing polyethylene glycol in the polymer backbone as an attempt to obtain a faster releasing particle system.

BIOT 20 DRAFT

Particle-in-particle nano-enabled approach for the topical eradication of dental biofilm without disturbing microbiota balance in vivo

Fatemeh Ostadhossein1, [email protected], Esra Altun2, Debapriya Dutta2, Dinabandhu Sar2, Indu Tripathi2, Shih-Hsuan Hsiao2, Valeriya Kravchuk2, Parikshit Moitra3, Dipanjan Pan2. (1) Stanford University, Palo Alto, California, United States (2) Bioengineering, University of Illinois Urbana Champaign, Urbana Champaign, Illinois, United States (3) University of Maryland Baltimore, Baltimore, Maryland, United States

Dental plaques are one of the most prevalent types of biofilm which can lead to dental caries due to the demineralization processes caused by acidogenic bacteria. These bacteria reside inside a protective sheath i.e. extracellular polymeric substances which makes any curative biofilm treatment challenging. Herein, we propose an antibiotic-free strategy to disrupt the biofilm by utilizing nanoparticles designed to function in the acidic environment of these biofilms. Carbon dot nanoparticles were decorated with phosphonium ions to endow them with antibiofilm properties and they were made stimuli-responsive by their entrapment in a pH-sensitive polymer shell. Mechanistic studies on Streptococcus mutans (S. mutans) revealed that the toxicity of these nanoparticles was mediated by their electrostatic interactions with the bacterial membrane, the generation of excess reactive oxygen species (ROS), and subsequent DNA fragmentation. In vitro and Ex vivo studies on mature biofilms revealed more than 90% biofilm inhibition. An in vivo examination in a rat model of dental biofilm further showed that the nanoparticles could effectively suppress the growth of S. mutans. The nanoparticles were biodegradable in artificial saliva after ten days and did not show any adverse effects on the surrounding oral tissues or on internal organs. Importantly, 16S rRNA analysis of the dental microbiota showed that the diversity and richness of bacterial species did not significantly change with nanoparticle treatment. Overall, this study presents a safe and effective, antibiotic-free approach to decrease the S. mutans population and dental biofilm formation without disrupting the ecological balance of the oral cavity.

BIOT 21

L cell stimulation: a novel strategy for oral peptide delivery in incretin- based diabetes treatment

Yining Xu1, [email protected], Matthias v. Hul2, Francesco Suriano2, Patrice D. Cani2, Véronique Préat1, Ana Beloqui1. (1) Advanced Drug Delivery and Biomaterials, Louvain Drug Research Institute (LDRI), DRAFT

Catholic University of Louvain, Brussels, Belgium (2) Metabolism and Nutrition Research group, Louvain Drug Research Institute (LDRI), Catholic University of Louvain, Brussels, Belgium

The delivery of therapeutic peptides via the oral route remains one of biggest challenges in the pharmaceutical industry. The current drug delivery systems act merely as a vehicle, and none of them has exploited their potential physiological effect. Enteroendocrine L cells have attracted particular interest because of the pleiotropic effects of their secreted peptides, such as glucagon-like peptide-1 (GLP-1). As therapeutic peptides, most GLP-1 analogs must be injected, which cannot perform all the GLP-1 physiological functions in type 2 diabetes mellitus (T2DM) patients. In this study, we exploited the physiology of L cells, developing and further modifying an innovative lipid-based nanosystem increasing the production of endogenous GLP-1 and the systemic absorption of a GLP-1 analog for T2DM treatment via oral route. We firstly discovered that empty 200 nm lipid nanocapsules could induce GLP-1 secretion from L cells mimicking the endogenous ligands that act physiologically in the body (Fig. 1A). Then, we developed a novel lipid-based nanosystem that synergizes its own biological effect with an increased GLP-1 analog bioavailability via oral route, which was found to be at least as efficient as the current marketed drug (subcutaneous) at treating T2DM (Fig. 1B). Bearing in mind that a foreseen clinical translation of the formulation might need prolonged GLP-1 plasma levels, we further provided the nanosystem with stealth properties (PEGylation) to increase the muco-penetration and to prolong the blood circulation. We were able to double the effect on endogenous GLP-1 secretion observed with plane nanocapsules. This modified nanosystem with increased GLP-1 secretion allowed us to decrease the administration frequency of drug-loaded formulations (Fig. 1C). In conclusion, this study maximizes the potential of the nanocarriers over current approaches in incretin-based diabetic treatment via oral route.

BIOT 22

Direct multiplex detection for nucleic acid biomarker profiles based on nucleic acid thermodynamics

Seung Won Shin, [email protected]. Chemical engineering, Sungkyunkwan University, Suwon, Gyeonggi-do, Korea (the Republic of)

Information on the quantitative distribution of biomarkers in tissues, cells or blood obtained from patients has accumulated through advanced detection DRAFT methods. Diagnostics based on multiplex biomarkers not only provide patients’ specific pathological information, but also enable personalized treatment. This valuable potential, based on bioinformatics, is attributable to advanced biomarker detection technology. However, the results produced from bioinformatics have been used as a final analysis tool to draw conclusions about the patient's condition or prognosis, and none have been used to mutually improve diagnostic techniques. Here we propose a new perspective for synthesizing optimized probe sequences that allow direct recognition of the profiles of multiplexed microRNAs. Probes, named as DIPS, were designed to achieve maximum hybridizations with multiple microRNAs simultaneously, and were optimized hybridization efficiency to specific expression profiles. the probe sequence was targeting the profile itself, rather than a single biomarker. The theoretically established nearest neighbor method and sequence alignment techniques were introduced for the probe- optimization genetic algorithm. Through iterative computation of genetic algorithms, random sequences have evolved into DIPS which is profile specific probe, which hybridizes target microRNAs profile. Based on computational analysis, we have defined a genetic algorithm-based probe retrieval process and confirmed that the cancer characteristics of patients can be classified in clinically reported databases. As far as we know, the concept of probe development for the entire biomarker profile has not been prop osed, and the accumulated clinical database is expected to contribute to the development of the promising diagnostic technology.

BIOT 23

Interpreting small-scale cell culture using noninvasive sensors for dissolved CO2, dissolved O2, and pH

Vida Rahmatnejad1, [email protected], Joel Tyson2, Xudong Ge3, Iordan V. Kostov4, Leah M. Tolosa5, Govind Rao6. (1) Chemical, biochemical and environmental engineering, University of Maryland Baltimore County (UMBC), Windsor Mill, Maryland, United States (2) UMBC, Baltimore, Maryland, United States (3) University of Maryland Baltimore County, Baltimore, Maryland, United States (5) Univ of Maryland, Baltimore, Maryland, United States (6) TRC Building, Ctr for Advanced Sensor Technology, Baltimore, Maryland, United States

Studies show that low levels of O2 or high levels of CO2 in cell culture can lead to inefficient cell growth. Furthermore, the pH value should be balanced and stable to achieve optimal results. Although the aforementioned factors play a major role, tissue engineering processes are currently conducted in single-use DRAFT vessels that are not equipped with systems monitoring them. As a result, tissue cultures are not usually studied from an analytical point of view, and this makes these studies less repeatable and reliable. The Center for Advanced Sensor Technology has developed noninvasive DCO2 sensor (featuring a sampler mounted outside the vessel), pH sensor and DO sensor (featuring a patch attached to the bottom of the vessel) to monitor cell culture environment. The low profile of the sensors makes them suitable for various kinds of bioreactors. In this study, cell cultures were conducted in standard flasks equipped with sensors for DO, pH, and DCO2, for continuous monitoring of process parameters. The results show that the use of these sensors improved understanding of the microenvironment. This ability to accurately measure the pH, CO2 and DO will be used as a measure of the cell viability as well as its metabolic activity.

BIOT 24

Ultra-sensitive platinum nanoparticle based digital assay toward point- of-care diagnostics

Hui Chen, [email protected], Zhao Li, Ping Wang. Pathology and Laboratory Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, United States

Quantitating ultra-low concentrations of protein biomarkers is critical for early disease diagnosis and treatment. However, most current point-of-care (POC) assays are limited in sensitivity to meet this clinical need. Herein, we introduce an ultra-sensitive and facile microbubbling digital assay readout method toward POC quantitation of protein biomarkers requiring only bright-field smartphone imaging. Picolitre-sized microwells together with platinum nanoparticle labels enable the discrete “visualization” of protein molecules via immobilized-microbubbling with smartphone. We also use computer vision and machine learning to develop an automated image analysis smartphone application to facilitate accurate and robust counting. Using this method, post- prostatectomy surveillance of prostate specific antigen (PSA) can be achieved with a detection limit of 2.1 fM (0.060 pg/mL), and early pregnancy detection using βhCG with a detection limit of 0.034 mIU/mL (2.84 pg/mL) . The results are further validated using clinical serum samples against clinical and research assays. This work provides the proof-of-principle of the microbubbling digital readout as an ultra-sensitive technology with minimal requirement for power and accessories, facilitating future POC applications.

BIOT 25 DRAFT

Use of accelerator mass spectrometry in the biomedical sciences

Bruce A. Buchholz1, [email protected], Graham Bench1, Ted Ognibene1, Gaby Loots2, Kelly A. Martin2, Edward Kuhn2, Heather A. Enright2, Daniel McCartt1, Jun Jiang1, Nicholas Hum2, Benjamin Stewart2, Michael A. Malfatti2, Kris Kulp2, Ken Turteltaub2. (1) Center for Accelerator Mass Spectrometry, Lawrence Livermore National Laboratory, Livermore, California, United States (2) Biosciences and Biotechnology Division, Lawrence Livermore National Laboratory, Livermore, California, United States

Accelerator mass spectrometry (AMS) was originally applied in the life sciences to overcome limitations in detection sensitivity for studying the molecular damage caused by exposure to low levels of environmental carcinogens and pollutants. For example, AMS can be used to conduct metabolite analysis at the picomole to the attomole level and is also being used to identify macromolecular targets for drugs and toxic compounds. Studies using 14C-labeled agents show that activities as low as a few nCi/person can be used to assess metabolism, and activities as low as 100 nCi/person can be used to address macromolecular binding in the study of candidate drugs or toxicants. This level of radioactive dose is less than that from a single day’s exposure to background ionizing radiation. In most cases the dose is less than that received during a cross-country commercial airline flight. The high sensitivity of AMS allows use of small samples of exfoliated tissues, isolated cell subpopulations, and precious tissues of human or animal origin. Sensitivity also enables quantitative study of ligand—macromolecule interactions at physiologically relevant concentrations, for studying effects such as hormones at low concentrations or where the receptor is present in low copy number, and for studying early events in the pathology of infection by labeling bacteria and viruses. The increased sensitivity also facilitates the use of compounds that are difficult to synthesize at high specific activity or cannot be used in large amounts. Examples from nutrition, pharmacology, toxicology, and cell biology will be described.

BIOT 26

Application of specifications for sterile drug products in-use stability: Industry perspective

Andrew Semple, [email protected], Jason Cheung, Douglas Watson, Robert Capen, Michel Chartrain, Monisha Dey, Rebecca Gentile, Sarita Mittal, Maya Salnikova, David Wylie. Merck & Co., Inc., Doylestown, Pennsylvania, United States DRAFT

There is relatively little in the published literature regarding industry best practices and regulatory expectations for product quality during the in-use period. There is also relatively little explicit regulatory guidance, including ICH and worldwide market-specific guidance, other than the recommendation that such stability studies must be performed. This presentation describes two strategies for controlling product quality during the inuse period, both of which are consistent with available regulatory guidance. The simplest and most straightforward approach is to set the same acceptance criteria for the in-use period as the stability specifications applied to the long-term storage condition. When this is not feasible, a different strategy can be considered whereby the in-use period is established based on alternate acceptance criteria that have been appropriately justified. For a particular program, the preferred approach is therefore based on considerations for the program, scientific rationale for specification setting, and engagement with regulators. A case study will be discussed that describes the stability studies that were performed, the data generated, and the statistical methodologies applied to support an in-use limit for a multi-use biologic drug product. For this study, the second strategy was selected due to the existence of product-specific USP and Ph Eur monographs that provided specifications for long-term storage. However, the application of either approach may be scientifically justifiable, and industry should consider both when evaluating product in-use stability limits.

BIOT 27

Elucidating the unusual reaction mechanism of glucuronyl C5- epimerase

Deepika Vaidyanathan1, [email protected], Xia Ke2, Yanlei Yu3, Robert J. Linhardt4, Jonathan S. Dordick5. (1) Chemical and Biological Engineering, Rensselaer Polytechnic Institute, Troy, New York, United States (2) Department of Chemistry and Chemical Biology, Rensselaer Polytechnic Institute, Troy, New York, United States (4) Chemistry, Rensselaer Polytechnic Institute, Albany, New York, United States (5) Center for Biotechnology, Renselaer Polytechnic Inst, Troy, New York, United States

Chemoenzymatic synthesis of heparin represents an emerging approach to provide a substitute for animal sourced heparan sulfate and heparin. D- Glucuronyl C5-epimerase is the first enzyme that acts on the heparin precursor, catalyzing the conversion of Dglucuronic acid (GlcA) to L-iduronic acid (IdoA). The absence of reliable assays for C5- epimerase has made understanding its reaction mechanism and its application in heparin and heparan sulfate biosynthesis quite difficult. The steady-state kinetic DRAFT parameters, for both the forward and the pseudo-reverse reactions of C5- epimerase have been determined for the first-time using full-length polysaccharide substrates directly relevant to the chemoenzymatic synthesis and biosynthesis of heparin. The forward reaction shows unusual sigmoidal Michaelis-Menten behavior and the pseudoreverse reaction displays non- saturating kinetic behavior. The atypical sigmoidal behavior of the forward reaction was probed using a range of additives. Surprisingly, the combination of 25 mM CaCl2 and 25 mM MgCl2 results in a forward reaction exhibiting conventional Michaelis-Menten kinetics. The addition of 2-O-sulfotransferase, the next enzyme involved in heparin/heparan sulfate synthesis, in the absence of cofactor 3’- phosphoadenosine 5’-phosphosulfate (PAP) also resulted in C5-epimerase exhibiting a conventional Michaelis-Menten kinetics in the forward reaction accompanied by a significant increase in Vmax. The activity and processivity of C5-epimerase as a function of polysaccharide sequence was also evaluated. The epimerase reaction is performed in deuterated environments leading to incorporation of a deuterium on the IdoA residues at the C5 position. LC-MS top down analysis is used to quantify activity based on the epimerase mediated deuterium incorporation and to map the distribution of GlcNS and GlcNAc within the polysaccharide and providing information on the activity and specificity of the enzyme as a function of the sequence of the polysaccharide. The data obtained in this study provides critical information towards understanding the mechanism of C5-epimerase and improving the chemoenzymatic synthesis of bioengineered heparin specifically accounting for variability in the GlcNS and GlcNAc content.

BIOT 28

Identification, characterization, and structural insights into the γ- lactamase activity and substrate enantioselectivity of a newly identified hydrolase from Microbacterium hydrocarbonoxydans

Shuaihua Gao, [email protected], Guojun Zheng. Beijing University of Chemical Technology, Beijing, China

2-Azabicyclo[2.2.1]hept-5-en-3-one (γ-lactam) is an important precursor of many carbocyclic nucleoside analogs and pharmaceuticals such as carbovir and abacavir. (+)-γ-Lactamase catalyzes the specific hydrolysis of (+)-γ- lactam out of the racemic γ-lactam (2-Azabicyclo[2.2.1]hept-5-en-3-one) to leave optically pure (−)-γ-lactam. In this study, a novel (+)-γ-lactamase was discovered from the gene library of Microbacterium hydrocarbonoxydans based on a colorimetric high-throughput screening method and it could be used to enantioselectively catalyze the bioresolution of DRAFT racemic γ-lactam with high enantiomeric excess (ee) (>99 %) and yield (>49 %). It displayed much higher (+)-γ-lactamase activity than any other biochemically characterized (+)-γ-lactamases. To date, no structural data has been reported on how the enzymes bind the γ-lactams and achieve their enantioselectivities. Here, we first discovered that this enzyme actually hydrolyzed both (+)- and (−)-γ-lactam, but with apparently different specificities. We determined the crystal structures of the apo-form, (+)-γ- lactam bound, and (−)-γ-lactam bound forms of the enzyme. Our results would facilitate the directed evolution and application of it in antiviral drug synthesis.

BIOT 29

Production of oxidized terpenes using P450 enzymes and their engineered enzyme fusions

Taek Soon Lee1,2, [email protected], Xi Wang1,2. (1) Joint BioEnergy Institute, Emeryville, California, United States (2) Lawrence Berkeley National Laboratory, Berkeley, California, United States

Terpenes are a large class of organic compounds, primarily produced by plants and constitute the main components of essential oils. Functionalization of the terpene carbon backbone using cytochrome P450 enzymes could derive more useful compounds that can be converted to higher value products. In this study, cytochrome P450 enzymes were investigated for oxidizing monoterpene (e.g. limonene, 1,8-cineole), a few oxidized terpenes were produced, including carveol, carvone, limonene 1,2-epoxide, 2-endo- hydroxy-1,8-cineole. Based on those oxidized terpenes, possible biosynthetic pathways were explored toward making biology-derived functionalized terpenes, such as monomer of the polymer.

On the other hand, terpenes are hydrophobic and volatile compounds that limit the accessibility of P450 enzymes. To optimize the production of oxidized terpenes, we developed an enzyme fusion strategy with linking terpene synthase and P450 enzyme. In an example of producing 2-endo-hydroxy-1,8- cineole, the engineered fusion showed higher efficiency from both in vitro and in vivo production results. Thus, engineering a fusion between terpene synthase and P450 presents a feasible strategy for producing oxidized terpenes, suggesting broad application during the production of terpene- based bioproducts.

BIOT 30 DRAFT

Simple, rapidly assembled saccharide-based sensor interface that enables interrogation of antibody glycosylation and titer for improved process analytical technologies

Dana Motabar1,2, [email protected], Jinyang Li1,2, Sally Wang3,2, Chen- Yu Tsao1,2, Gregory Payne2, William E. Bentley4,2,5. (1) Bioengineering , University of Maryland College Park, Gaithersburg, Maryland, United States (2) Institute for Bioscience and Biotechnology Research, University of Maryland, College Park, Maryland, United States (3) Bioengineering, University of Maryland, College Park, Maryland, United States (4) Fischell Dept of Bioengineering, University of Maryland, College Park, Maryland, United States (5) Robert E. Fischell Institute for Biomedical Devices, University of Maryland, College Park, College Park, Maryland, United States

Recombinant monoclonal antibody (mAb) therapeutics constitute a large segment of the biologics market as they have shown success in many therapeutic areas. N-linked glycosylation on the Fc- portion of mAbs is a common post-translational modification that can have a significant impact on the mAbs therapeutic efficacy, in vivo half-life, and safety. As a result, N- linked glycan composition is considered a critical quality attribute and the glycoform profile is required for mAb characterization. Current methodologies to assess glycosylation involve complex techniques that couple extensive separation methods with mass spectrometry or nuclear magnetic resonance spectroscopy. Although these techniques provide detailed structural information about the mAb’s glycosylation state, they can be costly and require extensive time and significant expertise. For process screening and monitoring (e.g., bioreactor and downstream process vessels), such complex structural information may not be required. Instead, analytical technologies that indicate levels of product and it’s glycosylation state may be sufficient, particularly when coupled with data analytic methods that provide discrimination and model-based control schemes. To meet this measurement need, we developed an electrochemical based interface that is rapidly electroassembled onto electrodes and when linked with glycan-recognizing lectins reveals levels of mAbs with the appropriate terminal glycan moieties. Further, when coupled with a protein G:HRP reporter this interface enables process control ready electrochemical output. Further, a new thiolated protein- G construct enables total mAb titre. That is, our interface is rapidly constructed using electrode actuated redox reactions enabling hydrogel assembly from thiolated-polyethylene glycol (PEG. Then, thiolated sugar groups (e.g., galactose) and sugar-specific lectins are layered on top of the PEG hydrogel. These surfaces then provide selective capture of mAbs based DRAFT on the traditional lectin-mAb binding. Our glycosensing approach is advantageous as it is simple, portable, inexpensive and can provide fast, quantifiable results for bioprocessing applications.

BIOT 31

Conformational dynamics due to cofactor binding courtney A. Clark2, Vincent DelFratte2, Erica L. Whittenberger2, J Rajendran Pandian4, Duilio Cascio3, GK Balendiran1, [email protected]. (1) Youngstown State University, Youngstown, Ohio, United States (2) Chemistry, Youngstown State University, Warren, Ohio, United States (3) UCLA-DOE, LOS ANGELES, California, United States (4) 2Department of Business Administration, College of Business Administration, Gulf University for Science and Technology, Kuwait, Kuwait

Aldose Reductase (AR) catalyzed reduction of glucose reaction that requires NADPH as the cofactor follows a sequential ordered kinetic mechanism. The catalytically active form of the enzyme.NADPH complex which needs to be assembled first preceding to the aldehyde substrate binding. The conformational change involving a hinge-like movement of residues in flexible loop is anticipated to take place upon the binding of the diphosphate moiety of NADPH in the enzyme.NADPH complex development step. The reorientation of this loop is expected to permit the release of NADP+ after the catalysis. The Translation, Libration, Screw analysis of apo AR crystal structure indicates that some of the residues in the flexible loop might move as a rigid group.

BIOT 32

DOZN™2.0 - a quantitative Green Chemistry evaluator

Ettigounder Ponnusamy, [email protected]. MilliporeSigma, Ballwin, Missouri, United States

MilliporeSigma created a unique web-based greener alternative scoring matrix, also known as DOZN™2.0 - A Quantitative Green Chemistry Evaluator based on the 12 principles of green chemistry. The 12 principles of green chemistry provide a framework for learning about green chemistry and designing or improving materials, products, processes and systems. DOZN™2.0 scores products based on metrics for each principle and aggregates the principle scores to derive a final aggregate score. The system DRAFT calculates scores based on manufacturing inputs, Globally Harmonized System (GHS) and Safety Data Sheet (SDS) information which provide a green score for each substance and is flexible enough to encompass a diverse portfolio of products ranging from chemistry to biology based products. The DOZN™2.0 system has also been verified and validated by a third party to ensure best practices are applied and, also published. Users can evaluate their processes and products in a safe and secure manner (data privacy). This new Greener Chemistry initiative offer customers’ an increased breadth of Greener Alternative products with confirmatory documentations to validate greener characteristics. Through MilliporeSigma’s DOZN™2.0 platform, customers can now calculate the green scores for their own processes and products. This free, web-based tool provides users with even more data so that they have more information to increase their sustainability.

BIOT 33

Rapid In silico screening strategy: development of QSAR model for antibody solubility

Xuan Han1, [email protected], Steven M. Cramer2, Qing Chai3. (1) chemical engineering, Rensselaer Polytechnic Institute, Troy, New York, United States (2) Ricketts Bldg, Rensselaer Polytechnic Inst, Troy, New York, United States (3) Eli Lilly, San diego, California, United States

Monoclonal antibodies(mAbs), being one of fastest growing classes of therapeutics, can suffer from poor solubility especially when high concentration formulation is needed for subcutaneous injection. In the early development stage, candidates screening based on experimental characterizations are both material and labor intensive. To facilitate the process, we developed a in silico predictive model using amino acid sequence as input to help researchers determine solubility level of molecules prior to experiments. A large pool of antibodies was used here for training the model and solubility data was acquired using PEG precipitation by high-throughput screening technique (HTS). Refined homology models of these mAbs had been built by taking protein dynamics into account. Molecular descriptors that depict surface properties of antibodies were calculated for each mAbs according to their sequences as well as 3D structures. Utilizing these descriptors as input, a classification model was developed to help distinguish mAbs that showed desirable solubility from those with poor behavior. Besides, we also evaluated our model using external published dataset for further validation. In all, this work not only provides a strategy for antibodies early screening based on solubility, but also sheds the light into the relationship DRAFT between protein surface properties and solubility where complicated protein- protein interaction is involved, giving important guidance for designing mAbs of better manufacturability.

BIOT 34

Development of targeted chemically self-assembling nanorings (CSANs) for cell-directed immunotherapy

Ellie Hofer, [email protected], Justine Delgado, Ozgun Kiliç, Carston R. Wagner. Medicinal Chemistry, University of Minnesota, Fridley, Minnesota, United States

Engineering cell-cell interactions has proven to be quite valuable due to the vast number of therapeutic applications that benefit from this technology. Although genetically engineering artificial receptors onto a patient’s cells has shown some success, preparation is costly, current applications are limited, and modifications are permanent. To address some of these concerns our lab has developed a non-genetic approach to facilitate selective cell-cell interactions using various targeted protein scaffolds. To date, we have employed commonly used targeting elements such as short peptides and single chain variable fragments (scfv), and alternative binding scaffolds such as affibodies and fibronectins that are engineered to have specificity for overexpressed cancer antigens. The diversity of the CSAN construct can be expanded by fusing two scaffolds onto the same ring that target different cellular antigens, creating a bispecific nanoring. Our lab has shown that functionalizing the CSAN construct with a cancer antigen-targeting scaffold and a T-cell targeting single chain antibody fragment (anti-CD3 scfv) forms a mixture of bispecific nanorings with the ability to traffic T-cells to the tumor. We are currently able to produce CSANs that facilitate T-cell interactions with: a) multiple solid tumor antigens such as EpCAM, EGFR, etc. b) multiple blood tumor antigens such as CD19, CD22, etc. c) cancer-initiating cell associated antigens such as CD133 and B7H3. Our rings have been shown to be stable in vitro and in vivo and can be used to deliver drugs or nucleic acids as well as for fluorescence and PET imaging. Finally, a distinct advantage of our construct over other bispecific immunotherapies is that we are able to disassemble our rings using the FDA approved drug trimethoprim, stopping the course of treatment.

BIOT 35 DRAFT

Impact of ammonia stress on glycosylation of the VRC01 monoclonal antibody produced in CHO cells

Claire McGraw1, [email protected], Kathryn Elliott3, Sarah W. Harcum4, Nicholas R. Sandoval2. (1) Chemical and Biomolecular Engineering, Tulane University, New Orleans, Louisiana, United States (3) Bioengineering, Clemson University, Clemson, South Carolina, United States

Glycosylation is a critical post-translational modification of proteins in mammalian cells. Glycosylation of monoclonal antibodies (mAbs) is of particular interest due to the effect it has on the effector function, structure and stability of the mAb, thus making it an important product quality attribute. To achieve a desired therapeutic effect, it is necessary to produce an antibody with a specific glycosylation profile. To produce this antibody, a better understanding of how changes to culturing conditions impact the glycoforms of mAbs is required. In this work we detect significant changes to glycosylation following ammonia stress of CHO produced VRC01- a broadly neutralizing antibody undergoing clinical trials to treat HIV. Ambr250 bioreactors were run in parallel with increasing concentrations of ammonia stress (0, 10 or 30 mM) on both young and aged CHO cells, with high and low gas flow rates. The determine the glycan profiles, the culture supernatant was collected over the course of the run and VRC01 was purified using gravity flow chromatography with protein A resin. Glycans were fluorescently labeled and released from the mAb and eluted from a BEH Amide column using a Waters AQUITY H-class UPLC. We found that increased ammonia concentrations measured in the culture for all conditions tested led to a decrease in total and terminal galactosylation, as well as a decrease in sialylation of the associated glycans. This observed reduction in sialylation of glycoproteins following ammonia stress is consistent with previously published results [1-3]. Future work will look into how the conserved Asn297 Fc glycan is impacted compared to Fab glycans under various culturing conditions

BIOT 36

Design and characterization of bispecific antibodies for in vivo imaging of pathological tau

Ghasidit Pornnoppadol1, [email protected], Alec A. Desai2, Charles G. Starr3, Peter M. Tessier4. (1) Pharmaceutical Sciences, University of Michigan, Ann Arbor, Michigan, United States (2) Chemical Engineering, University of Michigan, Ann Arbor, Michigan, United States (3) Pharmaceutical Sciences and Chemical Engineering, University of Michigan, Ann Arbor, DRAFT

Michigan, United States (4) Chemical Engineering and Pharmaceutical Sciences, University of Michigan, Ann Arbor, Michigan, United States

Early detection of pathological tau is vital for proper diagnosis of tauopathies including Alzheimer’s disease (AD). In vivo imaging would require imaging agents that can effectively cross the blood-brain barrier and specifically target pathological tau. Tau undergoes hyper-phosphorylation which leads to aggregation and formation of neurofibrillary tangles, a hallmark of AD. Although tau proteins have multiple phosphorylation sites, only a subset of them are exclusive to AD. While antibodies specific to phosphorylated tau proteins have been identified, little is known about their ability to cross the blood-brain barrier (BBB). Biologics such as antibodies have been shown to cross the BBB via receptor-mediated transcytosis. Transferrin receptors are the most highly expressed receptors on the BBB that have been exploited for delivery of large molecules. Few antibodies that target these receptors and facilitate receptor-mediated transcytosis have been identified previously. We are currently testing bispecific combinations of anti-tau and anti-transferrin antibodies in multiple formats for their ability to retain binding to their respective targets and mediate transcytosis. Our goal is to develop bispecific antibodies that can efficiently reach the brain and can be used for in vivo imaging pathological tau.

BIOT 37

Metabolic engineering challenges to extending N-glycans in CHO cells

Qiong Wang1,2, [email protected], John F. Cipollo1, Michael J. Betenbaugh2. (1) CBER/DBPAP, FDA, Bethesda, Maryland, United States (2) Department of Chemical and Biomolecular engineering, Johns Hopkins University, Baltimore, Maryland, United States

In mammalian cells, N-glycans are capped with sialic acids with one to three repeated N-acetyllactosamine units (LacNAc). LacNAc involved in various cellular functions in differentiation, metastasis, immune response and tumorigenesis. Moreover, LacNAc also affect therapeutic protein’s enzymatic activity and circulatory half-life. Although ploy-LacNAc is of significance for various cellular functions and the properties of therapeutic recombinant proteins, limited understanding in the literature regarding to the biosynthesis and influence of poly-LacNAc on N-glycans is available.Here we employed multiple genetic strategies including glyco-gene knockouts and knockins to achieve multi-dimensional modification of the glycosylaiton in CHO cells. Especially, previous studies indicated that β-1,3-N- DRAFT acetylglucosaminyltransferase 2 (B3GNT2) and β-1,4-galactotransferase 1 (B4GALT1) are two of the primary glycosyltransferases involved in generating LacNAc units. In the current study, knocking out sialyltransferase genes slightly enhanced the LacNAc content (≥4 repeats per glycan) on recombinant EPO protein. Next, the role of single and dual-overexpression of B3GNT2 and B4GALT1 was explored in recombinant EPO-expressing CHO cells. While overexpression of B4GALT1 slightly enhanced the levels of large glycans on recombinant EPO, overexpression of B3GNT2 in EPO-expressing CHO cells significantly decreased the recombinant EPO LacNAc content, resulting in N- glycans terminating primarily with GlcNAc structures, a limited number of Gals, and nearly undetectable sialylation, which was also observed in sialyltransferases knock-out cell lines. Furthermore, B3GNT2 overexpression enhanced intracellular UDP-GlcNAc and CMP-Neu5Ac content while slightly lowering UDP-Gal content. Furthermore, we were unable to overexpress B3GNT2 at either the transcriptional or translational levels following initial B4GALT1expression. Expression of B3GNT2 following initial expression of B4GALT1 was also problematic in that transcriptional and translational analysis indicated the accumulation of truncated B3GNT2 missing a section of the B3GNT2 trans-Golgi lumen domain while transmembrane and cytoplasmic domains were present.The addition of one or more recombinant glycosyltransferase genes may have an unexpected influence on the expression and activities of glycosyltransferases, which can disrupt the nucleotide sugar levels and lead to unexpected modifications of the resulting N-glycan patterns.

BIOT 38

High-throughput evaluation of antibody developability using flow cytometry

Emily K. Makowski1, [email protected], Lina Wu2, Alec A. Desai2, Peter M. Tessier1,2. (1) Pharmaceutical Sciences, University of Michigan, Ann Arbor, Michigan, United States (2) Chemical Engineering, University of Michigan, Ann Arbor, Michigan, United States

There are many chemical and physical requirements that must be met for antibodies to be suitable for use as therapeutics. One of the most important requirements for therapeutic antibodies is high specificity for their target antigens because off-target binding is correlated with rapid antibody clearance rates and suboptimal pharmacokinetic profiles. Therefore, we have sought to develop high-throughput methods for rapidly evaluating antibody specificity using minimal amounts of purified protein and widely accessible experimental DRAFT methods. Our flow cytometry approach – which uses multiple types of polyspecificity reagents – is well correlated with other methods for evaluating antibody specificity as well as with assays used for identifying antibodies with abnormal clearance rates in vivo. We expect that these results will improve the rapid evaluation of key antibody biophysical properties linked to their developability and thereby improve the drug development process.

BIOT 39

Machine learning techniques in the chemical information for drug discovery

Yu-Chen Lo2, Gui Ren3, Hiroshi Honda1, [email protected], Russ B. Altman2. (1) NPU, Fremont, California, United States (2) Bioengineering, Stanford University, Stanford, California, United States (3) School of Economics, Shanghai University, Shanghai, China

Demonstrate the use of machine learning techniques in the chemical information from large compound databases to design drugs with biological properties

BIOT 40

Rational design of an amylosucrase — cyclodextrin glucanotransferase fusion to produce cyclodextrins in a unique reaction mixture

Leidy R. Pico1, [email protected], Hermínsul Cano Calle2, Jorge Hernández Torres1. (1) Biology, Universidad Industrial de Santander, Bucaramanga, Santander, Colombia (2) Chemistry, Universidad Industrial de Santander, Bucaramanga, Colombia

Cyclodextrins (CDs) are cyclic oligosaccharides composed of different numbers of α-(1,4)-linked glucopyranose subunits. CDs are commercially produced from starch, through the action of cyclodextrin glucanotransferase (CGTase), which generally catalyses the cyclization of starch and related α- (1,4)-glucans via intramolecular transglycosylation reactions. A recent work developed a novel enzymatic process, utilizing sucrose as raw material instead of corn starch, to reduce time and energy consumption of CDs production at industrial level. In this process, a glucosyltransferase named amylosucrase (AS) catalyses the synthesis of α-(1,4)-glucan polymers from sucrose molecules, opening the possibility of exploring other commercially available sugar-rich substrates. The aim of this project is to test the combined DRAFT action of AS from Neisseria polysaccharea fused to the CGTase from Bacillus macerans, to produce CDs from cocoa mucilage, a substrate rich in sucrose and glucose. The first step involved the use of bioinformatics tools for the design and analysis of the AS-CGTase fusion protein. Wild type Protein Data Bank (PDB) structures were linked with tandem repeats of GGGS and GGGGS linker motifs, by using the Synlinker server. As a result, 3D models of AS-CGTase and CGTase-AS fusions were obtained, which were analyzed with bioinformatic tools to select the most stable ones for laboratory testing. Model selection was based on molecular dynamics simulations (GROMACS), Root Mean Square Deviation (RMSD) and Gyration Radius (RD) values as well as other structural validation parameters. The model with the highest stability was the AS-G3S-CGTase fusion, linked by GGGS motifs, with a RMSD value of 0.31 nm and a RD of 3.75 nm. Currently, synthetic genes are being constructed for the AS-CGTase and CGTase-AS fusions, for cloning into pET vectors and overexpression in Escherichia coli. A one-pot dual enzyme reaction will be implemented using the fused proteins with cocoa mucilage as a sugar-rich substrate.

BIOT 41

Developing a genetic engineering toolbox for anaerobic fungi

Ethan T. Hillman, [email protected], Casey Hooker, Jake Englaender, Kevin Solomon. Agricultural & Biological Engineering, Purdue University, West Lafayette, Indiana, United States

Anaerobic fungi are potentially powerful platforms for biotechnology that remain unexploited due to a lack of genetic tools. These fungi have the largest repertoire of CAZymes for degradation of lignocellulose, making them particularly attractive for biofuel production, and are predicted to possess novel biosynthetic capabilities. However, a genetic toolbox that would enable further study and exploitation of these capabilities for synthetic biology remains underdeveloped. No plasmids, promoters, terminators, selection markers, reporter genes, or gene-editing systems have been characterized. Here, we present our efforts to develop this toolkit. Anaerobic fungi are naturally competent during phases of their lifecycle, which we have exploited to introduce and transiently express heterologous genes. We have leveraged existing transcriptomic resources to identify putative promoters and terminators that regulate gene expression at different that appear conserved across 6 anaerobic fungal isolates. Using our transformation protocols, we have confirmed the ability of at least one promoter candidate to promote gene expression of a flavin-based fluorescent reporter and developed a pipeline to DRAFT characterize candidate gene regulatory elements. Similar experiments have also successfully introduced antibiotic selection systems, which we will extend to express alternative selection markers, fluorescent reporter genes, and CRISPR-based endonucleases. Other elements such as origins of replication and centromeric binding sequences are being developed in parallel to create a stable, plasmid-based expression system in anaerobic fungi. Ultimately, this toolbox will allow us to express genes of non-native pathways, selection markers, and engineering systems, which will unlock a variety of genome engineering possibilities for the study of fungal genetics and strain development for various synthetic biology applications.

BIOT 42

Ammonia stress triggers genome instability in CHO cells

Dylan Chitwood1, [email protected], Qinghua Wang2, Kathryn Elliott3, Aiyana Bullock4, Dwon Jordana5, Zhigang Li7, Cathy Wu2, Sarah W. Harcum6, Christopher Saski7. (1) Bioengineering, Clemson University, Seneca, South Carolina, United States (2) Center for Bioinformatics & Computational Biology, University of Delaware, Newark, Delaware, United States (3) Bioengineering, Clemson University, Clemson, South Carolina, United States (4) Department of Biological Sciences, College of Agriculture, Science & Technology, Delaware State University, Dover, Delaware, United States (5) Department of Biological Sciences, Grambling State University, Grambling, Louisiana, United States (7) Department of Plant and Environmental Sciences, Clemson University, Clemson, South Carolina, United States

As bioprocess intensification has increased over the last 30 years, yields from mammalian cell processes have increased from 10’s of milligrams to over 10’s of grams per liter. Most of these gains in productivity have been due to increasing cell numbers in the bioreactors, and with those increases in cell numbers, strategies have been developed to minimize metabolite waste accumulation, such as lactate and ammonia. Unfortunately, cell growth cannot occur without some waste metabolite accumulation, as central metabolism is required to produce the biopharmaceutical. Inevitably, metabolic waste accumulation leads to decline and termination of the culture. While it is understood that the accumulation of these unwanted compounds imparts a less than optimal culture environment, little is known about the genotoxic properties and the influence of these compounds on global genome instability. In this study, we examined the genotypic effects on Chinese hamster ovary (CHO) cells due to exposure to elevated ammonia levels. We identified genome-wide de novo mutations, in addition to variants in functional regions DRAFT of certain genes involved maintaining genome stability, such as RPA1, BRCA1, POLE and MLH3, which led to loss-of-function and eventual genome instability. Critical MMR genes were observed to be impaired, at the same time as major metabolism shifts were observed. Pathway analysis of impacted genes revealed high enrichment in pathways corresponding to ammonia metabolism as well as genome stability. Additionally, we characterized the presence of microsatellites against the CHO PICR assembly and discovered certain loci are not replicated faithfully the presence of elevated ammonia, which leads to microsatellite instability (MSI). We examined these microsatellite loci in order to identify those with dose-dependent responses to elevated ammonia that may be suitable biomarkers to gauge genome stability in CHO cultures.

BIOT 43

Multi-target ligand-guided selection (LIGS) to generate aptamers against B-cell biomarkers

Nicole Williams1, [email protected], Sana Batool2, Mohammad Jamal2, Gernan Sosa2, Hasan Ekrem Zumrut3, Prabodhika Mallikaratchy1,2,3. (1) PhD Program in Biology, Graduate Center, CUNY, New York, New York, United States (2) Chemistry, Lehman College, CUNY, Bronx, New York, United States (3) PhD Program in Biochemistry, Graduate Center, CUNY, New York, New York, United States

Nucleic Acid Aptamers (NAAs), are single-stranded ribo- and deoxyribo- oligonucleotide molecules, which fold into complex functional three- dimensional structures. These three-dimensional structures have the ability to bind to target molecules with high affinity and specificity. Nucleic Acid Aptamers are selected through an iterative process called Systematic Evolution of Ligands by Exponential enrichment (SELEX). Here, we implemented Ligand-Guided Selection (LIGS), a variant of SELEX, to elute aptamers against multiple biomarkers expressed in B-cells i.e., CD19 and CD20. During LIGS step, specific aptamers against CD19 and CD20 were competitively eluted using two specific monoclonal antibodies against the same target at 25°C. The eluted multiple libraries were then sequenced using Illumina high-throughput (HT) DNA sequencing platform. The resulting sequences were analyzed using a novel bioinformatics workflow, which was designed using GALAXY, an online bioinformatics server. Two major sequence families consisting of 18 hit sequences against CD19 and 34 hit sequences against CD20 were identified based on defined enrichment values. The potential hit sequences were subsequently synthesized and screened DRAFT against biomarker positive and negative cell lines to identify aptamers candidates against respective markers. Collectively, this study establishes LIGS as a state-of-the-art screening technology that can be used to generate highly specific aptamers against multiple receptor-proteins in their native state expressed on one cell population.

BIOT 44

Development of thermostable carbonic anhydrases using structure- guided recombination for use in CO2 removal systems on spacecraft

Michael Dougherty1, [email protected], Justine Richardson2, Darrell Jan2, Grace Belancik2, Jonathan Galazka2. (1) KBR, Inc., Moffett Field, California, United States (2) NASA Ames Research Center, Moffett Field, California, United States

Carbon capture and storage has been a research area of great interest in recent years as a method for mitigation of CO2 emissions, due to the effects of climate change. The development of technologies for the efficient capture of CO2 are also of great interest for human spaceflight applications. One of the most promising technologies in this area is CO2 scrubbing using liquid amines, unfortunately, liquid amines with low heats of desorption tend to have slow CO2 binding kinetics. One potential solution to this problem is to use the enzyme carbonic anhydrase (CA) to enhance the kinetics of CO2 binding to liquid amines, allowing the overall process to be more energy efficient. Interest in using carbonic anhydrase as a biocatalyst has led to a number of efforts to improve the thermostability and solvent tolerance of several distinct carbonic anhydrase enzymes. In the work described here, we screened through a diverse set of natural carbonic anhydrases to identify candidates for protein engineering aimed at increased stability and activity in various liquid amines. We then used SCHEMA structure-guided recombination to develop a set of chimeric carbonic anhydrases with high thermostability and activity. These chimeras were used as the starting points for further protein engineering work targeting activity in liquid amine systems. Our ultimate goal is to test the engineered enzymes in a liquid amine system for cabin air revitalization on ISS or other spacecraft.

BIOT 45

Engineering the secretory pathway for increased enzyme production in Yarrowia lipolytica DRAFT

Weigao Wang1, [email protected], Mark A. Blenner2. (1) Chemical and Biomolecular Engineering , Clemson University, Clemson, South Carolina, United States (2) Chemical Biomolecular Engineering, Clemson University, Clemson, South Carolina, United States

Eukaryotic cells cultures are preferred for the production of complex enzymes and biopharmaceuticals due to their ability to form post-translational modifications and the quality control system inherent to the quality control system of the endoplasmic reticulum. A non-conventional yeast species, Yarrowia lipolytica, has attracted attention due to its native protein secretion capacity and advanced secretory pathway. Conventional means of improving protein secretion includes codon optimization, amplification of the gene copy number, inducible expression, and secretory tag engineering; however, improvements to the secretory pathway are less common. In our study, secretion of the model enzyme T4 lysozyme was used to determine effective strategies for increasing enzyme secretion. By engineering a native secretion signal lip2prepro, we have successfully improved extracellular level of T4 lysozyme by 50%. Such a improvement in the secretion was also found in other heterologous proteins including hrGFP and -amylase. Further improvement in T4 lysozyme secretion was achieved by co-expression of important enzymes in the secretion tag processing pathway and expansion of ER.

BIOT 46

Bioinspired design of adeno-associated viral capsids with expanded sizes

Xiaozhe Ding, [email protected], Viviana Gradinaru. California Institute of Technology, Pasadena, California, United States

Adeno-associated viral (AAV) capsid is emerging as a powerful gene delivery vehicle. However, the 25-nm icosahedral nanocage’s conserved small size imposes a restrictive cargo capacity. Inspired by the geometry and assembly mechanisms observed in larger viral capsids, we report two design strategies – tandem multimerization and guide peptide grafting – to genetically alter capsid size. The two strategies result in the formation of eXtra Large AAV capsids (XL-AAVs) with 35-60 nm diameters and the ability to package recombinant AAV genomes. These design principles can be applied to multiple AAV serotypes. The design of XL-AAVs may open the door to AAV vectors with expanded cargo capacity and provide an exemplar for the rational expansion of protein nanocages. DRAFT

BIOT 47

Specificity engineering of clinical-stage antibodies

Lina Wu1, [email protected], Alec A. Desai2, Yulei Zhang3, Matthew D. Smith4, Peter M. Tessier5. (1) Chemical Engineering, University of Michigan, Ann Arbor, Michigan, United States (5) Chemical Engineering and Pharmaceutical Sciences, University of Michigan, Ann Arbor, Michigan, United States

The ability of therapeutic antibodies to recognize their target antigens with high specificity in the complex in vivo environment is crucial to their function as safe and efficacious drugs. However, therapeutic antibodies, especially those discovered or engineered using in vitro methods such as phage or yeast surface display, often display variable levels of non-specific interactions that lead to drug development challenges such as low bioavailability and fast antibody clearance. Therefore, we have developed methods for predicting sites in antibody variable regions linked to off-target binding and for designing antibody libraries that softly randomize such sites with residues predicted to improve antibody specificity. We have evaluated this approach for improving the specificity of multiple clinical-stage antibodies that display high levels of off-target binding. To accomplish this, we have displayed our designed antibody libraries on the surface of yeast, sorted them based on their affinity for multiple types of reagents (including their target antigens and several polyspecificity reagents), and used deep sequencing methods to identify rare variants that retain high target affinity and display increased specificity. Importantly, we find that a subset of antibody sites that mediate non-specific binding do not overlap with those involved in antigen binding. We will discuss our identification of such sites and how we are exploiting them to engineer antibodies with drug-like specificity.

BIOT 48

Enrichment of trace pollen-free DNA for next-generation sequencing to determine honey origins

Dimple Chavan1, [email protected], Jay R. Adolacion2,3, Mary Crum2, Suman Nandy2, Binh V. Vu2, Katerina Kourentzi2, Aniko Sabo5, Richard C. Willson1,2,4. (1) Department of Biology and Biochemistry , University of Houston, Houston, Texas, United States (2) Department of Chemical and DRAFT

Biomolecular Engineering, University of Houston, Houston, Texas, United States (3) Department of Chemical Engineering, University of the Philippines, Diliman, Quezon City, Philippines (4) Escuela de Medicina y Ciencias de la Salud ITESM, Monterrey, Mexico (5) Human Genome Sequencing Center, Baylor College of Medicine, Houston, Texas, United States

A large fraction of commercial honey is misidentified as to its origins. Pollen microscopy, the current gold standard for identifying the geographical origins of honey, is laborious and slow and requires extensive training. In addition, filtered honey cannot be identified by pollen examination and can be spiked with pollen from a more favorable plant to disguise its origins. We targeted the nuclear ribosomal ITS2 region of plant DNA, which is known to support species-level discrimination. We have successfully derived ITS2 sequences from pollen-free, filtered honey. We purified pollen-free DNA from filtered honey using three different methods: (i) anti-dsDNA antibodies coupled to magnetic nanoparticles; (ii) batch adsorption on Q Sepharose anion exchanger; and (iii) batch adsorption on ceramic hydroxyapatite, CHT Type I. The ITS2 region of the captured pollen-free DNA was then sequenced using next-generation sequencing. Enrichment of trace pollen-free DNA from filtered honey samples opens a new approach to identify the true origins of filtered honey samples.

BIOT 49

Incorporation of bio-inspired polymeric coatings for enhanced Schwann cell proliferation and stem cell differentiation

Jesse Roberts1, [email protected], Shannon L. Servoss2, Jorge L. Almodovar1, Luis Pinzon Herrera1, Harris Blankenship3. (1) Chemical Engineering, University of Arkansas, Fayetteville, Arkansas, United States (3) Biology, University of Arkansas, Fayetteville, Arkansas, United States

Human embryonic stem cells (hESCs) and Schwann cells provide a compelling component for tissue engineering and regenerative medicine. hESCs and Schwann cells have potential for treating many neurodegenerative diseases and nerve trauma by bridging structural gaps that occur within the brain and the peripheral nervous system, respectively. The extracellular matrix (ECM) is a non-cellular scaffold composed of collagen, heparin, laminin, and other macromolecules to provide a structural support system for cell adhesion and communication, along with tissue development. The physical, chemical, and topological composition of the ECM plays a major role in Schwann cell proliferation and the differentiation of hESCs into functional neurons. Both DRAFT hESCs and Schwann cells are extremely sensitive to properties of the matrix in that even small changes to the ECM can drastically affect cell growth. Promising studies have focused on the development of scaffolds with nano scale features that can incorporate neural differentiation enhancers. In this work, we have fabricated peptoid microsphere surfaces with multilayered films (layer-by-layer) consisting of type I collagen and heparin bi-layers. Previous studies show that the morphology of the peptoid microsphere coatings enhance cell adhesion and differentiation of hESCs to neurons and supports neurite outgrowth. The layer-by-layer coatings with collagen and heparin have shown promising results in Schwann cell adhesion and proliferation without the need for growth factors. By combining the morphology of the peptoid microspheres with the composition of the collagen/heparin bi-layer depositions, we have created a novel biopolymeric material that directly mimics the natural ECM with enhanced proliferation of Schwann cells and the differentiation of hESCs into functional neurons. This work is being extended to determine the mechanism of action, as well as incorporation of other cell proliferation and differentiation modifiers.

BIOT 50

Engineering protein scaffolds for targeting metalloproteinases with high selectivity

Maryam Raeeszadeh Sarmazdeh, [email protected]. Chemical and Materials Engineering, University of Nevada, Reno, Reno, Nevada, United States

Enzymes of the metalloproteinase (MP) family, including matrix metalloproteinases (MMPs), ADAMs play a key role in developing and progression of several diseases. For instance, MMP-9 has shown to play a significant role in progression of several cancers and neurodegenerative disease such as Alzheimer’s disease (AD), while other MMPs and ADAMs are found to be anti-tumorigenic and neuroprotective. Broad spectrum MMP inhibitors have proven ineffective, so inhibitors with greater selectivity are desired. Protein scaffolds such as natural tissue inhibitors of metalloproteinase (TIMPs) offer great options for targeting MPs with high selectivity. TIMPs bind to several MMPs and ADAMs with various ranges of affinities and offers a great scaffold for engineering of selective MP-targeted therapeutics. To overcome the challenges of wide multi-specificity of TIMPs for different classes of MPs, the state-of-the-art directed evolution and yeast surface display techniques recruiting high-throughput library screening technology were used to fully evolve novel protein-based drugs with high DRAFT selectivity for specific MMPs. We have previously developed a counter- selection strategy to screen TIMP variants that bind selectively to a specific MMP. This strategy has improved TIMP binding selectivity, highlighting the significant potential of this approach for development of inhibitors with single- MP selectivity. In ongoing studies, we combine rational design and directed evolution to develop novel protein scaffolds based on TIMPs to target specific MPs with high affinity and selectivity. These studies and methodology developed will lay the foundation for developing novel protein engineering strategies for generating therapeutics for MP-related diseases and understanding protein-protein interaction between MPs and their inhibitors.

BIOT 51

Chemical preparation and characterization of kinsenoside

Dong-Chun Ren3, [email protected], Yu-Chen Lo4, Edward J. Parish1, Hiroshi Honda2, [email protected]. (1) Auburn Univ, Auburn, Alabama, United States (2) Bioengineering, Northwestern Polytechnic University, Fremont, California, United States (3) Pharmacy, China Pharmaceutical University, Nanjing, Jiangsu, China (4) Bioengineering, Stanford University, Stanford, California, United States

Novel discovery of new approaches to the chemical preparation and characterization of kinsenoside

BIOT 52

Optimization of transcription factor and bacterial promoter-based biosensors in cell-free systems for defense applications

Kathryn Beabout, [email protected]. 711th Human Performance Wing, Air Force Research Laboratory (UES, Inc), Beavercreek, Ohio, United States

Cell-free biosensors have the potential to address a variety of sensing applications for the US Air Force and Department of Defense due to their high sensitivity and specificity for performance biomarkers and analytes of interest. Additionally, the ability to immobilize cell-free protein expression systems on paper allows for the establishment of ruggedized sensors for field applications. Here, we show the adaptation of several transcription factor- based biosensors to a cell-free platform optimized for expression of bacterial promoters. One analyte of interest is deoxycholic acid (DCA), a bile acid, indicator of fecal pollution in water, and a biomarker for certain disease states. DRAFT

We demonstrated the functionality of a DCA sensor dependent on BreR, a TetR-like repressor from Vibrio cholorae, in our cell-free system. The sensor responded to DCA in a dose-dependent manner in the cell-free system but lacked reproducibility when the BreR and sfGFP reporter proteins were expressed simultaneously in the same reaction. Generation of extract pre- enriched with the BreR transcription factor provided a more robust and reproducible sensing platform. The ability to detect water pollutants, such as heavy metals, in the field is critical. Therefore, we have developed a mercury (II) sensor dependent on MerR, a repressor/activator transcription factor from Shigella flexneri, in a bacterial RNA polymerase-dependent cell-free system. Combined our data shows the versatility of cell-free systems for the development of transcription-factor based sensors derived from a variety of bacterial species. Ongoing work includes the optimization of the sensors for increased sensitivity to detect biologically relevant concentrations of analytes and adaptation of these sensors for use on paper-based systems for field applications.

BIOT 53

High-throughput non-affinity purification methodology to enable bioreactor characterization and small-scale model qualification for a recombinant enzyme

Tridevi Dahal-Busfield, [email protected], Brian Murray, Brad Whitaker, Jiuyi Lu, Jason Walther. Purification Development, Sanofi, Framingham, Massachusetts, United States

High-throughput, small-scale methodologies are crucial to ensure efficient, rapid and robust development of biological processes. In particular, significant advances have been made in purification process development, allowing parallel experiments to be conducted at the microliter scale, leading to dramatic minimization of resource and material requirements.

This study describes the application of such high-throughput approaches to enable the bioreactor characterization of a recombinant enzyme process. A high-throughput method for rapid and small-scale purification of bioreactor harvest was required to enable cell culture characterization and small-scale model qualification. Because no affinity chromatography solutions exist for the recombinant enzyme molecule, this effort required a novel approach. To this end, we have developed an automated high-throughput, micro-scale, two-step purification method employing resins used in the at-scale purification process that generated purified product suitable for product quality testing, requiring DRAFT only milliliters of bioreactor harvest. The high-throughput scale-down Tecan method was developed and optimized to resemble the current purification process to ensure comparable product quality measured using glycosylation profiles. Similar methodologies could be applied for future programs also requiring non-affinity purification processes. In future, additional high- throughput automation techniques could be applied to completely automate this two column small-scale purification process.

BIOT 54

Evaluation of an Inline UFDF Unit Operation for Continuous Downstream Bioprocesses

Ehsan Borujeni1, Alexander Helling2, Martin Leuthold2, Mark Brower1, [email protected]. (1) Merck, Kenilworth, New Jersey, United States (2) Sartorious Stedim, Goettingen, Germany

Intensified and continuous bioprocessing has drawn significant attention in the field of biomanufacturing in recent years. Continuous ultrafiltration and diafiltration (UFDF) are key technologies in downstream processing both of which are based on single-pass tangential flow filtration approach. The available inline diafiltration (ILDF) technologies contain separate filtration subunits where diafiltration buffer is combined with the feed streams entering each subunit and replace the salts and low molecular weight impurities while passing through the module. However, this design of multi filtration units, separated from each other, makes the diafiltration set-up complex and in some case with a large footprint. In addition, there is a lack of proper scale- down model for the available ILDF technologies. Having a proper small-scale model for this unit operation is important and economically beneficial; process development and characterization as well as investigation of long duration runs, as a key element of continuous processes, requires significant amount of material and diafiltration buffer. In this work, a self-contained, multistage tangential flow filter module was evaluated. The module contains stacked pairs of flat sheet membrane creating one specific channel for diafiltration and two others for feed/retentate and permeate streams. This allows only one diafiltration port into the module which makes the diafiltration set-up less complex with a small footprint. Diafiltration experiments were conducted using model protein and two monoclonal antibodies with various operational conditions like feed concentration, feed flow rate and buffer to feed flow rate ratios applied. The efficiency of salt removal was assessed through monitoring the conductivity of feed and retentate streams compared to that of the diafiltration buffer. Applicability of this module for simultaneous inline DRAFT concentration and diafiltration processes (ILCDF) was also investigated through a set of limited experiments. Ultimately, an 8X scale-down prototype of this module with 0.025m2 membrane area was evaluated. Results indicated that this module can be successfully utilized in ILDF and ILCDF bioprocesses. In addition, the scale-down model could mimic the large-scale process performance with high confidence.

BIOT 55

Robust parvo virus (MVM) clearance on Sartobind® membrane adsorbers

Sherri Dolan, [email protected]. Sartorius Stedim North America, Williston, South Carolina, United States

The purpose of this poster is to guide users on the best practices when optimizing and sizing membrane adsorbers for virus clearance. AEX chromatography is a powerful method to clear all types of virus based on charge. Membrane chromatography devices offer many advantages over conventional chromatography resins in terms of virus binding/clearance. Compared to resins, membranes offer higher binding capacity for large molecules such as viruses, show equivalent virus clearance capabilities, have >10 fold faster flow rates, use >75% less buffer and present a flexible single use disposable solution which does not require validation. These qualities make membrane adsorbers a perfect fit for flow through chromatography operations where virus clearance is expected. However several considerations must be addressed when using membrane adsorbers for virus clearance, such as: Buffer conditions, loading capacity for virus and purity of virus spike, air removal/wetting of small scale devices, protein (or impurity) binding to membrane.

This poster also presents data to support parvo virus (MVM) clearance on Sartobind membranes under challenging process conditions. MVM parvo virus is considered to be the most difficult to remove virus based on its small size and high pI. Challenging conditions tested include: pressure pause, high loading, non-wetted filter, non-purified virus spike, scalability.

BIOT 56

Enhanced selectivity of ion exchange resin for ADC polishing

Annika Holzgreve, [email protected]. Millipore Sigma, Darmstadt, Germany DRAFT

Antibody drug conjugates (ADCs) have a great potential to increase the biopharmaceutical drug efficacy. Unfortunately, this great potential currently is explored at high costs due to the complexity of the manufacturing process. Especially the purification of immunoglobulins carrying different ratios of cytotoxic drugs requires the use of non-efficient chromatographic steps at narrow operation windows.

The exploration of an innovative ion exchange resin enabled us to reach an efficient The purification process for ADCs can be a challenging task. We have explored The exploration of an innovative ion exchange resin for the downstream purification of ADC’s. This has enabled us to reach an efficient DAR (drug antibody ratio) species separation without applying high conductivity and/or organic solvents. Moreover, we could improve the quality of commercially available ADC’s by preparative separation of each individual DAR species, regardless their manufacturing origin.

The application of ion exchange chromatography showed significant economic advantages compared to hydrophobic interaction chromatography, assuring high product recovery rates without influencing ADC’s efficiency. Additionally, we were able to upscale such separation from analytical column format to preparative mode at minimal process development efforts defining a broad window of operation and reflecting the later ease of use and templatability of such method.

BIOT 57

Evaluation of pressure fluctuation in constant pressure and constant flow rate control and effects on viral clearance

Victoria Kaloudis1, [email protected], Esha B. Vyas1, Mike Bobbie1, Daniel Strauss2,1, Takayuki Sohka1. (1) Science and Technology, Asahi Kasei Bioprocess, Glenview, Illinois, United States (2) Asahi Kasei Bioprocess, Glenview, Illinois, United States

Viral filtration is an integral part of downstream processes for biopharmaceutical products in which viruses and other contaminants are effectively removed by nanofiltration. Planova filters are operated in dead-end mode by either constant pressure or constant flow rate control. In cases where strict pressure control is required, Asahi Kasei Bioprocess recommends using constant pressure control, either by house air or compressed air tanks. DRAFT

However, little is known about how pressure fluctuation during filtrations affects virus removal. In this study, we examined the effects of pressure fluctuation on viral clearance when using constant pressure control and constant flow rate control by peristaltic pump with one or two pump heads. A total of 16 filtrations using a protein feed of 5 g/L BSA with porcine parvovirus (PPV) spike (virus challenge, 10.5 logs TCID50/m2) were conducted on 0.001 m2 Planova 20N filters according to manufacturer recommendations with pressure readings taken every 2 s. Filtrations were conducted in duplicate using constant pressure control by compressed air and using constant flow rate control by a peristaltic pump with one or two pump heads. Constant pressure runs were conducted at 8 psi (low pressure), <14.2 psi (manufacturer recommended maximum operating pressure) and 18 psi (high pressure). Target pressure for constant flow rate control with two pump heads was 6 – 8.5 psi, <14.2 psi and 16 – 18 psi, respectively. Constant flow rate control with one pump head was conducted in the range of 7 – 18 psi and <14.2 psi. The constant flow rate control runs represent exaggerated conditions with pressure fluctuation of >9 psi for constant flow rate with one pump head and relatively limited fluctuation of <4 psi for constant flow rate with two pump heads. The process volume was 150 mL, collected in two equal fractions, and PPV assay was conducted to determine viral clearance for each run. Pressure fluctuation during filtration runs did not cause marked decreases in PPV LRV under the conditions tested.

BIOT 58

Efficient clarification using a depth filter simplified the purification process of a therapeutic protein from the Escherichia coli system

Mark Geng, [email protected]. Bristol-Myers Squibb, Brentwood, New Hampshire, United States

Continuous centrifugation has been widely used by the biopharmaceutical industry to remove host cell protein (HCP) precipitates in the biological downstream purification processes. Although a continuous centrifugation followed by depth filtrations can achieve effective clarification at stainless steel bioprocess facilities, a single-use continuous centrifuge is not feasible at large scale. Therefore, centrifuge-free clarification using depth filters is of great interest to bioprocessing facilities that are undergoing a single-use transformation. However, depth filters’ general susceptibility to fouling presents a major challenge to maintaining a satisfactory throughput. This presentation presents an alternative clarification strategy using a single-stage depth filtration to simplify traditional clarification processes that use DRAFT centrifugation and two-stage depth filters. Multiple filters were evaluated in terms of the product quality, impurity removal, and pressure. The results showed the use of a Clarisolve® 20MS media grade filter had no impact to the yield and product quality (turbidity, monomer purity and removal of HCP and a process associated impurity protein of interest). The new filter had a minimum throughput of 130 L/m2 at 20 PSI pressure limit, which provided enough capacity with less depth filter footprint. The clarification efficiency was supported by characterization of the precipitates’ particle size distribution. The process robustness was demonstrated by varying load quality and process parameters. The intensified clarification strategy using a single-stage depth filter to remove HCP precipitates greatly simplifies the process with high productivity and process robustness.

BIOT 59

Clearance of host cell protein at elevated conductivity through synthetic depth filtration media during downstream processing

Chris Nieder1, [email protected], Camilla Oxley2, Vitali Stanevich2, Mark McInnes1. (1) MilliporeSigma, Burlington, Massachusetts, United States (2) Janssen, Malvern, Pennsylvania, United States

Host cell proteins (HCP) are a process-related impurity common to biological therapies that have the potential to elicit an immune response in patients and, because of this, regulatory bodies require <100ppm HCP in the drug substance. Downstream purification steps and, in particular chromatography resins, are heavily leaned on to accomplish the necessary clearance. HCP can be a complex process-related impurity in that there can be subtypes with differences in chemistry that can present a challenge to clearance using process-platform chromatography resins. Purification processes are at a point where depth filtration is commonly used for haze reduction and/or impurity clearance with product, buffer conditions commonly at a neutral pH and a conductivity of <10mS/cm. This work screened several media types e.g. cellulose diatomaceous earth, activated carbon, and synthetic media at a pH of 6.0 with varying conductivity to maximize HCP clearance and product yield. Millistak+® HC X0SP, a synthetic media depth filter, at elevated conductivity, was identified as the optimum depth filter.

BIOT 60 DRAFT

Application of an anion exchange membrane adsorber in late-stage bioprocesses

Tom Klimek1, Chris Nieder2, [email protected], Thomas Parker2, [email protected]. (1) Eisai, Exton, Pennsylvania, United States (2) Technology Management, MilliporeSigma, Raleigh, North Carolina, United States

Membrane adsorbers have become a popular choice in single-use centric and next generation bioprocesses, where some of the benefits are ease of use i.e. plug and play, and increases in productivity (g/L*h); however, these membranes are often replaced by traditional, packed bed chromatography operation for late-stage clinical and commercial processes. Packed bed chromatography often provides a better fit for late-stage processes, as columns are packed to scale with increased process volume and can be cycled as production needs rise.

BIOT 61

Process-scale ceramic hydroxyapatite columns: Packing methods, column efficiency evaluation, and post-packing handling

Louisa Vang, [email protected], Irene Chen, Xuemei He. BioRad, Hercules, California, United States

Ceramic Hydroxyapatite (CHT) has been widely employed in the production of biopharmaceuticals, including viruses and virus-like-particles for vaccines and gene/cell therapies. This mixed-mode chromatography media has demonstrated unique capability of resolving biomolecules that are not separable by alternative means, owing to the synergistic effects of the chelating calcium ions and the negatively charged phosphoryl groups in its matrix. Multiple process and product-related impurities, such as aggregates/fragments of drug molecules, host cell proteins and nucleic acids, endotoxins, and contaminant viruses, can be effectively removed in a single chromatography operation. The rigid spherical macroporous particles of CHT offer high permeability for biomolecules and a linear pressure-flow relationship irrespective of column diameter, which allows easy production scale-up with the robustness and consistency demanded by bioprocessing. In the present study, we have investigated the critical attributes in packing process-scale columns. CHT media can be packed by the axial movement of column piston or a constant flow of mobile phase in the linear velocity range of 100 ~ 400 cm DRAFT per hour, without significant effect on final column quality. On the basis of present results, we recommend testing column efficiency using the mobile phase flow rate as column packing speed. Movement of a CHT column after packing may lead to initial change of plate count (N/m) and/or asymmetry (As). However, subsequent movement does not seem to cause additional column bed settling or performance change. Other practical tips for packing CHT columns are also provided in the poster presentation.

BIOT 62

Development of protein A membrane chromatography devices for rapid cycling

Gerado Cedrone1, Hua Chen2, Sean Foley1, Igor Gonzalez1, Lloyd Gottlieb1, Chanis Harpin1, Brad Kachuik2, Rushd Khalaf1, Elena Komkova2, Gregory Molica1, Graham Morse2, Amro Ragheb2, Kevin Rautio1, Xiaojiao Shang2, Navneet Sidhu2, Gary Skarja2, Matthew Stone1, [email protected], Paul Turiano1, [email protected], Joaquin Umana1, Melissa Vandersluis2. (1) Chromatography R&D, MilliporeSigma, Bedford, Massachusetts, United States (2) Chromatography R&D, MilloporeSigma, Burlington, Ontario, Canada

Absorptive membrane chromatography offers many advantages over traditional chromatography resins in the downstream purification of therapeutic proteins. A chromatography membrane device can be operated at much higher flow rates enabling the length of a bind/elute chromatography step to be reduced from hours to minutes. Rapid cycling of a membrane device allows the same amount of protein to be processed with a significantly reduced volume of membrane in a given time period. In addition, rapid cycling enables single-use per batch operation, which eliminates sanitization and storage requirements as the full lifetime of a membrane device is exhausted within one batch. However, membrane chromatography for the downstream purification of proteins has been largely limited to flow-through steps due to the low binding capacities of traditional filtration membranes and the large device hold-up volumes that decrease the efficiency of a bind/elute step. Herein we report the development of Protein A (PrA) membrane devices optimized to facilitate the efficient chromatographic capture of a monoclonal antibodies from clarified cell culture via a rapid cycling process. Efficient bind/elute chromatography is possible due to the novel properties of the PrA membrane and the optimized design of the devices. The PrA membranes are constructed from a hydrogel that provides significantly higher binding DRAFT capacities than are possible with traditional filtration membranes. The devices were specifically designed in a flat sheet configuration to minimize the hold-up volume and facilitate efficient bind/elute chromatography processes. All the membrane device sizes utilize a flat sheet format to facilitate scaling of a bind/elute chromatography process.

BIOT 63

Development of a rAAV clarification platform for a viral gene therapy process

Albert S. Reger, [email protected]. MilliporeSigma, Burlington, Massachusetts, United States

Current trends in the biotechnology realm show the rapid growth of the use of viral vectors for gene therapy as one of the next generation of life changing medicines. rAAV (recombinant Adeno-associated virus) has become a popular viral vector that is able to deliver targeted genetic information to a patient and is considered non-pathogenic and safe for these patients to be treated with. Due to the life changing nature of these therapies to the patient population, the need for rapid and robust purification and manufacturing methods is a must. A key step in the purification and manufacturing of rAAV viral vectors is the clarification step. A typical rAAV manufacturing process will lyse transfected cells, releasing the rAAV viral vectors from the cell, producing a wide particle range of cellular debris that must be removed prior to moving forward in the purification process. During the early stages of developing a clarification method for a rAAV manufacturing process, it is crucial to select the correct depth filtration media that will maximize rAAV yield and provide reproducible quality filtrate. The work to be presented, focuses on the development of a clarification step by testing combinations of MilliporeSigma’s Millistak+® HC D0HC and Millistak+® HC Pro D0SP as primary depth filters and Millistak+® HC C0HC and Millistak+® HC Pro C0SP as secondary depth filters with a lysed cell culture feed containing rAAV9. A comparison of pressure profiles, volumetric throughput, turbidity of filtrate, and product yield was used to determine the optimal combination of primary and secondary depth filters. All filtration trains tested provided high quality filtrate which had high capacity on Express SHC® sterile grade filter, though total yield of product (>80%) suggests Millistak+® HC D0HC and Millistak+® HC Pro D0SP as good choices for a primary depth filter followed by Millistak+® HC C0HC as the secondary depth filter as the best option for the feed tested.

BIOT 64 DRAFT

Influence of low pH treatment on the elution behavior of monoclonal antibodies in protein A chromatography

Carolin Stange1, [email protected], Christian Dr. Frech2. (1) Institute of Biochemistry, University of Applied Sciences Mannheim, Mannheim, Germany (2) Biotechnology, University of Applied Sciences Mannheim, Mannheim, Germany

Low pH treatment is a commonly known factor for the aggregation of antibodies and Fc-Fusion proteins. During downstream processing of these proteins, low pH conditions are used as an effective way for virus inactivation at pH 3.0-3.5, as well as for elution from a previous Protein A chromatography step. Binding to Protein A material is induced through multiple interactions between the Fc-part of the antibody with the immobilized Protein A domains and is sensitive for conformational changes. The objective of this work was to use low pH treatment to study its effect on aggregation and structural changes of monoclonal antibodies. Therefore, a Protein A purified antibody solution was acidified (pH 2.8 to 3.5) and incubated for a period of time. Afterward, the solution was neutralized and reloaded onto a Protein A chromatography column. This resulted in only a partial binding of the sample; furthermore, when analyzing the resulting elution peak by size exclusion chromatography, a double-peak elution behavior at the monomer size was observed. Both phenomena are independent of each other but occurred during the same experiment. Additionally, these effects could not be nullified through the addition of excipients to the buffer system. The results show that multiple low pH treatments can significantly change the structure and binding behavior of monoclonal antibodies and should, therefore, be kept at a minimum during purification.

BIOT 65

Comparison of protein A resins and novel protein A matrices

Talaial B. Alina1, [email protected], Paul R. Randolph2, John Schreffler2. (1) Chemical and Biological Engineering, Drexel University, Philadelphia, Pennsylvania, United States (2) Active Pharmaceutical Ingredient - Large Molecule, Janssen Pharmaceuticals, Malvern, Pennsylvania, United States

The use of higher productivity protein A capture matrices for the purification of monoclonal antibodies (mAbs) is becoming essential due to higher productivity bioreactor technologies. Six different Protein A resins and 3 novel DRAFT protein A matrices were studied to characterize Dynamic Binding Capacity (DBC), sodium hydroxide (NaOH), resin, and matrix lifetime. Additionally, batch loading, intensified batch, and intensified batch with dynamic flow loading productivities were modeled to compare costs and productivities of these intensified processes to the theoretical costs and productivities of the novel matrices. These data demonstrate novel protein A matrix productivity gains, although substantial in comparison to the current state of batch chromatography, are not as large when compared to existing protein A resins using intensified processes.

BIOT 66

New Tool for the next generation of antibody purification: Case study using novel protein A affinity chromatographic membrane technology

Graham Temples1, [email protected], Tom Caldwell2, Jinxiang Zhou1, Daniel Henn1, Scott M. Husson1,3, Sarah W. Harcum2. (1) Purilogics, LLC, Greenville, South Carolina, United States (2) Bioengineering, Clemson University, Clemson, South Carolina, United States (3) Chemical Biomolecular Engineering, Clemson University, Clemson, South Carolina, United States

The success of monoclonal antibody (mAb) drugs has encouraged the pharmaceutical industry to pursue high-productivity biologics manufacturing processes. The demand is driven by multiple factors, particularly growing competition and market uncertainty. In addition, there is increased demand to lower the cost of mAbs for the treatment of chronic conditions. While continuous chromatography technology has been developed to improve productivity using resin-based media, mAb purification still remains a critical bottleneck in the manufacturing process due to the slow processing speed of resins.

This presentation highlights Purilogics’ recent progress on development of novel Protein A membrane chromatographic technology (PurexaTM-A). Beyond high binding capacity and fast processing speeds, we aim to deliver a differentiated product enabling a robust process over multiple cycles. In this presentation, we will demonstrate purification performance of our prototypes using a suite of CHO-derived monoclonal antibodies. Key quality attributes such as: host cell protein, DNA, and aggregate content of the purified antibody pool will be characterized along with mAb recovery. In addition, we will demonstrate reusability of the membrane with multiple-cycle bind-and-elute studies. DRAFT

BIOT 67

V3: Tentacle-based ion exchange resins for viral vaccines, virus-like particles, and viral vector purification

Helena Marie1,2,3, [email protected]. (1) sigmaaldrich, Darmstadt, Germany (2) University of Applied Sciences Mittelhessen, Germany, Giessen, Mittelhessen, Germany (3) University of Natural Resources and Life Sciences, Vienna, Austria, Vienna, Austria

Virus-based therapies have raised significant interest in cancer treatment, prophylactic vaccination, and gene therapy. The features of viruses are leveraged to prevent and treat diseases. Outstanding benefits of virus-based therapies are personalized medicine and the generation of novel therapeutic targets and universal vaccines. Current production processes generate significant amounts of product-related impurities which need to be removed during the purification process. These impurities do not have a therapeutic benefit and are considered a risk for a successful therapy by triggering an immune response. Chromatography has become the method of choice for challenging separations. MilliporeSigma has developed a variety of chromatography materials for the purification of molecules such as monoclonal antibodies, plasma proteins, and novel therapeutic targets. The unique tentacle chemistry of MilliporeSigma Fractogel® and Eshmuno® resins is beneficial for the purification of larger molecules such as viral vaccines, virus-like particles (VLPs) and viral vectors. Here we present data generated with our tentacle-based ion exchange resins for purification of viral vectors e.g. oncolytic measles virus and HIV-1 gag VLP.

BIOT 68

Small-scale model for studying resin interactions with chemical sanitants

Rachel Straughn, [email protected]. Just - Evotec Biologics, Seattle, Washington, United States

In connected continuous antibody manufacturing, the downstream capture system can be considered part of the upstream process. Single-use components and assemblies help to extend the sterile barrier of the bioreactor, but columns and resin remain a challenge. While single-use DRAFT products such as gamma-irradiated columns are available, they are costly and not conducive to low-cost antibody manufacturing. Therefore, one must rely on chemical sanitants to treat reusable materials such as resin and column housings. Just successfully implemented a sanitization procedure for Protein A chromatography capture in large-scale (500 liter) productions using a peracetic acid solution. After changing the Protein A resin used in this step, we experienced a series of contaminations at large-scale. To investigate these contaminations, we developed a small-scale model probing sanitant concentrations, hold duration, and potential resin-sanitant interactions. The data generated using this small-scale model informed concurrent large-scale column sanitization work, resulting in a sanitization procedure that successfully enabled a bioburden-free GMP continuous capture process.

BIOT 69

Process development strategies for high cell density monoclonal antibody harvest

Hong Zhang, [email protected], Stijn H. Koshari, Robert Luo. GlaxoSmithKline, King of Prussia, Pennsylvania, United States

The purpose of the harvest step in a monoclonal antibody downstream purification process is to remove biomass, particulates, and other suspended material from the cell culture broth to generate a product stream suitable for purification by liquid chromatography. Removing biomass and other particulates eases the burden on the capture column and extends its lifetime. A traditional harvest process employs a two-step approach using centrifugation combined with depth filtration. In this process, coarse solids and intact cells are first removed using a centrifuge, and the resulting centrate is further clarified by depth and membrane filtration in series. Direct filtration has the potential of greatly simplifying the overall harvest process by directly loading the harvest material on high-capacity depth filters and consequently removing the need for additional preparative steps such as centrifugation or flocculation. In this poster, we highlight process development strategies for direct filtration harvests that can increase the filter loading capacity, particularly with regard to high cell density feed streams (> 40 million cells/mL). Filter loading and facility fits were informed by metrics such as the depth filter volumetric capacity for solids and filtrate turbidity. Different types of depth filters, a wide range of fluxes and depth filter ratios, as well as different types of sterile membrane filters were explored to ensure a robust operation and scalability. The strategies discussed here can aid in transforming a trial- and-error development approach to one that is more methodological and DRAFT predictive. The strategy for tech transfer including facility fit will also be discussed.

BIOT 70

Small scale modeling of disk stack centrifugation with mammalian cell culture capillary shear for a more predictive harvest process transfer

Michael R. McGarrah, [email protected], Steve Timmick, Brian Crump, Antonio R. Ubiera. Downstream Process Development, GlaxoSmithKline, Philadelphia, Pennsylvania, United States

Advances in cell culture technologies and methods in recent years have resulted in increasing solids densities at harvest. This results in a higher number of cell culture harvest processes requiring a centrifugation step at pilot and commercial scale. The traditional scaling approach used for disk stack centrifugation is to use Q/Σ theory to transfer equivalent centrifugal separation conditions between different centrifuges. However, several data sets have been generated showing that this model can be too ideal to be used in practice. Some limitations to Q/Σ theory include that it assumes unbreakable solid spheres are being separated and there is perfectly efficient separation. In practice, disk stack centrifuges vary in efficiency by design and are affected by quality of feed material. Also, for shear sensitive mammalian cell types in particular, when intact cells enter a disk stack centrifuge they are broken down by shear forces into many smaller particles. In an attempt to generate a more realistic model of expected disk stack centrifugation performance, a scale down model using capillary shear constructed of common lab parts was performed. Presented here are the results of the capillary model experiments which successfully demonstrated that sheared broth material had turbidity and particles size distribution profiles more representative of process streams generated by disk stack centrifugation compared to traditional batch centrifuged broth. Additionally, observations on the difference in shear response for PER.C6 and Chinese Hamster Ovary (CHO) mammalian cells will be discussed.

BIOT 71

Affinity membrane for rapid, high-recovery concentration and purification of lentivirus vectors

Daniel Henn1, [email protected], Hui Ding3, Yanzhang Wei3, Jinxiang Zhou2, Scott M. Husson1,4. (1) Purilogics, LLC, Greenville, South Carolina, DRAFT

United States (3) Department of Biological Sciences, Clemson University, Clemson, South Carolina, United States (4) Chemical Biomolecular Engineering, Clemson University, Clemson, South Carolina, United States

Gene and cell therapies have the potential to dramatically expand the treatment possibilities for patients with conditions that have no other cures. Viral vectors, and particularly lentiviral vectors (LVs), are among the fastest growing types of gene-delivery vehicles. However, scalable production of LVs with high purity and high yield is a major challenge due to the fragility of LVs and a lack of efficient purification tools.Robust affinity chromatography products are desired to simplify the purification process and increase recovery. Here, we present a new affinity chromatographic membrane technology for the rapid purification of LVs with high binding capacity at short residence time and high recovery using mild elution conditions. The LVs were grown from 293FT cells, and the culture supernatant with a virus titer of approximately 10^6 virus particles per mL was loaded directly onto Purilogics scaled-down syringe filter-like affinity membrane column prototypes. The affinity interaction is based on reversible binding of the LV envelope protein. LVs were eluted under mild conditions to enable high recovery of active virus. Activity was assessed by transduction efficiency of green fluorescent protein (GFP) into HEK293 cells. LV purity was determined by SDS-PAGE and ELISA assays. Additionally, we will show a side-by-side performance comparison to ion-exchange membranes in terms of binding capacity, yield, and purity.

BIOT 72

Process intensification feasibility studies using novel multimodal anion- exchange membrane chromatography columns

Jinxiang Zhou1, Daniel Henn1, Graham Temples1, Mohamed Hashim1, [email protected], Scott M. Husson2. (1) Purilogics, LLC, Greenville, South Carolina, United States (2) Chemical Biomolecular Engineering, Clemson University, Clemson, South Carolina, United States

During monoclonal antibody (mAb) purification, high molecular weight species are normally removed using one or more additional chromatography steps, such as a hydrophobic interaction chromatography step. Existing multimodal ion-exchange resin-based chromatography columns combine hydrophobic interactions and columbic forces together. However, they often fail to provide satisfying capacity to remove host cell proteins (HCPs), aggregates and virus particles simultaneously, particularly in a two-chromatography-step purification DRAFT scheme. Here, we present feasibility studies applying Purilogics’ multimodal anion-exchange membrane chromatography columns (PurexaTM-MQ) to remove HCP and aggregates from mAb streams purified by Protein A chromatography. The data indicates that a very high volume of antibody stream could be loaded onto PurexaTM-MQ columns to obtain purified mAbs with low aggregate and HCP levels. In addition, we conducted virus loading challenges with spiked protein and DNA in various buffer conditions. The results suggest more than a 4 log reduction of virus can be achieved. PurexaTM-MQ columns offer promising results to simplify mAb purification and achieve process compression.

BIOT 73

Isoporous block copolymer membrane platform for downstream virus filtration of recombinant proteins

Chris Crock, Spencer Robbins, [email protected], Jack Wilson, Jiawei Zhu, Rachel M. Dorin, Marty Siwak. TeraPore Technologies, Inc., South San Francisco, California, United States

Traditional filter technologies and materials are constrained by a tradeoff between selectivity and permeability. In critically important and challenging separations such as downstream virus filtration, this tradeoff translates to sacrificing either patient safety or filtration efficiency. By carefully engineering and exploiting the self-assembly behavior of block copolymers, a new type of virus filter platform has been developed to allow high liquid permeability in conjunction with virus elimination. The self-assembly of block copolymers allows them to be processed into filters with very uniform pore sizes and high pore densities, enabling very high selectivity along with high liquid permeability. In conjunction with a prefilter, these isoporous membranes are part of a powerful new technology platform for biopharmaceutical downstream virus filtration.

BIOT 74

Generalizing purification of viruses and virus-like particles using aqueous two-phase system

Pratik Umesh Joshi1, [email protected], Dylan Turpeinen2, Caryn Heldt3. (1) Chemical Engineering, Michigan Technological University, Hancock, Michigan, United States (3) Michigan Technological University, Houghton, Michigan, United States DRAFT

The challenge to develop a high throughput platform method for virus-based therapeutics has been stimulating the biomanufacturing community for decades. The potential of alternate technologies that can meet the current requirements might be better than the conventional methods for larger and more complex yet fragile moieties. One of those candidates is an aqueous two-phase system (ATPS), formed by two partially miscible aqueous solutions. ATPS are known for being inexpensive, mild, environmentally- friendly, and can be easily scaled-up and processed in a continuous mode. However, the challenge of not fully understanding the partitioning mechanism had resulted in a large experimental setup to determine optimal compositions specific to different viruses. Our DOE using a tie-line framework and study of virus surface characteristics provides a method to discern the interaction between virus and phase compositions. The current experimental design in a polyethylene glycol (PEG)-citrate system at physiological pH with the porcine parvovirus (PPV, a non-enveloped virus) facilitated in achieving a high yield of 80% in the PEG-rich phase. The resulting preferential partitioning supported the hypothesis of major influence of hydrophobic and electrostatic (salting-out) interactions between virus and relative phases. The framework was then utilized to create a database for five more viruses having different characteristics of presence of envelope, pI, and size. This multivariate data is being employed to empirically predict and reduce experimental space. This work demonstrates the purification efficacy of herpes simplex virus (HSV-1), human rhinovirus (HRV-14), human immunodeficiency virus-like particle (HIV- gag VLP), and ways to generalize optimization of target virus in ATPS. The ATPS approach will not only help generalize virus purification but also provide a high throughput technology in virus-based therapeutics DSP.

BIOT 75

Downstream process development strategy for reducing residual lipases in monoclonal antibody products

Michael Iammarino1, [email protected], Divya Chandra1, William McKechnie1, Sunitha Kandula1, David J. Roush2, Jennifer Pollard1. (1) Downstream Process Development & Engineering, Merck & Co., Inc., Kenilworth, New Jersey, United States (2) Biologics Process R&D, Merck & Co., Inc., Kenilworth, New Jersey, United States

Host cell proteins (HCPs) are a major constituent of the process-related impurities during biologics production. The amount of residual HCP in the final drug product is a critical quality attribute due to its potential to affect both product safety and efficacy. It is therefore a regulatory requirement to monitor DRAFT the removal of HCPs during process development. The downstream process for monoclonal antibodies (mAbs) typically uses multiple chromatography and filtration steps to remove HCPs to below acceptable levels. However, for several mAbs it was found that specific HCPs known as lipases, despite being present at relatively low levels, were causing enzymatic degradation of polysorbates in the formulated drug product over time. In addition to the possible immunogenicity of these residual lipases, polysorbate degradation can potentially lead to a reduced shelf life of the drug product due to increased aggregation and particle formation. To overcome this issue, a focused effort has been made during downstream process development to track and remove these lipases, but that task is challenging as the lipases can associate or co-purify with the mAb and can be highly active even at sub-ppm levels. Process development case studies for two mAbs will be presented to demonstrate the development and analytical strategies used to address lipase clearance. Specifically, process parameters for different unit operations were investigated for their ability to remove lipases from the mAb process streams. Clearance of lipases was monitored via multiple analytical assays to enable a detailed understanding of process performance as well as deliver robust, final processes. The successes and challenges with lipase removal observed across the unit operations for the different mAbs will be highlighted along with the general strategy and approach used for tackling this downstream bioprocessing challenge.

BIOT 76

Evaluation of multimodal anion exchangers for protein recovery

Sushmita Koley1, [email protected], Mark A. Snyder2, Xuemei He2, Steven M. Cramer3. (1) RPI, Troy, New York, United States (2) Bio-Rad Laboratories, Hercules, California, United States (3) Ricketts Bldg, Rensselaer Polytechnic Inst, Troy, New York, United States

Multimodal chromatography has increasingly been utilized in several bioprocess based platforms due to its distinctive selectivity. Previously, our lab had reported a series of multimodal anion exchangers (MMAEX) with varying aromatic ring substitutions and its effect on protein selectivity. The nature of hydrophobicity and the ligand geometry of the bulky groups were found to play a major role in driving protein’s selectivity on the adsorbents. Probing further, we wanted to understand the recovery issues that are generally observed to be a concern when dealing with the MMAEX ligands in the industries. In regard to these, two other MMAEX ligands containing guanidine groups were evaluated. The chromatographic experiments consisted of a series of linear DRAFT salt gradients that were carried out with a set of model protein library which has demonstrated to have recovery issues. The role of ligand density and ligand structure was evaluated for binding of proteins to the adsorbents and the study was compared with the parent ligand structure, Prototype 13 and a commercial adsorbent, Capto Adhere. Interesting results were observed with these MMAEX adsorbents as it was seen that while a single guanidine moiety exerted a hydrophobic type of interaction leading to recovery issues, the other structure with two guanidine moieties did not pose these problems. It was also demonstrated that a linear pH gradient could alleviate the recovery problems so observed. These results indicated that the recovery issues can be circumvented by careful modification of the ligand structure and/or meticulous use of experimental conditions to employ the multi-mode interaction of a multimodal chromatography for improvement in selectivity.

BIOT 77

Novel membrane process for highly concentrated pharmaceutical solutions without heating and high pressure

Yuji Katayama, [email protected], Daisuke Hotta, Fumiyoshi Inoue, Mitsuru Fujita, Masato Mikawa, Noboru Kubota. Performance Materials Technology Center, Asahi Kasei Corporation, Fuji-city, Shizuoka, Japan

Concentration of pharmaceuticals is an important process toward the production of pharmaceutically active components. Conventionally, reverse osmosis (RO) membrane or distillation systems have been used for the concentration. However it is still challenging to concentrate the solutions to a high concentration without deteriorating the quality. Operation of RO membrane systems needs larger pump pressure than the osmotic pressure difference between the pharmaceutical solutions and permeated water. Therefore there is a concentration limit in RO membrane systems, and it causes a severe membrane fouling. In addition, high pressure of the solutions also gives the negative effects on some pharmaceutical molecules. On the other hand, distillation processes require heating, which can decompose pharmaceutically active components. Toward the process for highly concentrated pharmaceuticals without heating and high pressure, we developed Forward osmosis (FO) membrane systems. FO membrane systems can concentrate aqueous solutions only by taking advantage of the osmotic pressure difference between Feed solution (FS) which contain pharmaceuticals and Draw solution (DS) which has higher osmotic pressure than FS. The water permeates from FS to DS with the DRAFT osmotic pressure difference as a driving force, causing the concentration of the pharmaceuticals in the FS. As FO membrane systems only require the osmotic pressure difference between FS and DS, they concentrate the pharmaceutical solutions with mild conditions, which can avoid high pump pressure and heating, having milder fouling on the membranes. We demonstrated that small molecule solutions were concentrated to high concentration with FO membrane systems under mild conditions, which indicates the potential of FO membrane systems for the concentration of various pharmaceutical solutions.

BIOT 78

Two-stage high-throughput process development (HTPD) identifying conditions for tandem process

Tim Zhou1, [email protected], Abhijeet Shirke2, Dharmesh Kanani1, Zhaoqing Zhang1, Lu Wang1. (1) Teva Pharmaceuticals, USA, West Chester, Pennsylvania, United States (2) Downstream Processing-CMC Biologics, Teva Pharmaceuticals, West Chester, Pennsylvania, United States

Continuous and integrated downstream processes have been the focus for manufacturing process development for biologics to reduce cost and operation complexity. In this work, we present a two-stage HTPD approach to identify and predict optimal conditions for polishing chromatography processes design, especially flowthrough mode CEX, AEX and HIC. Customized resin plates are prepared by slurry dispensing. The plate screening is performed to compare the performance of multiple chromatography resins under variant process conditions (pH, conductivity) in terms of yield and aggregate clearance. The optimized conditions of each individual resin are selected and further verified using robocolumns, where yield vs. load, as well as monomer vs. HMW breakthrough curves are plotted for frontal analysis. The study results obtained from the resin plate screening and robocolumns have good agreements; furthermore, based on the experimental data, a prediction model is established to identify the potential conditions for a tandem purification process in flowthrough mode, where two chromatography columns are operated under the same pH and conductivity. By connecting two polishing columns tandemly and operated as one unit step, much processing time and cost can be saved for better efficiency during the mAb manufacturing.

BIOT 79 DRAFT

Streptavidin-ELP-based macromolecular ligand fusions for the purification of mAbs

Kabir Dhingra1, [email protected], Akshat Mullerpatan1, Ronit Ghosh1, Pim Hermans2, Pankaj Karande1, Steven M. Cramer1. (1) Department of Chemical and Biological Engineering, Rensselaer Polytechnic Institute, Troy, New York, United States (2) BAC BV, Leiden, Netherlands

The smart biopolymer elastin-like polypeptide has been successfully employed previously for the purification of mAb and non-mAb biologics using affinity precipitation. However, for each target, this involves recombinantly cloning and expressing a fusion of ELP and the respective affinity ligand. In this work, we have examined the utility of streptavidin-ELP fusions as a platform for affinity precipitation. While streptavidin usually expresses as inclusion bodies that are relatively difficult to purify, fusions of streptavidin with other proteins can enable soluble cytoplasmic expression and therefore easier purification. A tetrameric streptavidin-ELP (SELP) construct was cloned, expressed and purified from the cytoplasm of BL21(DE3) E. coli, attaining yields of > 10 mg/L of culture. Biotinylated CaptureSelect™ affinity ligands developed by ThermoFisher, were non-covalently conjugated with the tetrameric SELP, resulting in the formation of a larger macromolecular ligand- SELP complex (~190 kDa). Reversed phase UPLC, SE-UPLC and SDS PAGE analyses were employed to confirm the formation of the macro-ligand. The formed ligand complex was then evaluated for the affinity precipitation of mAbs in pure form and from a complex mixture. The results indicated 100% mAb capture in pure component as well as complete selective capture from a harvest supernatant, with greater than 80% recovery of highly pure mAb. A molar ratio of 16:1 of the ligand-SELP complex to mAb was required to achieve complete capture in from the complex feed. Importantly, no SELP was observed in the mAb eluate, indicating the preservation of the streptavidin-biotin interaction througout the process. Further, we showed that the purification efficiency of the ligand-SELP construct was maintained for multiple rounds of affinity precipitation. This proof-of-concept study demonstrates that SELP can be used in conjunction with biotinylated affinity ligands for mAb purification.

BIOT 80

Case study in raw material change control: Evaluating an alternative virus filtration post use pre-integrity testing flush solution for a monoclonal antibody manufacturing process DRAFT

Morgan Ayres1, [email protected], Melani Stone2, Scott A. Tobler3. (1) Merck, Doylestown, Pennsylvania, United States (2) Biologic Process Development and Commercialization, Merck, Harleysville, Pennsylvania, United States (3) Biologics Manufacturing Science and Commercialization, Merck and Co., Inc., Hoboken, New Jersey, United States

Evaluating the impact of vendor initiated raw material changes are a key activity for ensuring a biopharmaceutical manufacturing process maintains a state of control in response to these changes. The evaluation of an alternative virus filtration post-use flush solution (0.25 M NaOH, 1% SDS) for a monoclonal antibody (mAb) manufacturing process is used as a case study to describe this process. Triton X-100 is currently used in this manufacturing process to flush the filters after product filtration to remove any residual proteins and other foulants, and prepare the filters for a post-use filter integrity test. After 2021, triton X-100 will no longer be permitted for use in the EU due to environmental impact and an alternative post-use flush solution will have to be used. To supplement filter vendor studies supporting the use of an alternative flush solution, small-scale virus filtration runs were executed and a product specific assessment of the alternative flush solution was performed. The results of the integrity test after product filtration were compared and enhanced the understanding of the impact of the post-use flush on the integrity test outcome and demonstrated the acceptability of the alternative post-use flush solution for a range of virus filtration operating conditions. Evaluation of the post-use flush conditions at the small-scale supported the manufacturing change control and will ensure continued supply in response to raw material changes.

BIOT 81

Using AI based descriptors to model protein chromatography

Imee Sinha, [email protected], Camille Bilodeau, Shekhar Garde, Steven M. Cramer. Dept of Chemical and Biological Engr, Rensselaer Polytechnic Inst, Troy, New York, United States

QSAR techniques are commonly employed for the understanding of protein chromatographic behavior. These techniques utilize knowledge-based descriptors that arise from the protein’s structure and physiochemical properties. Knowledge-based descriptors rely on clever ways of quantifying the protein’s surface properties, like using software aided potential maps, measures of hydrophobicity/hydrophilicity, molecular shapes and partial charges. A new technique of descriptor calculation is the use of powerful DRAFT neural networks to understand structure-based activity. A variational autoencoder network can be used to learn protein surface properties which is independent of pre-existing knowledge. A variational autoencoder is trained to efficiently reconstruct input data by passing it through an information bottleneck. The latent space of a variational autoencoder encodes information of the input in a compressed format in the form of low dimensional vectors. These vectors can be extracted and used as descriptors which accurately represent the input data. In this work, a variational autoencoder is trained to reconstruct protein data. When the autoencoder is jointly trained with a property prediction network such as retention time predictor network, the latent space of an autoencoder encodes meaningful information in the form of vectors. These latent space vectors can be exploited as meaningful descriptors which can accurately represent protein’s structural and biophysical properties. This work demonstrates the use of data driven descriptors to perform QSAR modelling and presents the opportunity to use artificial intelligence to build descriptors for property prediction.

BIOT 82

Overcoming cell shear in mAb harvests using novel harvesting technology

Masa Nakamura1, [email protected], Alexei Voloshin2. (1) 3M, St. Paul, Minnesota, United States (2) 3M Company, St. Paul, Minnesota, United States

Mammalian cell culture continues to be the workhorse for expression of high value recombinant therapeutic protein modalities. As the requirements for the manufacturing process performance, product quality and safety continue to become more stringent, mammalian cell shear during clarification operation has become a critical consideration in preventing product quality degradation and in purification train design. Shearing of the cells during centrifugation and depth filtration exposes the complex therapeutic products to intracellular enzymes, such as the reductases and proteases. Additionally, cell breakage dumps difficult-to-remove proteins, such as lipases, into the process fluid. Cell shear in the primary clarification stage also produces an insoluble fragment “tail” that interferes with membranes, capture chromatography, and polishing train performance. While cell shear can be attenuated by a proper depth filter selection and running the primary clarification stage in a “true chromatography mode”, this approach is not very robust due to the non-chromatographic nature of depth filtration media and variability of cell size depending on cell line and on culturing strategy. DRAFT

Here we present a novel harvesting approach that utilizes chromatographically defined AEX non-woven bed that is capable of capturing cells and cell fragments using charge. Chromatographic approach to clarification enables capture of insoluble particles at very low pressures and is equally effective over very wide range of particle sizes. The result is that cell shear can be very precisely controlled and minimized to background level regardless of cell line, cell density, or cell culture conditions. Furthermore, since the clarification media is chromatographically defined, the system is highly reproducible and scalable from high-throughput screening, to manufacturing scale. This enables one to accelerate process development from discovery to commercial deployment.

BIOT 83

Development of an efficient purification process for a recombinant E. coli protein

Andrew Lees2, [email protected], Samson Gebretnsae2, William H. Rushton1, [email protected], Chelsea Pratt1. (1) Process Chromatography, Bio-Rad Laboratories, Hercules, California, United States (2) FINA Biosolutions LLC, Silver Spring, Maryland, United States

Chromatography is the heart of protein purification and the development of the chromatography process is facilitated by automated chromatography systems. Purification protocols often follow the general three step scheme of capture, intermediate and final polish. In this case study, we present the process development for a recombinant E. coli protein using new generation chromatography resins. The target protein has an acidic pI and is expressed at high levels as a soluble, properly folded intracellular protein. Thus, the lysate is full of nucleic acids and cytosolic contaminants, which are also generally acidic. Using an automated chromatography system, we screened four anion exchange resins and compared their capture capacity verses a dextran/agarose resin. A hydrophilic epoxide AEX resin with large pores provided the highest binding capacity and recovery. The capture step was then scaled up from 1 ml resin to 100 ml resin. The intermediate purification step was developed by screening HIC resins and the final polish step was developed using a second anion exchange chromatography resin.

BIOT 84 DRAFT

Evaluation of viral clearance using a high resolution CEX resin and a ceramic apatite resin

William H. Rushton1, [email protected], Xuemei He2, [email protected], Akunna Iheanacho3, [email protected], Irene Chen2, Coral Fulton3, Jenifer Dean3. (1) Process Chromatography, Bio- Rad Laboratories, Hercules, California, United States (2) Research & Development, Bio-Rad Laboratories, Hercules, California, United States (3) Research & Development, Texcell - North America, Inc., Frederick, Maryland, United States

As part of the ICH Q5A guidelines, evaluating the removal of potential endogenous or adventitious viruses is done by performing viral clearance studies. These studies spike a known amount of model viruses into individual downstream process steps. A high resolution cation exchange resin and a ceramic apatite resin were tested to determine their ability to remove Minute Virus of Mice (MVM) and Xenotropic Murine Leukemia Virus (XMuLV). These viruses are a standard virus panel for an early phase viral clearance studies. The operating conditions were chosen to approximate the typical running conditions used with these chromatography resins. Results for the study provide potential viral clearance values for each media under the evaluated conditions. This can assist with future development work by providing guidance as to the ability of each media to remove viruses.

BIOT 85

Platform for purification of VHH-type antibody fragments

Tomas Bjorkman, [email protected], Eva Heldin, mats ander. Cytiva, Uppsala, Sweden

Protein A affinity chromatography resins have for a long time been used to purify monoclonal antibodies (mAbs) by the bioprocess industry at all scales. The binding of protein A mainly takes place in the interphase between CH2 and CH3 in the Fc region of the mAb (constant domains 2 and 3 of the Ab heavy chains However, it is known that protein A also can bind to VH3 sequences located on the variable heavy chain of the Fab region.. Thus, this binding may also be utilized for the purification of target molecules derived from the variable domain of the heavy-chain antibodies of camelids, the so called VHHs.

DRAFT

When developing the high capacity, highly caustic tolerant protein A resin, MabSelect™ PrismA, the VH3 interaction was re-introduced during the maturation of the affinity ligand. Here we will show examples of purification of mAb derived fragments including Fab and VHH, focusing at binding characteristics e.g. binding capacities, elution conditions and purification performance. It will be demonstrated how the VHH can be purified directly from an E. coli harvest with good yield and at high load

BIOT 86

Impact of disulfide reduced LMW on HMW formation rate and long term stability of monoclonal antibodies

Vivekh Ehamparanathan1, [email protected], Zhijun Tan2, cheng du3, Angela T. Lewandowski3, Sanchayita Ghose3, June Kuang2, Letha Chemmalil2. (1) Process Development, BMS, Devens, Massachusetts, United States (2) Bristol-Myers Squibb, Devens, Massachusetts, United States (3) Process Development, Bristol-Myers Squibb, Hopkinton, Massachusetts, United States

High molecular weight (HMW) species and low molecular weight (LMW) species are two critical quality attributes (CQAs) for all monoclonal antibody (mAb) Drug Substance (DS) and Drug Product (DP) with well-defined specification criteria. During the long-term stability study for a mAb, we observed one DP batch to be out of specification (OOS) for HMW. Subsequent investigation found that the LMW of that DS batch was higher than normal, but within specification limits. This study was aimed at investigating the correlation of initial LMW levels in DS to the rate of HMW formation during stability studies under various stressed conditions. Interestingly, It was determined that light exposure and initial LMW level can have a significant impact on the rate of HMW formation during long-term stability. This novel study can provide a unique dimension to pursue a holistic specification strategy for both DS and DP

BIOT 87

Split intein technology: Universal purification template for non-mAb proteins

Tatjana Best, [email protected], Oliver Rammo, Vanessa Kohl. MilliporeSigma, Darmstadt, Germany DRAFT

Currently, the purification of “non-standardized” or non-mAb proteins remains challenging and is associated with spending a significant effort in developing a downstream scheme for each new molecule individually. With the split intein technology we will bring a change to the market while introducing a generic downstream template to simplify this purification procedure. Due to the cleaving ability of inteins, the N-terminal intein is immobilized as the affinity ligand on the chromatographic support whereas the C-terminal intein is fused to the target as a tag. The high affinity between both inteins enables the tagged target to be purified from the lysate. The intein-mediated C-terminal cleavage reaction triggered by a change in buffer conditions, causes tagless release of the target into the eluate. The tag can be regenerated from the column and does not need to be further removed from the target such it is often the case using conventional tag purification schemes e.g. His-tag. In numerous case studies, we will present on the integrity of the C-terminal intein tag in upstream and discuss its impact to various microbial and mammalian expressed molecules (e.g. expression titer, solubility and biological activity). Further we will focus on the efficiency of the split intein downstream scheme in different process modes leading to a purity >90% of the tagless eluted target from crude feed stocks.

BIOT 89

Process development: A case study with a recombinant E. coli protein

Andrew Lees, [email protected], Natalia Oganesyan, Samson Gebretnsae. FINA Biosolutions LLC, Silver Spring, Maryland, United States

Chromatography is the heart of protein purification and the development of the chromatography process is facilitated by automated systems such as the Bio- Rad NGC chromatography system. Purification protocols often follow the general three step scheme of capture, intermediate and final polish. In this case study, we present the process development for a recombinant E. coli protein using new generation Bio-Rad chromatography resins. The target protein has an acidic pI and is expressed at high levels as a soluble, properly folded intracellular protein. Thus, the lysate is full of nucleic acids and cytosolic contaminants, which are also generally acidic. Using the NGC chromatography system, we screened four Bio-Rad anion exchange resins and compared their capture capacity verses a dextran/agarose resin. We found that Bio-Rad Nuvia Q, a relatively large bead (120µ) with large pores provided the highest capacity binding and purification. The capture step was then scaled up from 1 ml resin to 100 ml resin. The intermediate purification step was developed by screening HIC resins and the final polish step was DRAFT developed using another anion exchange resin. Additional tips and tricks for process development will be presented during the seminar.

BIOT 90

High productivity protein A process chromatography using novel convecdiff chromatography membranes in order to enable membrane- based single-batch use antibody affinity processes also for large scale commercial operation

DANIELA SOLUK, [email protected]. Application Seervices, Sartorius, Hackettstown, New Jersey, United States

Considerable efforts are being made to identify alternative chromatographic materials, devices, systems and related processes in order to solve current short-comings of resin based column chromatography processes. The main short-coming being, the high diffusional resistance that large molecules like e.g. antibodies experience in these materials, leading to long processing cycles (4-6 h). Targeted life-times of 75 -150 bind and elute cycles, taking many months or even years to be achieved, requiring their re-use, cleaning and respective validation, storage, sometimes column un- and repacking. Membranes, due to intrinsically high mass transfer rates, have long been recognized as a possible solution to overcome these limitations, enabling residence times of seconds rather than minutes, where full life-time utilization (75-150 cycles) can be achieved in 1-2 shifts. So far, mainly purely convective membranes are available on the market, that offer very high mass transfer rates, but limited binding capacity. In order to achieve levels of dynamic binding capacities acceptable for capture processes the convectively accessible inner surface of those membranes needs to be maximized, leading to small convective pore sizes (0.2-0.3 µm) and in turn to materials with high hydrodynamic resistance, high fouling propensity and shallow bed heights with excessive axial dispersion. We have developed a novel “convecdiff” membrane material, using materials that are well established in the bioprocess chromatography space, offerin

BIOT 91

Monitoring of antibody throughout bioprocessing by fluorescence reporter with applications for protein A column process control

Ujwal Patil1, [email protected], Mary Crum2, Binh V. Vu2, Atul Goyal2, Katerina Kourentzi2, Richard C. Willson2,1, [email protected]. (1) Biology and DRAFT

Biochemistry, University of Houston, Houston, Texas, United States (2) Chemical and Biomolecular Engineering, University of Houston, Houston, Texas, United States

Measurement of product titer is pivotal in bioprocess development. A typical monoclonal antibody production process involves multiple unit operations from upstream to final product release. We have developed a target-agnostic assay to measure the concentration of IgG or Fc-fusions during the discovery, cell culture development, and in-process phases of the biopharmaceutical process train. An antibody in complex, UV absorbing culture fluid cannot readily be detected, limiting the ability to take quick process decisions, and reducing productivity. Our fluorescence-based monitoring technology overcomes some of the current barriers by allowing homogeneous and near-instantaneous (< 30 seconds) IgG-measurements. We demonstrate the application of this technology for batch and continuous IgG detection. In our approach, IgG- containing cell culture fluid is mixed with fluorescein-labeled Fc-binding ligands to produce a shift in fluorescence polarization, which can be detected in a batch or continuous-flow format. While screening various Fc-binding ligands for monitoring of human IgG, we also identified ligands which signal the presence of IgG by changes in fluorescence intensity, which is readily detectable with inexpensive hardware. We observe significant shifts in both polarization and intensity in CHO cell culture fluid spiked with human IgG and can detect 0.005-0.1 g/L human IgG in protein A column breakthrough at a flow velocity of 80 cm/hr. We have successfully evaluated IgG-monitoring at a flow rate of 900 ml/hr without significant delay in fluorescence intensity signal. The fluorescence intensity information was applied to automatically switch columns when IgG-breakthrough of 0.005 g/L was reached. This technology provides a fast and reliable way to detect IgG in a flowing stream (or culture or batch samples) without elaborate sample preparation. Similar approaches can be envisioned for the detection of host cell proteins, DNA, and non-mAb products.

BIOT 92

Freezing and thawing scale-down model to characterize the sensitivity of biologics bulk drug substance to freeze-thaw processing

Cody Simmons, [email protected]. Manufacturing Sciences and Technology, Bristol Myers Squibb, Devens, Massachusetts, United States

Many therapeutic proteins are stored frozen at bulk drug substance stage to improve stability and shelf life prior to drug product conversion. Although DRAFT frozen storage can simplify manufacturing logistics such as transportation and inventory, physical and chemical stresses present during bulk freezing and thawing often require stringent control of process steps and conditions. The stresses incurred may include cryoconcentration, cold denaturation, liquid-ice interface, pH shifts, and crystallization of excipients. Many of which are affected by the freezing and thawing process conditions. Therefore, characterization of freeze-thaw sensitivity for desired applications should be performed to understand any impact. However, performing such studies at manufacturing-scale is costly; they require large quantities of drug substance and are logistically cumbersome. Here we examine a freeze-thaw scale-down model that has been developed to mimic manufacturing-scale freeze-thaw processes. Characterization shows that the model is capable of inducing cryoconcentration and controlling the duration of liquid-ice interfacial stresses that are consistent with manufacturing-scale. Studies using the model have successfully predicted the sensitivity of a drug substance to various manufacturing-scale technologies such as blast freezing, room temperature thawing, and active plate freeze-thaw. Further, the model has demonstrated effectiveness across multiple modalities such as Fc fusion proteins and fully human monoclonal antibodies.

BIOT 93

Investigation into the pH effects in multimodal cation exchange chromatography using NIST mAb and its domains

Ronak B. Gudhka1, [email protected], Kabir Dhingra1, John Schiel2, Steven M. Cramer1. (1) Chemical and Biological Engineering, Rensselaer polytechnic institute, Troy, New York, United States (2) Bioanalytical Science Group, National Institute of Standards and Technology, Gaithersburg, Maryland, United States

Recent work in our lab has shown the importance of certain domains of a monoclonal Antibody (mAb) to be important for binding to different Multimodal (MM) chromatographic systems. Identifying the binding regions on the protein surface is of very high importance for designing efficient bioseparation processes for biologics products. Interaction of certain mAbs was governed by the (Fab)2 domain whereas other mAbs interacted via the FC region of the mAb. In this work, we extend this investigation to study the effect of pH on the binding of mAb to these MM chromatographic systems and evaluate the selectivity trends in these systems. For this study, we used NIST mAb that has been extensively studied by many research labs across the country. First, linear salt gradient chromatography experiments were performed with NIST DRAFT mAb on various MM cation exchange chromatographic systems. To further investigate on the regions on the protein surface that were involved in binding, the NIST mAb was enzymatically digested to get the Fab, (Fab)2 and FC domains. The retention of these domains was evaluated on MM chromatographic resins as a function of pH. There was a shift in the important domain of interaction from pH 5 to 7. While the FC domain was important for binding at pH, the Fab domain appeared to be important for binding at pH 7. By comparing these results with Electrostatic Potential (EP) and Surface Aggregation Propensity (SAP) maps under different conditions, we developed a heuristic understanding of how these shifts were a function of protein surface properties. This work has enabled us to develop a deeper understanding of how different regions on a protein surface interact with the MM chromatographic resins under various conditions and how the pH can be used to modulate the selectivity in these MM systems.

BIOT 94

Optimization of final filtration operations using single-use systems

Andrew Koch, [email protected]. MilliporeSigma, Burlington, Massachusetts, United States

Final sterilizing filtration is the last operation in downstream processing to assure sterility of medicinal products for human use. There is extensive regulatory guidance for final filtration encompassing both expectations for microbial retention, as well as approaches for confirming filter integrity. As single-use systems are increasingly being adopted in final filtration, there is a focus on implementing a more holistic approach to assuring sterility of the system that encompasses both filters and components. This presentation will highlight new developments in filter technology that reduce the risk of introducing contamination in pre-sterilized redundant filtration systems. They also offer opportunities for:

Simplified pre-use integrity testing of critical sterile filters Maximized product recovery of formulated product Streamlined single-use assembly design with fewer connections These developments in assembly design reduce risk, improve operation and increase robustness of this critical filtration step. In an industry focused on driving efficiency, they offer an important advancement in process risk mitigation and yield improvements.

BIOT 95 DRAFT

Impact of antibody disulfide bond reduction during harvest on downstream processing and product stability

Sandra Rios1, [email protected], Sketa Patel2, InKwan Han1, Hong Li1. (1) Merck, Kenilworth, New Jersey, United States (2) Celgene, Warren, New Jersey, United States

Therapeutic monoclonal antibodies (mAbs) are primarily produced in mammalian cell culture systems, which can be susceptible to disulfide bond reduction during harvest. This reduction, which is dependent on cell line and cell culture process, has been correlated with the release of reducing components from the cells and depletion of dissolved oxygen before, during and after harvest. Consequently, these factors can lead to disulfide reduction during storage at room temperature prior to initial purification via Protein A chromatography. This work seeks to understand the impact of disulfide reduction on product quality and on the downstream unit operations. mAb disulfide bond reduction appears to be reversible as the redox potential changes during the purification, however, we observed that the levels of fragment and aggregate increase during the low pH treatment proportional to the amount of free thiol present. The increase of such product related impurities impact the process yield and in process intermediate stability. Furthermore, the drug substance produced from clarified culture with disulfide reduction clearly showed a higher susceptibility to temperature induced changes on stability, potentially indicating incomplete assembly. In addition, for several mAb cases where disulfide bond reduction has been observed, we compared it with DTT induced mAb disulfide reduction. This comparison showed a different mAb sensitivity order to reduction by DTT than the observed at harvest indicating a potential different mechanism for disulfide reduction.

BIOT 96

Gas to liquid continuous flow bioreactor for conversion of biogas and waste methane to organic acids with hydrogel-encapsulated methanotrophs

Nathan C. Ellebracht1, [email protected], Fang Qian1, Samantha Ruelas1, Joshua DeOtte1, Hawi Gemeda1, Jennifer Knipe1, Michael Guarnieri2, Sarah Baker1. (1) Lawrence Livermore National Laboratory, Livermore, California, United States (2) National Renewable Energy Laboratory, Golden, Colorado, United States DRAFT

Underutilized methane sources continue to grow with increasing biogas generated from waste processing, including digestion of solid wastes and waste-water treatment. Biological upgrading of methane and/or biogas by methanotrophic microorganisms is a promising utilization approach. Metabolic engineering of methanotrophs like Methylococcus capsulatus has been successful in converting methane to a variety of organic acids including muconic, lactic, and succinic acids at relatively high titers. However, liquid fermenters utilizing methanotrophs are fundamentally mass transfer limited due to the poor solubility of methane in water. As agitated fermentation broths with sparged gases are ineffective for these bioconversions, an alternative bioreactor design is presented to overcome these limitations herein.

Whole cell methanotroph biocatalysts are encapsulated within a fixed hydrogel support to contact a continuous gas phase in a flow reactor. Methanotrophs encapsulated in biocompatible hydrogels such as cross-linked PEGs are shown to be viable and function as effective biocatalysts for months. Here, hydrogel biocatalysts are mechanically supported within 3D printed lattice scaffolds, allowing for controlled catalyst geometries for diffusion and reaction. This gas flow reactor was utilized for the conversion of simulated and real biogas feedstocks with these hydrogel-encapsulated biocatalysts. Early work demonstrated that reducing feature sizes to sub-mm scale achieves more than an order of magnitude increase in volumetric productivity of methanotrophs. Initial demonstrations at mL-scale were scaled up by several orders of magnitude in flow reactor operation. Wild type methanotrophs were utilized to study methane consumption and process parameters, and engineered strains were studied for organic acid productivity. Various 3D printed scaffold geometries were tested and tradeoffs between mass transfer improvements and overall biocatalyst loading were assessed. The effects of process variables including feed composition, pressure, and gas flow rates on flow bioreactor performance were evaluated. Finally, product recovery approaches and reactor configurations are addressed.

BIOT 97

3D printing of flexible ceramics via sol-gel method

Javier Lopez Navas1, [email protected], Liu Zhang2, King Lun Yeung1,2. (1) Department of Chemical and Biological Engineering, The Hong Kong University of Science and Technology, Hong Kong, Hong Kong, Hong Kong (2) Division of Environment and Sustainability, The Hong Kong University of Science and Technology, Hong Kong, Hong Kong, Hong Kong DRAFT

Additive manufacturing (AM) technologies have gained increasing interest in biomedical engineering during the last decades due to their versatility and capacity to synthesize materials with specific properties. The layer by layer fabrication process allows for a high degree of customization in the composition and geometry of the fabricated structures. Among all the materials used in AM, ceramic-based materials have proven to be effective in promoting bone tissue regeneration because of their similar chemical composition with the bone tissue. In terms of bone tissue regeneration, both micro and macro porosity features are desirable to promote cell differentiation and tissue regeneration. By using the sol-gel process to form ceramic structures, a high level of micro-porosity can be achieved. On the other hand, additive manufacturing technologies allow to control the macro-porosity of the structure. In the present research, a combined approach of both additive manufacturing and sol-gel technologies is used to create flexible ceramic materials with controlled micro and macrostructure. A conventional 3D printer system has been modified to create three-dimensional structures of a sacrificial material which is used as a mold for the sol-gel process. The results show how different 3D patterns of the flexible ceramic material can be created while preserving the mechanical properties of the original material via the sol- gel process inside a container.

BIOT 98

Development and deployment of bioreactor scale-up modeling software in Janssen biotherapeutics development-API

Christopher Canova, [email protected], Jeffrey Cohen, Kevin Clark, Eugene J. Schaefer. Janssen Pharmaceuticals, Malvern, Pennsylvania, United States

A combination of higher cellular growth processes, shortened timelines, and an ever-increasing array of biomolecules and manufacturing sites necessitated a need for increased internal bioreactor modeling capabilities in Janssen Biotherapeutics Development. In response, a first-principals based bioreactor model was created. The model probes the operating space and sensitivities for a unique facility and project combination. Data that is readily available at the laboratory scale is input into the model to make its operational predictions, which can then be transferred to the manufacturing site prior to the batch manufacture. This model was then packaged as a standalone application using Matlab and deployed for departmental use on the JnJ Appstore; This process allows for controlled deployment of model improvements and ensures that all scientists are working with the same set of DRAFT assumptions and equations. A training curriculum was developed to increase departmental knowledge of bioreactor modeling and drive accountability for bioreactor scale-up to the project teams. Modeling the operational space can help to identify equipment limitations prior to manufacture and may allow for right-the-first-time processes with no need for engineering trial runs.

BIOT 99

Flexible intensified manufacturing strategies to support existing and facilities of the future

Hemanth Kaligotla1, [email protected], Melisa Carpio2, Mandar Dixit3. (1) Sartorius, Bohemia, New York, United States (2) Cell Culture Technologies, Sartorius, Bohemia, New York, United States

Abstract With drug pipelines becoming diverse, manufacturing in biotech is evolving to require the use of different cell lines and the production of multiple types of complex proteins in the same facility. As the products vary, the manufacturing modes and volume requirements change along with the regulatory landscape. The need for scalable, flexible, and single-use manufacturing strategies is becoming increasingly important. The poster will discuss case studies on 1) demonstrate the intensification of upstream process development to an accelerated seed train and scale to a PAT enabled bioreactor production platform. Various combinations of rocking motion (RM) and stirred tank bioreactors (STR) allow for multiple seed train and manufacturing scenarios using fed-batch, intensified, or perfusion cultures 2) demonstrate the development of counterflow single-use centrifugation on a CHO fed-batch process with ~5g/l product titer, ~8% biomass, and ~ 90% viability to double the depth filtration performance for cell culture clarification.

BIOT 101

END-to-end data paradigm for cell line development

Jasmine Tat, [email protected], Kim Le, Christopher Tan, Huong Le, Kristine Daris, Fides Lay, Jonathan Diep, Natalia Gomez, Jennitte Stevens. Amgen, Thousand Oaks, California, United States

Cell Line Development holds a critical role in creating and selecting cell lines that become the Master Cell Banks for clinical and commercial biomanufacturing. Critical decisions are made during cell line development DRAFT processes, such as the final subclone selection which typically will be locked throughout the life of a project. Datasets used to drive these decisions traditionally requires resource-intensive manual cell culture experiments and analytics to generate hundreds to thousands of datapoints per experiment. However, these datasets have largely been composed of discrete or endpoint measurements in which critical decisions are dependent upon. Furthermore, data manipulation and analysis are often conducted on a per-project basis. We have identified the need to shift from processes governed by empirical data siloes to a data-centric platform that provides holistic insight and predictive power. Our approach to entering an end-to-end data paradigm in Cell Line Development involves adopting more high-throughput, data-rich technologies, establishing an Enterprise Data Lake (EDL), creating visualization tools, deploying automation, and applying off-the-shelf machine learning algorithms.

High-throughput technologies and methods were implemented, such as Berkeley Lights Beacon, ddPCR, RNA-seq that increase the scale and granularity of data we can now interrogate. Laboratory and data automation tools were deployed to reduce resources required to generate and manipulate data. Amgen’s EDL was established to house historical and novel big data and serves as the central source to enable digital tools and the ability to effectively manage and mine data. Real-time visualization dashboards were created and deployed to create instant trends and tables—eliminating hours of data extraction, manipulation, and analysis. Multi-variate data analysis was integrated into our decision-making process to leverage the availability and granularity of increased datasets and provide holistic insight across historical projects. Predictive models using open-source python-based machine learning algorithms were created to predict downstream features and provide future insight.

Altogether, in our new paradigm, data will be the central driver to ensure clones of high quality are produced from the moment a cell is transfected to when a clonal bank is created. This will open opportunities to shrink timelines and connect disjointed processes from research to commercialization.

BIOT 102

Ionic liquids-based antimicrobial coating for air disinfection and purification

Xizi SONG1, [email protected], Wei Han1, Yutang LUO1, King Lun Yeung1,2, Joseph Kai Cho Kwan1,3. (1) Environment and Sustainability, Hong DRAFT

Kong University of Science and Technology, Hong Kong, KWL, Hong Kong (2) Department of Chemical and Biological Engineering, Hong Kong University of Science and Technology, Hong Kong, Hong Kong (3) Health, Safety and Environment Office, Hong Kong University of Science and Technology, Hong Kong, Hong Kong

Higher requirement of disinfection and purification technology has been proposed due to frequent outbreaks of infectious diseases and long-time indoor activities. Comparing current air purifiers with multilayer filters, an effective coating on single substrate is more economic and energy-saving. Ionic liquid (IL) is a category of highly tunable salts which can maintain their liquid state at room temperature, thus being applied on the surface of different substrates. Among all ILs with various applications, imidazolium-based ILs are one of the most common ILs which have been proven good antimicrobial activities.

This study develops imidazolium-based IL coatings supported on porous substrates. By manipulating surface adhesion, uniform IL-based antimicrobial coatings were formed for air disinfection and purification. By adjusting cations and anions, as well as introducing other disinfectants, the IL-based antimicrobial coating can be formulated to exhibit synergistically antimicrobial activities. The result indicates that the IL-based antimicrobial coating not only effectively inactivates the microorganisms contacting the substrates or suspend in the air, but also enhances the filtration efficiencies for particulate matter of porous substrates without increased flow resistance.

BIOT 103

Rapid production of glycoengineered proteins via CHO transient transfection

Barry F. Wolf, [email protected], Ping Huang. Abbvie , Redwood City, California, United States

Therapeutic protein N-linked and/or O-linked glycoform characteristics can impact ADCC, CDC, immunogenicity, and other important drug properties. The ability to produce proteins with a wide variety of glycosylation patterns is crucial for the testing and selection of optimal candidates to move from Discovery to Clinical development. This work demonstrates that it is possible to control a variety of glycosylation patterns, through growth medium supplementation, in a CHO transient transfection system. Through chemical supplementation, galactosylation, fucosylation, mannosylation, and DRAFT arabinosylatoin were altered, quantified, and compared with an identical protein expressed in a stable CHO cell line and altered through similar medium supplementation. The ability to quickly and controllably express glycoengineered proteins by means of a transient CHO process can aid in speeding candidate selection and time to clinic.

BIOT 104

Homodimeric variant of an aptamer generated from LIGS activates TCR- CD3ε complex

Lina Freage1, [email protected], Deana Jamal1, Prabodhika Mallikaratchy1,2,3. (1) Chemistry, City University of New York-Lehman College, Bronx, New York, United States (2) Chemistry and Biochemistry, Graduate Center, PhD Program in Chemistry and Biochemistry, New York, New York, United States (3) Cellular and Developmental Biology, Graduate Center, PhD Program in Molecular, Cellular and Developmental Biology, New York, New York, United States

Nucleic acid aptamers represent a broad class of nucleic acids that reveal specific binding properties towards various ligands, including low molecular- weight substrates as well as macromolecules. Through binding specific classes of molecules, aptamers can be targeted towards cells displaying such substrates of interest on their surface. We recently introduced an aptamer named ZUCH-1 against the target T-cell receptor-cluster of differentiation three epsilon (TCR-CD3ε), expressed on human T cells utilizing the method called Ligand Guided Selection (LIGS). The aptamer ZUCH-1 showed high affinity and specificity towards the desired target, T-cell receptor Complex- CD3ε. In our current work, herein we report systematic truncation followed by modification utilizing synthetic nucleic acids, notably locked nucleic acid (LNA) and a 2’OMe RNA base in anti-TCR-CD3ε to improve the aptamer’s affinity without compromising specificity. Furthermore, dimerization of the modified aptamer showed higher avidity, and the observed avidity is comparable to corresponding monoclonal antibody. Functional studies using dimeric anti- TCR-CD3ε aptamers against TCR-CD3 expressed in cultured cells demonstrated that dimeric variants can activate TCR-CD3 and this activation is comparable to its corresponding monoclonal antibody. Thus, we will introduce a first-of-a-kind aptamer-based TCR-CD3ε activator that provides a synthetic tool to investigate TCR-CD3 mediated immunological mechanisms and synthetic immunotherapeutic development.

BIOT 105 DRAFT

Metalloporphyrins functionalized micellar artificial organelles with nitric oxide synthase activity

Xiaomin Qian, [email protected], Isabella Westensee, Catarina Fernandes, Edit Brodszkij, Brigitte Stadler. Interdisciplinary Nanoscience Center (iNANO), Aarhus University, Aarhus, Denmark

Artificial organelles (AOs) are nanosized compartments that are created with synthetic materials to mimic specific functions inside cells, and one of them being performing enzymatic reactions. While natural enzymes usually suffer from low stability in ambient environment and high cost in preparation and purification, artificial enzymes (AEs) aim to achieve similar catalytic activities as their natural counterparts but circumventing the abovementioned drawbacks. Inspired by nature, metalloporphyrins and their derivatives have been extensively studied as potential candidates in AEs as they are ubiquitously found in some enzymatic hemeproteins, for example, Cytochrome P450s, which function as monooxygenase and are mainly responsible for catalyzing oxidative reactions in living organisms. In this work, we presented micellar artificial organelles functionalized with metalloporphyrins to mimic nitric oxide synthase (NOS) intracellularly (Scheme 1). Water-soluble porphyrins were synthesized and coordinated with three different metal ions, Fe, Mn, and Co. Their capability of mimicking NOS by converting L-arginine to L-citrulline and nitric oxide, the latter is a vital signaling molecule in various physiological and pathological processes, in the presence of H2O2 were first examined and compared. Further, the metalloporphyrins were conjugated to micelles assembled from an amphiphilic block copolymer poly(cholesteryl methacrylate)-block-poly(2- carboxyethyl acrylate) (PCMA-b-PCEA) following an EDC/NHS coupling reaction. The metalloporphyrin-containing micelles showed negligible cytotoxicity at low concentrations, and good cellular uptake efficacy was observed. Finally, the catalytic activity of the metalloporphyrin seemed to have retained inside the cells. Taken together, this work represents a paradigm of immobilizing a small organometallic on an AO to perform enzymatic reactions in cells, which could be potential for treating deficient cells with lost functions for therapeutic purposes.

BIOT 106

Pall Xpansion® bioreactor supports progenitor cell growth to >1 million cells/cm2 and proper cell differentiation DRAFT

Andrew Laskowski, [email protected], Siddharth Gupta. Pall Corporation, Westborough, Massachusetts, United States

With the recent increase in early phase cell therapy clinical trials, there is a need for a manufacturing platform which can be implemented in research labs and easily scaled up to expedite process development studies, pre-clinical testing, and large-scale expansion of adherent human cells. Although various platforms such as traditional flatware and stirred tank bioreactors with microcarriers exist and have been well characterized, traditional flatware often does not allow for tight control of the cellular environment and is not scalable. Pall’s Xpansion single-use bioreactor offers a tightly-controlled, scalable manufacturing platform for cell therapy applications, allowing for expedited process development, pre-clinical testing and large scale, high-density expansion of progenitor cells while maintaining their differentiation capacity.

We have previously demonstrated efficient expansion of both epithelial cells (Vero, HEK293), human mesenchymal stem/stromal cells (hMSC), and progenitor cells in the Xpansion single-use bioreactor. Here we extend the progenitor cell findings by successfully expanding a proprietary progenitor cell to extremely high cell densities (> 1 million cells/cm2) reproducibly for 8 batches. In summary, all batches resulted in 1) cell expansion to > 1 million cells/cm2 and equivalent cell densities to the traditional flatware process control; 2) equivalent differentiation to a more mature cell fate as indicated by expression of key cell surface markers; and 3) self-assembly into characteristic three-dimensional structures.

BIOT 108

Considerations during working cell bank manufacturing for long term robustness and reliability

Jie Xu, [email protected]. Merck, West Point, Pennsylvania, United States

Working Cell Banks (WCBs) are a critical starting material for biologics manufacturing, yet the process by which WCBs are produced is often less characterized than the rest of the manufacturing process. The objectives of this project were to further characterize a typical WCB production process using a mammalian monoclonal antibody (mAb) cell line and identify areas of consideration during WCB process development and technology transfer/commercialization to ensure long term robustness and reliability. Case studies demonstrating enhanced WCB process understanding and characterization will be discussed, specifically: temperature sensitivity of cells DRAFT during expansion stages, impact of cryoprotectant contact time during the pre- freeze stage, optimizing growth for yield enhancement, and new technology introduction including an automated cell counter to replace manual cell counting. Together, consideration of these factors during initial WCB process development will help ensure that more robust and reliable WCBs are in place for subsequent technology transfer/commercialization and mitigate potential WCB supply risks for biologics manufacturing.

BIOT 109

Evaluation of retrovirus-like particle expression in fed-batch and perfusion as a function of cell density

Arshan Nazempour, [email protected], Douglas Rank, Joe Orlando, Michael Phillips, Audrey Chang, Herb Lutz. MilliporeSigma, Westford, Massachusetts, United States

Viral safety is a key concern in the growing field of biologics produced by mammalian cells as viral contamination could have serious clinical consequences. Viral contamination can arise from different sources including but not limited to cell bank with latent or persistent virus infection, viruses expressed endogenously by the cells, raw materials, contaminated reagent from an operation unit such as affinity chromatography and handling. To ensure the safety of human therapeutics, it is not only important to understand the sources of viral contamination but also required by regulatory agencies to ensure prevention, detection and removal of such contaminants.

Retrovirus-like particles (RVLPs) are expressed endogenously by Chinese hamster ovary cells and are considered as one of the sources for viral contamination. As so, it is integral to biologic manufacturing to investigate the life cycle and to quantify the level of RVLPs throughout bioprocessing unit operations. Considering today’s industrial trend toward continuous bioprocessing, there is a need to carry out studies into the life cycle of RVLPs in newly developed solutions/unit operations. One such area of interest is perfusion bioreactor processes. Compared to traditional fed-batch approaches, perfusion mode provides a favorable environment to the cells by continuous metabolite removal and nutrient addition. Rapid removal of toxic byproducts along with continuous supplement of fresh medium enable achievement of high cell density within perfusion bioreactors. However, the impact of increased cell density on RVLP expression has not been studied. In this poster, we present our investigation on evaluating the level of RVLPs measured by RT-qPCR as a function of cell density in fed-batch and perfusion DRAFT bioreactors. To the best of our knowledge, our study is the first direct comparison of retrovirus content between the two different modes of bioreactor operation. This has implications for differences in the risk from therapeutics produced by each mode.

BIOT 110

Characterization of hyperosmotic adapted CHO host cells for fed-batch and intensified perfusion process

Xiaowen Wang, [email protected], Jack Huang. Process Research Development, Merck & Co., Inc., Kenilworth, New Jersey, United States

Increased osmolarity of medium has immediate and prolonged impacts on cell culture performance of Chinese Hamster Ovary (CHO) cells expressing monoclonal antibodies (mAbs), including cell growth, cell-specific productivity, and product quality attributes. Cells under hyperosmotic challenge are subjected to multiple stresses, such as mitochondrial depolarization, reactive oxygen species production, and DNA damage, which can lead to cell cycle arrest and eventual cell death. However, hyperosmotic pressure correlates to high specific productivity in CHO cells, which is often observed in fed-batch culture. In addition, concentrated media have been recently implemented to maximize the efficiency and productivity of perfusion culture, which also exposes cells to these high osmolarity challenges. Therefore, cell lines that are resistant to hyperosmotic stress are needed to enable robust, productive, and flexible (intensified or non-intensified) cell culture processes. In this work, we evaluate the impact of adapting CHO cells to hyperosmotic medium. The adapted cells are characterized and compared to the parental cell lines for phenotypic changes. In addition, adapted mAb-producing cell lines are compared to parental lines for product quality attributes such as N- glycosylation and sialyation. Comparison of the cell culture performance and product quality attributes between adapted and parental cells are investigated in both fed-batch and perfusion cultures. Omics approaches are applied to the characterization of adapted cells to fully understand the cellular hyperosmotic response at the genetic and molecular levels, and to guide the design of rational cell engineering strategies to develop highly productive and hyperosmotic-resistant cell lines.

BIOT 111

Use of PAT to facilitate development and tech transfer of a cell culture process DRAFT satish cheepala1, [email protected], George X. Zhou2. (1) Merck & Co Inc, West Point, Pennsylvania, United States (2) Merck Co Inc, Rahway, New Jersey, United States

Development of process analytical technology (PAT) tools in parallel with cell culture development provides a mechanism to design, analyze and control biological processes through measurement of process parameters which affect critical quality attributes (CQAs). Some examples of PAT tools used in cell culture processes include capacitance probes and Raman probes, which can provide real-time monitoring of performance parameters such as metabolites, titer, and amino acid concentrations. The goals of this project are to 1) develop methods for online monitoring of metabolites and titer, using in- line Raman probe, and 2) develop in-line capacitance probe enabled real-time monitoring of VCD. Capacitance and Raman data is collected from multiple bioreactor runs and used to develop partial least square (PLS) calibration and prediction models for real time monitoring of cell culture process parameters. For example, multiple data points for VCD and metabolites derived from Raman models will enable optimization of the cell culture process to increase productivity and control the variability of the CQAs. In addition, real time monitoring of VCD using capacitance probes will help to streamline the transfer of cells to production bioreactor. In conclusion, PAT tools play a critical role in developing robust cell culture processes. Furthermore, experience gained in developing the PLS models and application at small scale will simplify the process of transferring PAT technology to manufacturing sites.

BIOT 112

Modulating cell line stability through media and process selection for a monoclonal antibody

Ashley Bui, [email protected]. K15-H4, Merck, Kenilworth, New Jersey, United States

Loss of productivity during the industrial production of monoclonal antibodies is sometimes observed in recombinant Chinese Hamster Ovary cells. This work highlights a case study demonstrating upstream process development efforts to address challenges associated with cell line instability and low production yield. Two molecular mechanisms known to cause production instability are loss of recombinant gene copies after subculturing over many cell generations and epigenetic gene silencing via DNA methylation. Genetic characterization of our cell line revealed no loss of gene copies and that DRAFT addition of a DNA methylation inhibitor, 5-Aza-2'-deoxycytidine (Decitabine), had a negligible effect suggesting that the change in phenotypic performance is not attributed to this typical genetic mechanism. Instead both clone stability and productivity are heavily dependent on passaging and production media, along with the feeds used as part of a fed-batch process. Evaluation of multiple commercial and in-house media and feeds led to the identification of an optimized combination, which resulted in minimal loss of stability over 60 population doublings while increasing titer in the fed-batch production cultures by 4-fold. Further development revealed that additional metal supplementation during the production step corrected the undesired metabolic profiles and phenotypes associated with cell age. Overall, we were able to develop a robust and manufacturable seed expansion and production scheme with stable and improved cell culture performance while maintaining comparable product quality attributes.

BIOT 113

CelliconTM perfusion solution

Allyson Fournier1, [email protected], Alison M. Dupont2. (1) R&D, MilliporeSigma, Bedford, Massachusetts, United States (2) EMD Millipore, Bedford, Massachusetts, United States

Heightened cost pressures are challenging biopharmaceutical facilities to operate with increased speed and flexibility. Process intensification has become a strategic focus aimed at efficiently utilizing existing facility assets to deliver increased productivity. A key example of this is utilizing perfusion in place of traditional batch or fed-batch processes in upstream. Adoption of perfusion in N-1 bioreactors yields a reduction in process time and/or increase in manufacturing capacity without increasing volume capacity. A key technology within the perfusion application is the mechanism for removing fluid from the bioreactor while maintaining the cells in the system. Historically, the technologies available for retaining cells in the bioreactor have had limitations either mechanically, or with separation performance aspects such as throughput, efficiency, product retention, or process scalability. Due to recent advances in the cell retention technologies, in parallel to advancements in cell line and media development, perfusion processes are becoming a more viable alternative to expand performance beyond what traditional batch/fed- batch platforms can attain. To address this market need, MilliporeSigma is developing a family of pre-sterilized, high-throughput perfusion filtration solutions/systems with ease-of-use in mind and scalable from bench to production. DRAFT

BIOT 114

Evaluating perfused seed trains through process modeling and cost analysis

Jeffrey Barna, [email protected], Habib Horry, Douglas Rank. MilliporeSigma, Acton, Massachusetts, United States

The bioprocessing industry is driving towards intensified processes to reduce the cost of goods and/or increase productivity. From an upstream perspective, there are two areas of focus - intensified seed and intensified production. Perfused seed is one specific upstream process that plays a role in both areas. To gain a better understanding of the benefits of incorporating perfused seed trains in mAb production, BioSolve process modeling software was used to perform a holistic cost analysis. Process models of several different perfused seed train options were generated, as well as a traditional seed train model to compare to. In this study, scenarios were run evaluating the effect of variables such as production/seed train ratio, number of production bioreactors, titer, and production duration. Results showed that under certain scenarios, perfused seed train options could deliver lower cost of goods, increase product throughput, or a combination of both.

BIOT 115

Identifying glucose feeding profiles to control growth rate in CHO cells using precision pumping in the ambr®250 bioreactor system

Stephanie Klaubert1, [email protected], Kathryn Elliott2, Madison Williamson1, Sarah W. Harcum3. (1) Clemson University, Clemson, South Carolina, United States (2) Bioengineering, Clemson University, Clemson, South Carolina, United States

Fed-batch bioreactors are the current industry standard for production using CHO cells due to ease of scale-up and compatibility with large-scale manufacturing. In fed-batch microbial cultures, nutrients are continuously or semi-continuously fed, while the effluent may be removed and if so, discontinuously. Typically, fed-batch culture conditions are used to adjust metabolic rates and overcome substrate inhibition by controlled substrate feeding. Currently, most CHO cell culture fed-batch feeding strategies utilize daily bolus feeding. This discontinuous feeding approach often relies on off- line daily measurement of a key substrate (e.g., glucose) that targets to maintain this metabolite at or above a desired concentration. With daily bolus DRAFT feeding, cells are typically not limited in growth and, therefore, exhibit an unrestricted exponential growth rate. By establishing an exponential feeding profile, a desired growth rate could be achieved by maintaining CHO cells in a state of substrate limitation according to a predetermined desired growth rate, much like already commonly practiced for microbial fermentations.

Preliminary data shows IgG production increases as cell growth begins to slow. A lower controlled growth rate may allow for redirection of cellular resources to protein production and reduce the burden of oxygen transport at high cell densities. Thus, it was hypothesized that slowing the growth rate earlier in the fed-batch phase for CHO cells would result in increased IgG production by allowing cells to redirect cellular resources to IgG production for a longer duration and ultimately result in higher IgG concentrations with improved protein quality. Pumping systems within the ambr®250 were used to control glucose feeding to achieve desired submaximal exponential cell growth rates. The results of this study show the impact of controlled glucose feeding on the growth profiles of CHO cells in fed-batch cultures and protein productivity.

BIOT 116

Substrate-functionalized single-walled carbon nanotubes as modular activity probes for optimization of lignocellulosic biomass pretreatment

Nathaniel Kallmyer, [email protected], Nigel Reuel. Chemical and Biological Engineering, Iowa State University, Ames, Iowa, United States

Conversion of biomass to precursor sugars and alcohols is complicated by the composite structure of lignocellulose which renders targeted cellulose inaccessible behind a matrix of hemicellulose and a coating of lignin. This structure necessitates pretreatment of biomass with chemical, physical, or biological means. Biological pretreatment with enzyme formulations or “cocktails” that degrade each component holds great promise as a sustainable method; however, current efficiencies of these formulations limit the cost- competitiveness of this strategy. While the need to improve enzyme efficiency has been recognized, methods of quantitatively measuring activity are limited by time-consuming or insensitive colorimetric and viscometric assays which require specialized equipment and often only report activity indirectly. Here, we demonstrate a library of substrate-functionalized single-walled carbon nanotubes (SWNT) as modular probes for characterization of cellulase, xylanase, and lignin-modifying enzymes. We also explore the effects of substrate substitution ratios and polymer lengths on consistency and DRAFT selectivity (exo- vs. endo- preference) of cellulases using carboxymethylcellulose-functionalized SWNT. Lastly, we demonstrate the utility of such probes by screening an array of enzyme formulations with respect to cellulytic, hemicellulytic, and lignin-modifying activities with a single- wavelength fluorescence detector. We also show progress towards such screening in micro-droplet reactors. These sensors will find utility as real-time activity probes for improvement of enzyme operating and formulation conditions as well as rapid-screening tools for directed enzyme evolution.

BIOT 117

At-line nutrient profiling from cell media in 7 minutes with a new integrated analyzer

Ji Young L. Anderson1, [email protected], Kathryn Elliott2, Kenion H. Blakeman1, Colin M. Gavin1, Sarah W. Harcum3, Glenn A. Harris4. (1) 908 Devices, Braintree, Massachusetts, United States (2) Bioengineering, Clemson University, Clemson, South Carolina, United States (4) 908 Devices Inc., Boston, Massachusetts, United States

Characterization of small molecules is essential when optimizing media formulations to improve process productivity. Simple biochemical analyzers are limited by small target lists, so LC systems are often used to perform more expansive analyses. These approaches suffer from large sample volume requirements, time consuming sample prep, and lengthy run times. Presented here is a new automated CE-MS benchtop analyzer for targeted media analysis in under 7 minutes. An analytical target list was developed spanning 33 metabolites including amino acids, biological amines, vitamins, and dipeptides. On-board, automated algorithms were used to generate quantitative calibrations and verify performance throughout data collection. The analyzer required no sample derivatization and a minimum sample volume of approximately 10 µL.

The system was first evaluated alongside the standard LC method by running a chemically-defined CHO media where the reported amino acid concentrations from both methods matched. Nutrient formulations were then analyzed with Dulbecco's Modified Eagle's Medium/Nutrient Mixture F-12 Ham (DMEM/F12) from five different vendors. Up to 2.5x differences between the reported formulation of the DMEM/F12 samples were observed and no formulation matched the expected component profile. Finally, the analyzer was tested doing spent media analysis of a batch fed process of CHO NIH VRC01 cell line grown under different ammonia stress concentrations on an DRAFT ambr®250. All bioreactors displayed net increases in the levels of Ala, Arg, Asp, Glu, Gly, His, Ile, Leu, Lys, Phe, Thr, Trp, Tyr and Val, with Gly levels showing increases in excess of 8-fold initial levels. Other media components either steadily decreased in concentration or were completely depleted by the end of culture. In the future, the rapid analysis capability of the analyzer combined with the low sample volume requirements would allow for more frequent spent media analysis to occur for real-time observation of these process changes and associated control strategies.

BIOT 118

Quantifying cell residence time in a perfusion system

Nick Marchand, [email protected], Michelle Olson, Todd Sanderson, Mike Collins. R&D, Pall Corp, Westborough, Massachusetts, United States

Most of the available perfusion technologies for mammalian cell culture involve removing the culture from the bioreactor, separating the cells from the spent media, and returning the concentrated culture to the bioreactor. While the environment in the bioreactor is tightly controlled (e.g. dissolved oxygen, temperature, etc.) there is less control of the environment in the external perfusion loop. The residence time of the cells in that loop can therefore be a critical factor when designing a cell culture process or associated perfusion system. Here we share a method for quantifying the cell residence time through a complex fluid pathway. We use populations of unmarked and marked (e.g. GFP-expressing) cells to equilibrate a perfusion system, and track cells as they pass through it. We will share work using this method to evaluate cell residence time in an alternating tangential flow (ATF) filtration system as well as a prototype acoustic separation system. We also show how the method can be used to understand the impact of system inputs such as flowrate and hold-up volume.

BIOT 119

Using RNA-Seq to characterize underlying gene expression changes in observed alanine metabolic shifts of ammonia-stressed Chinese hamster ovary (CHO) cell cultures

Benjamin Synoground1, [email protected], Kathryn Elliott1, Christina Leuze2, Tom Caldwell1, Cameron Schnabel1, Sarah W. Harcum1. (1) Department of Bioengineering, Clemson University, Clemson, South Carolina, DRAFT

United States (2) Department of Molecular Biotechnology, Heidelberg University, Heidelberg, Germany

Recombinant cell cultures used in the production of biopharmaceuticals naturally produce and accumulate ammonia (NH3) as a metabolic waste byproduct, largely as a result of glutamine metabolism. At the conclusion of a typical 14-day Chinese hamster ovary (CHO) cell culture, ammonia levels have the potential to reach concentrations as high as 10-20 mM. This accumulation of ammonia in the surrounding media is toxic and inhibitory to mammalian cellular metabolism and has been shown to decrease cell growth, protein productivity, and protein quality. Preliminary metabolite data from parallel fed-batch CHO cells cultured in ambr®250 bioreactors, treated with either 10 mM ammonia or untreated, identified distinctive alanine metabolic profiles due to the treatment. To further examine the changes in the alanine profiles under ammonia-stressed conditions, samples from 0, 3, 6, and 9 days were sequenced via RNA-Seq. The subsequent read data was processed through a bioinformatics pipeline to determine differentially expressed genes and enriched gene sets to characterize transcriptome changes related to alanine metabolism. Since ammonia is also well-known to affect critical quality attributes, glycosylation transcripts, as well as other major differentially expressed genes and pathways were identified.

BIOT 120

Amino acid cross-feeding enables inter- and intra-species cooperation

Alexandria Fischer1, [email protected], Cynthia H. Collins2. (1) BIOL, Rensselaer Polytechnic Institute, Troy, New York, United States (2) Rensselaer Polytechnic Institute, Albany, New York, United States

Naturally occurring bacterial communities are able to cooperate to complete complex tasks by splitting these tasks between organisms and reducing the metabolic burden placed on each individual strain. The ability to distribute tasks and the metabolic load is of interest for biotechnology applications where dividing labor between populations can allow one to take advantage of the natural strengths of a specific strain and increase production. The potential benefits of creating synthetic microbial communities capable of sharing metabolic load for biotechnology are challenged by the difficulty in co- culturing microbes. One way to combat this challenge is to use pairs of auxotrophic microbes that are able to trade essential nutrients, or cross-feed. To explore the use of cross-feeding we have assessed the growth of two industrially relevant bacterial hosts in both inter- and intra-species pairs of DRAFT amino acid auxotrophs and examined transcriptomic profiles of successfully cross-feeding pairs. Cross-feeding success was evaluated in three different culture types: semi-solid agar plates, batch, and continuous culture. We observed that cross-feeding occurs both in both inter- and intra-species cross- feeding pairs of bacteria and identified pairs that cross-feed for inter- and intra-species pairs in all three culture types. Using this growth data we have identified two “super donors”, strains of both E. coli and B. megaterium that grow with many other amino acid auxotrophs in both inter- and intra-species pairs. Transcriptomic samples were taken from three cross-feeding pairs. Preliminary evaluation of differential gene expression between E. coli co- cultures and monocultures show that strains react differently to co-culture. We also found an upregulation of amino acid transporter and biosynthesis genes in co-cultured E. coli strains. This work shows that using cross-feeding pairs of bacteria can enable us to build functional synthetic microbial communities by addressing difficulties in co-culture growth.

BIOT 121

Tracking bioactive compounds produced by genetically engineered yeast cells using in situ Raman imaging

Mohammed Ibrahim1, [email protected], Nosa Agbonkonkon2, Rui Chen1, Michael Leavell2. (1) Thermo Fisher Scientific, San Jose, California, United States (2) Amyris, Emeryville, California, United States

In recent years, bioengineering of yeast cells has made it a “biological factory” for producing a wide variety of chemicals, including biologically active compounds, through fermentation process. Bulk chemical analysis, such as GC-MS or NMR can be used for screening fermentation cultures, assessing product yield and process optimization; these methods, however, are destructive, time-consuming and do not provide cellular level specificity. Fluorescence microscopy-based imaging is capable of visualizing large or small molecules at the cellular level; however, the success of this approach largely hinges on the photostability and brightness of fluorescent labels, their possible perturbations to the biological system, and the methods by which the labels are attached. Confocal Raman microscopy-based imaging has emerged as a great alternative. The Raman spectroscopy provides chemical annotations to the constituents within the cell, and the high spatial resolution (sub-µm) intrinsic to confocal microscopy provides cellular level information in situ. In contrast to the fluorescence microscopy-based imaging approach, Raman imaging is non-destructive and requires minimal sample preparation. DRAFT

In this work, several strains of engineered yeast cells were analyzed by Raman imaging. Two biologically active compounds produced within the strains were readily identified through library match. Raman images of the individual cells showed the locales as well as the relative abundance of those compounds. The results demonstrated that the strains produce the two compounds with different efficiencies, and suggested an optimal process for product yield.

BIOT 122

Clone ranking methodology for expedited process development using parallel workflows

David Chang, [email protected], Hongxia Chen, Richard Harper, Yanjing Li, Dane Grismer, Niket Bubna, Gautam Nayar, Sigma Mostafa. Process Development, KBI Biopharma, Durham, North Carolina, United States

Despite demonstration of clonal origin of cell lines, the mutations rates in CHO cells with increasing age in culture result in genetic diversity. Investigation of generational stability to fulfill regulatory expectations adds significant timelines to cell line development, which may bottleneck expedited workflows. We have established a comprehensive methodology to minimize relative instability of top clones identified for process development and establishment of a robust manufacturing process. The procedure for selecting a top clone includes tracking maintenance of rank order from static to shaken stages during cell line development, using standardized methods for cloning, development of culture medium sensitive for appropriate selection pressure, mapping flow cytometry-based diversity in subpopulations, and implementation of cell culture fed batch conditions. The fed-batch bioreactor study performed using a platform cell culture process is used to assess manufacturability and product quality. Clone performance is tested in a seed train followed by production in ambr15 or ambr250 bioreactors. A clone ranking system using weighted rank orders has been developed and successfully utilized for multiple cell lines. Outputs from this study are measured in a quantitative and qualitative approach and will be described. This integrated strategy for clone selection provides the basis for determining the top candidate clone while allowing for flexibility with different cell line types, molecule types and product filing status.

BIOT 123 characterization of media precipitate and determination of precipitation kinetics DRAFT

Duc Hoang1, [email protected], Shaun Galbraith2, Seongkyu Yoon3. (1) Chemical Engineering, University of Massachusetts Lowell, Lowell, Massachusetts, United States (2) University of Massachusetts Lowell, Lowell, Massachusetts, United States

Process intensification of mAb production is leading to more concentrated feed media causing issues with precipitation of solids from the media solution. This results in processing problems since components in the precipitate are no longer in solution and bioavailable, this changes the media composition leading to variability in cell culture performance. The goal of this work is to characterize the feed media precipitate, and in particular to identify the components so that mitigation strategies can be developed. The precipitate was predominately found to be organic, and was analyzed with LC-MS (Liquid Chromatography – Mass Spectrometry) and ICP-OES (Inductively Coupled Plasma – Optical Emission Spectroscopy) to identify the constituent components. Up to 10 amino acids were identified with tyrosine and phenylalanine being the most prevalent amino acids. Elemental analysis with ICP-OES (Inductively Coupled Plasma – Optical Emission Spectroscopy) revealed that inorganic components were accounted for up to 1 weight percentage of the solid precipitate with metal sulfates being the predominant inorganic components. Precipitation experiments informed by this analysis are being used to determine the formation kinetics, which in turn can be used to better design feed media and better process scheduling. This work is critical for the current trends of process intensification for mAb production to continue.

BIOT 124

Metabolic engineering of carbon source utilization for sustainable chemical production

Min-Kyu Oh, [email protected]. DEPT CHEM ENG KOREA UNIV, Seoul, Korea (the Republic of)

The development of industrial strains that can efficiently utilize various carbon sources has been emphasized in metabolic engineering because cheaper carbon source should be utilized in biorefinery. In case of glucose, overflow metabolism is the main problem in because high amounts of byproducts are produced. To reduce the overflow metabolism, phosphotransferase system for glucose (ptsG) and other glucose transporters are deleted in Escherichia coli. The resulting strain provided retarded growth in glucose minimal medium, but showed little production of acetate and much higher biomass yield from DRAFT glucose. Transcriptome and 13C metabolic flux analysis (13C MFA) analyses were conducted with the strain, showing that higher expressions of pentose pathway and citric acid cycle genes. The results proved that glucose transporter mutations in E. coli is helpful for the yield improvement of several metabolites.

BIOT 125

Bioconversion of nonanoic acid and its esters by whole cell biocatalyst of candida tropicalis

BYUNGHYUN JANG, Yeong-Eun Cho, Kyung Un Kim, Hyeon-Cheol Lee, Min woo Jun, Ji-Young Kim, Haeseong Park, Yong-Cheol Park, [email protected]. Kookmin University, Seoul, Korea (the Republic of)

Azelaic acid is an α,ω-dicarboxylic acid with nine carbons and has multiple applications in plastic and cosmetic industries. Chemical oxidation of oleic acid with ozone (called ozonolysis) allows the production of azelaic acid and a major byproduct of nonanoic acid. To increase the total yield of azelaic acid in the ozonolysis, in this study, sustainable biotransformation process using a whole cell biocatalyst was developed to directly convert nonanoic acid and its esters to azelaic acid. Candida tropicalis ATCC20962 immediately cleaved ethyl nonanoate to nonanoic acid after ethyl nonanoate addition, and then converted nonanoic acid into azelaic acid with the aid of nonane addition and continuous glucose supply. Finally, a fed-batch biotransformation by continuous feeding of pure nonanoic acid resulted in the production of 30.1 g/L azelaic acid with 0.30 g/L-h productivity and 90% molar yield. By combination of the ozonolysis and our process, a maximum of 95% molar carbon yield of azelaic acid from oleic acid was estimated. This is the first report that nonanoic acid and its esters were directly and biologically transformed to azelaic acid with over 90% yield, and would be a groundwork for the biotransformation of fatty acids with under nine carbons to the corresponding α,ω-dicarboxylic acids.

BIOT 126

Oncolytic virus production using MRC5 cells in Pall’s iCELLis® nano bioreactor is equivalent in high and low compaction beds

Julia Sable2, Ashab Alamgir2, Chris Daniels2, Nathan Hazi1, Andrew Laskowski1, Kaitlynn Bayne1, [email protected]. (1) Pall Biotech, DRAFT

Westborough, Massachusetts, United States (2) Merck & Co., Inc., West Point, Pennsylvania, United States

In the past five years, Pall’s iCELLis bioreactors have established themselves in the industry for the development and production of viral products. The bioreactor’s characteristics, mainly scalability, fixed-bed cell retention, closed- processing, advanced process control, single-use ease-of-implementation and footprint reduction, make the iCELLis bioreactor well suited for viral production. In this poster, we present our work on the use of the iCELLis Nano bioreactor to successfully culture MRC5 cells and to produce a high-titer proprietary oncolytic virus. Since the iCELLis Nano bioreactor is available in 6 different fixed bed geometries, questions arise if equivalent productivity can be maintained in the largest bed size, a height of 10 cm and a carrier compaction factor of 1.5x. This bed geometry is thought to be the most challenging in terms of obtaining a uniform cell distribution, even and efficient infection, and subsequent high productivity. In these experiments, we demonstrate with an n=3 that the virus productivity in the largest, high compaction, 4 m2 fixed bed is equivalent or higher than the 2.7 m2 low compaction fixed bed and legacy flatware process.

BIOT 127

Implementation of the Aber Futura Biomass probe in Pall’s iCELLis® nano bioreactor provides a robust and reproducible method to assess cell density

Randy Alfano2, Atherly Pennybaker2, Todd Lundeen1, [email protected], Andrew Laskowski1. (1) Pall Biotech, Raleigh, North Carolina, United States (2) InVitria, Aurora, Colorado, United States

In the past five years, Pall’s iCELLis bioreactors have emerged as the leading bioreactor technology for clinical manufacturing of viral vectors and vaccines. Pall and InVitria have previously demonstrated the successful implementation of OptiPEAK HEK293T blood-free media formulation in the iCELLis bioreactor, including rapid and uniform cell attachment to the iCELLis fixed- bed bioreactor and robust cell expansion kinetics found to be equivalent to those observed in serum-containing media. However, a major challenge in assessing cell growth is accurately and reproducibly measuring cell density. One legacy method is nuclei counting, but this method can have up to 20% variability, requires open aseptic bioreactor handling, and only provides a discrete set of data points. Another method is metabolic analysis, such as glucose consumption and lactate generation. While these measurements are DRAFT less variable and can be performed sterilely, they are also discrete, and extremely sensitive to culture conditions. The Aber Futura Biomass Probe provides a third method to measure cell density, one that provides an accurate, continuous reading, while maintain sterility. In this poster, we demonstrate the implementation of the biomass probe to accurately and reproducibly count HEK293T cells, overcoming past implementation challenges, such as the interfering effect of transient air bubbles. Finally, we correlate the biomass measurements with legacy nuclei counting and glucose consumption methods.

BIOT 128

Scale-down model qualification using ambr250 high-throughput bioreactor system for process characterization

Shaunak Uplekar1, Brandon Brino1, [email protected], Tim Broderick1, Jameel Joyner1, Derek M. Ryan2, Gautam Nayar1, Sigma Mostafa1. (1) Process Development, KBI Biopharma, Durham, North Carolina, United States (2) Analytical Development, KBI Biopharma, Durham, North Carolina, United States

Cell culture process characterization conducted during late phase process development of biopharmaceuticals is an extremely critical but time consuming and labor-intensive stage. Traditionally, bench scale bioreactor systems have been used for scale-down model qualification (SDMQ) for this purpose. High-throughput mini bioreactor system such as ambr250 bioreactors can significantly improve the time and effort required to conduct process characterization studies. Here, we describe the approach taken to qualify ambr250 as a scale-down model of a GMP scale manufacturing cell culture process to produce a monoclonal antibody. The seed train was conducted using shake flasks and wave bioreactors up to the N-2 stage. Ambr250 system was used to establish a scale-down model for the N-1 seed stage and the production (N) stage. It was found that the cell growth, viability, nutrient/metabolite and titer profiles generated using ambr250 bioreactors were found to be comparable to the pilot and the GMP scale process. Product quality attributes such as size exclusion chromatography (SEC) profile, non- reduced and reduced capillary electrophoresis (NR/R CE) profile, charge heterogeneity using ion exchange chromatography (IEX) and N-glycan were also found to be comparable. The results indicated that ambr250 bioreactor system can be successfully used for scale-down model qualification of N-1 and N stage of cell culture process and the subsequent process characterization studies. DRAFT

BIOT 130

Kinetic modeling for enhanced understanding of a fed-batch mammalian cell culture process

Steve Provencher1, [email protected], Greg J. Nierode1, [email protected], Sarah Fadda2, Edward Close2, Diana B. Ritz1. (1) MCCD, GlaxoSmithKline, Philadelphia, Pennsylvania, United States (2) Process Systems Enterprise, London, United Kingdom

Development of cell culture processes for production of biopharmaceuticals such as monoclonal antibodies (mAbs) requires extensive experimental work. For instance, media and feed composition, feeding strategy, and an array of bioreactor set-points (pH, temperature, agitation, etc.) must be optimized to maximize production of high-quality product. To reduce this experimental burden, one attractive strategy is to use mechanistic models that capture key process dynamics to streamline experiment design and accelerate acquisition of process knowledge. In this work, we developed an unstructured, segregated kinetic model of a Chinese Hamster Ovary (CHO) fed-batch mAb production process using gPROMS FormulatedProducts, a commercially-available flowsheet-based modeling platform. We fit experimental data collected using a high-throughput bioreactor platform to the model and demonstrated that the model was able to capture the dynamics of several key cell culture parameters including glucose and lactate concentration, viable cell density, and mAb titer. Global Systems Analysis (GSA) was applied to refine the model parameters based on the order of parameter sensitivity, which also provided insight into the impact of process parameters on process performance. Finally, we evaluated the utility of the model to optimize various process conditions (e.g. feeding strategy, initial seeding density) to maximize the process yield. Efforts to evaluate the model accuracy for additional CHO lineages and mAb assets are ongoing.

BIOT 131

Perfusion enables increased lentivirus production using the iCELLis® bioreactor system

Odette Becheau1, [email protected], Todd Lundeen2. (1) SLS, Pall Biotech, Cambridge, Massachusetts, United States (2) Pall Biotech, Raleigh, North Carolina, United States DRAFT

Piriya Yoganathan1, Ann Catherine Lee1, Natalie Hyatt1, John C. Bell1, Odette Becheau2, Todd Lundeen2, Siddharth Gupta2, Andrew Laskowski2

1Ottawa Hospital Research Institute (OHRI), The Ottawa Hospital, 501 Smyth Road, Ottawa, ON, K1H 8L6, Canada. 2Pall Biotech, 20 Walkup Drive, Westborough, Massachusetts 01581, USA.

Historically, there have been no manufacturers using Good Manufacturing Practices (GMP) to produce lentivirus in Canada, which has hampered gene therapy and CAR-T clinical programs in this country. The Ottawa Hospital recently developed a lentivirus manufacturing process using adherent cells in a flatware culture system, and used this process to produce sufficient amounts of clinical-grade lentivirus in a GMP manner for a Canadian CLIC- 1901 CD19-CAR T Phase I/II clinical trial. While our previously established lentivirus manufacturing process used a flatware culture system, we plan to develop a larger-scale production process for clinical-grade lentivirus using the iCELLis bioreactor system. The iCELLis bioreactor system provides a large surface area for adherent cell growth, has a small physical footprint, and recently emerged as the leading scalable, single-use bioreactor technology for clinical manufacturing of viral vectors and vaccines. Our initial estimates suggest we could produce lentiviral vectors for approximately 2000 clinical doses for CAR-T cell therapy applications in a full-scale iCELLis 500+ bioreactor system; however, observed experimental titers in the first several iCELLis Nano batches were low. We hypothesized that this low titer was due to product instability in the bioreactor for long exposure times, and therefore implement a perfusion-based feeding system to remove product from the bioreactor for storage at more stable conditions. In this poster, we demonstrate that this perfusion strategy successfully resulted in bioreactor titers greater than the legacy flatware process for n=4 batches.

BIOT 132 iCELLis® nano bioreactor provides a reliable method to produce animal vaccines with high titer

Odette Becheau1, [email protected], Todd Lundeen2. (1) SLS, Pall Biotech, Cambridge, Massachusetts, United States (2) Pall Biotech, Raleigh, North Carolina, United States

In the past five years, Pall’s iCELLis bioreactors have emerged as the leading bioreactor technology for clinical manufacturing of viral vectors. In the DRAFT

Americas, most of this technology adoption has occurred in the United States and within the human gene therapy niche of biopharmaceuticals. However, the advantages of the iCELLis bioreactor – scalability, fixed-bed cell retention, closed-processing, advanced process control, single-use ease-of- implementation, and footprint reduction – that make the iCELLis bioreactor well suited for viral vector production also make it well suited for production of live-virus vaccines. In this poster, Pall has partnered with a Brazilian animal health company to demonstrate successful viral vaccine production in the iCELLis bioreactor for rabies virus. For each of these case studies, the titer observed experimentally in the iCELLis Nano bioreactor was either equivalent to or higher than the legacy flatware process, with little to no process development effort required.

BIOT 133

Exploring efficiency gains with the implementation of FlowVPE into accelerated downstream process development

Thomas Lindsey, [email protected], Rashmi Bhangale, David Brown, Nate Nicholes, Heather Horne, Leslie Wolfe. Downstream Process Development, KBI Biopharma Inc, Raleigh-Durham, North Carolina, United States

FlowVPE (C Technologies) is a variable path length UV-VIS spectrophotometer that can be used to accurately measure protein concentrations across a much larger range than typical inline UV detectors on FPLC instruments. The sensitivity of the instrument is sufficient to monitor the capture and breakthrough of product using clarified harvest as the feed stream on Protein A columns. KBI Biopharma has generated data for multiple molecules demonstrating the comparability of FlowVPE UV breakthrough data to offline ProA-HPLC titer analysis for determination of Protein A resin dynamic binding capacity. Application of FlowVPE enables real-time data generation, reducing analytical support requirements for sizing ProA capture columns. This capability results in both time and cost savings in an accelerated downstream purification process development workflow.

BIOT 134

Impact of mobile phase modulators on capacity and impurity clearance in protein A chromatography: Insights from mechanistic models DRAFT

Vijesh Kumar1, [email protected], Xuankuo Xu2, Steven J. Traylor3, Sanchayita Ghose2, Abraham M. Lenhoff1. (1) Chemical and Biomolecular Engineering, University of Delaware, Newark, Delaware, United States (2) Biologics Process Development, Bristol-Myers Squibb, Devens, Massachusetts, United States (3) Biologics Process Development, Bristol- Myers Squibb, Hopkinton, Massachusetts, United States

Protein A affinity chromatography is the dominant capture step in monoclonal antibody manufacture. Since it is the first unit operation in a succession of chromatography steps, process understanding using a mechanistic model can be efficacious. Although mechanistic models based on Langmuir isotherms are widely used, the physical significance of the model parameters is not always considered in detail. In this work, we have calibrated such a model using complete breakthrough curves in the presence of various mobile-phase modulators at different pH values. Both modified Langmuir and colloidal- based models accommodating pH effects were incorporated in the model. The isotherm models were coupled with the general rate model, including pore and surface diffusion effects. The dynamic binding capacity was shown to be impacted by the load concentration (0.5-10 mg/mL), kosmotropic salts (sulfates, phosphates, and citrates) and organic phase modulators (ethanol, isopropanol). The maximum binding capacity within the Langmuir isotherm and the protein-protein repulsion within the colloidal isotherm led to significant differences in the protein-surface binding affinity extracted from the models under these conditions. The parameters obtained from breakthrough curves were used to model the effects on product recovery of wash and elution buffers for mobile phase modulators and pH. The pH buffer titration on the Protein A column is one of the challenges in modeling elution peak shape and thus pool volume. This was overcome by modeling the column with a buffer titration curve run without protein load. The impact of the removal of host-cell protein and protein aggregate with mobile phase modulators and its pH was found to be important. Overall, this work demonstrates that mechanistic models can provide broad insight into protein binding behavior inside the column and hence utilize the knowledge to develop an optimal process.

BIOT 135

Influence of excipients during antibody purification on protein A chromatography

Carolin Stange1, [email protected], Supriyadi Hafiz2, Tanja Henzler2, Christian Dr. Frech3. (1) Institute of Biochemistry, University of Applied Sciences Mannheim, Mannheim, Germany (2) Merck KgaA, Darmstadt, DRAFT

Germany (3) Biotechnology, University of Applied Sciences Mannheim, Mannheim, Germany

The production of antibodies and Fc-fusion proteins involves several downstream processing unit operations. The widely used purification template with Protein A chromatography, virus inactivation at low pH, and subsequent ion exchange chromatography steps is mostly able to remove impurities like aggregates, host-cell proteins, and viruses, which could affect the safety and efficacy of the product. The low-pH elution during Protein A chromatography, as well as during virus inactivation may induce aggregation. Preventing protein aggregation during these unit operations instead of removing the multimeric forms during subsequent polishing steps would be a more efficient strategy. Excipients have shown that they can minimize aggregation levels in the final product formulation. The objective of this work was to apply excipients, like polyols, in buffer systems, to study their impact on several aspects during Protein A chromatography and virus inactivation. Therefore, a pH gradient system on three different Protein A materials was applied. The results include their effect regarding elution patterns, stabilization of the monomer antibody, host-cell protein removal, and virus inactivation rates, as well as the effect on binding capacities in following cation exchange chromatography. The results show that the addition of excipients can have a beneficial effect for the purification during Protein A chromatography and virus inactivation, without harming the subsequent chromatographic steps.

BIOT 136

Protein A fiber chromatography from research and early development to scale-up and manufacturing of monoclonal antibodies

Anna Graanberg1, [email protected], Martin Hall1, Lars Rosengren1, Anders Nilsson1, Ian Scanlon2, Matthew Townsend2, Oliver Hardick2. (1) R&D, Cytiva, Uppsala, Sweden (2) R&D, Cytiva, Stevenage, United Kingdom

A new Protein A fiber chromatography format has been developed as a complement to traditional column chromatography. The Protein A fiber matrix has an open pore structure where mass transfer is governed by convective flow which allows high mAb binding capacities at very short residence times. This results in cycle times of minutes instead of hours as compared to traditional column chromatography. The short cycle times allows for purification of many samples in short time which makes the pre-packed, small- scale Protein A fiber unit a versatile tool during e.g. clone selection and up- stream cell-culture media development. In the downstream process the time DRAFT savings in process development will be significant where screening of different buffer conditions and lifetime studies including hundreds of cycles can be performed in less than 24 hours. In larger scale, the Protein A fiber operated in a rapid cycling manner allows full utilization of the Protein A lifetime during one batch. The pre-packed and pre-sanitized single-batch disposable format eliminates column packing and reduce the risk of bioburden incidents. Examples will be given from small scale, where automation and a combination of capture and desalting steps will increase the throughput in up-stream process development as well as wash and elution buffer optimization in downstream process development. However, the focus will be on larger scales where multiple manifolded fiber units of different sizes, and number of cycles during operation will match the mass of mAb from the bioreactor. The compatibility of fiber units with current chromatography systems regarding system flow rates, back pressure limitations and dead volumes will also be demonstrated. Productivity, throughput, buffer consumption and purification performance will be compared to traditional and rapid cycling Protein A column chromatography.

BIOT 137

Improvement in protein A resin lifetime by implementing poly- diallyldimethylammonium chloride (pDADMAC) flocculation coupled with gradient depth filtration

Philip Szymanski, [email protected], Akshat Gupta, Dana Kinzlmaier. MilliporeSigma, Burlington, Massachusetts, United States

The economics of the protein A capture step in mAb purification rely on effective resin utilization over multiple cycles. To this end, pDADMAC-based flocculation and subsequent filtration with gradient depth filters was evaluated as a method of cell culture clarification to reduce the impurity burden on protein A resin. The effects on column performance, measured using yield, binding capacity, backpressure, and impurity clearance, were investigated over 160 cycles. Results were compared with a lifecycle assessment of protein A capture of the same feed stream clarified using traditional depth filters without flocculation. Improvements in maintenance of dynamic binding capacity, column backpressure evolution over number of cycles, and host cell protein (HCP) levels in protein A eluate were observed. Process consistency in terms of product concentration in the eluate and eluate volume was also maintained over the 160 cycles. Cost of goods analysis was performed to quantify the economic benefits of the enhanced clarification strategy on the DRAFT holistic clarification and capture process. Additional wash optimization to further improve the HCP clearance is currently being investigated.

BIOT 138

Fc gamma receptor as an affinity ligand for antibody glycovariant separation

Maria Livanos2, [email protected], Elizabeth Edwards2, Anja Krueger1, Stuart Haslam1, Anne Dell1, Daniel G. Bracewell2. (1) Imperial College London, London, United Kingdom (2) UCL Dept Biochemical Engr, London, United Kingdom

Biopharmaceuticals continue to grow as a fraction of pharmaceutical sales, the search for new products with increased efficacy often means the detailed glycobiology of the product and its interactions with the immune system are critical. Current products including antibodies and erythropoietin are heterogeneous in their glycosylation state, even though it is known that for example non-fucosylated antibodies are more potent in raising antibody- dependent cell cytotoxicity (ADCC) than fucosylated glycoforms. Current products using CHO cell lines express high levels of fucosylation. Processes capable of homogeneous glycosylation have therefore generated considerable interest and investment. These approaches have generally been upstream focused e.g. to knock out fut8 in CHO and create a fucosylation deficient host cell or engineer human glycosylation in yeast.

Here we propose to take a downstream processing based solution by designing affinity ligands capable of glycoform recognition for use in chromatography. This concept is based on the naturally occurring interaction between antibodies and immune system which is mediated by the Fc gamma receptor. This protein ligand has been used in affinity chromatography to perform both preparative separations and characterise the strength of the Fc gamma/Ab interactions. To be a useful basis for manufacturing such ligands need to have appropriate separation properties.

The wild-type Fc gamma 3a ligand has been shown to have selectivity between fucosylated and non-fucosylated antibody glycoforms. A cloning, expression and purification strategy for this Fc gamma affinity ligand is presented. Characterization of the ligand’s separation properties first required a strategy to create defined glycovariants of a monoclonal antibody product. This was done using an enzyme based strategy to remodel the antibody’s glycosylation and create homogeneous preparations e.g. non-fucosylated DRAFT structures. The Fc gamma ligand and resulting glycoengineered antibodies were analysed by mass spectrometry. The affinity between the ligand and the defined antibody glycovariants were then characterized and resulting separation properties analysed.

BIOT 139

Protein A fibro chromatography as a platform affinity capture of the next-generation manufacturing process of antibody therapeutics

Hendri Tjandra1, [email protected], Ryan Zolyomi1, Ian Scanlon2, Kierston Shill1, Florence Rusly1, Anna Gronberg3, Oliver Hardick2, Ashley Hesslein1. (1) Biologics Development, Bayer, Union City, California, United States (2) Stevenage Bioscience Catalyst, Cytiva | Puridfy, Stevenage, SG1 2FX, United Kingdom (3) Cytiva, Uppsala, Sweden

With an increasing number of antibody therapeutics and higher cell culture titers from intensified upstream processes, manufacturing batch sizes are smaller and more diverse. Consequently, single-use multi-product manufacturing facilities are utilized as they provide the flexibility necessary to meet these changing demands. These driving forces motivated us to evaluate ProteinA-derivatized Fibro, a single-use chromatography technology, in meeting the goals of our future manufacturing platform. Implementing a fully single-use manufacturing platform can be challenging as bead-based ProteinA resin chromatography has anchored industrial production of antibodies. ProteinA resins have an excellent record of clearing impurities while consistently delivering high quality products. Typically, large columns are used during clinical manufacturing to reduce processing times and are thus rarely exhausted of their entire lifetime capacity. This results in under- utilization, high initial costs and storage concerns. Fibro ProteinA enables rapid purification of monoclonal antibodies by offering faster mass transfer, high capacity, and fuller utilization within a single production batch. It offers the potential of delivering the same product quality as that of ProteinA resin with a potential cost- and time saving during clinical productions. Bench-scale Fibro studies performed across a range of monoclonal antibodies indicate its feasibility as a platform approach. To assess its scalability, pilot-scale case studies were conducted. They demonstrated its single-use facility fit, productivity benefits, and product-quality comparability to those of resin-based processes. Additionally, these pilot-scale studies addressed concerns over their adoption due to the disadvantages of membrane-like technologies, namely higher buffer consumption, larger elution pool volumes, and shorter device lifetimes. As Fibro units were scaled up, the buffer consumption, DRAFT elution volume and lifetimes were found to be comparable to those of resin- based affinity chromatography. We evaluated the feasibility of implementing FibroA chromatography into a future mAb manufacturing platform. While there remain uncertainties surrounding long-term process economics at a manufacturing scale, this technology allows for productivity increases, cost- savings, and flexibility. We envision its potential for a paradigm shift in manufacturing antibodies in the future.

BIOT 140

From sequence to process - in silico DSP development based on quantitative structure-property relationships

David Saleh1,2, [email protected], Federico Rischawy1,2, Gang Wang1, Simon Kluters1, Joey Studts1, Juergen Hubbuch2. (1) Late Stage Downstream Process Development, Boehringer Ingelheim, Biberach, Germany (2) KIT, Karlsruhe, Germany

During the last years, in silico process optimization and characterization have become a key technology in early and late stage development and the focus is turning more and more to mechanistic models for chromatography processes. Nevertheless, the complex and time-consuming model calibration inhibits the implementation of mechanistic modeling workflows in current development strategies. Quantitative structure-property relationship (QSPR) models have the potential to mitigate the risks related to parameter estimation by inverse curve fitting to experimental data. Here, we will present our multi- dimensional modeling pipeline, leading to the digital twin of a cation exchange (CEX) chromatography process from the primary sequence of the monoclonal antibody (mAb). Starting with the primary sequence, homology models were built for 21 IgG1, IgG4, and complex antibody-formats. Protein surface analysis resulted in >1000 structural descriptors per molecule, including shape-, patch-, and region-specific descriptors. Multiple high-throughput and lab-scale experiments were conducted for all mAbs. The comprehensive dataset enabled the calibration of a transport-dispersive model and the steric- mass action isotherm. QSPR models correlated structural properties to mechanistic chromatography model parameters. The established models were then used for the prediction of isotherm parameters and entire chromatograms of mAbs not involved in the training data set. Further, the correlations found between certain descriptors and isotherm parameters increased the understanding of the binding-mechanism in CEX chromatography. Feature selection revealed which subdomains of the mAb contribute to the adsorption process. Our work demonstrates a bottom-up DRAFT approach for building mechanistic chromatography models for therapeutic antibodies.

BIOT 141

Modeling host-cell protein retention to expedite ion-exchange resin screening for flow-through polishing

Chase Herman2, [email protected], Xuankuo Xu1, Steven J. Traylor1, Sanchayita Ghose1, Abraham M. Lenhoff2. (1) Biologics Process Development, Bristol-Myers Squibb, Devens, Massachusetts, United States (2) Univ of Delaware, Newark, Delaware, United States

The effectiveness of flow-through anion-exchange polishing steps requires retention of a variety of impurities that differ in their strength of adsorption, and the relatively weak adsorption of some host-cell protein (HCP) impurities may lead to inadequate clearance. Such effects may be amplified by intrinsic retention differences among different resins. Models that can capture this sensitivity would be useful in directing experimental design efforts. To that end, the present study seeks approximate models to predict HCP adsorption equilibria from protein sequences and to integrate them into column models for flow-through chromatography. In the absence of HCP structural information, coarse-grained calculations of molecular-level interactions are performed using a colloidal representation of the protein and the resin, making use of HCP sequence and other structural information. Adsorption energetics are estimated from numerical solutions to continuum electrostatics equations. The resulting retention predictions are compared to those from colloidal models, and their accuracy is assessed by comparison to column experimental data.

BIOT 142

Case study: AEX resin age impact on chromatographic separation of fusion protein charge isoforms

Timothy Henderson, [email protected]. Bristol-Myers Squibb, Syracuse, New York, United States

In Imaged Capillary Isoelectric Focusing (iCIEF) method, a BMS molecules protein charge variants are separated using high voltage based on their isoelectric point (pI), the pH at which the glycoproteins have a net charge of zero. The distribution of the charge isoforms is primarily driven by the DRAFT molecules sialic acid content with contributions from C-terminus lysine removal and asparagine deamidation. The protein molecules with more sialic acid moieties are acidic and therefore have an overall low pI compared to the protein molecules with less or no sialic acid groups.

The iCIEF method electropherogram profiles contains three main groups expressed as area percentages and distributed over a pH gradient: Group 1: High degree of sialyation; Group 2: Main Group which represents the majority of charge variants observed in BMS-224818; Group 3: Limited sialyation.

Process Monitoring for drug substance manufacturing processes at Bristol- Myers Squibb Company is performed to monitor process variation to ensure the parameters stay within the established control limits. During routine manufacturing of the BMS molecules shifts in the iCIEF Groups 2 and 3 were noticed beginning with the same drug substance Lot. This shift coincided with the retiring of expired resin and use of virgin resin in the AEX chromatography column.

Through lab scale experimentation and statistical analysis, it was identified that AEX resin age had statistically significant effects on iCIEF Groups 2 and 3. As the column ages, the amount of low sialyated species decreases resulting in low iCIEF Group 3 and high iCIEF Group 2. Therefore, the inverse is also true. Low sialylated species are more predominant in lots manufactured using newer AEX resin resulting in lower iCIEF Group 2 and higher iCIEF Group 3.

BIOT 143

Evaluating the virus reduction potential of a mixed-mode chromatography media during early phase downstream development

Akunna Iheanacho1, [email protected], Coral Fulton1, Katelyn Pritchard1, Xuemei He2, William H. Rushton2, Irene Chen2, Louisa Vang2, Jenifer Dean1, Kaye Peden1. (1) Texcell North America Inc, Frederick, Maryland, United States (2) Bio-Rad Laboratories, Hercules, California, United States

To ensure the safety of biopharmaceutical products, downstream unit operations should be evaluated to determine their viral clearance capacity during process development. In this case study, we utilized a DoE approach to investigate the effects of buffer pH and conductivity on the removal of product and process-related impurities by a new mixed-mode chromatography media. Results from this study have offered insights on the interactions DRAFT between this chromatography media and virus particles, as well as the design space for the removal of viral contaminants.

BIOT 144

Mechanistic understanding of aggregate reduction in a monoclonal antibody downstream process

Lee Bink, [email protected], Gerald Terfloth. UE0551, GlaxoSmithKline, King of Prussia, Pennsylvania, United States

High molecular weight aggregates are considered a product-related impurity due to their potential immunogenicity in patients. Aggregates are typically removed from a monoclonal antibody product stream via various modes of chromatography, including cation exchange (CEX) and hydrophobic interaction chromatography (HIC). However, formation of aggregates during downstream processing due to various stresses (shear, interfacial stress, pH, etc.) can also be regulated and/or prevented in order to minimize the total levels in bulk drug substance. Using both differential scanning fluorimetry and a measurement of aggregation kinetics, it was determined that the Affinity Chromatography step was contributing towards the denaturation of the mAb and causing a subsequent aggregation event during the low pH virus inactivation step. Implementation of a milder elution pH buffer resulted in less formation of aggregate across these unit operations. This work demonstrates how the optimization of the existing two column process can be achieved to reduce aggregation while negating the need for a third chromatography step and the necessary capital expenditure in the New Product Introduction (NPI) facility.

BIOT 145

Identifying preferred binding regions in multimodal chromatography using covalent labeling and liquid chromatography/mass spectrometry

Ronak B. Gudhka1,2, [email protected], Kabir Dhingra1,2, Dimitri Zagorevski2, Steven M. Cramer1,2. (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

Recent biophysical investigations into protein-ligand interactions carried out in our lab have helped us to develop a deeper molecular level understanding of DRAFT the binding of biological products to Multimodal (MM) chromatographic systems. These biophysical investigations employed Nuclear Magnetic Resonance (NMR) spectroscopy to get the residue specific binding information for various proteins. However, NMR spectroscopy requires the use of 15N-labeled and assigned proteins that can be challenging for large proteins like monoclonal antibodies (mAbs) or viral capsid proteins, which are generally greater than 50kDa. Hence, there is a need to employ a different technique that can be applied to a broader range of proteins as well as binding conditions. To address this, we used a combination of covalent labeling and Liquid Chromatography/Mass Spectrometry (LC/MS) to identify the regions on the protein surface that are involved in binding to the chromatographic resin. In this study, we evaluated the binding region on small model protein Ubiquitin (Ubq) using Diethyl pyrocarbonate (DEPC) as a covalent label for binding to various MM chromatographic resin systems. DEPC is a versatile covalent label that can readily react with several nucleophilic amino acid residues on the protein surface. First, a DOE study was designed to identify the optimum DEPC:Protein ratio as well as the reaction time for covalent labeling of Ubq in solution. These optimized conditions were then used for the on-resin labeling experiments of Ubq. By a comparative study of the in-solution and the on-resin labeling reaction, the residues involved in binding with the chromatographic resin were identified. The binding region identified using this technique was confirmed with the results obtained from NMR spectroscopy. This combined covalent labeling and LC/MS approach allowed us to develop a molecular level understanding of binding of Ubq to different MM chromatographic systems. This work, which requires very small amounts of sample, has potential applications for identifying the residues on surfaces of complex biological products like mAbs, bispecific antibodies, viral capsid proteins etc. for binding to various chromatographic systems. This rich binding region information would guide us to design efficient separation processes a priori for new biological products that are associated with product related variants.

BIOT 146

Mechanistic modeling of ligand density variations in anion exchange chromatography

Gabriela Sanchez Reyes1, [email protected], Heiner Graalfs3, Christian Dr. Frech2. (1) Biotechnology, Mannheim University of Applied Sciences, Mannheim, Baden-Württemberg, Germany (2) Biotechnology, University of Applied Sciences Mannheim, Mannheim, Germany (3) Chromatography Life Science, Merck KGaA, Darmstadt, Germany DRAFT

Ion exchange chromatography (IEC) is a widely used unit operation in the purification of therapeutic proteins. However, the performance of an ion- exchange process depends on a complex interrelationship between several parameters, such as protein properties, mobile phase conditions, and chromatographic resin characteristics. Consequently, batch variations of IEC resins play a significant role in the robustness of these downstream processing steps. Ligand density is known to be one of the main lot-to-lot variations, affecting protein adsorption and separation performance. The use of a model-based approach can be an effective tool for comprehending the impact of parameter variations (e.g., ligand density) and their influence on the process. The objective of this work was to apply mechanistic modeling to gain a deeper understanding of the influence of ligand density variations in anion exchange chromatography. To achieve this, 13 prototype resins having the same support as the strong anion exchange resin Fractogel® EMD TMAE (M), yet only differing in ligand density, were analyzed. Linear salt gradient elution (LGE) experiments were performed to observe the elution behavior of a monoclonal antibody and BSA. An extended stoichiometric displacement model for ion-exchange chromatography, describing the dependence of ligand density variations on protein retention, was successfully applied. Additionally, a lumped rate model was used to perform in silico chromatograms, which proved the reliability of the model.

BIOT 147

Impact of multimodal ligand density on free energy of adsorption of proteins to ligand-immobilized SAM surfaces: Molecular dynamics study

Mayank Vats1, [email protected], Shekhar Garde2, Steven M. Cramer3. (1) Chemical and Biological Engineering, Rensselaer Polytechnic Institute, Troy, New York, United States (2) Dept of Chemical and Biological Engr, Rensselaer Polytechnic Inst, Troy, New York, United States (3) Chemical and Biological Engineering, Rensselaer Polytechnic Inst, Troy, New York, United States

Multimodal chromatography presents unique selectivity in bio-separations, but a fundamental understanding of differences in multimodal (MM) ligands’ behavior still eludes us. Recent simulations work in our lab on MM-ligand immobilized self assembled monolayer (SAM) surfaces has demonstrated that ligand flexibility and density differences can result in surfaces with relatively large patches of hydrophobicity and charge. These self-associated ligand DRAFT patterns have also been found to be persistent in chromatographically relevant higher salt concentrations and could govern protein-surface interactions. Here we develop that premise using enhanced sampling in molecular dynamics simulations of these systems. First, INDirect Umbrella Sampling (INDUS) was used to study the hydrophobic nature of surfaces where immobilized MM-ligands were found to aggregate and create hydrophobic and charged patches, and then compared with low ligand density surfaces without ligand self-association. To further understand the importance of ligand clustering in protein adsorption, free energies of binding of model proteins to a series of SAM surfaces with varying MM-ligand densities were calculated from potential of mean force (PMF) measurements performed using umbrella sampling. The possibility that water-mediated interactions during binding and unbinding at the protein-SAM interaction interface show hysteresis-like behavior was also investigated. Correlations between the calculated free energies and corresponding patch distributions for those ligand surfaces, inform us about the role of ligand density driven self-association in adsorption. Our study presents an understanding of hydrophobic behavior in multimodal systems at different ligand densities and connects multimodal resin surface properties with density dependent behavior during protein adsorption, which in turn has implications for surface property driven multimodal resin design and selection.

BIOT 148

Orthogonal pre-use and post-use in situ integrity testing for single-use anion exchange chromatography

Jonathan F. Hester1, [email protected], Daniel Lu1, Jacob Calhoun1, Rebecca Hochstein2. (1) 3M Bldg. 236-1C-14, 3M Separation and Purification Sciences Division, Saint Paul, Minnesota, United States (2) 3M, St. Paul, Minnesota, United States

Flow-through anion exchange (AEX) chromatography is frequently used in biopharmaceutical purification processes for the reduction of negatively charged host cell proteins and viral reduction as part of a validated viral clearance strategy. While AEX column chromatography is the technology most frequently used at commercial scale for electrostatic viral clearance, the introduction of single-use AEX technologies in recent years has illuminated the opportunity for reduced capital, regulatory, and operational costs. Physically resembling and operated like filters, single-use AEX devices feature improved specific capacity and enhanced flow rates compared with columns due to the replacement of diffusive kinetics with convective flow. DRAFT

These features create the potential for simpler operation, decreased processing times, and reduced buffer consumption leading to improved economics relative to columns. Additionally, their single-use nature obviates validation costs associated with cleaning and performance over repeat use cycles, including viral clearance performance. With respect to the deployment of single-use AEX devices in large-scale manufacturing, however, many investigators have noted the need for sensitive, in situ test methods, analogous to HETP and peak asymmetry studies conducted on packed columns, capable of detecting device defects that might result in reduced viral clearance. While the bubble point or pressure-hold type tests frequently used to test the integrity of filters can reduce the risk of undetected bypass due to mechanical defects, such tests alone would fail to detect defects in AEX capacity or in the density or uniformity of the AEX functional chemistry on the media. While a few in situ integrity tests enabling assessment of the chemical functionality of AEX single-use media have been published, they involve exotic analytes, require complex analytical procedures, and/or are only qualitative. In this talk, we present an orthogonal set of two non-destructive, in situ pre-use integrity tests and a companion post-use integrity test. When employed in combination, the tests characterize both the mechanical and chemico-functional integrity of a single-use AEX device, strongly reducing the risk of undetected viral clearance failures. Correlations between integrity test performance and bacteriophage clearance are presented in the context of a new family of biomimetic single-use AEX chromatography devices with robust viral clearance performance.

BIOT 149

Evaluating cleaning and re-use feasibility for an anion exchange membrane adsorber

Thomas Elich1, [email protected], Jaime De Souza1, Matthew Johnson2, Elizabeth M. Goodrich1, Mochao Zhao3. (1) MilliporeSigma, Burlington, Massachusetts, United States (2) MilliporeSigma, Bedford, Massachusetts, United States (3) MilliporeSigma, Burlington, Ontario, Canada

Many of today’s bioprocesses require flexible manufacturing strategies in order to maintain a diverse multiproduct facility. Membrane adsorber chromatography is an effective tool to improve manufacturing flexibility by enabling rapid mass transfer and eliminating the need for column packing. This technology also offers opportunities for increased purification productivity as compared to resin based technologies. While membrane adsorbers provide DRAFT value as a single-use device, in some applications the ability to regenerate the device for multiple re-use cycles may be desired. In this work, a commercially available high capacity membrane adsorber featuring quaternary amine chemistry was evaluated for re-use feasibility in a monoclonal antibody polishing application. Results are presented to evaluate device performance over consecutive processing cycles with a focus on product yield, host cell protein purification, viral clearance, and device backpressure. Considerations for implementing membrane adsorber chromatography in both single-use and re-use applications are discussed.

BIOT 150

High resolution membrane chromatography: Modeling and simulation of mAb charge variant separation

Jan Hedrich2, Manuel Phillipe Díaz Carreño2, Alina Wulff2, Romas Skudas1, Christian Dr. Frech2, [email protected]. (1) EMD Millipore, Darmstadt, Germany (2) Biotechnology, University of Applied Sciences Mannheim, Mannheim, Germany

Due to the increased performance of novel upstream processes, continuous downstream processes, which allow higher throughput, are gaining more importance in industry, since they increase yield and decrease process time manifolds. Hence, further efficient technologies are being developed recently. Among new emerging technologies in downstream processing, membrane chromatography holds excellent potential for the purification of proteins. Their high level of downstream productivity and versatility is achieved by combining the functionality of resin-based columns with the high flow rates of filter membranes. The major applications of membrane chromatography are the purification of large biomolecules like viruses, VLPs and protein complexes, and the removal of contaminants like DNA, HCPs etc. in flow-through mode. In this study, a prototype ion-exchange membrane was used for high resolution separation of protein variants in a bind and elute mode. As a midsize model protein, a monoclonal antibody with three charge variants was used at low and high protein loadings. To elucidate the elution behavior of the three monoclonal antibody charge variants, mechanistic modeling was applied using linear pH and salt gradient elution experiments. Utilizing the model parameters in in-silico simulations, chromatograms are predicted, and the separation has been optimized. The performance and binding capacity of the membrane is compared to a commercially available resin-based column.

BIOT 151 DRAFT

Development of a robust and high throughput bispecific process which includes controlled Fab arm exchange and hydrophobic interaction chromatography

Joseph A. Sakyiama, [email protected]. Active Pharmaceutical Ingredient Large Molecule, Janssen Pharmaceutical Company, Lower Gwynedd, Pennsylvania, United States

Janssen has several Bispecific programs at various clinical phases. Some are Bispecific antibodies (BsAb) generated through a DuoBody® controlled Fab Arm Exchange (cFAE) process from different parental monoclonal antibodies. We have developed a highly efficient cFAE process, but residual parental contaminants remain and typically removed using bind-elute chromatography. Amongst the various options to remove residual parental mAbs, hydrophobic interaction chromatography (HIC) can be a powerful tool but generally has low capacity. This presentation explores the development of a robust and high throughput Bispecific process which employed cFAE and HIC to remove residual parental mAbs. The cFAE was robust for bispecific yield and non-reduce cSDS purity and high throughput. Development of the cFAE required an understanding of the chemistry and kinetic principles of the reaction to optimize critical parameters. The cFAE step is able to achieve bispecific yields of 93% with 35g/L IgG concentration and >97% non-reduced cSDS Bispecific purity. At this step, the sum of contaminant parental mAb levels are approximately 7%. The HIC had to be efficient at removing the parental residues and have high capacity. To optimize the HIC step, a strong understanding of the properties of different hydrophobic resins and the influence of salts on protein hydrophobicity were utilized. An HIC step capable of achieving 80g/L and over 99% bispecific purity was developed. Characterization data shows the BsAb is correctly assembled by analysis via peptide map, free sulfhydryl, intact mass, cIEF, and preserved binding activity of each parental arm.

BIOT 152

Multimodal membrane adsorbers for clearance of minute virus of mice

Xianghong Qian1, [email protected], Tanmoy Patra1, Shu-Ting Chen1, S R. Wickramasinghe2. (1) Biomedical Engineering, University of Arkansas, Fayetteville, Arkansas, United States (2) Raplh E Martin Department of Chemical Engineeeing, University of Arkansas, Fayetteville, Arkansas, United States DRAFT

During purification, manufacturers of mAbs and other therapeutic proteins must demonstrate reduction of 103-105or more virus particles than is estimated in a single dose equivalent of the unprocessed bulk. The level of virus clearance is calculated by summing the clearance obtained from individual unit operations having different mechanisms of action (e.g. size exclusion and ion exchange). Depending on the antibody titer, estimates of the number of virus particles in a single dose equivalent could be as high as 1010-1015retrovirus-like particles per mL. In addition, removal of adventitious virus using a model parvovirus is required. In recent years, several membrane adsorbers with high binding capacity have evolved for the removal of contaminants for the polishing steps during downstream purification of biologics. In particular, multimodal membrane adsorbers with tunable pore sizes and ligand properties have generated tremendous interest due to their many possible applications. Here the removal of Minute Virus of Mice (MVM), host cell proteins (HCPs) and DNA using multimodal membrane adsorber Puri™-MQwas studied. The membrane performance was evaluated over a range of feed and elution conditions. The effects of pH and HCP concentration on MVM clearance were investigated. Comparison with other membranes at similar conditions demonstrates that multimodal membranes are highly effective for virus clearance.

BIOT 153

New multimodal anion-exchange membranes for polishing of biologics

Joshua Osuofa1, [email protected], Daniel Henn4, Jinxiang Zhou3, Scott M. Husson2. (1) Chemical and Biomolecular Engineering, Clemson University, Clemson, South Carolina, United States (2) Chemical Biomolecular Engineering, Clemson University, Clemson, South Carolina, United States (4) Purilogics, LLC, Greenville, South Carolina, United States

This presentation will highlight findings from a study using new multimodal membranes for high throughput polishing at elevated solution conductivities. Multimodal membranes combine Coulombic interactions, hydrophobic interactions and hydrogen bonding in binding bioprocess impurities, thereby enabling salt-tolerant performance under conditions where conventional ion- exchange media fail. In this study, we evaluated the performance of Purilogics’ multimodal anion-exchange membranes (PurexaTM-MQ) under a variety of industry relevant conditions. We will present dynamic binding capacity (DBC) measurements of bovine serum albumin (BSA), human immunoglobulin G (hIgG), and salmon sperm DNA (ssDNA) for a range of conductivities (0 -30 mS/cm), salt types, flowrates, and pH values (6-9). Using DRAFT

1 mg/ml BSA, measured DBC values were greater than 90 mg/ml for the highest conductivities but showed a maximum of 130mg/ml at around 6 mS/cm. DBC values showed only 3% decrease over 100 bind-elute cycles with regeneration. The membranes had exceptionally high ssDNA DBC of 50 mg/ml under high conductivity conditions, easily outperforming all other commercial products. Purilogics’ PurexaTM-MQ membranes can rapidly purify bioprocess pools without dilution or desalting of feed, and may also hold potential applications for DNA production in the gene therapy industry.

BIOT 154

Bridging high-throughput screening and mechanistic modeling for the development of multimodal chromatographic processes

Scott Altern1, [email protected], Jessica Yang3, Chris Williams3, John Welsh4, Jake Klockowski1, Jamie Peyser5, Vijesh Kumar2, Abraham M. Lenhoff2, Steven M. Cramer1. (1) Chemical and Biological Engineering, Rensselaer Polytechnic Institute, Troy, New York, United States (2) Univ of Delaware, Newark, Delaware, United States (3) Purification Development, Genentech, South San Francisco, California, United States (4) Biologics Process Development and Commercialization, Merck, Kenilworth, New Jersey, United States (5) Bioprocess Development, Repligen, Waltham, Massachusetts, United States

In recent years, high-throughput screening techniques have become more prevalent in the development of chromatographic processes. During initial development stages, it is often challenging to identify optimal combinations of resins and process conditions. Traditional trial-and-error based approaches require extensive experimentation to explore the available design space. High-throughput techniques have since been applied to expedite this time- consuming process. The introduction of robotic liquid handling systems, for instance, has greatly accelerated early stage development. Such capabilities have enabled fast screening of the design space to identify potentially useful process contenders. In addition to high-throughput screening, mechanistic modeling of chromatographic processes has also been shown to expedite early stage process development. Although process modeling can reduce experimental burden during later stages of development, the large amount of experiments necessary for model design presents a challenge for creating an integrated workflow. Since it is increasingly time consuming to model many processes within the design space, it is vital to identify top process candidates from high-throughput screens for further development. As such, an ideal workflow would combine high-throughput screening and mechanistic modeling DRAFT to quickly develop a process of interest while minimizing the use of material.

In this paper we present an efficient workflow for combining high throughput screening and data generation with isotherm formalisms and column modeling to both accurately predict complex behavior in industrial multimodal chromatographic processes and to expedite process development. Examples are presented for both multimodal cation and anion exchange systems operated in bind-elute and flow-through modes of operation, respectively. Using this combined approach, we were able to capture the unique salt and pH dependent chromatographic behavior obtained for mAb preparative separations in multimodal chromatography.

BIOT 155

Iterative mapping approach for identifying next generation multimodal chromatography resins

Lalita Kanwar, [email protected]. Downstream, Cytiva BioPharma, Uppsala, Sweden

Multimodal chromatography with its multi-functional ligands plays a key role in the purification of complex cell culture harvests obtained from upstream processing. The goal of the present study is to develop an iterative framework for the development and selection of new multimodal resins to provide novel selectivity for future purification challenges. Here, we study the adsorption mechanisms of a monoclonal antibody (mAb) starting with a small chemically diverse virtual library of multimodal resins based on 100 prototypes of CaptoTM MMC ligand variants. This chemical diverse library is characterized by a broad array of chemical descriptors calculated in silico, and principal component analysis (PCA) is used to map the chemical diversity w.r.t the influence of new ligands designed towards varied properties such as hydrophobicity, H-bonding, π - π, and cation- π interactions etc. From this chemical diversity map, a smaller library of Capto MMC ligand prototypes was selected for synthesis and coupling to the ImpRes agarose base matrix. For faster screening of these novel ligands, a high-throughput plate-based study of the binding capacity at low protein load for six model proteins including one monoclonal antibody was performed in 6µL PreDictorTM 96-well filter plates with respective novel multimodal cation exchange resins. The binding capacity was determined at four different binding pH, and eight different NaCl salt conditions allowing the construction of a chromatographic diversity map using PCA. We have used the preferential interaction adsorption model to provide DRAFT estimates of: (a) the total number of water molecules and counter salt ions released during mAb adsorption, and (b) hydrophobic contact area (HCA) between the mAb and ligand for various isocratic chromatographic retention conditions. Further, the separation resolution between the mAb and Fab fragments, as well as high molecular weight (HMW) aggregates were studied using linear gradient elution. Promising new multimodal ligands were identified and the modeling points to the importance of secondary interactions (e.g. H- bonding, π - π, and cation- π interactions, etc.). The improved performance of new multimodal ligands in separation applications will be discussed. The iterative mapping approach of chemical and chromatography diversity maps described in the study proves to be a promising method for identifying new ligand chemistries for biopharmaceutical purification challenges.

BIOT 156

Augmenting dissolved oxygen control in intensified cell culture processes

Jonathan Wang1, [email protected], Kai Hoeffner2, Brooke Tam2, May Chin2, Jiuyi Lu1, Jason Walther1, Armin Optiz2, Franqui Jimenez2, Jean McLarty1. (1) Upstream BioProcess Development, Sanofi, Framingham, Massachusetts, United States (2) MSAT, Sanofi, Framingham, Massachusetts, United States

Sanofi’s continuous biomanufacturing platform affords numerous advantages over both legacy perfusion processes and traditional fed-batch systems. Process intensification allows for an order of magnitude higher cell density resulting in increased volumetric productivity. Intensified processes rely heavily on automation with fewer manual operator inputs in order to ensure robust operations.

Dissolved oxygen probes in intensified high density cell culture processes are plagued by gas bubble adherence related noise. Excessive noise may lead to suboptimal oxygen supply resulting in hypoxic cell culture conditions that impact growth and product quality. A root cause analysis was performed to identify factors such as raw materials, equipment, and automation strategies that might influence ability to control dissolved oxygen. We propose a strategy to address common problems observed in dissolved oxygen control and present the proof of concept obtained in an at-scale production vessel.

BIOT 157 DRAFT

Manufacturing of a trivalent subunit vaccine in a single campaign

Laura E. Crowell1, [email protected], Neil Dalvie2, Joseph Brady3, Mary K. Tracey1, Kerry Love1, John C. Love4. (1) Koch Institute at MIT, Cambridge, Massachusetts, United States (2) Chemical engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts, United States (3) Massachusetts Institute of Technology, Cambridge, Massachusetts, United States (4) Building 76 Room 261, Massachusetts Inst of Technology, Cambridge, Massachusetts, United States

Vaccines, along with most other protein biologics, are currently manufactured in large-scale, centralized facilities. Such manufacturing paradigms lack the flexibility and agility required to quickly address global epidemics and have led to stagnant coverage and missed immunization targets due to failures in campaigns or supply chains. Multivalent vaccines are particularly challenging since each subunit is manufactured individually, requiring multiple campaigns to produce a single batch of drug product. New manufacturing methods, and associated process development strategies, are needed to improve the flexibility and dependability of vaccine manufacturing at low costs. We have previously demonstrated a small-scale, integrated and automated manufacturing platform capable of rapidly producing a variety of recombinant proteins with phase-appropriate quality for early development. Our platform comprises three modules for fermentation via perfusion, straight-through chromatographic purification, and formulation combined with a holistic strategy for process design. Here, we demonstrate the application of our platform to manufacture a trivalent rotavirus vaccine in a single campaign through co-expression and co-purification. We first developed and deployed processes for the production of each individual subunit antigen, while maintaining similarity in process operating conditions. Our initial processes showed a 40% reduction in the number of steps as compared to the original manufacturing method for these molecules. While process-related variants were removed, product-related variants including N-terminal truncation, glycosylation, and aggregation persisted. In order to maintain process conditions compatible with all three subunits, we addressed product-related variants at the molecular level by strategically altering specific amino acids. We then created a single Pichia pastoris strain that expressed all three engineered subunits. Finally, a single manufacturing campaign was executed on our system. Successful co-expression and co-purification of all three subunits was confirmed by RPLC, SEC and LC-MS. Process-related impurities were reduced below typical regulatory guidelines. We believe that this work demonstrates the feasibility of manufacturing a multivalent vaccine DRAFT in a single campaign. We hope that such process intensification will enable more flexible and agile manufacturing of vaccines to meet global needs.

BIOT 158

Implementation of continuous processing for the downstream purification of vaccines

Tiago Matos, [email protected], Ayobami Adegbite, David Hoying, Adam Kristopeit, Marc Wenger. Vaccine Process Development, Merck & Co, Phoenixville, Pennsylvania, United States

The increasing worldwide demand for vaccines along with intensifying economic pressure on health care systems underlines the need for further improvement of vaccine manufacturing. In addition, regulatory authorities are encouraging investment in continuous manufacturing (CM) processes to ensure robust production, avoid shortages, and ultimately lower the cost of medications for patients. As of January 2019, the FDA has approved five products from manufacturers using CM. All were, however, small molecules. Our goal is to develop continuous downstream processing for viral vaccines, using strategies such as simulated moving bed chromatography (SMB), countercurrent chromatography, single-pass tangential flow filtration (SP- TFF), and inline filtration and diafiltration (IF/DF). Recently, these techniques have been successfully implemented for smaller biopharmaceutical molecules to address product instability, improve process efficiency, and reduce facility footprint and capital cost. However, the large size of viruses, virus-like particles, and other biological macromolecules has complicated the application of these methods for vaccines. Membrane chromatography is often presented as an optimal approach for viral vaccine processing due to its large pore sizes and convective mass transport properties. We therefore explored continuous processing with membrane chromatography for a viral vaccine candidate by implementing simulated moving bed (SMB) chromatography in a 4-column mode. SMB improves the efficiency of the chromatography step by enabling higher loads and multiple cycles of column use. The separation performance of SMB was compared to a conventional batch chromatographic capture step used at full manufacturing scale. The product step yield while operating in SMB was comparable to batch mode. Furthermore, the total amount of membrane area required for SMB was a tenth of that needed for batch operation, while keeping total processing times similar. Since SMB uses smaller column sizes, it can take advantage of the high flow rates achievable for membrane chromatography that are not typically possible at larger column scales due to the flow rate limitations of DRAFT many chromatography skids. To achieve a fully continuous purification train, we subsequently integrated these continuous chromatography platforms with SP-TFF and IL/DF, taking advantage of larger pore membrane options available to viruses.

BIOT 159

Process Characterization Strategy for Upstream Continuous Manufacturing

Erin Franco1, [email protected], Jack Huang2, Nelson Chavez1, Madiha Khurshid1. (1) Amgen, Inc, Thousand Oaks, California, United States (2) Merck, Kenilworth, New Jersey, United States

Continuous Manufacturing (CM) offers many benefits over traditional fed batch bioreactor processes. The continuous perfusion of CM allows for increased culture productivity and an extended harvest period. However, CM cell culture runs can be significantly longer than traditional processes. The extended length of the process can prove challenging for completing Process Characterization (PC) studies efficiently. Understanding the steady-state of the process allowed us to greatly reduce the time needed to evaluate each condition. By implementing a creative PC design strategy that tests multiple conditions within the steady-state portion of a single bioreactor run, we were able to reduce the PC study timeline by >60%.

BIOT 160

Economics of intensive/continuous downstream manufacturing

Paul A. Sinclair, [email protected]. Biopharm Services Ltd, Chesham, Buckinghamshire, United Kingdom

Within Bioprocessing there is a keen interest in the intensification of BioPharm manufacturing. The key drivers are assumed to be improved flexibility, reduced risk and reduced cost. It is an evolving area and key to gaining insight is the analysis of the downstream requirements from upstream. Advanced bioprocess models are used to understand the drivers and resource requirements of different modes of intensified downstream (DSP) operation. Using as the basis Biophorum technology roadmap models the three areas addressed are. Connected processing: Batch operations that are linked directly without product hold points. DRAFT

Continuous processing: Material flows between unit operations, two scenarios are considered partial (some batch) and fully continuous operation. Solution management: How this is impacted by connected and continuous processing. By considering the total resource requirements the models have provided insights into the impact of new modes of operation and, how these are affected by scale of operation and the influence upstream feed condition(fed batch versus perfusion). Focusing on continuous manufacturing the economic benefits in terms of Cost of Goods (CoG) in the downstream are dependent on scale ranging between 0% to 22% savings for capacities up to 1800 kg/yr. of mAb production. Many of the benefits can be realized from partial integration of continuous technology. The models identify those areas where the new emergent continuous/connected technology can optimised further. At capacities of 900kg/yr. DSP costs are split 60% process and 40% solutions, the bulk of the costs are consumables (chromatography resins, filters etc.). This means that continuous technologies should be optimised for reduced resin usage, and that solution management is a key area to be addressed. Intensified processing can impact negatively on solution management and the options to reducing its impact through use of concentrates and buffer preparation skids is explored.

BIOT 161

Protein A column sanitization: Enabling continuous antibody production for GMP manufacturing

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

Continuous manufacturing of biologics offers the promise of lower capital outlay, improved productivity and reduced cost of goods. Continuous operation can be facilitated by extending the sterile envelope from the bioreactor to the capture step and into the downstream process through the use of sterile, single use components and equipment. However, chromatography columns remain a weak link since in-house packed columns are not sterile, and gamma irradiated columns are expensive and not widely available commercially. Here, we present a case study in the development and implementation of an inexpensive chemical sterilization process for in- house packed columns that permits extended perfusion durations and continuous capture. Initial introduction at the 500L scale was met with mixed success due to contaminations during two of four runs. To address this issue, DRAFT we performed a series of experiments looking at several different Protein A resins, sanitant types and sanitization techniques. We found that the resin and sanitant can interact and impact column sanitization. With this information, we developed a technique to successfully sanitize process scale columns in the PD lab and for two 500L scale fully continuous production runs. A robust sanitization step will be a key step in enabling continuous manufacturing at process scale.

BIOT 162

Optimisation of continuous crystallisation conditions to produce protein crystals of desirable physical properties: The case of lysozyme

Ian Rosbottom, [email protected], Huaiyu Yang, Wenqian Chen, Xiaoyu Li, Jerry Heng. Department of Chemical Engineering, Imperial College London, London, United Kingdom

Crystallization of proteins could be a potentially cost effective and robust method to purify protein molecules in the pursuit of the next generation of protein based biopharmaceutic therapies. However, the step from traditional small-scale high throughput screening methods, to crystallizers which can produce high yields of pure crystalline protein material, is still in its infancy. Here, we present a platform for continuous oscillatory flow crystallization (COFC) (Figure 1) and its application for the crystallization of lysozyme. A workflow is developed from µl screening experiments, to scaled up batch oscillatory flow crystallization (BOFC) and COFC experiments. The induction time for lysozyme is inversely proportional to the initial protein concentration, with the same relationship found between the induction time and the frequency and amplitude of the oscillation. However, the decrease in induction time comes at a cost to the crystal size and quality. In turn, the high shear environment from increased amplitude and frequency of the COFC can also damage the fragile protein crystals, resulting in the formation of more fine particles and an undesirable size distribution. This led to further investigation of slug flow mixing in the continuous crystalliser, where the crystal size and shapes are compared to the particles produced from the osciallatory flow reactor. Further work around optimization of the crystallization parameters and crystallizer geometry using the gCrystal platform to find the optimum scale up conditions are also discussed. The use of seeding and potential templating materials is also covered with a view to expanding the use of this crystallizer to less well characterized protein materials.

DRAFT

BIOT 163

Intensification of buffer preparation using concentrates and a pump- based dilution system

Adam Sokolnicki, [email protected]. MilliporeSigma, Burlington, Massachusetts, United States

This presentation will discuss the use of buffer concentrates and pump-based dilution to intensify buffer preparation operations. Although next generation manufacturing facilities are smaller, buffer demand remains proportional to drug product mass. The higher throughput of these facilities still requires large buffer preparation operations. However, the intensification of in-house buffer preparation to reduce tank capacity and cleanroom storage footprint as well as labor costs, may be achieved through the application of buffer concentrates plus a pump-based dilution system. Furthermore, buffer concentrates eliminate weighing of powders, environmental dust, and keep processing closed, attributes beneficial for next gen facilities. To facilitate implementing such a system, several aspects were demonstrated. On the concentrates side, the development and validation of the concentrates included generating stability data on concentrates stored in single-use assemblies, prior to dilution. Achieving pH and conductivity target specifications required accurate titration of the concentrates. On the dilution system side, pump-based dilutions provided automated buffer makeup to within 0.1 pH unit and 1 mS/cm, at greater than 20x. The overall buffer preparation time for a full suite of mAb buffers at different scales, not including QC release testing, will be compared.

BIOT 164

Integrated continuous bioprocessing with CHO cells and ultra-high productivity for biologics manufacturing

Weichang Zhou, [email protected]. WuXi AppTec Co Ltd, Shanghai, China

Ultra-high productivity continuous bioprocesses have been developed for production of monoclonal antibodies, fusion proteins and bispecfic antibodies using CHO cells. These enable, for example, 2,000L disposable bioreactors to achieve comparable productivity as traditional 20,000L stainless bioreactors, and to significantly reduce manufacturing cost of goods. This process DRAFT technology platform consists of continuous CHO cell culture and continuous direct product capture. Several case studies, which achieved a cell culture productivity of 2-3 g/L/day, and a similar or better purification yield of the traditional purification process, will be discussed to highlight advantages of this integrated continuous bioprocessing platform in terms of productivity gains and quality improvements.

BIOT 164

Integrated continuous bioprocessing with CHO cells and ultra-high productivity for biologics manufacturing

Weichang Zhou, [email protected]. WuXi AppTec Co Ltd, Shanghai, China

Ultra-high productivity continuous bioprocesses have been developed for production of monoclonal antibodies, fusion proteins and bispecfic antibodies using CHO cells. These enable, for example, 2,000L disposable bioreactors to achieve comparable productivity as traditional 20,000L stainless bioreactors, and to significantly reduce manufacturing cost of goods. This process technology platform consists of continuous CHO cell culture and continuous direct product capture. Several case studies, which achieved a cell culture productivity of 2-3 g/L/day, and a similar or better purification yield of the traditional purification process, will be discussed to highlight advantages of this integrated continuous bioprocessing platform in terms of productivity gains and quality improvements.

BIOT 165

Novel bacterial toxin enables CRISPR-free, RNA-free base editing

David R. Liu, [email protected]. Harvard Univ, Cambridge, Massachusetts, United States

Bacterial toxins represent a vast reservoir of biochemical diversity that can be repurposed for biomedical applications. Here, we report the discovery of a novel bacterial toxin that catalyzes the unprecedented deamination of cytidines within double-stranded DNA. All previously described cytidine deaminases operate on single-stranded DNA and thus when applied to genome editing require unwinding of dsDNA by macromolecules such as CRISPR-Cas9 complexed with a guide RNA. Engineering this novel bacterial toxin resulted in RNA-free programmable cytosine base editors that catalyze DRAFT

CG-to-TA conversions efficiently and with high DNA sequence specificity and product purity at a variety of targeted sites in human cells. CRISPR-free base editing enables precise manipulation of DNA when RNA delivery is not possible.

BIOT 167

Developing an empirical media and feed strategy for intensified production of AAV using the Sf9/Baculovirus expression system

Krishanu Mathur, [email protected], Charles Ly, Anastatia Neuman, Peter Slade. Voyager Therapeutics, Cambridge, Massachusetts, United States

Explosive growth in the field of gene therapy has enabled numerous therapeutic candidates to accelerate into the clinic and advance towards commercialization. This rapidly increasing clinical demand necessitates manufacturing processes with ever-increasing productivity that also consistently meet a high standard of product quality. The current Baculovirus/Sf9 platform, despite its many advantages over mammalian systems for rAAV production, is used in batch mode with little intensification being used since its introduction over 20 years ago. Commercially available media solutions cannot support high cell density infections; which causes steep drop in AAV productivity. This productivity drop is attributed to nutrient deficiency in cell culture at high densities. However, a lack in understanding of the Sf9 cell metabolism has inhibited the implementation of strategies which can sequentially promote Sf9 cell growth, baculovirus infection, and AAV production. This presentation highlights Voyager Therapeutics’ efforts at intensifying the Baculovirus/Sf9 platform. Spent media analysis combined with empirical DOE and mixing design studies has led to identification of key nutrients involved in different phases of Sf9 cell growth and AAV production. These learnings have enabled development of a robust basal media formulation and feed concentrate that can be used together for a high cell density infection, fed-batch manufacturing process. Work is continuing to further intensify the process as well as understand the effects of an intensified process on AAV product quality.

BIOT 168

Achieving a potent AAV using the insect cell/baculovirus system DRAFT

Shamik Sharma, [email protected], Jeffrey Slack, Andrade Hendricks, Harvir Grewal, Peter Slade. Voyager Therapeutics, Cambridge, Massachusetts, United States

The baculovirus/insect cell system has proven to be a robust production method for the manufacture of adeno-associated virus (AAV) for gene therapy. It is an attractive method when compared to the triple transfection mammalian process. The baculovirus/insect cell system does not require the production of DNA plasmids, produces a high percentage of full capsids, has similar over all productivities, and is easily scalable. Although the current baculovirus process is adequate for drug manufacturing, it has seen little optimization since it was first introduced to produce recombinant proteins over 20 years ago. Recently there has also been evidence of decreased vector potency some AAV serotypes. Our work has focused on redeveloping the baculovirus/insect cell system to create a highly productive process, producing potent AAV. This has been accomplished by first engineering the baculovirus genome to predictably control AAV viral protein (VP) ratios, resulting in a baculovirus expression system that has the potential to dial-in AAV potency based different protein ratios. We have also focused on increasing overall productivity by addressing bioreactor parameters and media for Sf9 host cell growth, baculovirus infection and AAV production. It is our goal that this work will result in a state-of-the-art process for AAV production on par with current IgG manufacturing in CHO cells.

BIOT 169

Ultrafiltration behavior of adeno-associated viral vectors: Understanding the implications of choice of process parameters and device format on commercial manufacturing

Abhiram Arunkumar, [email protected], Nripen Singh. Technical Operations, Voyager Therapeutics Inc., Cambridge, Massachusetts, United States

Recombinant adeno-associated viral vectors (rAAV) are increasingly used as the vector of choice in gene therapy. The purification process of rAAVs involves cell lysis, clarification of the lysed cells to separate the supernatant containing the rAAV, benzonase® treatment, followed by several purification steps. The final step is an ultrafiltration/diafiltration step to concentrate the viral vector and place it in the formulation buffer matrix. While the ultrafiltration behavior of proteins and monoclonal antibodies are very well understood, rAAV transport in ultrafiltration membrane modules are DRAFT not. This work provides data to demonstrate the impact of membrane molecular weight cut-off, transmembrane pressure, crossflow-rate and the choice of constant TMP-mode or constant permeate flux-mode on the process performance and product quality (recovery of rAAV and aggregation). Several process runs using pressure-dependent ultrafiltration and pressure- independent ultrafiltration are compared to identify the most robust operating condition. Finally, the choice of reusable ultrafiltration modules is compared against single-use ultrafiltration modules keeping product quality, cost of goods and process performance as performance indicators for commercial rAAV manufacturing.

BIOT 170

Scalability and competitor benchmarking of anion exchange membrane chromatography devices for adeno-associated virus purification using a DoE approach

Kurt Boenning, [email protected], Aydin Kavara, Adam Hejmowski, Julio Huato, Mark Schofield, Rene Gantier. Bioprocess R&D, Pall Corporation, Westborough, Massachusetts, United States

Adeno-Associated Viruses (AAVs) are increasingly receiving approval for the in vivo treatment of acquired and inherited genetic diseases. However, purification challenges remain as AAV production typically results in a significant proportion of empty capsids which may invoke an immune response, without the ability to deliver the gene therapy payload. Here we propose a scalable purification process, based on anion exchange (AEX) membrane chromatography for final polishing and empty/full capsid separation. Despite widespread application of AEX membrane chromatography, there has been little published effort to understand the scalability of membrane devices when transitioning from filter plate experiments to flow capsules. A Design of Experiments (DoE) approach was implemented for the optimization of binding and elution conditions of AAVs via high throughput screening on Mustang Q filter plates. These conditions were transferred to Mustang Q capsules along with alternative membrane chromatography devices of comparable scale. Gradient and step elution experiments were carried out to understand the role of salt and additives in desorption of AAV from the membrane surface and the separation of empty and full capsids. In summary, a scalable and platform-ready AAV purification protocol was developed to facilitate the transition from Mustang Q filter plates to XT acrodiscs. DRAFT

BIOT 171

Navigating the unique manufacturing and supply chain challenges in cell and gene therapy and what we can learn from our past experiences

Rolf A. Ramelmeier, [email protected]. Sangamo Therapeutics, Brisbane, California, United States

Gene and cell therapies present a new wave of modalities to treat and potentially cure disease. They are considerably more complex compared to therapeutic proteins and small molecules; while we can leverage much of our knowledge and experience in that space, cell and gene therapies will require a significant investment in technical and scientific understanding as well as a different approach for manufacturing and supply chain to meet the needs of the patient. This presentation will address the unique challenges to process and analytical development, which is still in the early stages of its evolution and frankly is like the wild west right now. This talk will also address why manufacturing and supply chain need a lot of attention, from starting materials, vector banking, vector and cell production, and frozen vials to chain of custody. Given the huge growth in gene and cell therapy clinical trials and need for clinical material, considerable pressure has been put on both CDMOs and innovator companies to find the right manufacturing capacity. These complex therapies demand more oversight and control via the technical experts which may not reside at the CDMOs, which further challenges our ability to reliably manufacture these compounds. Finally, Sangamo is one of the early pioneers in the gene and cell therapy space; their approach to address these challenges will be highlighted.

BIOT 172

Understanding the effect of Benzonase endonuclease on downstream purification of adeno-associated viral vector (AAV) manufacturing

Blake Hotz, [email protected], Jian Lin, Abhiram Arunkumar, Nripen Singh. Technical Operations, Voyager Therapeutics Inc., Cambridge, Massachusetts, United States

Adeno-associated viruses (AAVs) are an emerging modality as promising vectors for gene therapy. The manufacturing process involves cell expansion, infection and subsequent purification processes, which involves adventitious viral inactivation, cell lysis, followed by several chromatography and ultrafiltration/Diafiltration steps. This work examined the use of benzonase® DRAFT endonuclease in the clarified lysate for nucleic acid (DNA) digestion and its impact on downstream processes that involve affinity purification and polishing chromatography. The optimized conditions for benzonase® endonuclease were chosen to maximize the enzyme activity on DNA digestion and the effect of benzonase® treatment on the performance of the affinity step was studied in combination with several wash and elution conditions. The implications of using benzonase® treatment on the lifecycle of the affinity resin was also evaluated and the implications of this treatment step on impurities at the drug substance level was assessed. A guideline for an optimized process that balances AAV recovery, AAV purity, processing time and manufacturability is provided based on the outcomes of this work.

BIOT 173

Scaled-down, high-throughput optimization of transfection for lentivirus production for use as gene therapy vectors

Sneha Gopal1, [email protected], Ronit Ghosh2, Steven M. Cramer3, Jonathan S. Dordick4. (1) Chemical and Biological Engineering, Rensselaer Polytechnic Institute, Troy, New York, United States (2) Department of Chemical and Biological Engineering, Rensselaer Polytechnic Institute, Troy, New York, United States (3) Ricketts Bldg, Rensselaer Polytechnic Inst, Troy, New York, United States (4) Center for Biotechnology, Renselaer Polytechnic Inst, Troy, New York, United States

Lentiviral vectors (LVVs) are emerging as a promising tool for gene and cell therapy. Due to the ability of LVVs to deliver genetic payloads to both dividing and non-dividing cells, they possess significant potential for introducing therapeutic genes to cells of interest. LVVs are typically produced by transfecting envelope, packaging and transfer plasmids into Human Embryonic Kidney 293 (HEK293) or other suitable cell lines. Nevertheless, several challenges remain in the production process including low titers and a lack of suitable scale-up production. To address these concerns, we developed a high-throughput microwell platform for upstream transfection optimization leading to higher yields of LVVs. Several commercially available cell media were evaluated for their ability to support high cell densities. BalanCD – a commercially available HEK293 suspension cell medium was able to support both high cell densities and generate high transfection efficiencies, and was, therefore, selected for further screens. To optimize transfection, screens were performed to establish optimal amount of DNA/cell, viable cell density at time of transfection, optimal ratios among packaging, DRAFT envelope and transfer plasmids, and the ideal transfection reagent to DNA ratio. Comparisons of optimal conditions across scales were then made for both batch scale-up modes.

BIOT 174

Use of mechanistic modeling in process development for lentiviral vector purification

Sushmita Koley, [email protected], Scott Altern, Ronit Ghosh, Steven M. Cramer. Department of Chemical and Biological Engineering, Rensselaer Polytechnic Institute, Troy, New York, United States

With the advent of gene therapy and its potential to provide a cure for a variety of diseases, there has been an increasing interest in lentivirus as a therapeutic vector as well as for the transfection of cell therapy agents. However, there is an urgent need to improve the yield of downstream processes for producing lentiviral vectors. The objective of this work is to generate mechanistic models that bridge the gap between high throughput screening and process scale chromatography to reduce time and experimental efforts for process generation with gene therapy agents. This work includes the development of a modeling strategy for predicting the elution behavior of lentivirus and removal of product and process-related impurities such as HCP and hcDNA using various ion exchange platforms. The paper presents an efficient workflow to facilitate process development using a combined high throughput screening and modeling approach and demonstrates its utility using a number of formats including resins, membranes, and monoliths.

BIOT 175

Enabling scalable, quality-by-design (QbD)-driven biomanufacturing of cell therapies: 10 year roadmap from the NSF ERC on cell manufacturing technologies (CMaT)

Krish Roy, [email protected]. Georgia Tech, Atlanta, Georgia, United States

With the FDA approval of the first two cell therapies and several in the pipeline, therapeutic cells have rapidly emerged as a critical product platform in the biotech and pharma industry. However, for most cell-based therapeutics and regenerative medicine products little is known about their mechanisms of DRAFT action (MoA), critical quality attributes (CQAs), and critical process parameters (CPPs) that could enable quality-driven manufacturing. In fact, understanding what “quality” means for a particular cell product and how to manufacture those cells with consistent and reproducible quality, at large scale, and at a cost significantly lower than the current approved products are the fundamental challenges facing the cell therapy field. The Center for Cell manufacturing Technologies (CMaT), and NSF-funded Engineering Research Center (ERC) is an eight university public-private consortium that is working closely with industry, academia, and government organizations to enable QbD driven manufacturing of cell therapies by developing new tools and technologies for CQA and CPP identification of complex cell products, real time monitoring of cell-manufacturing processes and products, rapid in-line or at-line analytics, feedback-controlled flexible bioprocessing, efficient supply chain management, scalable manufacturing of the most potent and safe cells, and the development of standards and best practices. In this talk I will provide an overview of our industry-led roadmap to achieve QbD driven cell manufacturing and show specific examples of projects and results from various CMaT funded projects in this area.

BIOT 176

T cells redirected with highly modular biepitopic and bispecific antibody mimic receptors for cancer Immunotherapy

Rihe Liu1,2, [email protected]. (1) Eshelman School of Pharmacy, UNC Chapel Hill, Chapel Hill, North Carolina, United States (2) Carolina Center for Genome Sciences, UNC Chapel Hill, Chapel Hill, North Carolina, United States

Tumors are inherently heterogeneous in antigen expression, and escape due to antigen loss remains one of the major limitations of targeted immunotherapy. Despite the remarkable clinical activity of adoptive therapy with chimeric antigen receptor (CAR) redirected T cells in lymphoblastic leukemia, treatment failure due to epitope loss occurs. One of the effective approaches to improving the outcome of CAR-T cell therapies is by targeting multiple tumor-associated antigens (TAA). To address current drawbacks in achieving multiple tumor targeting using scFvs, we used single domain antibody mimics with small size, simple structure, and high stability that can be modularly engineered on the surface of immune cells to acquire multi-functional TAA recognition. We demonstrated that protein domains targeting EGFR and HER2 of the ErbB family can be efficiently assembled into antibody mimic receptor (amR) molecules, and efficiently redirect T cells DRAFT for biepitopic or bispecific TAA recognition in tissue culture and in soild tumor mouse models.

BIOT 177

Effect of key cytokines on the growth of different Tcell phenotypes

Canaan Coppola1,2, [email protected], William J. Kelly2, Zuyi Huang2, Brooks Hopkins2. (1) Chemical Engineering, Villanova University, Philadelphia, Pennsylvania, United States (2) CHE, Villanova, Schwenksville, Pennsylvania, United States

A significant fraction of CART batches fail every year due to slow Tcell growth rates during the expansion step of the process. Furthermore, the starting material for every batch differs in many ways including the distribution of phenotypes in the Tcell population, yet the final product specification is fixed. Tcell growth rates can be effected by levels of cytokines and different Tcell phenotypes respond differently to these growth factors. In this study, the growth rates of important CD4+ and CD8+ Tcell phenotypes (i.e. Tnaive, Teff, Tcm) were evaluated and modelled as a function of cytokine (Il2, Il7 and Il15) level. Cell health was assessed via an Annexin V-propidium iodide assay, and phenotype was determined via Flow Cytometry. The results indicate significant variation between the different Tcell phenotypes in terms of how their growth and health is affected by different levels of the different cytokines. Interestingly, all of the Tcell phenotypes in varying degrees exhibited a reduction in cell numbers in the first day or two following vial thaw and bead- induced activation. Immediately following that period, effector memory cells in general grew faster than the other phenotypes. Interestingly optimal growth rates were observed at intermediate cytokine levels, a result that is helpful in keeping process costs down. A fuzzy model employed effectively predicted how the Tcell growth rates varied with the different combinations of cytokines, and transcriptomics data helped to inform the model on metabolic activities relating to cell growth. Some of the experiments involved mixed culture of different Tcell phenotypes, to assess how interaction between Tcells effects their growth.

BIOT 178 DRAFT

Characterization of the effect of extended K562 stimulation on natural killer cell functionality for emergent off-the-shelf allogeneic therapy applications

Jennifer One1, [email protected], Samira Azarin4, Wei Shou Hu3, Frank Cichocki2. (1) Biomedical Enginneering, University of Minnesota, Minneapolis, Minnesota, United States (2) Department of Medicine Division of Hematology, Oncology and Transplantation, University of Minnesota, Minneapolis, Minnesota, United States (3) Department of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, Minnesota, United States (4) Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, Minnesota, United States

Natural killer (NK) cells, which have both cytotoxic and cytolytic effector functions to kill cancer cells, are an ideal allogeneic therapy as they do not induce Graft-vs-Host disease and can be expanded by several orders of magnitude through multiple rounds of activation with K562 artificial antigen presenting cells. However, the changes that extended stimulation of NK cells might incur in cell cycle status, metabolism, and functionality necessary for production of sufficient cell quantities are not well understood and thus were the focus of this study. We observed that repeated activation allowed for 106- 108 fold expansion, although a reduced growth rate was observed from approximately day 21 onwards, potentially signifying the presence of senescent cells in later stages of culture. Flow cytometry was used to further evaluate expression of various markers for exhaustion while functional cell cytotoxicity assays were used to determine the presence of important secreted factors for NK cell functionality. IncuCyte assays measured the rate of killing during the different growth phases. Together, this allowed us to explore the relationship between NK proliferative capacity and cytotoxic ability. RNA-sequencing analysis was used to explore activation-associated gene expression dynamics, while ATAC-sequencing was utilized to unveil the dynamic nature of epigenetic/transcriptional modifications during NK cell activation. Activation was found to be accompanied by an increase in oxygen consumption and upregulation of important metabolic pathways, demonstrating that metabolism is an important indicator of activation. Extensive characterization of prolonged K562 stimulation through transcriptome analysis, dynamics of chromatin accessibility, metabolic characterization, and phenotypic analysis of exhaustion and senescence allows us to evaluate the activation and expansion process of NK cells over continuous long-term culture. Insights into mechanisms underlying activation and expansion will provide a path to develop strategies for large-scale DRAFT biomanufacturing while maintaining functionality, thus laying the groundwork as a potential off-the-shelf allogeneic cellular therapy.

BIOT 179

Engineered IL13 variants direct specificity of IL13Rα2-targeted CAR T cell therapy

Lawrence A. Stern1, [email protected], Xin Yang1, Ignacio Moraga Gonzalez2,3,4, Brenda Aguilar1, Joseph R. Cohen1, Renate Starr1, Stephen J. Forman1, K. C. Garcia2,3,4, Christine E. Brown1. (1) Beckman Research Institute, City of Hope, Duarte, California, United States (2) Howard Hughes Medical Institute, Stanford University School of Medicine, Stanford, California, United States (3) Department of Molecular and Cellular Physiology, Stanford University School of Medicine, Stanford, California, United States (4) Department of Structural Biology, Stanford University School of Medicine, Stanford, California, United States

Chimeric antigen receptor (CAR) design incorporates a binding moiety to direct T cell activation toward an individual cancer biomarker that a T cell would not normally recognize. Although antibody variable fragments are commonly used for their versatility in antigen recognition, naturally occurring ligands are also capable of directing specificity. One such ligand, IL13, has been incorporated as the binding moiety for CAR T cell therapy of IL13 receptor α2 (IL13Rα2) expressing glioblastoma multiforme (GBM). This IL13- CAR T cell therapy has shown notable safety and success in early stage clinical trials via local delivery. IL13Rα2 is an attractive target for many malignancies and has rare normal tissue expression, motivating application of systemic delivery of IL13-CAR T cells. However, IL13Rα1, the other receptor for IL13, has low level ubiquitous expression on normal tissue. Notably, IL13Rα1 and ILR4α, a high affinity receptor pair for IL13, are co-expressed in lung tissue. This normal tissue expression raises concerns for systemic application of IL13-CAR T cells including 1) impedance of CAR T cell trafficking, 2) early exhaustion of CAR T cells due to antigen experience on route to tumors, and 3) on-target off-tumor toxicity. Herein, we describe the engineering and investigation of CARs incorporating IL13 mutants (termed muteins) with markedly enhanced selectivity toward IL13Rα2-binding. Structure-guided mutagenesis of IL13 yielded two mutants that retain binding affinity for IL13Rα2 while diminishing affinity for IL13Rα1 >1,000-fold (termed C4 and D7). CARs incorporating these muteins direct T cell-mediated cytotoxicity against IL13Rα2-expressing tumors identically to IL13-CAR T cells both in vitro and in an orthotopic GBM model. Importantly, both muteins DRAFT display significantly diminished T cell activity against IL13Rα1-overexpressing tumors in vitro, but only the C4 variant spares IL13Rα1 and IL4Rα co- expressing tumors in vitro and in vivo. T cell trafficking studies illuminate potential advantages of decreased IL13Rα1 interactions in the mutein-CAR context.

BIOT 180

Genetically engineered immunity for cancer: The development and global regulatory approvals of chimeric antigen receptor (CAR) T cells and what’s coming next

Bruce Levine, [email protected]. University of Pennsylvania, Philadelphia, Pennsylvania, United States

Since the 1990’s, we have conducted clinical trials of gene modified T cells. Gene editing has created T cell resistant to HIV infection. Chimeric antigen receptor (CAR) T cells targeting CD19 on B cells leukemias and lymphomas have induced durable complete responses in patients who are relapsed or refractory to all other available treatments. This synthetic biology technology has now undergone global multi-center clinical trials and recently received FDA, EMEA, Canada, Switzerland, Japan, and Australia approvals (KymriahTM, Novartis) in relapsed/refractory acute lymphoid leukemia in children and young adults as well as in diffuse large B cell lymphoma. Translation of these technologies from research bench to clinical application requires integrated scientific, engineering, clinical, and regulatory expertise. New designs for genetically engineered T cells include switches and potency enhancements that will be required for targeting solid tumors. The road forward for wide patient access to these uniquely personal cellular therapies depends not only on scientific progress in targeting, gene modification and cellular manipulation, but also on meeting automation, engineering, clinical site onboarding, and health policy challenges.

BIOT 182

Accelerated development of processes for the integrated, straight- through purification of biopharmaceuticals

Laura E. Crowell1, [email protected], Sergio Rodriguez2, Kerry Love1, Steven M. Cramer3, John C. Love4. (1) Koch Institute at MIT, Cambridge, Massachusetts, United States (2) MIT, Cambridge, Massachusetts, United States (3) Ricketts Bldg, Rensselaer Polytechnic Inst, Troy, New York, United DRAFT

States (4) Building 76 Room 261, Massachusetts Inst of Technology, Cambridge, Massachusetts, United States

Processes for the manufacturing of biopharmaceuticals typically vary widely in the type of and number of unit operations, particularly for molecules other than monoclonal antibodies. This leads to custom facility designs and unique strategies for process development for every new molecule. To enable flexible, multi-product manufacturing facilities and to reduce the speed to clinic for new molecules, streamlined manufacturing processes and associated strategies for process development are needed. We have developed a bench- scale, integrated and automated manufacturing platform capable of rapidly producing a variety of pre-clinical quality recombinant proteins. This system utilizes straight-through purification, where the eluate of one column is loaded directly onto the next column without any conditioning (pH or salt changes), removing the need for hold tanks or additional unit operations and significantly reducing processing time, buffer requirements, and manufacturing footprint. Development and optimization of straight-through chromatographic processes is a challenge, however, as conventional high throughput screening methods optimize each chromatographic step independently, with little consideration towards the connectivity of steps. Here, we demonstrate a method for the development and optimization of fully integrated, multi-column processes for straight-through purification. Resin selection was performed using an in- silico tool for the prediction of straight-through purification processes based on a one-time characterization of host-cell proteins combined with the chromatographic behavior of the product. A two-step optimization was then conducted to maximize yield while minimizing process- and product-related impurities. This optimization included a series of range finding experiments on each individual column, similar to conventional screening, followed by the development of a model for the fully integrated, multi-column process using design of experiments (DoE). We use these methods to produce pre-clinical quality products in as few as six weeks after obtaining the product sequence. Further, we present a method for predicting the optimal operating conditions for a new molecule within the same class based on its biophysical characteristics. Rapid process development techniques for integrated processes, such as those presented here, could enable the use of multi- product manufacturing facilities and reduce the speed to clinic for new molecules.

BIOT 183

Integration of normal flow filtration into downstream continuous bioprocessing DRAFT

Jessica R. Molek, [email protected], Mehdi Ghodbane, John Starcevich, Patrick Thompson, Antonio R. Ubiera. GSK, Springfield, Pennsylvania, United States

Normal flow filtration (NFF) is ubiquitous throughout the purification process for biopharmaceuticals. Common applications of NFF within the downstream platform include removal of cell and cell debris, of precipitated particulates, of microbial contaminants, or of virus and virus-like particles. In NFF the pressure-driven feed stream flows perpendicular to the membrane, where process- and product-related impurities are retained primarily due to size exclusion while allowing the product to pass through unretained. While NFF is relatively straight forward to introduce into a batch process, implementation within an integrated processing environment is not immediately obvious due to the semi-continuous nature of the operation. This work explores the use of a duty/standby strategy to operationalize and allow NFF unit operations to fit within the integrated semi-continuous downstream processing framework. Using this approach, failure modes and thus links between quality attributes and process parameters must be reconsidered to develop robust control strategies that are informed by process analytical technologies (PAT). Processes for each NFF unit operation are defined based on small scale characterization of the system, as well as by the requirements and limitations of the manufacturing environment, such as overall process run time, desired filter change out rate, and capability of available PAT and automation systems. A comprehensive analysis of a proposed duty/standby-based strategy for normal flow filtration within an integrated continuous process, relative to traditional batch processing, will be discussed, highlighting the benefits, efficiency, and level of robustness that can be achieved with this approach.

BIOT 184

Process intensification of mAb purification for efficient impurity removal

Hirohisa Takeuchi1, [email protected], Jeonghyun Son2, Kyo Izumida2, Masato Suenaga2, Olga M. Paley1, Norbert Schuelke1. (1) Takeda Pharmaceuticals International Co., Lexington, Massachusetts, United States (2) Takeda Pharmaceutical Company Limited, Fujisawa, Japan

The present work is a case study in applying process intensification principles to purification development of a commercially suitable manufacturing process for a monoclonal antibody at Takeda. A previously developed process for the antibody faced significant challenges such as insufficient impurity removal DRAFT efficiency and low throughput. Hence, the existing process was intensified to remove impurities more efficiently and establish an economically and environmentally friendly process by increasing the load capacities of each step and reducing the associated number of cycles and buffer consumption demands. In addition, by combining the flow through mode of hydrophobic interaction chromatography with a reduced bed height, resin consumption and operation time were reduced, while impurities were nonetheless removed more efficiently than the existing process. The present result from this study will be used as a basis for a future assessment for a semi continuous interconnected process.

BIOT 185

Integrated and continuous biomanufacturing: Technical, quality and regulatory opportunities

Nihal Tugcu, [email protected]. Sanofi, Framingham, Massachusetts, United States

This keynote lecture will discuss Integrated and Continuous Biomanufacturing: Technical, Quality and Regulatory opportunities.

BIOT 186

Continuous downstream processing: A case study for enhanced monoclonal antibody purification

Jungmin Oh2, Josh Sumoski1, Calvin Cheah1, Andrew Kalinovich1, Jonathan Fura1, [email protected], Nandu Deorkar1. (1) Avantor, Macungie, Pennsylvania, United States (2) Avantor, Bridgewater, New Jersey, United States

An integrated downstream process is needed to efficiently manufacture and support the more than one thousand antibody therapeutics currently in clinical development. Although technological advancements such as chromatography resins with higher binding capacities, in-line process monitoring, and digitization has improved antibody manufacturing, downstream processing is still a major bottleneck due to disconnected unit operations. Herein, we will present an integrated continuous downstream purification process that utilizes a novel approach to connect unit operations such as affinity chromatography resin, viral inactivation and cation exchange polishing steps. Additionally, we demonstrate the use of novel buffers and additives for column washing and DRAFT generation to increase impurity clearance in a single step to meet quality requirements while simultaneously reducing overall process time. We have screened various buffer additives and identified few for use in affinity chromatography step that can remove at least 50% more host cell protein impurities and antibody aggregates in a single step as compared to traditional approaches. In totality, these additives utilized in conjunction with this newly developed continuous purification process represents a highly efficient and advantageous approach to obtaining purified monoclonal antibodies.

BIOT 188

Evolution of intensified downstream processes

Ashley Hesslein, [email protected]. Bayer HealthCare, Berkeley, California, United States

Over the past decades, the Biotechnology industry has designed a robust processing platform for monoclonal antibodies and similar therapeutic proteins. With few exceptions, a well-established toolbox can be used with small variations. However, industry scientists and businesspeople are consistently pushing to improve efficiencies in manufacturing including increasing productivity, decreasing costs and maximizing facility usage. Upstream cell biologists and engineers have developed incremental productivity improvements that increase the demand on cell clarification, chromatography and other downstream unit operations. This talk will summarize the upstream and downstream intensification trends in the industry starting with mAbs, and moving in to the more recent field of gene therapy and viral vectors.

BIOT 189

Process development for continuous flow-through polishing purification for mAb processing

Sandeep Mora, Junyan Zhang, [email protected], Meghan Higson. BioContinuum™ Platform , MilliporeSigma, Cambridge, Massachusetts, United States

Over the last several years, the biopharmaceutical industry has had a significant focus on connected and continuous processing to improve both process economics and plant utilization. As opposed to the traditional polishing trains comprised of bind-elute chromatography operations, an all DRAFT flow-through polishing train easily enables connected and continuous processing while simultaneously improving process economics, flexibility, and productivity. Leveraging commercially available and novel prototype chemistries and devices, we investigated how a properly designed flow- through polishing train can be used to meet the stringent demands expected for mAb polishing purification. A streamlined methodology will be described with two monoclonal antibody molecules to investigate the performance of individual units as well as synergies between technologies. For both individual technologies and connected processes, results will be discussed on their ability to meet purity and yield targets robustly. Finally, we will show how leveraging the integrated combination of unit operations can result in improved performance over the standard batch, segregated processing paradigm including economic benefits of the flow-through polishing toolbox using BioSolve modeling.

BIOT 190

Nanoparticle assisted protein crystal nucleation: Towards crystallization in downstream protein purification

Caroline McCue, [email protected], Henri-Louis Girard, Kripa K. Varanasi. Mechanical Engineering, MIT, Cambridge, Massachusetts, United States

Slow protein crystal nucleation and the need for high protein concentration are two major barriers to using crystallization as a separation and purification strategy in protein manufacturing. This work investigates the use of functionalized nanoparticles in solution to act as templates for initiating crystal nucleation, decreasing the time required for crystallization and lowering the concentration of protein required. We used lysozyme, a protein with well- characterized crystallization conditions, to evaluate the probability of crystal growth at a range of protein concentrations and to measure nucleation rates and induction times of crystals grown on the functionalized particles. We demonstrated the use of these nanoparticles to enable undersaturated nucleation using a microbatch technique. The nucleation rates and induction times were measured using an emulsion based setup and an image recognition algorithm to count the number of crystals that formed in drops over time. The functional groups used include bioconjugates, which bind covalently with biomolecules, and are commonly used in diagnostic imaging and targeted drug delivery. This work suggests that functionalized nanoparticles enhance crystal nucleation, and could help enable crystallization to be used as a DRAFT purification step in protein manufacturing by reducing both the protein concentration and the time required to nucleate protein crystals.

BIOT 190

Nanoparticle assisted protein crystal nucleation: Towards crystallization in downstream protein purification

Caroline McCue, [email protected], Henri-Louis Girard, Kripa K. Varanasi. Mechanical Engineering, MIT, Cambridge, Massachusetts, United States

Slow protein crystal nucleation and the need for high protein concentration are two major barriers to using crystallization as a separation and purification strategy in protein manufacturing. This work investigates the use of functionalized nanoparticles in solution to act as templates for initiating crystal nucleation, decreasing the time required for crystallization and lowering the concentration of protein required. We used lysozyme, a protein with well- characterized crystallization conditions, to evaluate the probability of crystal growth at a range of protein concentrations and to measure nucleation rates and induction times of crystals grown on the functionalized particles. We demonstrated the use of these nanoparticles to enable undersaturated nucleation using a microbatch technique. The nucleation rates and induction times were measured using an emulsion based setup and an image recognition algorithm to count the number of crystals that formed in drops over time. The functional groups used include bioconjugates, which bind covalently with biomolecules, and are commonly used in diagnostic imaging and targeted drug delivery. This work suggests that functionalized nanoparticles enhance crystal nucleation, and could help enable crystallization to be used as a purification step in protein manufacturing by reducing both the protein concentration and the time required to nucleate protein crystals.

BIOT 191

Removal of host-cell proteins and mAb aggregates using cation- exchange multi-column displacement chromatography

Ohnmar Khanal2, Vijesh Kumar1, [email protected], Abraham M. Lenhoff3. (1) Chemical and Biomolecular Engineering, University of Delaware, Newark, Delaware, United States (2) University of Delaware, Newark, Delaware, United States (3) Univ of Delaware, Newark, Delaware, United States DRAFT

Host-cell proteins (HCPs) and monoclonal antibody (mAb) aggregates are mostly removed during Protein A affinity and anion-exchange chromatography in downstream platform processes for mAbs. The remaining trace amounts of HCPs and aggregates are, however, difficult to remove in a typical cation- exchange (CEX) chromatography polishing step. HCP removal in CEX has not been studied in detail and depends largely on the pH of the wash buffers and the choice of resin chemistry. Aggregate clearance is achieved mainly by peak cutting and peak pooling in gradient elution, leading to yield loss. In the present work, we demonstrate the use of multi-column displacement chromatography for removal of HCPs and mAb aggregates. HCPs can be placed into two broad categories based on their isoelectric points (pI): HCPs with pI values higher than that of the mAb will flow through during product loading, while HCPs with pI values higher than that of the mAb will bind more strongly and displace the mAb. The displacement profiles of HCP versus mAb samples were constructed using analysis of elution fractions at different axial positions in the column. Based on ELISA analysis and capacity calculations for HCPs, it was found out that a column of bed height ~ 15 % of the main column prior to the main column would be required to displace the mAb sufficiently so as to obtain less than 10 ppm of HCP in the final elution pool. In a similar way, mAb aggregates, owing to their higher binding affinity, also showed displacement of monomer on the CEX column. The displacement profile was obtained by analyzing fractions of breakthrough load on a single column at different pH values. The optimal displacement condition was obtained at a pH where the mAb did not bind too weakly or too strongly to the column. To this end, a multi-column set-up was used to obtain product with less than 0.2% aggregate and 90% yield. These multi-column displacement methods can easily be incorporated into a continuous chromatography system to obtain product of high purity without compromising product recovery.

BIOT 192

Replacing process characterization studies by applying mechanistic model simulations: Smart way to accelerating BLA submission?

Gang Wang, [email protected], David Saleh, Federico Rischawy, Simon Kluters, Joey Studts. Late Stage Downstream Process Development, Boehringer Ingelheim Pharma GmbH & Co KG, Biberach, Baden-Württemberg, Germany

On the way to submitting a biologics license application (BLA), there are thousands of lab hours required. The largest and most critical work package in late stage downstream process (DSP) development are process DRAFT characterization studies (PCS). With the PCS results, one should be able to identify the critical process parameters (CPPs) and to define the proven acceptable ranges of the input parameters (PARs), with the final goal of designing a robust manufacturing process that is able to deliver a drug with constant quality. Currently, the industrial standard is to carry out lab scale experiments with multivariate process parameter adjustments and to use response surface models to approximate the univariate influences of process parameters on the critical quality attributes (CQAs). This practice still requires a large number of wet lab experiments. Furthermore, a gap between different scales and low resolution of statistical models used due to multivariate experimental design would deliver unreliable results. In the presented work, we investigated the feasibility of using mechanistic modeling to identify CPPs and PARs of the polishing chromatography step of an on-going late stage project in DSP. A calibrated and validated mechanistic model was used to predict PCS experiments carried out in the wet-lab. With this knowledge, dependencies between potential CPPs and CQAs were determined and the PARs could be defined using this mechanistic model.

BIOT 193

Cytosolic delivery of therapeutically-relevant proteins by site-specific incorporation of endosomolytic peptides to treat breast cancer cells

Rachel M. Lieser1, [email protected], Daniel Yur1, Wilfred Chen2, Millicent O. Sullivan3. (1) Chemical & Biomolecular Engineering, University of Delaware, Newark, Delaware, United States (2) Chemical Engineering, University of Delaware, Newark, Delaware, United States (3) University of Delaware, Newark, Delaware, United States

Efficient intracellular delivery of proteins remains a significant challenge despite its immense therapeutic potential for enzyme therapies and genome- editing. Proteins often require active strategies for intracellular delivery due to their membrane impermeability. In addition, accessing the cytosol or subcellular compartments requires internalized proteins to escape the endosome to evade lysosome degradation or exocytosis. Past efforts to deliver proteins to intracellular targets have often relied on cell-penetrating peptides and lipid nanoparticles, which have shown promise, but often suffer from low potency, limited in vivo activity, and a lack of generalization. Endosomolytic peptides, which are able to fuse with the endosomal membrane following a pH-triggered conformational change, have shown promise in nanoparticle systems, but only a handful of these peptides have DRAFT been tested for cytosolic protein delivery. Further studies are necessary to understand the full potential of these peptides for protein delivery and determine how to use these peptides more generally for delivering proteins that vary in size and composition. Our previous work site-specifically conjugated epidermal growth factor (EGFR)-targeting peptides to proteins to explore the effect of ligand valency and spacing on protein internalization in EGFR-overexpressing breast cancer cells. High EGFR ligand valency and clustering was associated with robust internalization as compared to unmodified protein, however, the internalized protein remained entrapped in the endosome leading to its eventual degradation. In this study, EGFR ligands and multiple endosomolytic peptides were site-specifically conjugated to the protein to enhance endosomal escape. Two endosomolytic peptides with differing mechanisms of endosomal disruption, e.g., the pore-forming and the carpet-like mechanisms, were explored for their ability to deliver a model maltose-binding protein to the cytosol. Both peptides decreased lysosomal protein degradation and enhanced endosomal disruption, although the carpet-like sequence was more effective. Furthermore, this approach was used to deliver p19 siRNA binding protein, for siRNA mediated gene silencing, and Granzyme A, for enhanced cell apoptosis, demonstrating the modularity of the approach. These results, combined with the growing body of literature, may provide insight for future delivery systems so that intracellular protein therapies become more clinically relevant.

BIOT 194

Roles of endocytosis and FcRn-mediated recycling in IgG transport across an in vitro blood-brain barrier

John S. Ruano-Salguero, [email protected], Kelvin H. Lee. Chemical & Biomolecular Engineering, University of Delaware, Newark, Delaware, United States

Immunoglobulin G (IgG)-based therapies for numerous neurodegenerative indications are being evaluated in clinical trials as first-in-class disease- modifying agents. Although promising, poor brain delivery of IgG (~0.01% of the administered dose) – attributed to the restrictive phenotype of brain endothelial cells (BECs) that form the blood-brain barrier – necessitates dosing regimens that have elicited severe adverse effects and raised questions regarding treatment cost and accessibility. Successful strategies to improve IgG brain delivery, such as antibody engineering, must first identify the processes regulating IgG transport across BECs. To overcome the DRAFT experimental complications of in vivo models/methods that have contributed to conflicting hypotheses, mainly concerning the neonatal Fc receptor (FcRn) that regulates IgG recycling and transport in peripheral endothelial cells, IgG transport and trafficking was systematically evaluated using an in vitro model of brain endothelium derived from human pluripotent stem cells. Complementary super-resolution and live-cell fluorescence microscopy were used to correlate intracellular distributions and transport rates for IgGs exhibiting disparate FcRn engagement. Relationships between processing and transport were further explored using modified IgG and other macromolecules, which revealed the impact of endocytoic pathways on transport rates. Collectively, our findings shed light on the factors influencing the rate of brain entry for conventional therapeutic IgG and also provide possible routes to engineer improved variants with enhanced brain delivery.

BIOT 195

Enhancing tissue repair via peptide-mediated gene and drug delivery

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

We have developed new strategies to employ extracellular matrix (ECM)- inspired peptides as drivers of gene and drug delivery during bone and skin repair. This talk will specifically highlight the use of collagen mimetic peptides (CMPs) as tethers that enable integration of DNA polyplexes and antimicrobial agent-loaded liposomes within collagen/fibrin hydrogels (“co-gels”) for applications in wound healing. Chronic wounds are a growing unsolved health problem with a prevalence rate of 1-2% of the general population in developed countries. Two key issues hindering repair include limitations in growth factor activity and accessibility, and the prevalence of infection and biofilm formation in the chronic wound environment. However, while a variety of ECM, growth factor, and antimicrobial treatments have received wide attention, clinical efficacy has been limited. Matrix-mediated delivery offers a lower cost pathway for growth factor delivery that addresses the confounding issues with recombinant protein therapies through its capacity to induce sustained, localized production of bioactive growth factor by endogenous cells; moreover, matrix represents an ideal depot to enable sustained delivery of antimicrobials. In our study, we demonstrate spatio-temporally-controlled PDGF gene transfer by conjugating CMPs to PDGF-encoding polyplexes, such that the polyplexes were stably tethered into the co-gel via CMP strand invasion into the collagen triple helix. In vitro results confirmed that the polyplexes were highly efficient in transfecting fibroblasts to induce PDGF DRAFT expression and subsequent cell proliferation and migration. Simultaneously, we demonstrate that CMPs also provide useful tethers for incorporation of sustained liposomal depots for antimicrobials that enable improved resistance to infection over prolonged time periods. In vivo excisional wound healing studies in mice established faster wound closure using both CMP deliveyr approaches: CMP polyplex-loaded co-gels induced faster wound closure with minimal scar formation as well as improved collagen production, myofibroblast activity, and collagen orientation. Meanwhile, CMP-tethered antimicrobial liposomes offered increased resistance to bacterial infection at the surface of the wounds, resulting in improved healing. These results are indicative of the potential of matrix-modified biomaterial scaffolds in effective treatment of chronic non-healing wounds.

BIOT 196

Peptide nanocluster vaccine formulation impacts immune response

Alexandra Tsoras1, [email protected], Julie Champion2. (1) Chemical & Biomolecular Engineering, Georgia Institute of Technology, Atlanta, Georgia, United States (2) Georgia Institute of Technology, Atlanta, Georgia, United States

Peptide subunit vaccines are desirable because they increase control over the induced immune response and safety of a vaccine by reducing the risk of off- target responses. The immunogenicity of most peptides, however, is often insufficient to produce robust and lasting immunity. Many materials and delivery vehicles have been developed to incorporate target antigens to improve the induced immune response while maintaining specificity and control. We have developed peptide nanoclusters (PNCs) as a subunit peptide vaccine formulation that has shown potential to increase immunogenicity of peptide subunit vaccines. PNCs are comprised of crosslinked peptide antigen, and have been successfully synthesized from several therapeutically relevant peptides as small as 8 amino acids in length. However, synthesis of the PNCs requires either adding residues to the peptide epitope sequence or engaging amino acids within the antigen epitope in conjugation chemistry to form a stable material with sufficient antigen incorporation. It is important to understand if and how these modifications and the stability of the crosslinking itself affect immune responses to inform the formulation of peptide vaccines for induction of the desired immune response with a wide variety of peptide antigens. To understand the effects of peptide sequence and crosslinking, several types of PNCs were synthesized by desolvation from the model peptide epitope DRAFT

SIINFEKL, which is derived from the immunogenic protein ovalbumin. SIINFEKL was altered to include extra residues on each end strategically chosen to enable more conjugation chemistry options, as well as to increase the size of the peptide beyond the minimal presented epitope. Additionally, both reversible and irreversible crosslinkers were used to form the PNCs. These variations were evaluated for immune cell responses in vivo. Dendritic cell antigen presentation and antigen-specific T cell activation increased at varying levels dependent on the type of peptide formulation. The knowledge gained in this work will be applied to disease-specific PNC synthesis to maximize immunogenicity and, ultimately, efficacy of peptide subunit vaccines.

BIOT 197

Amphotericin-B loaded compressed nanofibrous oral tablets as a potential drug delivery vehicle for Leishmaniasis

Mrunalini K. Gaydhane, [email protected], Anindita Laha, Chandra Shekhar Sharma, Saptarshi Majumdar. Chemical Engineering, IIT Hyderabad, Medak, Telangana, India

Amphotericin B is a drug of choice for life threatening disease Leishmaniasis called “kala azar” and severe mycotic infections. Every year Leishmaniasis affects lives of 12 million people whilst fungal infections affect 1.2 billion worldwide. The Amphotericin B molecules are amphipathic in nature and comes with the lipid formulations to be only administered parenterally. Low bioavailability (<0.9%) and high toxicity inside the oral track due to its macromolecular and self-aggregative nature limits oral administration of Amphotericin B. The current research unveils an efficient route to address the toxicity, stability and release kinetic of Amphotericin B, in a controlled manner at realistic pH condition inside the Gastro-intestinal tract. The Gelatin nanofibers have chosen as an excipient for the drug molecule which are cross-linked with a chemical cross-linker, glutaraldehyde for only 8 minutes to improve its water resistance and thermo-mechanical properties. The as spun drug loaded gelatin nanofibers are then compressed to form oral tablets and the release kinetic are studied. The compressed nanofibrous oral tablets (CNOT) containing 20 mg drug / 500 mg weight, has shown the zero order release profile for 10 consecutive days. The antifungal assay against Candida albican species showed resistance towards the fungi till seventh day. The cytotoxicity assay was investigated to find out if there was any toxic effect of glutaraldehyde and amphotericin B onto L929 mice fibroblast cells, which inferred 100% cell viability for 1st, 3rd and 5th day culture. Thus, CNOT offers DRAFT prolong release of Amphotericin B, which will surely enhance patient’s compliance. It protects the drug from premature degradation in high acidic pH condition in stomach. Also, it reduces the infusion related toxicities of lipid based amphotericin B in much more cost efficient manner.

BIOT 198

Modelling sustained release through non-destructive computed tomography

Xutao Shi1, [email protected], Koshari Stijn1, Abraham M. Lenhoff2, Norman J. Wagner3. (1) Chemical and Biomolecular Engineering, University of Delaware, Newark, Delaware, United States (2) Univ of Delaware, Newark, Delaware, United States (3) Dept of Chemical Engineering, University of Delaware, Newark, Delaware, United States

Sustained-release systems improve the bioavailability of biopharmaceutical products and eliminate the necessity for a frequent administration schedule of the drug. A better understanding of such delivery systems, and therefore a more efficient system design, can be achieved by using mathematical models capable of predicting the drug release profiles a priori. However, these models are often formulated phenomenologically, and morphological details such as the early-stage pore formation of the sustained-release system are often neglected or assumed without proper experimental characterization. In this work, we utilized non-destructive X-ray computed tomography (CT) to quantitatively characterize the internal structure of a PLGA-based sustained release system. We modified an existing mechanistic sustained-release model and incorporated the structural characteristics obtained from segmentation of the three-dimensional CT reconstructions. The model predictions were compared to the release profiles obtained from experimental studies. The results demonstrate the importance of structural porosity in sustained release and the potential of a CT-aided mechanistic model to predict the drug release profiles of a PLGA-based sustained-release system.

BIOT 199

Biomaterial approaches for treatment of metastatic ovarian cancer

Samira Azarin, [email protected]. Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, Minnesota, United States DRAFT

Ovarian cancer is typically diagnosed at advanced stages and consequently has poor clinical outcomes. Standard of care is surgical debulking followed by chemotherapy; while many patients initially respond to treatment, 20-30% of patients are resistant to the chemotherapy, as indicated by recurrence within six months of completing treatment. Various immune-based therapies are being explored as alternative therapies for ovarian cancer, including natural killer cells, chimeric antigen receptor T cells, and dendritic cell vaccines. However, these approaches require extensive ex vivo manipulation and expansion of cells. The combination of in vivo focal ablation of solid tumors with immune checkpoint blockade has shown promise in establishing an antigen-specific cytotoxic T cell response. As ovarian cancer is typically detected after it has spread throughout the abdominal cavity, ablation of the primary tumor is likely insufficient for establishing a therapeutically relevant response. To that end, we have developed biomaterials to achieve ablation of metastatic ovarian cancer cells. One such approach involves addition of metal components to polymeric biomaterial scaffolds that have previously been shown to recruit metastatic cells in order to facilitate non-invasive thermal ablation of infiltrating cells in the abdominal cavity. While this approach would be suitable for patients with disseminating tumor cells circulating throughout the abdominal cavity that can be captured by the scaffold, for patients with extensive metastatic nodules already established throughout the abdomen, we instead sought to deliver a biomaterial to those metastatic sites to facilitate ablation. Thus, we have also developed graphene-based biomaterials that bind to ovarian cancer spheroids. Incorporation of a sonosensitizer within the biomaterial enables killing of attached tumor cells via ultrasound treatment. Current efforts are focused on evaluating the immune response following ablation for both approaches and identifying strategies for local or systemic immune modulation to promote a robust cytotoxic T cell response.

BIOT 200

Engineering smart nanotechnologies to harness cytosolic immune surveillance

John T. Wilson, [email protected]. Chemical & Biomolecular Engineering, Vanderbilt University, Nashville, Tennessee, United States

Therapeutics that can harness the immense power and specificity of the immune system have enormous potential to improve human health. Cells of the immune system sense invading pathogens or pathologic issue using a network of receptors that act cooperatively to enhance and shape the type and magnitude of an immune response. Hence, regulating the delivery of DRAFT immunologic cues the cells and pathways of the immune system is fundamental to immunological decision-making. While most therapeutic strategies to modulate immunity target cell surface or endo/lysosomal receptors, the cytosol is also rich in targets for immunomodulation. However, the development of therapeutics that harness these immune surveillance pathways is hindered by major drug delivery barriers, most notably inefficient delivery to cytosolic targets. This talk will focus on research in my laboratory focused on the molecular engineering of nanostructured, supramolecular polymeric materials that enhance intracellular delivery of immunomodulatory proteins, nucleic acids, and small molecules. I will discuss our efforts in developing pH-responsive, endosome-destabilizing “smart” materials for enhanced cytosolic delivery of antigens and immunostimulatory nucleic acids and the application of these approaches in cancer immunotherapy and vaccines for infectious disease. In particular, I will discuss our recent work in the following areas. First, I will describe the development and preclinical evaluation of polymersomes for efficient cytosolic delivery of cyclic dinucleotide STING agonists and describe how we are leveraging this technology to enhance responses to immunotherapy in multiple ways in a diversity of cancer types. Second, I will discuss the development of nanoparticle-based vaccines for installing tissue resident memory T cells at mucosal sites and the capacity of these materials to protect against respiratory virus infection. Collectively, these results highlight the potential to design and leverage self-assembled, pH-responsive materials to open access to cytosolic immune surveillance pathways with potential to lead to next generation immunotherapies and vaccines.

BIOT 201

Enabling targeted cargo delivery to hematopoietic stem and progenitor cells with biomimetic membrane-wrapped nanoparticles

Jenna Harris, Samik Das, Eleftherios T. Papoutsakis, Emily Day, [email protected]. University of Delaware, Newark, Delaware, United States

Enabling targeted cargo delivery to hematopoietic stem and progenitor cells (HSPCs) is a long-standing problem whose solution would transform the treatment of diverse blood disorders. HSPCs are notoriously difficult to transfect, so there is a substantial need for a potent delivery system that will overcome biological barriers to efficiently deliver cargo to HSPCs. Here, we report the development of biomimetic membrane-wrapped NPs that meet this need.

DRAFT

The NPs we developed consist of poly(lactic-co-glycolic acid) (PLGA) cores, loaded with desired cargo, that are wrapped with megakaryocyte (Mk) cell- derived membranes that provide targeted recognition of HSPCs. Either hydrophobic (DiD fluorophores) or hydrophilic (small interfering RNA; siRNA) cargo can be loaded in MkNPs by synthesizing the PLGA cores through single- or double-emulsion solvent evaporation, respectively. These PLGA NPs can then be wrapped with Mk membranes by co-extruding them through a porous membrane. We confirmed successful wrapping by examining NP size, zeta potential, and morphology by dynamic light scattering, nanoparticle tracking analysis, and transmission electron microscopy. Wrapped MkNPs are monodisperse and spherical, with diameter 10-20 nm larger than bare NPs and zeta potential matched to that of the source Mk membranes.

DiD-loaded MkNPs’ interaction with targeted HSPCs versus non-targeted HUVECs (human umbilical vein endothelial cells) and MSCs (mesenchymal stem cells) was evaluated by flow cytometry and confocal microscopy. HSPCs internalized MkNPs within 24 hr of incubationat a much higher rate than than the other two cell types, confirming the Mk membrane coating facilitates HSPC-specific binding.

To test delivery of functional cargo, we prepared MkNPs carrying siRNA targeting CD34 (siCD34) or negative control siRNA (siNeg). These were cultured with CD34+ HSPCs for 24, 48, 72, or 96 hr, then CD34 expression was analyzed by flow cytometry after exposing the samples to fluorophore- labeled anti-CD34 antibodies. MkNPs carrying siCD34 significantly decreased CD34 expression in HSPCs by >15% over 96 hr, while MkNPs carrying siNeg had no impact on CD34 expression.

In summary, MkNPs can be successfully loaded with and deliver either hydrophobic or hydrophilic cargo to HSPCs in vitro, while exhibiting minimal binding to non-targeted cell types. This exciting advance paves the way for further optimization and in vivoevaluation of MkNPs as HSPC-specific delivery vehicles.

BIOT 202

Evolution-guided design of pharmaceutical leads

Ankur Sarkar, Edward Y. Kim, Akarawin Hongdusit, Jerome M. Fox, [email protected]. Chemical and Biological Engineering, University of Colorado, Boulder, Boulder, Colorado, United States DRAFT

Natural products and their derivatives represent a longstanding source of pharmaceuticals and medicinal preparations. These molecules—perhaps, as a result of their biological origin—tend to exhibit favorable pharmacological properties (e.g., bioavailability and “metabolite-likeness”) and can exert a striking variety of therapeutic effects (e.g., analgesic, antiviral, antineoplastic, anti-inflammatory, cytotoxic, immunosuppressive, and immunostimulatory). Recent advances in metabolic engineering and applied enzymology have suppled new approaches for the efficient biosynthesis and functionalization of known, pharmaceutically relevant natural products; complementary methods for the discovery and optimization new products with targeted, therapeutically relevant activities, however, remain poorly developed. In this presentation, I describe an approach for using microbial systems to evolve natural products that inhibit known drug targets. Our group has implemented this approach to build new inhibitors of protein tyrosine phosphatases (PTPs), a class of enzymes that has long eluded drug design, and we have characterized those inhibitors with detailed kinetic, biostructural, and in vitro cell studies. Our results provide evidence of convergent evolution in the biosynthesis of functional metabolites and illustrate how microbial systems can find surprising solutions to sophisticated biochemical objectives. The technology described in this presentation provides a new platform for the discovery and optimization of precisely targeted, readily synthesizable leads compounds.

BIOT 203

Surpassing thermodynamic, kinetic, and stability barriers to biocatalytic isomerization of galactose to tagatose

Nikhil U. Nair, [email protected], Josef R. Bober. Chemical & Biological Engineering, Tufts University , Medford, Massachusetts, United States

Bober & Nair (2019) Nature Communications

D-Tagatose is a rare ketohexose sugar with sweetness similar to that of sucrose. However, its glycemic index and caloric value is much lower because of low bioavailability, making it an attractive GRAS (generally regarded as safe) sugar substitute. Recent studies have also demonstrated that it is anti- hyperglycemic and prebiotic, which promotes gut health.

The enzyme L-arabinose isomerase (LAI) that isomerizes L-arabinose to L- ribulose can also isomerize D-galactose to D-tagatose. A variety of LAIs from different microorganisms have been isolated and have reported optimal activity at a range of temperatures and pH. Some of the limitations of tagatose DRAFT biosynthesis using LAI that may be hindering commercial viability are, 1) unfavorable enzymatic kinetics since galactose is not the native substrate of LAI, 2) low enzyme stability, and 3) low equilibrium constant for galactose to tagatose isomerization.

Few previous reports have been successful at engineering enzymatic properties of LAI for industrial application; often addressing only one of the bottlenecks to productivity. To address the kinetic issue, several groups have used enzyme engineering methods to enhance activtity of LAI toward galactose. To counter low-stability issues, many groups have tested the utility of thermophilic enzymes. However, high reaction temperatures (≥ 80 °C) result in significant caramelization, which is undesirable. Finally, the thermodynamic limitations of isomerization arguably, the most recalcitrant issue since ΔG°rxn ≈ +1.2 kcal/mol.

There have currently been no studies that look to systematically analyze all three limitations – kinetic, thermodynamic, and enzyme stability. This work clearly demonstrates the presence of these three limitations and provides a novel approach to balance their advantages and limitations. We engineered the probiotic bacterium Lactobacillus plantarum as expression host in this approach that enabled ~50 % conversion of galactose to tagatose in 4 h (productivity of ~38 mmol tagatose L-1 h-1) ultimately reaching ~85 % conversion after 48 h at high galactose loading (300 mM) in batch culture. This is among the highest conversions and productivities reported to date for tagatose production using a mesophilic enzyme. Such an approach is expected to be applicable to other biocatalytic systems where similar trade- offs between kinetics, thermodynamics, and/or stability pose hurdles to process development.

BIOT 204

Quantifying noncanonical amino acid incorporation events in yeast: Laying the groundwork for engineering more druglike proteins

Kelly Potts, Jessica Stieglitz, Ming Lei, James A. Van Deventer, [email protected]. Chemical and Biological Engineering, Tufts University, Medford, Massachusetts, United States

Genetically encoding noncanonical amino acids (ncAAs) in proteins is a promising strategy for enhancing the properties of biological therapeutics. There are now tantalizing hints of the power of this approach in the literature, including antibody-drug conjugates and cyclic peptides with properties that DRAFT would be difficult to access using other approaches. However, not all protein engineering platforms are yet compatible with ncAAs. This is especially true in the yeast S. cerevisiae, where the underdeveloped ncAA incorporation toolkit limits integration with powerful yeast display technology. One major barrier to effective integration is a lack of quantitative understanding of ncAA incorporation events. Here, we describe the evaluation of a number of reporters of ncAA incorporation efficiency and fidelity in response to the amber (TAG) codon in yeast that may help shed light on this question: single- fluorescent protein reporters, dual-fluorescent protein reporters, and cell surface displayed protein reporters. Across these varying reporter architectures, we found that flow cytometry-based evaluations with dual reporters yielded consistent quantitative trends across numerous incorporation conditions. In addition, using dual-fluorescent protein reporters, we found that ncAA incorporation efficiency and fidelity is influenced by both stop codon location and N- and C-terminal fluorescent protein identity. Finally, we investigated the feasibility of using several reporters to quantify the effects of single-gene knockouts on ncAA incorporation events. Both dual-fluorescent protein reporters and a modified “drop-in” version of yeast display revealed that several gene knockouts enhance ncAA incorporation efficiency without compromising fidelity. These findings highlight the important roles that quantitative reporter systems play in identifying high efficiency, high fidelity ncAA incorporation events, and suggest opportunities for improving ncAA insertion using genome engineering. Using these reporters to facilitate the encoding of a diverse set of ncAAs in yeast is expected to set the stage for engineering more “druglike” proteins that leverage the broad set of chemistries present in noncanonical amino acid side chains.

BIOT 205

Enhancing protein crystallization screening results using engineered nucleation features

Kyle Nordquist1, [email protected], Tiffany Kinnibrugh1,2, Kevin Schaab1, Jessica Johnson2, Youngchang Kim2, Gyorgy Babnigg2, Andrew H. Bond3. (1) Denovx, Chicago, Illinois, United States (2) Argonne National Lab, Lemont, Illinois, United States (3) DeNovX, Hoffman Est, Illinois, United States

Protein structure determination remains a critical facet of understanding cellular functions and developing treatments for disease, and macromolecular X-ray crystallography is still the benchmark method for determining these structures at atomic resolution. However, time- and sample-consuming DRAFT bottlenecks in the processes between final protein purification and crystallization still exist. Despite advances in high throughput crystallization workflows, < 20% of crystallization screens produce positive results, and fewer crystals from these positive outcomes are of diffraction quality. DeNovX develops crystallization platforms that utilize surface energy modifications to improve crystallization outcomes while retaining diffraction quality of the crystals. Overall, we have demonstrated 1.1-8.7-fold increases in crystallization hit percentages while reducing time to crystallization by an average of 40% for a group of benchmark proteins (e.g., lysozyme, BPT, etc.) and proteins with less characterized crystallization behavior. In addition, samples prepared using engineered nucleation features generate an average of 2.5-fold more crystalline material, which is beneficial where protein supplies are limited and for emerging techniques like X-ray Free Electron Laser Serial Femtosecond Crystallography, which are acutely sample-destructive and can require hundreds or more crystals per study.

BIOT 206

Engineering of virus-derived nanoparticles with controlled architecture for diverse applications

Kok Zhi Lee1, [email protected], Yu-Hsuan Lee2, Corren Scott2, Rachel Susler2, Sue Loesch-Fries4, Michael Harris2, Kevin Solomon3. (1) Agricultural and Biological Engineering, Purdue University, West Lafayette, Indiana, United States (2) Chemical Engineering, Purdue University, West Lafayette, Indiana, United States (3) Agricultural & Biological Engineering, Purdue University, West Lafayette, Indiana, United States (4) Botany and Plant Pathology, Purdue University, West Lafayette, Indiana, United States

Nanoparticles are attractive materials for engineering applications in catalysis, sensing, electronics, photonics, and medicine as material properties are controlled by particle size and morphology. For example, increasing battery electrode surface area via nanorods increases its charge capacity, which advances essential and ubiquitous portable technologies such as pacemakers, cell phones, and other devices. However, uniform nanostructures with defined properties are challenging to produce. One strategy to improve structure uniformity and reduce synthesis costs is with metal-coating of rod-shaped biotemplates derived from plant viruses such as Tobacco Mosaic Virus (TMV) and Barley Stripe Mosaic Virus (BSMV). BSMV is particularly attractive as it is able to bind twice as much metal as TMV, increasing capacity further. However, current plant-based production platforms are non-ideal as viral architecture is coupled to viral replication and DRAFT success, making it challenging to control structure for improved battery performance. Here, we develop E. coli as a production platform for BSMV virus-like particles (VLPs), independent from virus replication. Initial heterologous expression of BSMV coat proteins only produced disc-shaped structures indicative of aborted nanorod self-assembly. To successfully create BMSV-VLP nanorods, we fused the gene for BSMV coat protein to an assembly-initiating DNA sequence from TMV, demonstrating heterologous expression of BSMV-VLPs in E. coli for the first time. We further optimized the production process by improving protein expression conditions, solubility, and purification conditions. Produced BSMV VLPs range in size from 20-160 nm. Collectively, our work creates a novel platform for the synthesis of BSMV biotemplates and enables future engineering opportunities for the development of nanomaterials.

BIOT 207

Synthetic biology for high-performance protein-based materials

Fuzhong Zhang, [email protected]. Energy, Environmental Chem Engineering, Washington University, Saint Louis, Missouri, United States

Advances in synthetic biology have enabled the quantitative engineering of biological processes for the production of various metabolites. Compared to small molecules, efficient and robust production of macromolecular biomaterials is strongly lagging. Microbially produced protein-based materials (PBMs) are particularly attractive to various applications due to their tunable side-chain chemistry, high-performance, and biodegradability. However, high- strength PBMs typically have high molecular weights (HMWs) and repetitive sequences that are difficult to produce microbially due to genetic instability and metabolic burden. We have developed multiple synthetic biology approaches for the synthesis of HMW PBMs using engineered bacterial systems, including a split intein based in vitro protein ligation strategy and an in vivo seeded chain-growth protein polymerization. These biosynthetic strategies were applied to the microbial production of high-performance protein fibers, resulting in synthetic spider proteins with MW reaching 556 kDa and synthetic fibers fully replicate the mechanical performance of natural spider silk by all common metrics, i.e. tensile strength (1.03 ± 0.11 GPa), modulus (13.7 ± 3.0 GPa), extensibility (18 ± 6%), and toughness (114 ± 51 MJ/m3). We also engineered microbes to produce strong underwater adhesive that can be processed to adhesive hydrogels for various underwater repair work. These developed strategies will DRAFT provide numerous novel materials with high-performance and tailored properties, enabling a broad range of new applications.

BIOT 208

Direct identification and counting of microRNAs in single cells by transient hybridization and kinetic fingerprinting

Karen Montoya1, [email protected], Lidan Li2, Greg Shelley3, Evan Keller3, Nils G. Walter4. (1) chemistry, university of michigan, San Diego, California, United States (2) College of Life Science and Technology, Beijing University of Chemical Technology, Beijing, Beijing, China (3) Urology, University of Michigan Ann Arbor, Ann Arbor, Michigan, United States (4) University of Michigan, Ann Arbor, Michigan, United States

The detection and quantification of microRNAs is important for a wide range of applications in biological research and clinical medicine. Current techniques used for quantification of microRNAs rely on amplification via PCR which results in errors emerging from heat-induced chemical damage, fidelity of the DNA polymerase, etc. Recent advances in microRNA detection include the use of imaging techniques to overcome potential bias and errors introduced by PCR. However, even current approaches are limited to relative quantification. To overcome these pitfalls, we are developing a single- molecule microRNA quantification method manipulating single molecule recognition through equilibrium Poisson sampling (SiMREPS), an amplification-free kinetic fingerprinting approach for digital single- molecule detection. This method has been successfully used to detect microRNA and DNA strands with a limit of detection around 1 femtomolar, but has not been tested on single cells and will require optimization and coupling with other methods. In this study, we investigate the detection of miRNA panels in single cells by combining microfluidics and single-molecule microscopy and intend to extend it to intracellular single molecule imaging. Our end goal is to develop a platform for parallel detection of multiple microRNA (miRNA) biomarkers, to be able to detect their abundance in single cells to help better understand human health and predict the likelihood of human disease, including cancer progression.

BIOT 209

Field guide to scientific entrepreneurship DRAFT

Jordi Mata-Fink, [email protected]. Entrepreneur and Inventor, Portland, Oregon, United States

Start-ups are our most efficient means of transforming innovations into products, but the process of converting a scientific advance into a valuable company is opaque and haphazard. This talk will present a systematic framework for successful innovation. We will explore how to: navigate the academic/entrepreneurial interface; define, perceive, and convey value at all stages of an idea or company; and ideate, experiment, and persuade with purpose.

Dr. Mata-Fink is a biotechnology leader, entrepreneur, author, and inventor. Based for many years at Flagship Pioneering in Cambridge, MA, his work has led to the creation of multiple valuable public and private biotechnology companies including: CYGNAL THERAPEUTICS, developing drugs that target neuronal pathways for the treatment of cancer and inflammation; RUBIUS THERAPEUTICS (NASDAQ: RUBY), developing engineered red blood cells as a safe, potent, and long-lasting cellular therapy; and EDITAS MEDICINE (NASDAQ: EDIT), developing the first CRISPR/Cas9-based genome editing therapies for human disease.

BIOT 210

Clonable metal nanoparticles: Applications in electron and fluorescence microscopies

KANDA BORGOGNONI, [email protected]. Chemistry, Colorado State University, Fort Collins, Colorado, United States

Biological microscopic studies, once revolutionized by the discovery of encodable fluorophores such as Green Fluorescent Protein (GFP), engender foundational evidence for cellular processes via in situ protein labeling and tracking by fluorescence. As fluorescence microscopy is inherently hindered by the diffraction limit of light, electron microscopy unveils details of cellular ultrastructure with atomic level resolution. The primary barrier preventing biological electron microscopy from overtaking fluorescence microscopy as the main imaging modality is creating target specific contrast. While methods have been developed to generate more specific contrast agents, none have achieved the specificity of an encodable tag. Our lab initiated the development of encodable contrast agents for electron microscopy by isolating a glutathione-like metalloid reductase from a selenophilic plant capable of reducing selenium precursors into nanoparticles. After confirming the DRAFT portability of the enzyme through maintenance of its activity in vitro and in vivo, modification of the wild type further enhanced its performance as a clonable selenium nanoparticle (cSeNP) tag. We then used the cSeNP tag in a proof of concept study that involved tracking filamenting protein FtsZ, our model system, in vivo using electron microscopy and elemental mapping. Once we fused the cSeNP to FtsZ, initial results generated using electron microscopy closely matched previous fluorescence microscopy studies of GFP tagged FtsZ in vivo. The next phase of this project relies on collaborating with experts in the field of 3D biological electron microscopy to generate tomograms of cSeNP decorated filaments. Ongoing work includes not only post-synthetic modification of cSeNPs to create fluorescent quantum dots (i.e. cCdSe) for applications in fluorescence microscopy, but also expansion of our clonable nanoparticle toolbox through the discovery of other metal reducing enzymes.

BIOT 211

We can do better than kcat/KM: Theory-based development of performance metrics for comparing multi-reactant enzymes

Scott Banta, [email protected]. Chemical Engineering, Columbia University, Fairfield, Connecticut, United States

Protein Engineering provides a toolbox for the creation of new biocatalysts with capabilities beyond what is found in the natural world. As new enzymes are designed, evolved and engineered, it is critical to develop figures of merit to benchmark and compare performances under different conditions. The ratio of kcat/KM is widely used in the evaluation of enzymes due to its mechanistic importance in uni-molecular (Michaelis Menten) enzymatic reactions. However, its applicability in the evaluation of multi-reactant systems is widely used but rarely justified. Therefore, we have derived new figures of merit for multi-reactant enzyme mechanisms based on the stabilization of transition states and/or the kinetic Haldane equilibrium relationships. For most bi- molecular reactions, the ratio of kcat/KiaKb (where Kia is the dissociation constant of A and Kb is the Michaelis constant of B) is best suited for comparing steady state enzyme performances. This approach can be extended to derive useful figures of merit for other more complex reaction mechanisms. The value of the new parameters for the comparison of a series of alcohol dehydrogenase mutants will be discussed. DRAFT

BIOT 212

Data-driven modeling of SHP2-regulated signaling to identify combination therapy approaches in glioblastoma

Evan Day1, [email protected], Matthew Lazzara2. (1) University of Pennsylvania, Philadelphia, Pennsylvania, United States (2) Unversity of Virginia, Charlottesville, Virginia, United States

Targeted therapy efficacy in cancer is influenced by context-dependent, multivariate cellular responses to inhibition of the drug target. These effects often determine whether targeting a particular protein will lead to durable responses in patients, irrespective of the affinity or selectivity of the drug. SH2 domain-containing phosphatase-2 (SHP2) is a proto-oncogene in some cancers for which potent and selective allosteric inhibitors have been recently developed. While SHP2’s most well-known role is to promote activity of the extracellular regulated kinase (ERK) downstream of receptor tyrosine kinases (RTKs), SHP2 can also negatively regulate proliferation and survival signaling (e.g., via the STAT3 pathway). These divergent roles make it difficult to predict the consequences of SHP2 inhibition. For example, we recently demonstrated that in glioblastoma cells SHP2 knockdown simultaneously slowed proliferation but also increased resistance to RTK inhibitors and temozolomide chemotherapy. Our studies also revealed a role for SHP2 in regulating the expression of hypoxia-inducible factors (HIFs) in low oxygen environments, further complicating our ability to predict the effects of SHP2 antagonism in a tumor. To develop quantitative understanding of the pathways and phenotypes regulated by SHP2 in glioblastoma, we implemented a data-driven modeling approach. In cells treated with RTK inhibitors or temozolomide, we measured dynamic changes in the phosphorylation of >40 proteins over 48 hrs using Luminex bead-based immunoassays. Experiments were repeated with or without SHP2 knockdown and cultured in 21% or 1% oxygen, for a total of ~3000 distinct signaling measurements. Parallel measurements of cell cycle progression, cell death, and HIF expression were made. The matched signaling and phenotypic measurements were subjected to partial least squares regression (PLSR) analysis to identify key, phenotype-determining SHP2-regulated signaling processes across different drug treatments and levels of oxygen exposure. One therapeutically relevant prediction from our model was the observation that temozolomide resistance in response to SHP2 inhibition was mediated by a non-canonical role of the tumor suppressor PTEN, which could be overcome by combinatorial RTK inhibition. DRAFT

BIOT 213

Deep learning to improve bacterial cell counting - implementation of classification-type convolutional neural networks (CNN)

Denis Tamiev1, [email protected], Paige Furman1, Nigel Reuel2. (1) BBMB, Iowa State University, Ames, Iowa, United States (2) Iowa State Unviersity, Ames, Iowa, United States

Accurate classification and counting of bacterial cells is important in various microbiome assessment experiments. Most common methods rely on flow cytometry or manual microscopy. Application of flow cytometers in bacterial cell typing is limited due to cell size (bacteria are up to 1000 times smaller than eukaryotes) leading to misclassification of multicellular clumps (biofilms), and cell debris for single cells. Automation of manual microscopy can be achieved with software. Most common software solutions for microscope image processing such as imageJ cannot accurately parse multicellular clumps into single cells or classify them into bacterial cell types. Recent progress in the field of big data and deep learning can be used to develop a classification-type Convolutional Neural Network (CNN) tool for bacterial cell counting. This presentation focuses on describing a custom CNN algorithm that identifies bacterial cell clusters on microscope images, and classifies them based on the number of cells found in each cluster (Fig 1A). Here, we developed a robust image preprocessing algorithm that improves the neural network’s confidence in classifying bacterial clusters by 13.96% (Fig 1D& 1E). This image preprocessing algorithm increases the size of the training database through image augmentation by 72 times (Fig 1B). Such augmentation allowed us to use a very small image dataset, and increased CNN’s accuracy by 13.83% (Fig 1C). In this work we also compare our algorithm to existing techniques and human counting performance. We conclude that classification type CNN can serve as a reliable tool for bacterial cell counting and sorting, and can be leveraged by groups that possess a simple camera-enabled microscope setup

BIOT 214 DRAFT

Predicting protein developability via a deep informatic/experimental hybrid

Alexander Golinski1, [email protected], Katelynn Mischler1, Matthew Fossing1, Nicole Neurock1, Sidharth Laxminarayan1, Hannah Pichman1, Benjamin Hackel2. (1) Chemical Engineering, University of Minnesota, Minneapolis, Minnesota, United States (2) Chemical Engineering and Materials Science, University of Minnesota - Twin Cities, Minneapolis, Minnesota, United States

Proteins must exhibit both a designated primary function (e.g., specific binding, enzymatic activity) and developability (stability, solubility, and producibility) to become lead candidates for downstream applications. While many methods enable high-throughput discovery of primary function, unless developability is additionally screened, translational challenges often occur. Evolving developability is difficult because of the immensity of sequence space and the limited throughput of measuring recombinant yield and thermostability. We hypothesize that high-throughput developability assays will empower deep data collection and analysis to improve the design and discovery of developable proteins and combinatorial libraries. We utilized the small protein scaffold GP2 with two diversified loops as a proof of concept for predicting developability from loop sequence. 105 mutants were subjected to 3 classes of assays previously validated to be representative of protein stability and expression: 1) A yeast-displayed stability assay differentiated mutants by the degree of proteolytic cleavage using a combination of proteinases, temperatures, and denaturants. 2) A bacterial split-GFP assay separated mutants by expression as measured by reconstituted GFP fluorescence. 3) A split β-lactamase assay separated mutants by solubility, stability, and expression, as the GP2-enzyme must be transported to the periplasm as a well-folded protein to remain active. The sorted populations were deep sequenced to measure clonal enrichment scores related to developability. Machine learning models were trained to predict the assay scores based upon sequence yielding an average Spearman’s correlation coefficient of 0.58 across the assays. Results confirm an increase in median stability of a disulfide stabilized loop and 2-9% increase in Spearman’s rho when utilizing an embedding and recurrent neural network over traditional modeling strategies. Also, SHuffle cells, engineered for increased expressability, were shown to increase clonal differentiation. Our model was able to reduce continued experimentation of poorly developable mutants by lowering the false positive rate of producible proteins by 5%. Continued progress will aim to predict additional developability metrics DRAFT including thermostability and self-aggregation to identify a more developable subset of mutants. Long term advancements will expand the assays to assess alternative protein scaffolds, enzymes, and anti-microbial peptides.

BIOT 215

Extending BioSolve with multi-objective Bayesian optimization for automated tuning of upstream and downstream decisions in a mAb process

Folarin B. Oyebolu1,2, [email protected], Richard Allmendinger1, Adriana G. Lopes2, Jonathan Shapiro3, Paul A. Sinclair2. (1) Alliance Manchester Business School, University of Manchester, London, United Kingdom (2) Biopharm Services Ltd, Chesham, Buckinghamshire, United Kingdom (3) School of Computer Science, University of Manchester, Manchester, United Kingdom

BioSolve Process is the biopharmaceutical industry’s most powerful software for bioprocess analysis. It enables users to investigate a wide range of process scenarios and parameters, and make informed and targeted process design decisions. However, due to the complexity of the bioprocess models and the large decision space, these simulations can be labour- and time- intensive. This challenge motivates the use of optimisation techniques, such as Bayesian optimisation, that require only a few evaluations to arrive at near optimal solutions. BioSolve Process functionality has therefore been extended to incorporate a Bayesian optimisation tool that automates the selection of the most efficient process parameters, based on pre-defined user criteria. This optimisation tool is customised to tune a combination of categorical and discrete variables, subject to black-box constraints and conflicting objectives. Firstly, a case study based on optimising upstream decisions over a range of target throughputs for a traditional monoclonal antibody (mAb) process was used to assess the performance of the optimisation tool. The upstream decisions considered included choosing between stainless-steel and single- use bioreactors, the number of installed vessels, and how many bioreactors were pooled together for further downstream processing. Compared to manual full enumeration, the optimisation tool showed a substantial reduction in evaluations. A second case study incorporated downstream decisions. In particular, the optimisation tool was able to select between conventional or pre-packed chromatography columns, and from a set of Protein A chromatography resins on offer. Finally, the tool was used to estimate the set of trade-off solutions DRAFT

(Pareto front) for multi-objective problems for the entire mAb process train with the aim of minimising cost of goods (CoGs) in combination with capital expenditure, water usage, and process mass intensity (PMI). The results of these case studies identified scenarios in which certain process decisions were attractive. In addition, they demonstrated the efficiency of the optimisation tool in presenting users with good solutions to their decision problems through BioSolve Process.

BIOT 216

Neural network correlation of protein retention in chromatographic separations

Schuyler R. Reed, Abraham M. Lenhoff, [email protected]. Univ of Delaware, Newark, Delaware, United States

Prediction of protein retention times in chromatographic elution has previously been attempted using models based on protein and resin structural properties and biophysical models. QSAR models that also employ detailed 3D protein structure information but in the form of structural descriptors have also been used. In the interest of correlating retention without using full protein structural information a simpler approach was investigated here by training a neural network based on the amino acid sequence of a protein. In doing so the prediction process can be accelerated and information about elution behavior can be gained with less information about the protein structure. Retention data sets were obtained from the literature and amino acid sequences were taken from on-line databases. These sets were trained via Matlab’s neural network training tool both using full amino acid sequences and using just the amino acid composition. For each system, networks were explored with different architectures, including different layer sizes and numbers of internal neurons. Networks were found that could correlate retention times of proteins with R2 values exceeding 0.95, without evidence of overfitting, a potential risk because of the small size of the data set is a cause for concern. Investigation using larger data sets would allow more extensive evaluation of the approach.

BIOT 217

Propagation of uncertainty improves the quality of microbiome data analysis

Jacob R. Price1, [email protected], Amy D. Willis2, Thea L. Whitman1. (1) Department of Soil Science, University of Wisconsin - Madison, DRAFT

Madison, Wisconsin, United States (2) Department of Biostatistics, University of Washington, Seattle, Washington, United States

The adoption of sequencing-based technologies to probe biological and ecological questions has become exceedingly common due to significant reductions in cost, improvements in the quality of the data produced, and the ability to efficiently and quickly characterize many samples at a time, as well as the concurrent development of numerous bioinformatics tools, software packages, and reference databases. While great advances have been made, there are a number of shortcomings that remain as barriers to microbiome analysis being a precise, replicable, and reproducible science.

One of these shortcomings results from how raw sequencing data are processed to create counts of observed microbial abundances. Typical microbiome analysis pipelines approach read-to-taxon assignment as a discrete classification problem, wherein each read is explicitly assigned to a single member of an operational taxonomic unit (OTU) set. These approaches treat the data as though there is zero uncertainty in the assignments once they are made and, thus, they overestimate the certainty we have in any given read-to-OTU assignment. This leads to the second major shortcoming: a lack of appropriate computational and statistical methods to interpret the OTU tables. In some cases, difficulties and imprecision arise from using analytical approaches which were not developed with the nuances of sequencing data in mind, such as the data being non-census and compositional in nature.

In this work, we illustrate a method for incorporating uncertainty into typical microbiome analyses in order to overcome these challenges via a new and fully integrated bioinformatics and statistical pipeline. We apply this pipeline to two disparate datasets (fire-affected soil and photobioreactor) and demonstrate results obtained from a collection of statistics and hypothesis tests commonly used in microbiome research, including several alpha- and beta- diversity measures, differential abundance testing, and estimates of relative abundance for use in community profiling. Quantifying and accounting for uncertainties in the generated results reduces the risk of false positives due to understated variability, and improves the sensitivity of the tests, while making microbial community analysis more replicable in the process.

BIOT 218

TLmutation: predicting the effects of mutations using transfer learning DRAFT

Diwakar Shukla, [email protected]. Chemical & Biomolecular Engineering, University of Illinois at Urbana-Champaign, Urbana, Illinois, United States

A reoccurring challenge in bioinformatics is predicting the phenotypic consequence of amino acid variation in proteins. With the recent advancements in sequencing tech- niques, sufficient genomic data has become available to train models that predict the evolutionary statistical energies, but there is still inadequate experimental data to di- rectly predict functional effects. One approach to overcome this data scarcity is to apply transfer learning and train more models with available datasets. In this study, we propose a set of transfer learning algorithms we call TLmutation, which implements a supervised transfer learning algorithm that transfers knowledge from survival data of a protein to a particular function of that protein. This is followed by an unsupervised transfer learning algorithm that extends the knowledge to a homologous protein. We explore the application of our algorithms in three cases. First, we test the supervised transfer on 17 previously published deep mutagenesis datasets to complete and refine missing datapoints. We further investigate these datasets to identify which mutations build better predictors of variant functions. In the second case, we apply the algorithm to predict higher-order mutations solely from single point mutagenesis data. Finally, we perform the unsupervised transfer learning algorithm to predict mutational effects of homologous proteins from experimental datasets. These algorithms are generalized to transfer knowledge between Markov random field models. We show the benefit of our transfer learning algorithms to utilize informative deep mutational data and provide new insights into protein variant functions. As these algorithms are generalized to transfer knowledge between Markov random field models, we expect these algorithms to be applicable to other disciplines.

BIOT 219

Seed train intensification by implementing WAVE perfusion bioreactors

Linda Hoshan, [email protected], Ashley Bui, Sara Rodriguez, Jack Huang. Biologics Upstream Process Development, Merck, Kenilworth, New Jersey, United States

Both WAVE bioreactors and N-1 perfusion bioreactors have been widely applied as separate steps in seed train expansion for mammalian cell culture. Although N-1 perfusion bioreactor is an effective tool for high seed density process, there are needs and interests to streamline the passaging scheme DRAFT and replace this complicated and labor-intensive unit operation. Hence, an optimized, highly-efficient and simple seed train process is highly desirable to improve this platform. In this case, leveraging WAVE perfusion bioreactors at bench top scale (50L) for seed train expansion can reduce the number of required expansion steps and can potentially be used for direct inoculation of production bioreactors at the manufacturing scale. In this work, we first optimized the cell-specific perfusion rate and operation parameters for both the N-2 and N-1 WAVE perfusion bioreactors using one CHO cell line expressing a recombinant monoclonal antibody. The optimized perfusion rate schedule is able to maintain similar daily cell growth rate at high cell density and is successfully applied to three CHO cell lines, with similar cell culture performance. Exponential growth was observed throughout the seed expansion reaching > 15 × 106 vc/mL and > 80 × 106 vc/mL for the N-2 and N- 1 perfusion bioreactors, respectively. This process was successfully scaled up from a 10-L WAVE perfusion bag to a 50-L WAVE perfusion scale. Cell cycle analysis also confirms that cell cycle phase profiles are comparable throughout the WAVE perfusion bioreactor step. Importantly, inoculation of cells from different stages of WAVE perfusion bioreactor demonstrated to have similar cell culture performance and product quality attributes, comparing to the platform N-1 perfusion bioreactor process. Using three different CHO cell lines, we showed that WAVE perfusion seed cultures can improve process efficiency by optimizing the number of seed expansion bioreactor steps with simplified operations.

BIOT 220

Fabrication and characterization of a 3D-printed plate insert for dynamic co-culture applications

Sharif Rahman1, Rachael Coates1, Jordan Remont2, Conner Allison2, David Quiring1, Emmaline Miller1, Elizabeth Martin2, Adam T. Melvin1, [email protected]. (1) Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana, United States (2) Biological and Agricultural Engineering, Louisiana State University, Baton Rouge, Louisiana, United States

Cellular communication has been studied extensively by scientists, clinicians and engineers, as it is a critical component in development and disease. In order to communicate, cells emit extracellular signaling molecules (e.g., growth factors) which regulates if a cell will proliferate, migrate, or undergo apoptosis. The most common method is the Transwell assay which has several weaknesses: (i) an inability to directly visualize the cells, (ii) the invasion of the cells in the top chamber into the polycarbonate membrane, (iii) DRAFT very low cellular yields for post-experimental analysis. To address this, we developed a novel co-culture approach that utilizes a 3D-printed plate insert to dynamically co-culture two different cell types in the same 10 cm petri dish by creating two distinct culture chambers above and below an agarose hydrogel. The agarose hydrogel physically separates, yet chemically connects, the two culture regions and is coated with poly-D-lysine to allow for the adhesion, spreading, and growth of adherent cell lines in the top layer. The approach design is user-friendly, affordable, and allows for dynamic and tunable co- culture conditions while maintaining similar growth patterns as compared to off-chip controls in tissue culture plastic (TCP). Experimental and COMSOL simulations show the rapid diffusion of biomolecules and oxygen across the hydrogel reaching a steady state within ~12-18 hours. The plate insert approach allowed for the direct visualization of cell lines in both the top and bottom chambers facilitating the study of independent cell growth and morphology. The plate insert was used to study the interactions between two different breast cancer cell lines, triple negative (MDA-MB-231) and ER+ (MCF7) co-cultured with adipose-derived stem cells (ASCs). Recent studies have shown that ASCs play a critical role in promoting metastasis and drug resistance in breast cancer patients. Using our novel co-culture approach, we identified enhanced cellular growth and overexpression of EMT- associated genes in the MDA-MB-231 cell line when they were cultured in the presence of ASCs. We also observed increased drug resistance in the MCF7 cell line in the presence of ASCs, evident through increased proliferation. Our approach allowed for the independent harvesting of both cell lines in numbers sufficient for downstream PCR and Western Blot analysis to identify changes in the genotypic and phenotypic signature of cancer cells.

BIOT 221

High-throughput toxicological screening on 3D hiPSC cultures and derived progeny

Andre L. Rodrigues1,4, [email protected], Sneha Gopal4, Tiago G. Fernandes2, Joaquim Cabral1,2, Jonathan S. Dordick3. (1) Department of Bioengineering , Institute for Bioengineering and Biosciences - Insituto Superior Tecnico, University of Lisbon, Troy, New York, United States (2) Department of Bioengineering, Instituto Superior Técnico, Universidade de Lisboa, Lisbon, Portugal (3) Center for Biotechnology, Renselaer Polytechnic Inst, Troy, New York, United States (4) Chemical and Biological Engineering, Rensselaer Polytechnic Institute, Troy, New York, United States DRAFT

Pluripotent stem (PS) cells, both embryonic and induced pluripotent stem cells (iPSC), have the capacity for prolonged self-renewal and differentiation into numerous distinct cell lineages. Human PS cells (hPSC), therefore, may be a reliable source of donor cells for transplantation therapies and for drug screening. These applications require tight control over the in vitro culture microenvironment to obtain the desired hPSC differentiated derivatives with a high efficiency and functional quality. Microscale high-throughput culture systems are useful tools for mimicking and controlling hPSC’s in vivo microenvironment unveiling the complex interactions between cell-cell and cell-soluble/immobilized factors. In this study, we intend to establish a robust and reproducible 3D microarray platform for expansion and differentiation of hiPSC into neural progenitors, ultimately enabling the analytical quantification of the effects of growth factors and small molecules that impact cell viability and phenotypic characteristics. This platform also enables to perform high-throughput studies on the toxicity of small molecules towards hPSC-differentiated derivatives and evaluate different cellular sensitivities to pharmaceuticals.

BIOT 222

Trace metal variability in culture media can impair CHO culture performance by enhancing mechanisms of oxidative stress

Ryan Graham1, [email protected], Ashli Polanco2, Yongsuk Lee3, George Liang1, Adil Mohammad4, Seongkyu Yoon3. (1) Chemical Engineering, University of Massachusetts-Lowell, Cambridge, Massachusetts, United States (2) Chemical Engineering, University of Massachusetts Lowell, Lowell, Massachusetts, United States (3) University of Massachusetts Lowell, Lowell, Massachusetts, United States (4) Division of Product Qulaity Research, Food and Drug Administration, Silver Spring, Maryland, United States

The variability of trace and bulk metals in culture media may subsequently lead to unpredictable behavior in CHO cell culture. In the fields of human health and nutritional sciences, trace metal deficiencies and surpluses have shown to significantly impact mechanisms oxidative stress in mammalian cells. However, the relationship between cellular metal consumption, nutrient metabolism, and oxidative stress interplay in CHO cell culture remains unclear. In this study, the total intracellular and mitochondrial trace metal content of CHO cells supplied with various treatments of zinc and iron content in culture media and isolated mitochondria were assessed via inductively coupled plasma mass spectrometry (ICP-MS). Mechanisms of oxidative DRAFT stress—peroxidase activity, superoxide dismutase activity, and hydroxyl radical generation, as well as mitochondrial membrane potential were also surveyed. Current results indicate that intracellular metal acquisition is most active during lag phase, after which minimal differences in cellular metal consumption are noticed. Apoptotic cultures operating under either a trace metal deficiency or overabundance also exhibit higher specific production of both glutamate and ammonium comparatively. Peroxidase activity and total superoxide dismutase activity are also optimized by these trace metal treatments, demonstrating cellular capability to combat oxidative stress under varying metal conditions. Moreover, hydroxyl radical generation and membrane potential are significantly impacted by mitochondrial iron content as well while enhancing mAb specific productivity. Here, we offer new insight into the impact of trace metal variations in culture media on overall CHO culture performance.

BIOT 223

Oxygen mass transfer coefficient with sparged bubble size analysis in stir tank bioreactors

Yusuke Tomioka, Takao Ito, [email protected]. MilliporeSigma, Tokorozawa, Japan

Stir tank bioreactors are widely used in mammalian cell culture process for various scales of biomanufacturing. One of the successful bioreactor scale-up is achieved from the understanding of an operation parameter for efficient mixing and gas transfer etc. Volumetric oxygen mass transfer coefficient (kLa) is widely used to determine gas transfer capacity of bioreactor operation, that is refracted on the bioreactor design, such as impeller, vessel height and diameter, gas sparger shape and capacity. A lot of engineering approach has been discussed to determine kLa value, but the size of the bubble and distributions has not been verified with the current model. Bioreactors have multiple type of sparger to supply oxygen to fit various culture condition, the generated bubbles also have different diameters and size distribution, that affects the specific interfacial area. The correlation of engineering parameters from modeling are not well characterized yet. In this study, the determination approach of mass transfer coefficient with bubble diameter and size distribution analysis was evaluated by using MobiusR single use bioreactor. We verified the performance of the bioreactor mass transfer coefficient with general mechanical models and analyze the correlation between mass transfer coefficient and sparged bubble DRAFT specifications, also discuss about the unique sparger design and characterization.

BIOT 224

13C metabolic flux analysis quantifies changes in metabolism during proliferation, differentiation, and under low oxygen conditions

Eleanor H. Oates1, [email protected], Maciek R. Antoniewicz2. (1) Chemical and Biomolecular Engineering, University of Delaware, Newark, Delaware, United States (2) University of Delaware, Newark, Delaware, United States

Intracellular metabolic fluxes provide important information about the internal operation of mammalian cells when exposed to different external conditions. In this work, we have quantified detailed metabolic flux profiles for proliferating, differentiating, and hypoxic mammalian cells. To accomplish this goal, we performed comprehensive stable-isotopic tracer experiments coupled with mass spectrometry and 13C metabolic flux analysis (13C-MFA). Specifically, we performed four parallel labeling experiments under each metabolic condition using the following optimal tracers: [1 -13C]glucose + [4,5,6 - 13C]glucose, [2 -13C]glucose + [4,5,6 - 13C]glucose, [3 -13C]glucose + [4,5,6 - 13C]glucose, and [U -13C]glutamine; subsequently, we collected intracellular mass isotopomer distributions as well as external growth and nutrient uptake and secretion rates and analyzed the data. With this information, we then applied 13C-MFA using the software METEAN to generate detailed metabolic flux maps of intracellular metabolism. Through these comprehensive investigations we identified how metabolic pathways were metabolically rewired in response to different growth phases and oxygenation conditions, including dramatic changes in glycolysis, TCA cycle, oxidative pentose phosphate pathway, and lipogenesis. For example, comparing proliferating to differentiating cellular metabolism, we observed reversal of glutamine flux (i.e. from net consumption to net production), more than 2-fold upregulation of glycolysis flux, more than 4-fold reduction in lactate secretion, and significant upregulation of lipogenesis. Comparing normoxic to hypoxic metabolism, we observed more than 4-fold increase in lactate production and more than 4-fold reduction in citric acid cycle activity. These results provide important new insights into the regulation of mammalian metabolism.

BIOT 225 DRAFT

13C metabolic flux analysis of pre-implantation bovine embryos: Surmounting challenges due to small sample size and complex media

Daniel Lugar2, [email protected], Jaewook Chung1, Carol L. Keefer1, Ganesh Sriram2. (1) Animal and Avian Sciences, University of Maryland, College Park, Maryland, United States (2) Chemical and Biomolecular Engineering, University of Maryland, College Park, Maryland, United States

13C-assisted metabolic flux analysis (13C MFA) is a powerful methodology to quantify intracellular fluxes through primary and intermediary metabolism. 13C MFA relies on feeding a 13C-labled carbon source to a cell culture, measuring the ensuing metabolite labeling patterns, principally by mass spectrometry, and employing isotopomer modeling to back-calculate the intracellular fluxes from the labeling data. We report the application of 13C MFA to preimplantation bovine embryos toward investigating carbon source usage and the effects of medium composition on metabolism. This problem presents unique challenges owing to the small size of the culture (< 200 cells) and the necessity to assess the metabolism of individual embryos. Previous studies have shown that the composition of culture medium strongly affects the viability of in-vitro fertilized embryos. Furthermore, such media are complex and contain multiple carbon sources, thus potentially obfuscating the labeling patterns of the 13C-labeled carbon source. Additionally, because of the small sample size, there is only a limited set of metabolites whose labeling patterns can be accurately measured. We have designed novel labeling experiments to surmount these challenges and obtain a metabolic flux map for this complex system. We will present MFA results comparing flux identifiability resulting from multiple types of 13C-glucose and 13C-fructose, and explain how these results may be collectively interpreted to provide snapshots of the metabolism of bovine embryos cultured in media of various compositions. For example, parallel labeling experiments that use differently labeled glucose and fructose have revealed that the embryos utilize substantially more glucose than fructose. Further experiments have been designed to probe the relative utilization of glycolysis and the pentose phosphate pathway, as well as the contributions of different carbon sources to metabolites originating at the pyruvate node. The methods developed by us demonstrate the extension of the MFA framework to complex mammalian systems.

BIOT 226

Feeding TCA cycle intermediates improves lactate consumption and antibody production in CHO cell cultures DRAFT

Xiaolin Zhang1, [email protected], Rubin Jiang1, Henry Lin1, Sen Xu2. (1) Merck & Co., Inc., Kenilworth, New Jersey, United States (2) Biologics Development, Bristol-Myers Squibb Company, Pine Brook, New Jersey, United States

Media components play an important role in modulating cell metabolism and improving product titer in mammalian cell cultures. To sustain cell productivity, highly active oxidative metabolism is desired. Here we explored the effect of tricarboxylic acid (TCA) cycle intermediates supplementation on lactate metabolism and productivity in Chinese hamster ovary (CHO) fed-batch cultures. Direct addition of 5 mM alpha-ketoglutarate (α-KG), malic acid, or succinic acid in the basal medium did not have any significant impact on culture performance. On the other hand, feeding α-KG, malic acid, and succinic acid in the stationary phase, either as a single solution or as a mixture, significantly improved lactate consumption, reduced ammonium accumulation, and led to higher cell specific productivity and antibody titer (~35% increase for the best condition). Delivering those intermediates as an acidic solution for pH control eliminated CO2 sparging and accumulation. Feeding TCA cycle intermediates was also demonstrated to be superior to feeding lactic acid or pyruvic acid in titer improvement. Taken together, feeding TCA cycle intermediates was effective in improving lactate consumption and increasing product titer, which is likely due to enhanced oxidative TCA metabolism in an extended duration.

BIOT 227

Early activation and late resolution of the unfolded protein response in fed-batch culture associated with high productivity CHO cells

Dyllan Rives1, [email protected], Tara Richbourg2, Sarah W. Harcum3, Mark A. Blenner1. (1) Chemical and Biomolecular Engineering, Clemson University, Clemson, South Carolina, United States (2) Genetics and Biochemistry, Clemson University, Clemson, South Carolina, United States (3) Bioengineering, Clemson University, Clemson, South Carolina, United States

Chinese hamster ovary (CHO) cells are common protein production platforms due to efficient post-translational modification machinery and endoplasmic reticulum (ER) quality control; however, secretion levels needed for industrial cell lines can lead to an imbalance in ER homeostasis. Accumulation of improperly folded proteins is a particularly challenging bottleneck in cell line engineering and initiates the unfolded protein response (UPR). This research aims to demonstrate higher specific productivity results in unavoidable ER DRAFT stress by measuring the UPR in cell lines engineered for high-levels of recombinant protein production. In fed-batch culture, we achieve high and low specific productivity with two CHO cell lines, one producing immunoglobulin G (IgG) and one producing erythropoietin (EPO-Fc), respectively. Western blot and quantitative polymerase chain reaction (qPCR) were used to investigate the UPR time course, correlated to product titer. In comparison to the parental cell line controls, both CHO cell lines exhibited early activation of the IRE1 and PERK pathways. By Day 3 of fed-batch culture, spliced XBP-1 mRNA expression increased up to 64 fold, and both ATF4 and CHOP mRNA expression increased up to 8 fold. Transcriptomic analysis was conducted to monitor UPR targets and identify additional signaling transduction pathways activated during fed-batch culture. To determine if highly productive lines have been selected for the ability to cope better with ER stress, the UPR was monitored in fed-batch cultures of CHO cell pools with reduced IgG production via shRNA interference. Lastly, we explore the impacts of combining chemicals known to enhance productivity or reduce ER stress on cell productivity, titer and the UPR.

BIOT 228

Engineering in mammalian cells to streamline development of antibodies, antibody-like molecules and other proteins of therapeutic interest

Jennifer Maynard, [email protected]. Chemical Engineering, University of Texas at Austin, Austin, Texas, United States

Engineering of antibodies and related receptors is most commonly performed using phage or yeast display, but mammalian cells are used for production. To streamline engineering of these and other membrane bound proteins, we developed a mammalian screening platform which allows for generation of large libraries and selection of variants with desirable characteristics in the manufacturing host. We initially demonstrated that this platform could be used to engineer antibody Fab domains and that it could anticipate issues associated with glycosylation that would be missed by other screening platforms. However, this system appears particularly well-suited to engineering of complex proteins that require mammalian components for folding or function: the mammalian membrane to support folding, post- translational modifications, or accessory proteins. Accordingly, we have used this platform to identify human T cell receptors with sub-nanomolar affinities and increased TCR stability, Fc variants with pH dependent binding to human Fc receptors and heavily glycosylated viral fusion proteins. Finally, we have DRAFT used this platform to reduce tonic signaling issues in chimeric antigen receptors which may lead to increased persistence in vivo.

BIOT 229

Understanding the molecular mechanism governing the correlation between specific productivity and glycosylation

Sri Madabhushi, [email protected], Alexei Podtelezhnikov, Nicholas Murgolo, Sen Xu, Henry Lin. Merck & Co., Inc., Kenilworth, New Jersey, United States

The therapeutic efficacy of an antibody is influenced by the product quality attributes of the protein that include post-translational modifications such as glycosylation, charge variants, aggregates and fragments profile. One of the challenges in process development is to find a balance between increasing productivity while maintaining product quality of the protein. Across the various monoclonal antibody (mAb) programs at Merck & Co., Inc., we have observed that increasing the specific productivity (qP) of the mAb results in a decrease in the % galactosylation (%Gal) levels on the protein. Galactosylation levels may influence the mechanism of action (e.g., effector function) of the therapeutic mAbs. In order to understand the molecular mechanism governing the correlation between qP and %Gal, we cultured cells under conditions that exhibited a range of qP and N-glycan distribution and performed transcriptomics analysis. Results showed a correlation between specific productivity and expression levels of enzymes and solute transporters that are involved in the N-glycosylation pathway. Interestingly, based on the cell culture condition, different functional gene sets were either upregulated or downregulated which correlated with the productivity of the cells. These included pathways involved in cell cycle, cholesterol biosynthesis, apoptosis, and cell adhesion. Furthermore, we demonstrated that supplementation of potentially limiting cofactors increased the %Gal levels while maintaining qP for the high qP conditions. Together, the study suggests that the reduced galactosylation levels for higher productivity conditions is governed by the availability of enzymes, substrates and cofactors involved in the glycosylation pathways and not just the residence time of the protein in the cell. The insights from this study can be applied to modulate cell culture conditions to obtain the desired product quality thus producing mAb therapeutics with greater efficacy.

BIOT 230 DRAFT

More efficient site-specific integration in CHO cells enabled by a CRISPR-mediated genome editing reporter system

Nathaniel K. Hamaker, [email protected], Kelvin H. Lee. Department of Chemical & Biomolecular Engineering, University of Delaware, Newark, Delaware, United States

Chinese hamster ovary (CHO) cells continue to be the biomanufacturing platform of choice for the production of therapeutic glycoproteins, including monoclonal antibodies. The traditional cell line development workflow involves random integration of the transgene encoding the recombinant protein of interest. Random integration inherently leads to high levels of clonal heterogeneity and necessitates the tedious screening of multiple clones to isolate those with suitable performance characteristics. In contrast, site- specific integration (SSI) results in transgene knock-in events at user-defined genomic target sites, including prevalidated hot spots capable of supporting highly active and stable expression. SSI has therefore emerged as a powerful cell line engineering technique by enabling the accelerated development of production cell lines with more predictable performance and product quality. Despite these benefits, attempts to perform CRISPR-mediated SSI in CHO cells are typically met with remarkably low efficiencies, hindering efforts to isolate clones in which precise, rather than random, integration has occurred. Numerous methods have been proposed to remedy similar shortcomings observed in other mammalian cell types, yet few have been systematically evaluated in CHO. To overcome this obstacle, we developed the SSIGNAL (Site-specific Integration and GeNome ALteration) reporter system to facilitate rapid screening of CRISPR-mediated genome editing strategies. The system uses fluorescent signals to indicate genotype states, enabling data acquisition via standard flow cytometry instrumentation without the need for single-cell cloning or deep sequencing. The system was applied in CHO cells to test several current approaches for targeted integration to determine the most effective double-strand break repair strategy with the goal of improving efficiency. Ultimately, the development of robust SSI protocols may be implemented to improve long-term stability of transgenes for better cell culture process control.

BIOT 231

Systems design and synthetic construction of influenza virus for vaccine production DRAFT

Thu Phan1, [email protected], Christopher Stach1, Ryan Langlois2, Wei- Shou Hu1. (1) Chemical Engineering and Materials Science, University of Minnesota, St Paul, Minnesota, United States (2) Department of Microbiology and Immunology, University of Minnesota, Minneapolis, Minnesota, United States

Traditional productions of flu vaccine rely on the rescue of influenza viruses through transient transfection of multiple plasmids encoding the virus genome. The viruses generated are then used to infect embryonated eggs or cultured cells in the manufacturing process. Because of the high mutation frequency of the replication of the virus’s RNA genome, a limited number of infection cycles is used to amplify the viruses. The manufacturing process thus relies heavily on the virus's innate replication. Not only does this reliance make the vaccine less effective but it also leads to long response time to a pandemic situation. We took a synthetic approach by dismantling virus replication regulation and controlling the kinetics of the expression of various viral components instead. To characterize virus replication kinetics, we performed a transcriptomic analysis of the three types of viral RNA components in a single-cycle infection. We developed a novel informatics pipeline using total RNA-Sequencing that allows for the quantification of all viral RNA species: the temporary template cRNAs, genomic vRNAs and mRNAs. The kinetics of the expression of all RNA components for different viral genes served as a guide in our design of the expression system for virus production. We constructed an expression platform that allows for the generation of virus components in mammalian cells, which can be employed for virus production without the use of seed virus. This synthetic platform not only allows for more efficient, high-fidelity generation of flu virus but also provides opportunities to optimize the kinetics of virus gene expression for higher virus titer.

BIOT 232

Next generation bioprocess development

Gregg Nyberg, [email protected]. Process Research & Development, Merck & Co, Inc., Southborough, Massachusetts, United States

Rapidly delivering diverse biotherapeutics to patients requires efficiently developing innovative, robust manufacturing processes and technologies. Successful bioprocess development organizations must simultaneously optimize for speed to clinic, speed to launch, robust supply and low cost. Flexible manufacturing platforms that are robust across diverse modalities are foundational. State of the art platforms are highly productive and deployable in DRAFT reconfigurable facilities that facilitate rapid, cost effective capacity scaling. Such platforms increasingly rely upon intensified and continuous manufacturing technologies. This talk will review enabling technologies and strategies for evolving legacy manufacturing platforms.

BIOT 233

New strategy for identification of DNA zip codes via targeted rewiring of endogenous loci in mammalian cells

Meng Zhang1, [email protected], Pankaj Chaturvedi2, Andrew Belmont2, Huimin Zhao1. (1) Chemical and Biomolecular Engineering, University of Illinois at Urbana-Champaign, Urbana, Illinois, United States (2) Cell and Developmental Biology, University of Illinois at Urbana-Champaign, Urbana, Illinois, United States

Genetic materials inside the nucleus are not randomly distributed but spatially organized. For example, certain genomic loci are preferentially located at the nuclear periphery, while others are more likely to be found at nuclear interior. However, the underlying mechanisms that regulate such spatial distribution of the genome remain largely unknown. Originally discovered in Saccharomyces cerevisiae, “DNA zip codes” are short DNA sequences that encode spatial information targeting towards defined compartments in the nucleus. Similarly, in higher eukaryotes such as mammalian cells, DNA sequences around 5- 10kb in size have been identified for spatial targeting from ~200kb bacterial artificial chromosomes (BACs) harboring human genomic sequences. However, previous strategies have two major drawbacks: 1) the screening relies heavily on random integration of candidate DNA fragments, which suffers from no control over the copy number, no prior knowledge about the landing sites, and potential bias towards open chromatin regions; 2) the construction of BAC libraries by traditional recombineering method is labor- intensive and time-consuming. To overcome these bottlenecks, here we present a new strategy that couples the specificity of CRISPR/Cas9 system together with the large-cargo capability of serine recombinases to achieve targeted rewiring of preselected endogenous loci in mammalian cells. Moreover, direct cloning techniques are applied to quickly generate BAC libraries (20~100kb) with desired deletions in order to accelerate the library construction for downstream screening. Using this strategy, the targeting function of each modified BAC can be evaluated at the same, defined locus, which excludes undesired chromosomal context effects associated with random integration. Therefore, our strategy enables the dissection to be carried out in a more consistent, reliable and efficient manner. Collectively, we DRAFT anticipate this strategy to expand the current paradigm of large-scale genome engineering in mammalian systems, leading to novel knowledge about gene regulation on an epigenetic level.

BIOT 234

Analytical and process development challenges and opportunities in cell and gene therapy

Susan Abu Absi, [email protected]. Pharmaceutical Development & Technology, bluebird bio, Cambridge, Massachusetts, United States

The manufacturing processes and analytical methods for engineered T-cell therapies (e.g. CAR-T) and autologous ex vivo gene therapies are complex. Cells are first collected from patients via apheresis, modified ex vivo with lentiviral vector (LVV) or other gene addition/gene editing methods, expanded in culture, and frozen as a living drug product. With the rapid growth of the cell and gene therapy field, the demand for LVV has increased exponentially. Current LVV production methods using transient transfection are robust but will ultimately be constrained volumetrically and temporally. To overcome these constraints, alternative production processes may be required to meet future demand. Autologous drug products present a tremendous opportunity for understanding critical quality attributes (CQAs), by correlation of drug product attributes with clinical results and via advanced cell characterization. Although most commercial manufacturers utilize a centralized manufacturing approach, there is a lot of interest in advancing manufacturing beyond decentralized manufacturing to point-of-care or even bedside manufacturing. This presentation will provide an overview of the analytical and process development challenges for cell and gene therapy and opportunities for advancing the technologies to enable these therapies to reach all patients who can benefit from them.

BIOT 235

Harnessing expertise for creating smart automated image analysis: Training an AI for segmentation and classification of pleomorphic particles in nsTEM for gene therapy

Max Pihlström, [email protected]. Vironova AB, Stockholm, Sweden DRAFT

Negative stain transmission electron microscopy (nsTEM) can provide an informative and unique insight into the characterization of viral gene delivery platforms. Traditionally, the analysis of such nsTEM images required an experienced microscopist to identify and categorize the particles of interest.

A neural network pre-trained on thousands of nsTEM images on a broad range of expert segmented viral morphologies is used as an adaptive backbone for segmentation and classification of new nsTEM material, specifically geared towards the gene and cell therapy field.

Using only a handful of image examples, a user can guide the pre-trained network and instantly obtain automatic and deterministic segmentation and classification of the new images.

Leveraging a solid backbone model, the network can adapt to both heterogeneous and pleomorphic new particles. This versatility enables investigatory analyses in a fast and continuous workflow. This is particularly useful when analyzing for example lentiviral vectors, where a pre-set definition guideline for detection is challenging.

This morphological characterization of gene therapy delivery platforms can support and speed up the drug development process.

BIOT 236

Rapid downstream process development of an in-licensed non-platform protein

Caitlin Windsor, Timothy Tully, Benjamin Huffman, Nicole S. Schonenbach, [email protected], Joseph Silhavy. Pfizer, Inc., Chesterfield, Missouri, United States

A recent trend in drug development by established biotech companies has been the acquisition (in-licensing) of promising drug candidates. An established biotech company plays the role of providing the resources and expertise needed to bring this drug quickly to patients. But, based on the development experiences of the innovator company, process development of in-licensed drugs can be challenging. This presentation will discuss the challenges and learnings from the rapid process optimization of an acquired early stage candidate to ensure Clinical trial material would meet Pfizer’s high quality standards. Increased complexity DRAFT arose due to the stage of development when the molecule was acquired and the candidate design where platform techniques were not applicable. Development was enabled by Pfizer’s high throughput screening lab to characterize the existing process as well as test alternative chromatography resins using batch binding and RoboColumns. Improvements were verified against the current process during lab and pilot scale confirmation runs. Optimization of this 4-column process was performed in 4 weeks and process changes were implemented immediately during the ongoing GMP manufacturing campaign.

BIOT 237

Crystallization of peptides: Case of glycine homopeptides

Ian Rosbottom, [email protected], Mingxia Guo, Jerry Heng. Department of Chemical Engineering, Imperial College London, London, United Kingdom

Amongst the biologic drugs family, peptides, as polymers of amino acids, play important roles in human physiology and are hence an important class of therapeutics. The unique selling point of peptides is that they are selective, which translates into them being efficacious, safe and tolerable for administration to humans. The economic bottleneck of peptide manufacture, analogous to many biomolecules, is the challenges associated with purification as chromatographic methods are most widely used. However, advances in crystallisation optimisation methods has resulted in this technique being a genuine alternative to chromatography for peptide purification. Here we present the nucleation kinetics, derived from induction time data, of a homologous series of mono-, di- and tri-glycine. The induction time is found to increase with increasing chain length of the peptide. However, the increase in the induction time between di- and tri-glycine is found to be much lower than between mono- and di-glycine. This is found to be due to the preferential nucleation of a tri-glycine di-hydrate structure, rather than pure tri-glycine. A single crystal of this form was isolated, where the coordinates of the crystal structure are solved for the first time. Molecular modelling examines the energetically favourable molecular conformation and intermolecular solute/solute and solute/solvent interactions, to further uncover the molecular mechanisms behind the experimental observations. In particular, the crystallisation of the tri-glycine hydrate is examined, with respect to why of the three molecules only the tri-glycine presents a hydrated structure. DRAFT

This study uncovers detailed molecular chemistry, which underpins the nucleation of peptides of different chain lengths, providing insight into how the self-assembly mechanism changes from amino acid, to peptide to protein. This information is fundamental to maturing crystallisation as a viable peptide and protein purification technique.

BIOT 238

Development of AVIPureTM - AAV9: The first affinity resin for AAV production that can be cleaned with sodium hydroxide enables high resin reusability and decreases annual resin costs for high scale production

Thomas C. Scanlon1, [email protected], Warren Kett2, Karol Lacki3. (1) Ligand Discovery, Avitide Inc, Lebanon, New Hampshire, United States (2) CSO, Avitide, Inc, Lebanon, New Hampshire, United States (3) Technology Development, Avitide, Inc, Lebanon, New Hampshire, United States

Purification of recombinant AAV9 from cell lysates remains a technical challenge and an economic burden for commercial manufacturing of novel gene therapeutics. Currently available affinity resins for AAV9 purification are plagued by poor reduction in Host Cell Protein (HCP) and the inability to use NaOH solutions for Clean-In-Place (CIP). To support the rapidly-expanding gene therapy industry we described the development of AVIPureTM – AAV9, a novel affinity chromatography resin with ability to perform single-step purification of AAV9 from cell lysates with excellent yield and purity. Key distinguishing features are the ability to clean with NaOH and being based on an agarose bead for improved non-specific absorption. The large diameter agarose bead enables high flow rates and excellent resistance to fouling, important features when a typical purification cycle involves loading the column with up to 200 column volumes of host cell culture fluid. The order of magnitude improvement in resin lifetime promises to dramatically decrease resin costs for AAV programs requiring annual production of very large amounts of capsids. The affinity ligand is a novel peptide with exquisite specificity for rAAV9, binding the virus with picomolar affinity. A long, flexible spacer molecule attaches the ligand to the resin, enabling efficient capture of virus. Elution of several variants of AAV9 capsids elution was achieved in the range pH 4.5 – 3. Overall, AVIPureTM – AAV9 resin provides excellent yield, purity and throughput for single-step purification of rAAV9 and enables the use of sodium hydroxide as a cleaning agent for multiple cycles. DRAFT

BIOT 239

Understanding scalability of mustang® Q Devices in lentiviral vector purification using DOE approach

Aydin Kavara1, [email protected], Kurt Boenning2, Julio Huato2, Adam Hejmowski2, Ana S. Moreira2, Rene Gantier2, Todd Sanderson2, manuel carrondo3, Mark Schofield4. (1) Bioprocess R&D, Pall Corporation, Westborough, Massachusetts, United States (2) Pall Biotech, Westborough, Massachusetts, United States (3) Animal cell technology, IBET, Oeiras, Portugal (4) Biopharm R&D, Pall, North Grafton, Massachusetts, United States

Lentiviral vectors (LVVs) are increasingly used in treatment of acquired and inherited genetic diseases. Production and purification of the LVVs remains challenging due to low titers from upstream processes and the low stability of the virus. A typical LVV purification strategy might include an AEX step positioned after clarification and before tangential flow filtration and final filtration. Thus, anion-exchange (AEX) chromatography has a key role in purification of LVVs. As the AEX step is positioned directly after clarification, relatively large volumes have to be processed. This makes membrane chromatography an ideal choice as it can be operated at and above 10 membrane volumes per minute. Despite widespread application of AEX membrane chromatography, there has been little published effort to understand the scalability of membrane devices when transitioning from filter plate experiments to small and then larger capsules. Here we implemented a Design of Experiment (DOE) approach to optimize the bind and elute conditions of LVVs in Mustang Q filter plate experiments and then transferred these conditions to 0.18 mL and 1 mL Mustang XT Acrodisc®Units. Gradient and step elution experiments were carried out to understand the role of salt and additives in desorption of LVV from the membrane surface. In addition, we compared the performance of Mustang Q devices to other AEX sorbents. Under optimized conditions, the observed capacity of Mustang Q capsules was 2.4 x 1011 to 1.5 x 1012transfecting particles/mL of membrane. In summary, a scalable LVV purification protocol was developed for facile transition from Mustang Q filter plate to larger XT capsules.

BIOT 240

PID control: Forgotten process parameter DRAFT

Sarah W. Harcum1, [email protected], Kathryn Elliott1, Tom Caldwell1, Bradley Skelton1, Benjamin Synoground2, Cameron Schnabel1, Stephanie Klaubert3, Madison Williamson1. (1) Bioengineering, Clemson University, Clemson, South Carolina, United States (2) Department of Bioengineering, Clemson University, Clemson, South Carolina, United States (3) Clemson University, Clemson, South Carolina, United States

In order to gain a better understanding of Chinese hamster ovary (CHO) cell genome instability, the ambr® 250 bioreactor platform was adopted as the highly-controlled environment for this project. Development of the baseline and stress conditions for the genome instability studies has highlighted the need for control of critical process parameters, including the Proportional, Integral, and Derivative (PID) control loops. Process parameters or factors that are considered scale-independent, dissolved oxygen (DO) and pH, were determined to be sensitive to the PID values for the ambr® 250 process. Unlike larger bioreactor control loops that can use cascade settings to control DO utilizing agitation, air, and oxygen concurrently, the ambr® 250 requires individual levels of control that are used sequentially to change a particular process parameter. For example, the Level 1 for DO control could have PID values for the gas sparge (Air) 0.5 mL/min to 2.5 mL/min (P = 0.1, I = 500, D = 0). Once the DO drops below the setpoint, Level 2 would be called. The Level 2 default increases the Stir rate from 300 rpm to 600 rpm (P = 3.5, I = 250, D = 0). The consequence of having multiple levels is that each level needs to have the PID values tuned individually. For mammalian cell cultures Sartorius provides with the ambr® 250 platform with three default DO Levels and a set of upper and lower pH PID values. However, even with minor adjustments these PID values resulted in high variability in DO, were too slow to respond to antifoam additions at high-cell densities, and allowed the pH to drift away from the setpoint that resulted in lactate accumulation. After consulting with industrial ambr® 250 users and Sartorius field specialists, and tuning the PID for each level, DO and pH control loops PID values were developed that minimized uncontrolled oscillation and over- and under-shooting, as well. These improved PID control values can support CHO cell growth, reproducibly to over 47 million cells/mL in a fed-batch process. This work highlights that PID control loop values can profoundly affect culture outcomes, yet are often neglected and rarely mentioned in publications.

BIOT 241

Control of glycosylation and titer in fed-batch monoclonal antibody production DRAFT

Yu Luo1, [email protected], Devesh Radhakrishnan2, Evan Wells3, Deepro Banerjee3, Liqing Song3, Varghese Kurian1, Anne S. Robinson3, Babatunde Ogunnaike1. (1) Chemical & Biomolecular Engineering, University of Delaware, Newark, Delaware, United States (2) BioMarin Pharmaceutical Inc., Novato, California, United States (3) Chemical Engineering, Carnegie-Mellon University, Pittsburgh, Pennsylvania, United States

Therapeutic monoclonal antibodies (mAbs) are currently produced commercially in fed-batch cultures using media recipes and protocols based primarily on heuristics. The main production objectives are maximizing mAb titer (productivity requirement) and achieving desired glycan distributions (quality attribute) that result from glycosylation—a post-translational process that attaches sugar molecules (glycans) to antibodies. The glycan profile is a critical quality attribute of mAb manufacturing because it can either promote or inhibit a mAb drug's therapeutic effects. Yet, controlling glycosylation remains challenging because it is a complex, non-template driven process that takes place within the cell; the manipulated variables are far fewer than the controlled variables; the process is difficult to model due to its complexity; on- line sensors do not exist for characterizing glycans frequently enough for effective on-line control. Nonetheless, recent advances in both process modeling and glycan assays have improved the possibility of implementing model-based control of glycosylation in place of the current heuristics-based approach.

We present an overall model-based strategy for meeting production requirements in the manufacture of mAbs. The proposed multivariable cascade control system consists of two nested control loops. The outer loop— responsible for controlling productivity and product quality attributes—uses a model predictive controller (MPC) to determine the setpoints for the inner loop. The inner loop employs multiple PI controllers to maintain nutrients and operating parameters at desired levels in the cell culture. Due to infrequent glycan assays, the outer loop incorporates a state observer that uses sensor measurements and model predictions to reconstruct the complete glycan profile, which the MPC uses, along with the process model, to determine the desired control policies.

As a first step toward implementing on-line control of industrial fed-batch mAb manufacturing processes, we tested the complete control system in a MATLAB simulation and investigated a variety of industrially relevant process conditions under several operating sensor configurations (frequencies, qualities, and delays). Subsequently, we integrated the control system with DRAFT the bioreactor through Open Platform Communications. The results offer insight into the optimal sensing strategy as well as the usefulness and limitations of the proposed control system.

BIOT 242

Binding of CHO host-cell proteins to monoclonal antibodies

Young Hoon Oh1, [email protected], Matthew Becker1, Kerri Mendola1, Leila Choe1, Kelvin H. Lee1, Abraham M. Lenhoff2, Yinges Yigzaw3, David J. Roush4. (1) Chemical & Biomolecular Engineering, University of Delaware, Newark, Delaware, United States (2) Univ of Delaware, Newark, Delaware, United States (3) Purification, Genentech, South San Francisco, California, United States (4) Process Development and Engineering, Merck, Sharp and Dohme, Colts Neck, New Jersey, United States

In the manufacture of biopharmaceuticals, individual host-cell proteins (HCP) may elude clearance and be present in measurable quantities in the final drug product, with the potential impact on the product quality and risk to patient safety. One cause of HCP co-purification is strong association with the product, such as a monoclonal antibody (mAb). We will present proteomic data, obtained by LC-MS, tracking HCPs in downstream processes for six different mAbs from two companies, showing the similarities and differences among the clearance patterns for different processes. In order to explore possible binding mechanisms and reasons for co-purification in these systems, a smaller group of CHO HCPs were expressed in prokaryotic hosts and purified. The dissociation constants, KD, for binding of these HCPs to the six mAbs studied above were measured; KD values in the tens of µM range were obtained for several HCP/Mab systems. The implications of the results for HCP clearance will be discussed.

BIOT 243

Enabling acceleration: A case study in leveraging platform data sets to establish a platform raw material control strategy

Erin O. Wilson1, [email protected], Lee Bink2, Robert W. Deitcher1, Diana B. Ritz3, Mark Lankford4. (1) Manufacturing Science & Technology, Glaxo Smith Kline, Collegeville, Pennsylvania, United States (2) UE0551, GlaxoSmithKline, King of Prussia, Pennsylvania, United States (3) UE0548, GlaxoSmithKline, King Of Prussia, Pennsylvania, United States (4) UE0548, GlaxoSmithKline, King of Prussia, Pennsylvania, United States DRAFT

The implementation of platform manufacturing processes enables new products to be accelerated from discovery to the clinic, and ultimately to the market. GSK has defined a platform drug substance manufacturing process for new monoclonal antibody (mAb) products that enables rapid transfer from the lab to clinical manufacturing, and subsequent transfer to commercial manufacturing with minimal changes in scale and equipment. Quality by Design enables further acceleration when knowledge gained from implementation of the platform process is used to develop platform process control strategies. The development of a platform raw material risk assessment is presented as a case study in leveraging larger data sets available from the implementation of the platform processes to address the challenges of robust mAb control strategy development in an accelerated environment. The benefits include: 1) significantly reducing the resources required to complete a key component of platform mAb control strategy, 2) identification and mitigation of raw material-related risks to the platform process and 3) prioritization of raw materials in support of raw material supply chain workstreams.

BIOT 244

Harnessing the power of a data-driven revolution to enable single-use and continuous processing technologies to meet the market need

Rene Reinbigler1, Michael Phillips2, [email protected]. (1) Merck KGaA, Molsheim, France (2) Technical Director, Next Generation Processing R&D, MilliporeSigma, Burlington, Massachusetts, United States

There are several challenges associated with next generation processing for bio pharmaceutical production. The ambitious goals set by the customers and the overall market are being addressed through the combination of technology, bioprocess intensification and data driven insights. A comprehensive strategy is covering new modular hardware approach, standardized communication interface, smart data management and additional self-service cloud based on big data and machine learning. The combination of those capabilities drastically increases flexibility and automation enabling the biopharmaceutical Facility of the Future. On the intersection of technology and business there are new opportunities leveraging collaboration power of integrated ecosystem of vendors, service providers and customers. We are performing a set of pilot implementations where many of our assumptions and technologies are being tested and confirmed. In this session we will present the findings and concepts covering: - Lab and Manufacturing sites connectivity – challenges and opportunities for DRAFT

OPCUA based, standardized connectivity and control integrating unit operations, consumables, operators and broader ecosystem - Data management leveraging GMP data lake – what is the real value of advanced data management including ontology and taxonomy models covering end to end process definition? - Advanced Analytics leveraging Machine Learning and AI – how ML and AI models can be used in GMP environment?

BIOT 245

Chemoinformatic-guided engineering of polyketide synthases

Amin Zargar1, [email protected], Ravi Lal1, Luis Valencia1, Jessica Wang1, Tyler Backman1, Pablo Crux-Morales1, Ankita Kothari1, Miranda Werts1, Andrew Wong1, Constance Bailey1, Arthur Loubat1, Yuzhong Liu1, Yan Chen1, Samantha Chang1, Veronica Benites1, Amanda Hernandez1, Jesus Barajas1, Mitchell Thompson1, carolina barcelos2, Rasha Anayah1, Hector Garcia Martin1, Aindrila Mukhopadhyay2, Chris Petzold1, Edward Baidoo1, Leonard Katz1, Jay D. Keasling3. (1) UC Berkeley, Berkeley, California, United States (2) Biological Systems and Engineering , Lawrence Berkeley National Laboratory, Berkeley, California, United States (3) Depts Chem Biomole Eng Bioeng, Univ of California Berkeley, Emeryville, California, United States

Polyketide synthase (PKS) engineering is an attractive method to generate new molecules such as commodity, fine and specialty chemicals. A central challenge in PKS design is replacing a partially reductive module with a fully reductive module through a reductive loop exchange, thereby generating a saturated β-carbon. In this work, we sought to establish an engineering strategy for reductive loop exchanges based on chemoinformatics, a field traditionally used in drug discovery. We first introduced a set of donor reductive loops of diverse genetic origin and chemical substrate structures into the first extension module of the lipomycin PKS (LipPKS1). These results demonstrated that chemical similarity between the substrate of the donor loops and recipient LipPKS1 correlated with product titers. Consequently, we identified donor loops with substrates chemically similar to LipPKS1 for further reductive loop exchanges, and we observed a statistically significant correlation with production. Reductive loops with the highest chemical similarity resulted in production of branched, short-chain fatty acids reaching a titer of 165 mg/L in Streptomyces albus J1074. Collectively, our work formulizes a new chemoinformatic paradigm for de novo PKS biosynthesis which may accelerate the production of valuable bioproducts. DRAFT

BIOT 246

Kinetic rationale for functional redundancy in fatty acid biosynthesis

Alex Ruppe, Kathryn Mains, Jerome M. Fox, [email protected]. Chemical and Biological Engineering, University of Colorado, Boulder, Boulder, Colorado, United States

Cells build fatty acids with biocatalytic assembly lines in which a subset of structurally unique components (e.g., enzymes or catalytic centers) often exhibit overlapping activities. Although the reactions necessary for the construction of different varieties of fatty acids are well resolved, the biochemical advantage of multiple enzymes that carry out the same reaction remains poorly understood. Our group has developed a detailed kinetic model of the fatty acid synthase (FAS) of Escherichia coli and paired that model with a reconstituted in vitro system to examine the biochemical repercussions of functional redundancy in fatty acid synthesis. Analyses of different FAS compositions indicate that systems with partially redundant dehydratases and/or ketoacyl synthases enable broader control over competing biochemical objectives—the fraction of unsaturated fatty acids, the total concentration of fatty acids, and the average length of fatty acids—than FASs with individual multi-functional enzymes. Maximal production of unsaturated fatty acids, for example, requires a dehydratase that is not essential for their synthesis. Broadly, the results of this study supply a kinetic rationale for the retention of redundant FAS enzymes in microbial genomes and provide a framework for exploiting functional redundancy in the design of robust oleochemical producing microbes.

BIOT 247

One-step microbial synthesis of chondroitin sulfates using engineered Escherichia coli

Abinaya Badri1, [email protected], Robert J. Linhardt2,1, Mattheos Koffas3,1. (1) Chemical Engineering, Rensselaer Polytechnic Institute, Troy, New York, United States (2) Chemistry, Rensselaer Polytechnic Institute, Albany, New York, United States (3) Center for Biotechnology and Interdisciplinar, Rensselaer Polytechnic Institute, Troy, New York, United States

Sulfated glycosaminoglycans (GAGs) like heparan sulfates and chondroitin sulfates constitute an essential component of the extracellular matrix in higher eukaryotes. Hence, they serve as important pharmaceuticals. Of these, DRAFT chondroitin sulfates (CS) are majorly used in the treatment of osteoarthritis. CS is also used to improve liver function, lower blood sugar and inhibit tumor metastasis. It is also used in drug delivery, food thickeners and preservatives. CS is commercially extracted from porcine, bovine, chicken, fish and shark tissues. There is a growing interest in animal-free CS production world-wide. Among such efforts, we have engineered for the first time, single strains of Escherichia coli that have the capability to produce CS in a single scalable fermentation.

CS production involves biosynthesis and sulfation of chondroitin polymer. We previously engineered E. coli K4 to produce chondroitin by deleting the fructosyltransferase gene. However, expressing chondroitin sulfotransferase in this strain does not enable CS synthesis. In this work, we show that in addition to sulfotransferase expression, accumulation of 3'- phosphoadenosine-5'-phosphosulfate (PAPS) is crucial for CS synthesis. We employ different genetic constructs for accumulation of PAPS and expression of sulfotransferase to produce CS in E. coli. We also study the effect of different PAPS levels and polysaccharide transport efficiencies on sulfation and yield of CS from our engineered strains. This is the first report of animal- free one-pot, one-strain microbial CS synthesis. We propose that our animal- free product will also enable pharmaceutical and cell-culture applications, thus opening avenues to understand the importance of specific CS types in biological processes.

BIOT 248

Genetically encoded enzyme inhibitors for post-translational dynamic control of yeast metabolism

Christopher Gonzalez1, [email protected], Cesar Carrasco-Lopez2, Jose L. Avalos3. (1) Chemical and Biological Engineering, Princeton University, Princeton, New Jersey, United States (2) Chemical and Biological Engineering , Princeton University, Lawrence Township, New Jersey, United States

The baker’s yeast Saccharomyces cerevisiae is one of the most widely-used microorganisms for industrial chemical production. However, economically- viable production of commodity chemicals with this organism is made challenging by the yeast’s overwhelming bias to produce ethanol. Because of the crabtree effect, fermentation is necessary for the yeast’s growth on glucose, making genetic knockouts of the ethanol pathway unfeasible for DRAFT growth on industrially-relevant carbon sources. With dynamic control of the ethanol pathway, production of competing chemicals of interest could be improved without sacrificing biomass accumulation, by splitting the fermentation into a growth stage and a production stage. Most efforts to induce the switch from growth to production have focused on transcriptional regulation of the ethanol pathway; while transcriptional circuits can allow for high-fold changes in gene expression, they do not affect proteins already accumulated within cells, creating an inherent lag in circuit response. In this talk, I present a novel genetically-encoded posttranslational inhibitor of a key enzyme in ethanol production, which we have biochemically and structurally characterized. The inhibitor exhibits low-nanomolar affinity for its target, with fast association on the order of 104 M-1s-1, and slow dissociation on the order of 10-5 s-1. The inhibitor shows potent, noncompetitive inhibition of its target, with one-to-one binding. Constitutive expression of the inhibitor in vivo shows up to a 10-fold reduction in the growth rate. Finally, inducible expression of the inhibitor shows improved production of chemicals of interest compared to purely transcriptional control of the ethanol pathway, demonstrating the advantage of posttranslational regulation of metabolic pathways: with faster dynamic control, essential pathways can be more rapidly shut down, leaving a greater pool of substrate to be converted to products of interest.

BIOT 249

13C Metabolic flux analysis and kinetic modeling of Clostridium thermocellum grown on cellobiose

Charles Foster1, [email protected], Veda S. Boorla1, Tyler Jacobson5, Ratul Chowdhury1, Saratram Gopalakrishnan1, Satyakam Dash2, Daniel G. Olson3, Daniel Amador-Noguez5, Lee R. Lynd3, Costas Maranas4. (1) Chemical Engineering, Pennsylvania State University, University Park, Pennsylvania, United States (2) Penn State University, University Park, Pennsylvania, United States (3) Thayer School of Engineering, Dartmouth College, Hanover, New Hampshire, United States (4) Penn State, University Park, Pennsylvania, United States (5) Bacteriology, Univeristy of Wisconsin, Madison, Wisconsin, United States

Clostridium thermocellum is a promising candidate for consolidated bioprocessing due to its ability to consume and ferment cellobiose to ethanol. Despite significant efforts, achieved yields and titers fall below industrially relevant targets. This implies that enzymatic, regulatory, and possibly thermodynamic bottlenecks that throttle back metabolic flow still exist. This DRAFT work aims to reveal some of these factors by elucidating internal metabolic fluxes in wild-type C. thermocellum grown on cellobiose via 13C-metabolic flux analysis (13C-MFA) using three distinct datasets. Subsequently, using the resultant flux distribution in conjunction with batch fermentation process yield data for various mutant strains, we constructed an updated kinetic model of C. thermocellum core metabolism (i.e. k-ctherm101). During kinetic parameterization, we introduced a systematic method for identifying a minimal set of putative regulatory mechanisms that helps to achieving good model agreement with experimental data using the K-FIT kinetic parameterization algorithm. Flux elucidation results confirmed that that both pyruvate, phosphate dikinase and the malate shunt are pyruvate contributors. 13C labeling data interpreted by 13C-MFA alluded to serine generation via the mercaptopyruvate pathway. The pathway was shown to be thermodynamically feasible and candidate gene Clo1313_0176 was identified for encoding this activity in C. thermocellum. Statistical analysis on k-ctherm101 confirmed that the C. thermocellum-specific regulatory mechanisms on lactate dehydrogenase (activated by fructose 1,6-bisphosphate), alcohol dehydrogenase (inhibited by ethanol and NAD+), and pyruvate, phosphate dikinase (activated by ammonia) were essential to accurate training data recapitulation.

BIOT 251

Redirecting metabolic flux via combinatorial multiplex CRISPRi- mediated repression for isopentenol production in E. coli

Taek Soon Lee1,2, [email protected]. (1) Joint BioEnergy Institute, Emeryville, California, United States (2) Biological Systems and Engineering, Lawrence Berkeley National Laboratory, Berkeley, California, United States

CRISPR interference (CRISPRi) via target guide RNA (gRNA) arrays and a deactivated Cas9 (dCas9) protein has been shown to simultaneously repress expression of multiple genomic DNA loci. By knocking down endogenous genes in competing pathways, CRISPRi technology can be utilized to re-direct metabolic flux toward target metabolite. In this study, we constructed a CRISPRi-mediated multiplex repression system to silence transcription of several endogenous genes in order to increase precursor availability in a heterologous isopentenol biosynthesis pathway. To identify genomic knockdown targets in competing pathways, we first designed a single-gRNA library with 15 individual targets, where 3 gRNA cassettes targeting gene asnA, prpE, gldA increased isopentenol titer by 18-24%. We then combined the 3 single-gRNA cassettes into two- or three- gRNA array and DRAFT observed up to 98% enhancement in production by fine-tuning the repression level through titrating dCas9 expression. Our strategy shows that multiplex combinatorial knockdown of competing genes using CRISPRi can increase production of target metabolite, while the repression level needs to be adjusted to balance the metabolic network and achieve the maximum titer improvement.

BIOT 252

Engineering TATA box to create a promoter library to balance heterologous pathways in Y. lipolytica

Vijaydev ganesan, [email protected], William Burnette, Patrick Dayton, Mark A. Blenner. Chemical Biomolecular Engineering, Clemson University, Clemson, South Carolina, United States

Metabolic engineering requires precise control of gene expression to improve flux through target molecules. Oleaginous yeast Yarrowia lipolytica has been emerging as an industrial workhorse due to its high flux lipids biosynthetic pathways; however, fewer tools to control gene expression are available compared to model yeast S. cerevisiae. To date, most approaches to control gene expression use endogenous promoters or hybrid promoters. Both types of promoters have difficult to anticipate responses to different substrates and genetic contexts. Here, we develop a tunable expression system in Y. lipolytica by mutating the TATA Box region in different promoters. The 8bp TATA Box region was randomized through degenerate bases and promoter libraries created to express hrGFP. We used FACS to isolate TATA Box mutants resulting in different expression levels and validated individual mutants through RT-qPCR. The correlation between fluorescence intensity, relative transcriptional level and relative translational level is evaluated over time in the context of five commonly used yeast promoters TEF, TEF-Intron, UAS1B8-TEF, EXP and POX3. To demonstrate the utility of finely tuned expression systems, we used the expression system to balance a multienzyme heterologous pathway for naringenin synthesis. This work generates a simple and reliable system for tuning the expression system in Y. lipolytica by modifying the TATA Box to balance a heterologous pathway

BIOT 253

Controlling pyocyanin synthesis in synthetic microbial consortia by regulating culture composition DRAFT

Kristina Stephens, [email protected], Kayla Chun, Sally Wang, Eric VanArsdale, William E. Bentley. Fischell Dept of Bioengineering, University of Maryland, College Park, Maryland, United States

Interest in use of microbial consortia or co-cultures by metabolic engineers has expanded due to limitations of single populations. These limitations include high metabolic burden on cells and difficulty optimizing cells to carry out complex pathways. Co-cultures allow for division of labor and options for plug and play. Partitioning a product pathway by splitting it among different populations also provides opportunities to control synthesis of each section of the pathway by controlling culture composition. Importantly, this provides an additional layer of control (as opposed to transcriptional or translational regulation) in increasingly complex synthetic biology systems. However, these systems could benefit from methods for engineering cell-cell communication between strains in the co-culture, enabling the cells to work together autonomously. Quorum sensing (QS) systems are commonly used to engineer cell-cell communication, but there are limited numbers of orthogonal QS systems. Here, we show that E. coli can be engineered to produce pyocyanin as an additional molecular cue for engineering cell-cell communication. Previously, it has been shown that pyocyanin causes activation of the sox promoter in E. coli allowing for pyocyanin-mediated induction of a gene of interest. Interestingly, because the pathway for pyocyanin synthesis requires several genes, there are opportunities to split the pathway between different strains to allow for more complex behavior within the synthetic consortia. That is, we designed an E. coli co-culture for the synthesis of pyocyanin and then demonstrate that pyocyanin synthesis and cell response to pyocyanin is dependent on culture composition. Population A expresses P. aeruginosa genes phzA1-G1 to synthesize an intermediate product phenazine-1-carboxylic acid, while Population B expresses genes phzMS to synthesize pyocyanin. We show that by varying the ratio of Population A to Population B, we vary the amount of pyocyanin produced. Then, we show that by adding a third population, Population C, which contains a pyocyanin responsive promoter, we control activity in Population C based on the combined activity (and composition) of Populations A and B. We further explore methods for manipulating the dynamics of the co- culture system. We believe our strategy could be useful for many applications including for coordinating co-cultures that are synthesizing a product and for engineering cooperativity in synthetic biology systems.

BIOT 254 DRAFT

Towards microbiome engineering using highly selective bacteriocin lysostaphin and phage endolysin PlyPH

Amala Bhagwat1, [email protected], Cynthia H. Collins2, Jonathan S. Dordick3. (1) Chemical and Biological Engineering, Rensselaer Polytechnic Institute, Troy, New York, United States (2) Rensselaer Polytechnic Institute, Albany, New York, United States (3) Center for Biotechnology, Renselaer Polytechnic Inst, Troy, New York, United States

Development of novel selective antimicrobials is gaining momentum due to increased bacterial drug resistance, microbiome dysbiosis caused by broad- spectrum antibiotics and the need to understand and precisely engineer human and environmental microbiomes. Cell lytic enzymes, including bacteriocins, autolysins and bacteriophage endolysins represent a unique, highly specific catalytic repertoire hydrolyzing bacterial peptidoglycan. Owing to their exquisite binding and/or catalytic specificity and lower probability of gaining microbial resistance, they are not only emerging as promising alternative to antibiotics for therapy and infrastructural resilience applications but also potentially for microbiome engineering and remodeling via selective pathogen elimination. We have assessed the ability of two different lytic enzymes, bacteriocin lysostaphin (Lst) and endolysin PlyPH targeting skin pathogens Staphylococcus aureus and Bacillus cereus, respectively, to control a community (mixture or co-culture) of S. aureus and B. cereus with a skin commensal Micrococcus luteus. Lst (100 mg/mL) was highly effective and selective, causing ~ 3-log reduction in S. aureus, however, 100 mg/mL PlyPH selective but less active, resulting in ~1.5-log reduction in B. cereus in an actively growing co-culture. Lst and PlyPH were limited by proteases secreted during growth, as suggested by their increased activity in culture supernatants in the presence of a protease inhibitor cocktail, highlighting effect of bacterial metabolism on enzyme activity. Since Lst and PlyPH activity is governed by cell wall binding, analysis of their binding is critical for elucidating activity in nutrient-rich environments mimicking natural microbiomes. Lst and PlyPH binding was measured in three increasingly complex media by evaluating binding isotherms and real-time binding kinetics of their isolated fluorescently-tagged binding domains using both flow cytometry and Surface Plasmon Resonance. Slower binding kinetics was observed with increased medium complexity, causing limited killing in nutrient rich media. In conclusion, we demonstrate the utility of lytic enzymes and fundamental considerations to enhance their robustness for microbiome engineering.

BIOT 255 DRAFT

Bacterial microcompartments as tunable nanobioreactors: Engineering cargo and pathway encapsulation

Taylor Nichols1, [email protected], Nolan Kennedy2, Danielle Ercek1. (1) Chemical and Biological Engineering, Northwestern University, Chicago, Illinois, United States (2) Northwestern University, Evanston, Illinois, United States

While metabolic engineering can enable the sustainable bioproduction of new materials, efforts are often impeded by pathway bottlenecks. To mitigate the effects of toxic or reactive intermediates and resource competition resulting from heterologous pathway incorporation, bacterial microcompartments (MCPs) have recently been considered for engineered compartmentalization in bacterial host organisms. MCPs are protein shells that form natively in certain bacteria to selectively encapsulate specific metabolic processes, functioning to sequester and colocalize pathway components to enhance flux. The use of MCPs as enclosed scaffolds for metabolic engineering applications requires the control of heterologous cargo encapsulation, with pathway selection designed to efficiently utilize the native benefits of MCP systems.

We use the 1,2-propanediol utilization (Pdu) MCP from Salmonella enterica LT2 as a scaffold for the encapsulation of heterologous cargo and pathway components. With this talk, I will detail how we developed a genomic integration platform to enable and control cargo encapsulation in Pdu MCPs. Specifically, we integrated different fluorescent reporter constructs into the operon that encodes the native Pdu MCP and subsequently assessed cargo expression and encapsulation levels. We were able to achieve encapsulation of multiple cargo, with different relative encapsulation levels observed. Importantly, the trends were reproducible and encapsulation efficiency was primarily determined by the targeting sequence and integration locus. Results from this platform have subsequently informed efforts to encapsulate an industrially-relevant non-native metabolic pathway within Pdu MCPs. This work established a genomic integration platform for heterologous cargo encapsulation in Pdu MCPs, further enabling their development as tunable nanobioreactors for the encapsulation of heterologous pathways.

BIOT 256

Commercial scale production of low carbon fuels and chemicals from waste gases DRAFT

Michael Koepke, [email protected]. LanzaTech, Skokie, Illinois, United States

Rapid population growth and climate change are posing some of the most urgent challenges to mankind and have intensified the need for low-cost manufacturing of fuels, chemical-building blocks, materials and food from sustainable resources. Gas fermentation using autotrophic microorganisms offers a sustainable path to these products from a range of local, highly abundant, waste and low-cost resources. LanzaTech has pioneered a gas fermentation process using anaerobic acetogenic microbes capable of fixing carbon oxides.

While 10 years ago, acetogens were considered genetically inaccessible and mass-transfer of gases was considered a major scale up hurdle, LanzaTech has since developed a suite of synthetic biology tools successfully scaled up the process from the laboratory bench to full commercial scale. In May 2018, LanzaTech successfully started up a world-first commercial scale (48k MTA) gas fermentation plant using emissions from the steel making process as feedstock for ethanol production.

The technology has been demonstrated with a diverse range of additional low- cost feedstocks including waste gases from other industrial sources (e.g., processing plants or refineries) or syngas generated from any biomass resource (e.g., unsorted and non-recyclable municipal solid waste, agricultural waste, or organic industrial waste) and additional plants are under construction.

In order to maximize the value that can be added to the array of gas resources that the process can use as an input, LanzaTech has established a unique biofoundry that enables automated strain engineering of anaerobic organisms and strain screening in context of flammable and toxic CO and H2 gases. Through this platform, LanzaTech has demonstrated direct production over 50 different products from gas. Several of these products have been scaled up to pilot scale and developed for commercial rollout.

BIOT 257

Bioreactor engineering for a type 3 secretion system powered protein production process DRAFT

Samuel Leach, [email protected], Danielle Ercek. Chemical and Biological Engineering, Northwestern University, Evanston, Illinois, United States

The revenue from protein products manufactured by microbial fermentation eclipsed $50 billion in 2015, which includes pharmaceutical products such as proinsulin and enzymes for food production and agriculture. While bacteria are advantageous as protein production hosts because of their tractability and high titer, the difficulty and expense of downstream purification prevents the market from growing. When proteins are expressed intracellularly in bacteria, product recovery typically requires lysis, protein refolding, and multiple purification steps, which must be carefully optimized for each product. A solution to this problem is to export the product from the cytoplasm to the extracellular space. Salmonella enterica natively employ a Type 3 Secretion System (T3SS) for exporting proteins outside the cell, making it a promising target for this purpose. Previously, we have shown that heterologous proteins can be exported through the T3SS at titers up to 400mg/L, but secretion titer is sensitive to many environmental factors which are critical to scale up. Therefore, determining how bioreactor operating conditions impact T3SS secretion titer is necessary for its future use as a protein production platform. We explored how oxygen transfer, pH, glucose feed rate, and growth rate affect secretion titer through cultivation in a sparged stirred tank bioreactor. In this talk, I will present the results of these optimizations, including the identification of a stir speed and air flow rate regime which maximizes titer, and the observation that a slower cellular growth rate increases secretion rate. These results provide insight into the optimal operating conditions to maximize secreted protein, furthering the development of a protein production platform based on the S. enterica T3SS.

BIOT 258

Connecting biology to electronics through a redox communication network of tyrosine, tyrosinase, and eumelanin

Eric VanArsdale1, [email protected], David Hornstrum2, Eunkyoung Kim1, Juliana Pitzer1, Gregory F. Payne3, William E. Bentley4. (1) Fischell Department of Bioengineering, University of Maryland, University Park, Maryland, United States (2) KTH Royal Institute of Technology, Stockholm, Sweden (3) Univ of Maryland Biotech Inst, College Park, Maryland, United States (4) Fischell Dept of Bioengineering, University of Maryland, College Park, Maryland, United States DRAFT

Systemic control of cellular physiology will allow for the design of synthetic electronic-biological communication networks, capable of coordinating physiology across large distances, disparate environments, and varied time scales. In order to achieve this connectivity, we have begun to develop cellular equivalents of network technology. To do this, we have engineered a coculture transducer system in which cells convert molecular communication into a redox signal by synthesizing both the catalytic and reagent components of an oxidative reaction. The “catalytic” transducer cell population integrates molecular cues to regulate the expression of a surface-linked tyrosinase. Similarly, the “reagent” transducer cell population regulates the synthesis of the enzyme-substrate L-tyrosine, which is converted by the catalytic transducer cell population into the redox signal L-DOPA. In cocultures, this system enables real-time electrochemical detection of molecular information by measuring the oxidative current of L-DOPA. This system was able to eavesdrop on cell-cell molecular communication, which we demonstrate by intercepting the quorum sensing signals of Pseudomonas aerugionosa. We have also adapted this system to receive electronic signals through “transmitter” cells that produce quorum sensing signals in response to electrochemically generated hydrogen peroxide. Finally, we have stored cellular and electronic information in cellularly-produced “hard drives” to store redox memory, through controlled oxidation and reduction of the catechol- quinone groups of eumelanin with redox mediators. The melanin memory units were capable of directing cellular signaling, either by attenuating or amplifying the oxidative signature of a mediator mixture. Together, these cellular units allow for regulated information exchange with biology to allow for rational, user-guided cellular communication and physiology.

BIOT 260

Genetic devices for controlled gene expression in Clostridium

Nicholas R. Sandoval, [email protected], Rochelle C. Joseph, Nancy M. Kim. Chemical and Biomolecular Engineering, Tulane University, New Orleans, Louisiana, United States

Control of gene expression on the transcription level is vital in synthetic biology for the construction of complex and sophisticated tools. To achieve such control, synthetic genetic parts must be developed that can enable such control. Although much recent progress has been made in transcription factor- based biosensors for workhorse organisms, non-model organisms such as Clostridium still lack tools for rapid and customizable gene expression control. Clostridium species are biotechnologically relevant, but do not have a large DRAFT number of transcription regulation devices, reducing the speed of their engineering. In this work, we show two systems for transcriptional control in bacterial systems. First, we demonstrate a Cas12a-based CRISPRi gene repression system for the tuning of biosynthetic pathways. Cas12a CRISPR effector proteins are better suited for use in Clostridium due to the genus’ AT-rich genomes and the corresponding simple and T-rich protospacer adjacent motif. We demonstrate in Clostridium pasteurianum tunable repression based on proximity to regulation elements, strand, and number of targeted sites through reporter genes, transcription levels, and redistribution of carbon flux. Second, we demonstrate optimization of transcription factor (TF) DNA binding sites through both a rationale and high-throughput naïve approach (sort-seq). We engineer a set of engineered product-responsive transcription factor regulated promoters (biosensors) for real-time single cell monitoring of butanol accumulation. We demonstrate tunability of gene expression though altering the number, location, and sequence of the transcription factor binding sites. Additionally, we show transcription factor-operator binding energy plays a key factor in the tunability of such systems. Biophysical characteristics between the TF-operator are evaluated using surface plasmon resonance (SPR). This work can enable rational strategies to edit the dynamic range of transcription factor-based biosensors.

BIOT 261

Metabolic footprinting as a tool to understand the metabolic flux pattern in Monascus purpureus and to model the biosynthetic pathway of lovastatin

Seenivasan Ayothiraman, [email protected]. Department of Biotechnology, National Institute of Technology, Tadepalligudem, Andhra Pradesh, India

Lovastatin is the first commercialized drugs for hypercholesterolemia and still serving as a precursor for the next generation drugs, such as, wuxistatin and simvastatin. At present, the production cost of fermentation derived statins is not competitive enough compared to synthetic statins due to low yield values. System level investigation may help to explore the complexity of biosynthesis pathway of lovastatin, controlling point and associated pathways. In the present work, the predominant flux pattern in Monascus purpureus was found through exometabolites (Footprinting) that helped in proposing the robust biosynthesis pathway of lovastatin. Metabolic footprinting techniques is a qualitative analysis of exometabolites of an organism. The same technique DRAFT has been used to find the predominant metabolic pathway of the organism that favoured the production of lovastatin. The flux distribution pattern of the organism is different in those culture conditions that gave an idea of developing a robust metabolic pathway for the biosynthesis of lovastatin. It was found that the culture condition and the morphology of the organism has also played an important role in the different level of expression of lovastatin in both static and shake cultures. Further, the observation was subjected to metabolic flux analysis and also developed into dynamic metabolic models to validate the proposed flux pattern in Monascus purpureus.

BIOT 262

Reaction compartmentalization leads to enhanced production of PHA in Yarrowia lipolytica

Michael Spagnuolo, [email protected], Meredith Bailey, Mark A. Blenner. Chemical Biomolecular Engineering, Clemson University, Clemson, South Carolina, United States

Making the most efficient use of the finite amount of carbon and energy present in feedstocks is a common goal for the microbial production of commodity chemicals, thus improving the yield of reactions and improving process economics. A well-studied method to improve reaction efficiency is to increase the local concentration of the reactants. To this end, we have focused on increasing the efficiency of reactions building off the beta-oxidation cycle of Yarrowia lipolytica as a means of improving production of the bioplastic polyhydroxyalkanoate (PHA). Local concentration changes are achieved by compartmentalizing key enzymes within the peroxisome, a membrane-enclosed organelle where beta-oxidation occurs. Import into this organelle is accomplished through recognition of some form of peroxisomal targeting signal (PTS) and mediated by a cascade of peroxisome-associated (PEX) enzymes. Manipulation of this tag and the associated enzymes resulted in a greater than three-fold increase in PHA production from the same carbon feed conditions. In an effort to further improve production, expansion of the total peroxisome volume was investigated through knockout and overexpression of specific PEX enzymes. Finally, improvements in carbon flux into the peroxisome were sought via traditional metabolic engineering strategies including carbon-sink knockouts, transporter manipulation, and pathway overexpression.

BIOT 263 DRAFT

Increasing ammonia (NH3) production in model diazotroph Azotobacter vinelandii using kinetic modeling for rational metabolic engineering of the biological nitrogen fixation (BNF) system

Anna M. Crumbley1, [email protected], Weitong J. Chen1, Susan Butler2, Michael Lee3, Ramon Gonzalez4,1. (1) Chemical and Biomolecular Engineering, Rice University, Houston, Texas, United States (2) Bioengineering, Rice University, Houston, Texas, United States (3) Biosciences, Rice University, Houston, Texas, United States (4) Chemical and Biomedical Engineering, University of South Florida, Tampa, Florida, United States

Nitrogen is an essential building block in biological processes. However, most plants and animals cannot use it in its most abundant atmospheric form, diatomic nitrogen (N2) gas. Instead, these organisms rely on limiting sources of fixed nitrogen primarily obtained from nitrogen-fixing microorganisms or the industrial Haber-Bosch process. Due to anticipated rising food demands globally, in situ use and synthetic expression of bacterial enzymatic processes for N2 conversion are being pursued as alternative and environmentally- friendly generators of ammonia (NH3) for agricultural fertilizer and industrial applications. However, layers of fastidious environmental regulatory controls and the complexity of expressing the 19+ genes necessary for functionality in aerobic environments have challenged heterologous expression and limited efforts to optimize its function for strategic applications. To develop a more thorough understanding of the system for rational metabolic engineering of nitrogen fixation, we developed kinetic models of the nitrogenase expression system in the aerobic, free-living, model soil bacterium Azotobacter vinelandii. Using the kinetic models, we identified five new potential targets for engineering that increased NH3 production in silico. Employing synthetic biology strategies to validate the model results, in vivo data indicates that NH3 accumulation can be synthetically tuned using lac inducible overexpression of targeted genes to produce statistically significant amounts of NH3. These results suggest that kinetic models should play a role in identifying indirectly associated enzyme targets that can be optimized to maximize NH3 production, not only in model organism A. vinelandii but also more broadly in other agriculturally-relevant microorganism and synthetic in planta nitrogen-fixation processes.

BIOT 264

Implementation of CRISPR-Cas9 system in Geobacter sulfurreducens to enable ammonium generation from nitrogen gas DRAFT

Mark Poole1, [email protected], Douglas F. Call2, Amy Grunden3. (1) Civil and Enviornmental Engineering, North Carolina State University, Raleigh, North Carolina, United States (2) Civil, Construction, and Environmental Engineering, North Carolina State University, Raleigh, North Carolina, United States (3) Plant and Microbial Biology, North Carolina State University, Raleigh, North Carolina, United States

+ Nitrogen gas (N2) conversion to ammonium (NH4 ) is an essential, but cost- and energy-expensive process, with the industrial Haber-Bosch process constituting the major form of artificial N2 fixation implemented today. Biological nitrogen fixation (BNF) is a more efficient process that has seen significant barriers to commercial implementation. Geobacter sulfurreducens is a free-living N2-fixing soil bacterium that can also respire on electrodes. Recent evidence from our group has indicated that the metabolic rate of G. sulfurreducens while respiring in microbial electrochemical cells + (MECs) is correlated to N2 fixation. However, NH4 is not secreted from cells due to tight regulatory control. The CRISPR-dCas9 system is a robust method for offsetting undesirable regulatory control. In this study, the CRISPR- Cas9/dCas9 system was implemented for the first time in G. sulfurreduencs. As evidence for this new approach, we first targeted acetate utilization in G. sulfurreducens via acetate kinase. Next, we targeted a protein that inhibits the function of nitrogenase in response to high extracellular + NH4 (DraT). In this study, DraT is knocked down with CRISPR-dCas9 to ensure continued function of nitrogenase in the presence of high extracellular + NH4 . Lastly, the knockdown mutants were run under N2-fixation conditions in + MECs to determine their ability to secrete and accumulate NH4 .

BIOT 265

Genome engineering for expanded inducible gene expression in lactic acid bacteria

Samuel Rothstein1, [email protected], Thomas J. Mansell2. (1) Chemical and Biological Engineering, Iowa State University, Ames, Iowa, United States (2) Chemical and Biomolecular Engineering, Iowa State University, Ames, Iowa, United States

Lactococcus lactis plays a vital role as a starter culture in dairy fermentations, exhibits remarkable protein secretion capabilities, and is Generally Regarded As Safe (GRAS). A limitation for the use of this organism for industrial and probiotic applications is the shortage of inducible gene expression systems, an essential tool for heterologous protein expression. Using a native DRAFT homologous recombination and a unique in trans antibiotic selection system, we have inserted several promoter-transcription factor pairs for orthogonal expression of target proteins. The scar-less integration of these inducible gene expression systems lends itself to iterative functionalization of different inducible gene expression systems. In particular enabled induced expression from three different orthogonal inducers. Furthermore, the methods used to introduce these gene expression systems uses a mechanism that which lends its use in lactic acid bacteria beyond L. lactis.

BIOT 266

Understanding fouling mechanisms of filters during the harvest of monoclonal antibody therapeutics

Stijn H. Koshari, [email protected], kusum solanki, Hong Zhang, Hiren D. Ardeshna, Antonio R. Ubiera, Robert Luo. Downstream Process Development, GlaxoSmithKline, King of Prussia, Pennsylvania, United States

The harvest step in a monoclonal antibody downstream purification process removes biomass, particulates, and other material from the cell culture to generate a product stream suitable for purification by liquid chromatography. In a widely employed, two-step harvest process, coarse solids and intact cells are first removed using a centrifuge, and the resulting centrate is further clarified by depth and membrane filtration in series. Conversely, direct depth filtration has the potential to greatly simplify the overall harvest process by directly loading the cell culture broth on high-capacity depth filters and consequently removing the need for centrifugation. We evaluated this potential of direct filtration for multiple biopharmaceutical assets to understand the underlying mechanisms that define fouling and the resulting capacity limitations in these filters. In this presentation, we show how each of the individual filters contributes to the overall capacity of the direct filtration train, and how the capacity is affected by characteristics of the cell culture broth and process parameters, including enhanced upstream processes such as high cell density perfusion. Our findings highlight the limitations of direct filtration when implemented in large-scale, high cell density facilities. Ultimately, this increased understanding of the fundamental principles that govern depth and membrane filtration can aid in moving from a trial-and-error to a more predictive and methodology-based approach for harvest development.

BIOT 267 DRAFT

Understanding membrane performance during combined tangential flow and depth filtration

Da Zhang1, Daniel Strauss2, Parag Patel2, Xianghong Qian3, S R. Wickramasinghe1, [email protected]. (1) Raplh E Martin Department of Chemical Engineeeing, University of Arkansas, Fayetteville, Arkansas, United States (2) Asahi Kasei Bioprocess, Glenview, Illinois, United States (3) Biomedical Engineering, University of Arkansas, Fayetteville, Arkansas, United States

Rapid advancement in upstream cell culture operations has led to a significant increase in product titers. However, this high-level of productivity is accompanied by a rather high cell density which places a significant burden on traditional downstream clarification and purification operations. This is especially true for the initial clarification steps. Depth filtration, centrifugation and tangential-flow filtration (TFF) are used to remove cells, cell debris and other impurities. Depth filters consist of a thick porous bed that can trap particles within the filter matrix. Screen type filters on the other hand largely reject particulate matter by surface filtration. In tangential flow filtration the feed flows parallel to the membrane surface while the filtrate flows perpendicular to the feed. This mode of operation tends to suppress cake growth. The BioOtimal MF-SL from Asahi KASEI represents a hybrid TFF- depth filter. The performance and especially fouling behavior of the BioOptimal filter is not well understood. Here a combined pore blockage and cake filtration model has been developed to describe the fouling behavior of this filter in normal flow mode whereas a resistance in series model has been used to describe the performance in TFF mode. Significant insights have been obtained into the fouling mechanism and optimal operation conditions of this filter.

BIOT 268

Performance characteristics of hollow fiber microfilters for purification of monoclonal antibodies by precipitation

Zhao Li1, [email protected], Qin Gu2, Jonathan Coffman3, Todd M. Przybycien4, Andrew L. Zydney5. (1) ChE, Penn State University, University Park, Pennsylvania, United States (2) Chemical Engineering, Carnegie Mellon University, Pittsburg, Pennsylvania, United States (3) Bioprocess Technology and Engineering, AstraZeneca, Gaithersburg, Maryland, United States (4) Chemical Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania, DRAFT

United States (5) Dept of Chemical Engineering, Penn State University, University Park, Pennsylvania, United States

Precipitation can be a highly effective step in the initial purification of monoclonal antibodies (mAb). High yield of the precipitates can be produced continuously in a tubular reactor using cross-linking and volume exclusion agents. A fully continuous precipitation process has been explored by integrating the precipitation step with hollow fiber microfiltration for dewatering and washing. The objective of this study was to investigate opportunities for increasing the performance of the membrane steps by increasing the critical flux/conversion and reducing the product loss due to re-dissolution of the antibody product. Experiments were performed using human serum Immunoglobulin G (IgG) as a model antibody with 0.2 µm pore size polyethersulfone membranes. IgG precipitates with >99% yield were produced using combinations of zinc chloride and polyethylene glycol in a tubular precipitation reactor with residence time of approximately 30 s. Critical flux experiments were performed using a total recycle system, with the inlet feed flow rate and precipitate concentration varied to determine their effects on the critical flux. Washing experiments were performed using different IgG precipitate concentrations and wash buffer conditions. The overall conversion in the hollow fiber module could be increased by adding a recirculation loop to the membrane module, providing more effective dewatering and washing of the protein product. Model calculations predict that conversion greater than 90% can be achieved with feed IgG concentration of 2.5 g/L, which can provide more than 100-fold impurity (host cell protein, HCP) removal using a single stage wash or more than 900-fold impurity removal after a two-stage counter-current wash step. Pre-concentration of the IgG gave precipitates that were more resistant to product re-dissolution, leading to an increase in product yield. These results provide important insights into opportunities for improving the design and operation of a precipitation – microfiltration process for the initial capture / purification of monoclonal antibody products.

BIOT 269

Evaluation of caprylic acid precipitation as a tool to increase clearance of potentially critical host cell proteins

Frederik Rudolph1, [email protected], Christopher Thamm1, Viktor Gross1, Joey Studts2. (1) Boehringer Ingelheim Pharma GmbH Co. KG, Biberach, Germany (2) Boehringer Ingelheim, Biberach an der Riss, Germany DRAFT

HCPs are a highly diverse range of proteins. While some HCPs can be degradative to the product itself, others could induce an unwanted immune response compromising the safety and efficacy of the biologic. The majority of host-cell proteins are removed through multiple chromatography steps and filtrations; individual host-cell proteins can be more challenging and can be carried through even into bulk drug substance. Hence reducing them to the generally accepted level can be challenging. A promising strategy to achieve this goal is to integrate a precipitation step during harvest operations. In this talk, we discuss caprylic acid (CA) precipitation, as a promising tool to increase clearance of potential critical HCPs and therefore reduce the burden on the subsequent purification and filtration steps. Operational conditions (e.g. pH and caprylic acid concentration) were evaluated with respect to yield, impurity clearance (e.g. HCPs) and impact on product quality attributes (e.g. HMW, LMW, charge pattern). Precipitated cell culture pools were further purified to investigate the impact on following purification steps. In summary, 4 different IgGs were used for this study to demonstrate the applicability of this technique as valuable option for increasing impurity clearance in mAb processes.

BIOT 270

Purification process including cell separation using protein A magnetic beads

Nils A. Brechmann3,1, [email protected], Kristofer Eriksson1,2, Veronique Chotteau3,1. (1) AdBIOPRO, VINNOVA Competence Centre for Advanced BioProduction by Continuous Processing, Stockholm, Sweden (2) Lab-on-a- Bead AB, Uppsala, Sweden (3) Industrial Biotechnology - Cell Technology Group, KTH - Royal Institute of Technology, Stockholm, Sweden

Magnetic bead-based purification from cell suspension has been used for decades for analytical purpose but their use in larger scale has not been possible. A purification process of mAb carried out with non-clarified cell suspension could provide significant savings of operation time and labor compared to the legacy procedure of cell clarification followed by column chromatography purification.

Recently, high capacity magnetic protein A agarose beads LOABeads able to support a purification process have become commercially available. To support such process, we have developed a magnetic separator with which the cell clarification and capture of antibody (mAb) can be performed in a single step. This was applied to harvest CHO cell fed-batch cultures in 20 L DRAFT pilot scale. A preliminary mAb capture step process was first developed with a batch of 26 L clarified harvest using 1 L magnetic beads in the magnetic separator. Two fed-batch runs of 16 L working volume were then purified by direct application on the non-clarified cell suspension of these runs. A rapid mAb adsorption larger than 96.6% was observed within one hour. The yield of this step was 86% and provided a 16 times concentration factor. The mAb purity was similar to a protein A capture by chromatography column performed on clarified harvest however the host cell protein content (<10 ppm) was significantly lower. This study demonstrated the feasibility of magnetic bead-based purification in absence of cell clarification.

BIOT 271

Fouling and yield during sterile filtration of virus-like particle vaccines

Neil Taylor1, [email protected], Wanli (Justin) Ma2, Adam Kristopeit2, Sheng-ching Wang2, Andrew L. Zydney1. (1) Dept of Chemical Engineering, Penn State University, University Park, Pennsylvania, United States (2) Vaccine Process Development, Merck & Co, Inc., West Point, Pennsylvania, United States

There is growing interest in the development of new classes of vaccines to protect against the spread of dangerous infectious diseases. Many of these new vaccines are based on virus-like particles that are in the 100 – 400 nm size range. This significantly complicates the use of sterile filtration to insure sterility of the final vaccine product. This study evaluated the performance characteristics of four commercially available, sterile-filters using a model nanoparticle system, consisting of polystyrene latex spheres, to simulate the behavior of virus-like particle vaccines. Experiments investigated the effect of filtrate flux, particle concentration, and operating mode on membrane fouling and particle yield. Confocal microscopy was used to directly visualize the location of particle capture within the depth of the different sterile filters. Surprisingly, the sterile filters were able to pass particles of 200 nm or greater in diameter under appropriate operating conditions. For example, one filter achieved more than 75% yield of 330 nm particles (average size determined by dynamic light scattering) with a loading capacity greater than 1000 L/m2 at a flux of 300 LMH. The particle yield was independent of filtrate flux at values above 200 LMH during constant flux filtration experiments but decreased at a flux of 100 LMH. The particle yield and fouling behavior varied significantly for filters with different asymmetry and pore morphology. Although these results DRAFT would need to be confirmed with actual biological samples, our work provides important insights into the factors controlling the transmission of relatively large particles through commercially available sterile filters.

BIOT 272

Fouling phenomena during sterile filtration of glycoconjugates

Parinaz Emami1, [email protected], Fatemeh Fallahianbijan1, Seyed Pouria Motevalian3, Brenda CarrilloConde4, Kelvin Reilly5, Andrew L. Zydney2. (1) Penn State University, University park, Pennsylvania, United States (2) Dept of Chemical Engineering, Penn State University, University Park, Pennsylvania, United States (3) Pfizer Inc, Andover, Massachusetts, United States (4) Pfizer Inc, Chesterfield, Missouri, United States (5) Pfizer Inc, Dublin, Ireland

In the past decades, there has been significant interest in the development of glycoconjugates for novel vaccines. Protein-polysaccharide conjugates have been developed for protection against diseases like meningitis, pneumonia, and otitis media. One of the challenges in the manufacture of these large biotherapeutics is in the sterile filtration step, with the high degree of membrane fouling leading to low capacity and potential yield lost. The overall objective of this study was to evaluate the fundamental phenomena governing transmission and fouling during sterile filtration of novel glycoconjugate vaccines. Experiments were performed using several glycoconjugates provided by Pfizer, Inc. Sterile filtration data were obtained during both constant flux and constant pressure operation using sterile filters with different pore morphologies and chemistries including asymmetric polyethersulfone (PES) and homogeneous polyvinylidene fluoride (PVDF) membranes. The location of fouling within the different filters was examined by confocal microscopy using fluorescently labeled glycoconjugates. Pressure data obtained during constant flux filtration showed a gradual increase in the transmembrane pressure (TMP) during the initial stage of filtration followed by a sharp rise in TMP as the membrane became more heavily fouled. A noticeable decline in transmission was observed for some of the glycoconjugates, typically corresponding to the time at which the sharp increase in TMP was observed. Confocal microscopy images showed that the glycoconjugates were captured primarily near the filter entrance, typically in a band penetrating several microns into the depth of the membrane. The volumetric capacity for most glycoconjugates was greater for the asymmetric PES membrane compared to that of the PVDF membrane, although the glyco- DRAFT conjugate with the smallest hydrodynamic radius showed greater capacity with the PVDF membrane. The capacity during constant flux operation was also a function of the filtrate flux, with greater capacity typically achieved when operating at higher filtrate flux. These results provide important insights into the fouling phenomena observed during sterile filtration of glycoconjugate vaccines.

BIOT 273

Three dimensional quantification of depth and membrane filter structure based on high resolution X-ray imaging with flow based simulation

Thomas F. Johnson1, [email protected], Paul R. Shearing2, John H. Welsh3, Daniel G. Bracewell1. (1) Biochemical Engineering, University College London, London, United Kingdom (2) Chemical Engineering, University College London, London, United Kingdom (3) Pall Biotech, Portsmouth, United Kingdom

High resolution 3D imaging has been applied to produce detailed representations of membrane and depth filter structure to analyse geometric characteristics and simulate flow through porous networks. Four depth filters and five membrane filters were selected for respective micrometre and nanometre scale imaging using X-ray computed tomography, enabling visual and quantitative comparisons between materials and to established literature values using non-imaging techniques to validate accuracy of 3D datasets acquired. Average pore size was the main basis of comparison to vendor-provided nominal retention, with other parameters such as porosity and surface area to volume ratio also evaluated. Variation within and between filters was also of interest, in particular the interface between the rejection and capacitance layers within a membrane that can also be symmetric or asymmetric. 3D simulation was also performed on these structures to quantify permeability and tortuosity through these materials to identify flow characteristics that could be related to confocal microscopy findings for both depth and membrane filters. This study enabled nanometre scale locational comparison within and between filtration structures that enhances understanding of the degree of heterogeneity and how geometrical characteristics inside of membrane and depth filters are linked to function and performance.

BIOT 274 DRAFT

Developing a simple automation platform to increase throughput for tangential flow ultrafiltration process development

Michael Homsy, [email protected], Michael A. Winters, Joseph G. Joyce, Matthew P. Watson. Merck & Co. , West Point, Pennsylvania, United States

Increasing productivity in process development can be achieved using small- scale models and automation. Breakthroughs in miniaturization and automation have facilitated high throughput (HT) process development for many downstream unit operations. For example, HT chromatography experiments are now routinely conducted at the microliter scale using small- scale columns with automated liquid handlers. Unfortunately, miniaturization of tangential flow filtration (TFF) systems is challenging due to larger liquid volume requirements. Recognizing challenges with miniaturizing TFF, an alternative strategy leveraging automation has emerged to increase throughput of TFF development. Various commercially available TFF systems now offer full automation capabilities with the ability to run several experiments simultaneously. Procurement of a commercial, automated TFF system, however, can be expensive and may require extensive customization. This work demonstrates an inexpensive and flexible automated TFF system that has been developed at Merck as a platform that can use existing lab equipment. Arduino microcontrollers were programmed to control equipment to run automated concentration and diafiltration functions. A LabVIEW application was designed to connect to multiple Arduinos wirelessly via Bluetooth for data acquisition and display and to initiate automation functions remotely. This remote operation capability along with built-in safety features are especially desirable for work involving potentially harmful biological agents and toxic chemicals. Implementing this system has allowed multiple small- scale experimental arms to be run simultaneously with minimal operator involvement. This presentation will describe features of the automated TFF system and will highlight a range of experiments the automated TFF system has been used for, including ultrafiltration development of a live virus vaccine. Although this approach has so far been used exclusively for ultrafiltration experiments, this technology is highly adaptable and can be used for any activity or unit operation that requires integration of different lab scale instruments.

BIOT 275 DRAFT

Development of isoporous block copolymer membranes for virus filtration: effects of structural and chemical features on throughput, LRV, and robustness

Marty Siwak, [email protected]. Terapore Technologies, South San Fransisco, California, United States

Virus filtration remains one of the most variable process steps in downstream mAb or protein purification. It is also a dominant expendable cost for DSP. Existing commercial VF membranes are based fundamentally on an extension of ultrafiltration technology. To achieve acceptable LRV values, this conventional technology leads to sacrifices in capacity due to broad pore size distributions. In contrast, ISOBLOCK self-assembling technology results in filters with periodic, ordered pore structures, enabling high throughput while maintaining high virus retention. Four key membrane properties can influence both capacity and virus LRV: pore size distribution, membrane asymmetry, surface chemistry, and membrane thickness. Data will be presented to show the effects of these properties and ultimately how they fit into the design of a downstream VF process.

BIOT 276

Effects of solution condition and process interruption on clearance of minute virus of mice during virus filtration of protein therapeutics

Xianghong Qian1, [email protected], Fnu Namila1, Steven J. Traylor2, Tung Nguyen2, Nripen Singh3, S R. Wickramasinghe4. (1) Biomedical Engineering, University of Arkansas, Fayetteville, Arkansas, United States (2) Biologics Process Development, Bristol-Myers Squibb, Hopkinton, Massachusetts, United States (3) Process Development, Bristol Myers Squibb, Acton, Massachusetts, United States (4) Raplh E Martin Department of Chemical Engineeeing, University of Arkansas, Fayetteville, Arkansas, United States

Viral safety is one of the major concerns in the production of mammalian cell and plasma derived biotherapeutic products. A demonstration of virus clearance is required by the regulatory agencies. Virus filtration can provide a robust removal of virus. Virus filtration unit operation has been integrated into many downstream purification processes. Here the effects of solution condition and process interruption on virus retention during virus filtration are systematically investigated for the filtration of Fc-fusion protein spiked with minute virus of mice (MVM) using three commercial parvovirus filters. Our results indicate that virus breakthrough occurs more frequently at a later stage DRAFT of filtration due to the increased virus loading resulting in the repartitioning of the virus particles at the solid-liquid interface. Pressure release from process interruption also leads to more severe virus breakthrough due to the back diffusion and redistribution of the virus particles in the membrane pores. Our study shows that virus breakthrough is complex, but strongly filter and solution condition dependent. Surprisingly, virus breakthrough appears to be most severe under the relatively low membrane fouling condition for one of the commercial filters investigated.

BIOT 277

Low flux impacts on virus filtration: The devil is in the details

Daniel Strauss, [email protected], Alice Butler, Naokatsu Hirotomi. Asahi Kasei Bioprocess America, Glenview, Illinois, United States

Virus filtration provide highly robust virus removal for many different virus types over wide ranges of operating conditions. One operating parameter that has been identified as potentially impacting viral clearance is transmembrane pressure. Using Planova 20N filters, previous data has shown that effective parvovirus removal can be expected under typical operating pressures of 0.5 to 1.0 bar but operating at lower pressures can impact parvovirus clearance under certain solution conditions. This phenomenon is hypothesized to occur due to low flow rates, which allow for diffusion of particles that would otherwise be trapped by hydrostatic forces. However, such a mechanism raises the question as to whether other mechanisms of flow rate reduction will also impact viral clearance. In this presentation, we discuss how other feedstock conditions, such as high product concentration and high aggregate content, can also reduce flow rate. Especially, we show that the mechanism by which flow reduction occurs can have different impacts on viral clearance capabilities.

BIOT 278

Effects of filtration flux on the clearance of minute virus of mice

Rong Fan3, Mi Jin4, Dharmesh Kanani4, Fnu Namila2, Xianghong Qian3, S R. Wickramasinghe1, [email protected]. (1) Raplh E Martin Department of Chemical Engineeeing, University of Arkansas, Fayetteville, Arkansas, United States (2) University of Arkansas, Fayetteville, Arkansas, United States (3) Biomedical Engineering, University of Arkansas, Fayetteville, Arkansas, DRAFT

United States (4) Teva Pharmecuticals, West Chester, Pennsylvania, United States

Viral safety is one of the major concerns in the production of mammalian cell and plasma derived biotherapeutic products. Demonstration of virus clearance is required by the regulatory agencies. Virus filtration is a size-based virus clearance method and is conducted industrially in dead-end mode. It is often operated in batch mode under constant pressure. Membrane performance is shown to be affected by feed condition (pH, ionic strength, and feed buffer components) as well as the operating conditions (flux, pressure, time). During continuous bioprocessing, constant flux is a more desirable mode of operation. Here the effects of filtrate flux and solution conditions on virus retention during constant flux filtration are systematically investigated. Feed streams consisted of monoclonal antibodies spiked with minute virus of mice (MVM). Three commercially available virus filters have been investigated. Surprisingly under low constant flux conditions virus breakthrough was observed for longer filtration periods. The interplay between virus loading on the filter, flux, membrane fouling and filtration time on virus retention will be discussed.

BIOT 279

Seeding strategies for the selective crystallisation of model proteins

Ian Rosbottom, [email protected], Xiaoyu Li, Wenqian Chen, Jerry Heng. Department of Chemical Engineering, Imperial College London, London, United Kingdom

The recognition of the selective nature of biopharmaceutical drugs and the advances in biotechnology have led to over 246 approved products with cumulative revenues reaching $140 billion. Upstream biomanufacture of proteins has advanced, such that the time and expense bottleneck in biopharmaceutics lies in the chromatographic purification steps. Crystallisation is a more economic and energy efficient alternative purification step, producing crystalline proteins with higher purity and stability. However, this technology is yet to mature to manufacturing scale and robustness. Here, we report our optimised crystallisation, additive and seeding strategies to crystallise model proteins and demonstrate selective crystallisation of target proteins from a binary mixture. Mesoporous silica nanoparticles (NPs) are shown to be effective heterogenous nucleants for the selective crystallisation of lysozyme from a DRAFT binary mixture of lysozyme and thaumatin. From hanging drop solution conditions which can crystallise both proteins, the ‘hard template’ SBA-15 NPs are found to not only promote the nucleation of lysozyme, but also suppress the crystallisation of thaumatin. These conditions are also effectively scaled up to the ml level. Further to this, the effectiveness of graphene oxides and DNA polymeric additives are also examined as seeds for protein crystallisation promotion. Here, we report on the use of several templating strategies for the crystallisation of proteins demonstrating the feasibility for selective crystallisation for downstream separations.

BIOT 280

SBE/AICHE/BIOT Wang award: Three continents, two biomanufacturing philosophies, multiple therapeutic modalities: Endless excitement of being a biochemical engineer. K. Konstantinov

Konstantin Konstantinov, [email protected]. Codiak, Cambridge, Massachusetts, United States

Biochemical engineering offers diverse research opportunities for the biotechnology professional, either in academia or industry. This diversity is a unique advantage of our field, which becomes better appreciated as our careers progress, and we keep entering new and exciting research areas, often in an unpredictable way. This is how I feel about my own professional path, which began in the late 80s after a somewhat random, but very lucky choice. Looking back, it is amazing to see the large diversity of research topics I have worked on, spanning from single cell protein to exosomes, from simple batch to advanced continuous processing, from wild type bacteria to highly engineered human cells. Along with this diversity, there are certain fundamental trends, concepts and challenges that remain unchanged with time. My presentation will review some of these topics from a historical/personal perspective, including the ever-growing complexity of the new therapeutic modalities, the complex dynamics of transferring technology from academia to industry, the re- emergence of familiar concepts in highly advanced new forms and, not the least, the rapid advance of integrated continuous biomanufacturing. I will also share some experiences on the soft side of the job, including a few episodes involving Professor DIC Wang. Lastly, I will also share my thoughts of the future of biochemical engineering and what it will take for the new generation of biotech professionals to drive the technology to new heights. With time, the opportunities in the field will only DRAFT get better. Biochemical engineering will continue to enrich our lives with joy and excitement, for which I will remain forever grateful.

BIOT 281

Investigating glioblastoma stem cell behaviors in three-dimensional hyaluronic acid hydrogels

Pinaki Nakod, [email protected], Yonghyun Kim, Shreyas Rao. Chemical and Biological Engineering, The University of Alabama, Tuscaloosa, Alabama, United States

Glioblastoma multiforme (GBM), an aggressive brain tumor, represents one of the most incurable human cancers due to its highly invasive nature. GBM tumors are highly heterogeneous that are composed of tumor cells and small portion of tumor initiating cells also known as glioblastoma stem cells (GSCs) that contribute to the therapeutic resistance and tumor recurrence following surgery. For improving the therapeutic efficiency, targeting GSCs and their surrounding microenvironment appears to be a promising approach. The tumor microenvironment provides physical, chemical and cellular cues that nurture the stemness properties of the GSCs and thus promote tumorigenesis, therapeutic resistance and tumor recurrence. A mechanistic understanding of the impact of these microenvironmental cues on GSCs is crucial in developing an efficient therapeutic approach. Here, we employed three dimensional (3D) hyaluronic acid (HA) hydrogels that allow incorporation of brain microenvironment cues to investigate GSC behaviors. In this study, U87 cell line and patient derived D456 cells were cultured adherently (in the presence of serum) and/or as suspension culture (serum-free) and were then encapsulated in HA hydrogels. We observed that all the seeded single cells expanded and formed spheres, and size of the spheres increased with time. Increasing the initial cell seeding density of cells influenced the sphere size distribution. Interestingly, clonal expansion of serum-free grown tumor cells in HA hydrogels was observed. Also, stemness marker expression of serum and/or serum-free grown cells was altered when cultured in HA hydrogels. Next, we hypothesized that the niche cells in the brain environment when cultured together with the GSCs enhance their stemness phenotype. To test this, we are currently investigating the impact of astrocytes and/or endothelial cells in our 3D HA model in different cell seeding densities. Overall, such biomimetic culture systems could further our understanding of microenvironmental regulation of GSC phenotypes.

BIOT 283 DRAFT

Lipid-functionalized single-walled carbon nanotubes as abiotic cell proxies to screen membrane-penetrating peptides for use in drug delivery and antimicrobials development

Nathaniel Kallmyer, [email protected], Nigel Reuel. Chemical and Biological Engineering, Iowa State University, Ames, Iowa, United States

Membrane-penetrating proteins play a key role in defense against foreign or competing microbes and have been used to design vehicles that can deliver therapeutics or combat antibiotic-resistant pathogens. The study of these naturally occurring proteins has led to the design of new, synthetic cell- penetrating peptides. These designer peptides have been used to deliver large molecules, such as GFP, into the cytoplasm and to lyse notoriously drug-resistant bacteria, such as S. aureus and P. aeruginosa. While these peptides demonstrate great potential, screening techniques to find and optimize their sequences are limited by the timescale of cell growth and require stringent safety protocols to prevent exposure of personnel to target pathogens. Herein, we demonstrate fluorescent single-walled carbon nanotube (SWNT) sensors as abiotic cell proxies to test penetrating peptides. These proxies are constructed by suspending SWNT in either phosphatidylcholine (PC) or lipopolysaccharide (LPS) to simulate the outer surface of an animal cell or Gram-negative bacterium respectively. Disruptions to this simulated cell surface, whether caused by surfactants, chaotropes, electrical potential gradients, or temperature changes, expose the SWNT surface to solvent and cause an immediate quenching of the fluorescent signal. When screening cell-penetrating peptides with these proxies, concentration thresholds are observed; below these thresholds, cell- penetrating peptides cause an increase to fluorescent signal (binding of the peptide to the SWNT membrane proxy), and above these thresholds, peptides cause a rapid decrease to signal (penetration of membrane proxy). The magnitude of the signal drop of PC-SWNT correlated inversely with the activation energy of peptide translocation. The observed concentration threshold similarly correlated inversely with published uptake efficiencies. A short response time (< 30 s) and reduced risk of pathogen exposure make this sensor a potentially valuable pre-screening tool for cell penetrating and antimicrobial peptide discovery and design.

BIOT 284

Liposomal azithromycin reduces post-ischemic cardiac inflammation DRAFT ahmed Al-Darraji2, Jarrod Creameans1, David Henson1, Dave Feola4, Ahmed Abdel-Latif2,3, Vincent Venditto1, [email protected]. (1) Pharmaceutical Sciences, University of Kentucky, Lexington, Kentucky, United States (2) Cardiovascular Research Center, University of Kentucky, Lexington, Kentucky, United States (3) Internal Medicine, University of Kentucky Healthcare, Lexington, Kentucky, United States (4) Pharmacy Practice and Science, University of Kentucky, Lexington, Kentucky, United States

Ischemic heart disease is the leading cause of morbidity and mortality in the US and is often caused by myocardial infarction (MI). The inflammatory response associated with MI leads to an increase in circulating bone marrow cells, particularly monocytes and macrophages. This response is crucial for removing dead tissue and initiating the healing process. However, exacerbated inflammation following MI can be detrimental and is associated with infarct expansion, poor cardiac remodeling and adverse clinical outcomes. Currently, there are no therapies to treat the inflammation that occurs after MI. Modulating inflammation represents a paradigm shift in the treatment of MI, but clinical translation has not been realized. Therapeutic transition of macrophages to a reparative phenotype, known as alternative macrophage polarization, represents a promising strategy in reducing scarring and heart failure after MI. To address this, we have prepared a formulation containing azithromycin (AZM), a clinically approved antibiotic that has been shown to induce alternative macrophage polarization. The lipid composition of the formulation enables 30 mol% AZM to be packaged in the lipid bilayer and a resulting formulation diameter of 203 ± 1 nm. Intravenous administration of liposomal azithromycin (L-AZM) to mice after MI indicates rapid uptake by macrophages and monocytes with significant accumulation in the cardiac lesion after injury. Notably, less than 1% of liposomes are present in cardiomyocytes. Using in vitro and in vivo assays, L-AZM is found to reduce the pro-inflammatory cytokine milieu (e.g. TNFα and IL-1β), while increasing the anti-inflammatory cytokine profiles (e.g. IL-10, TGF-β). Finally, L-AZM is shown to preserve cardiac function, reduce lesion volume post MI, and improve overall survival. These data indicate that L-AZM is a promising candidate for continued pre-clinical evaluation and translation to treat patients experiencing a myocardial infarction.

BIOT 285

Nano-scale temperature measurements using anisotropy-based nanothermometers DRAFT

Sebastian Thompson, [email protected]. IMDEA Nanoscience, Madrid, Spain

Temperature is a crucial parameter in biology, medicine, and physics. Because of that, in the last years, several methods have been developed and presented to measure nanoscale temperature. Optical methods excel because they are non-invasive, spatially accurate and can measure real-time local changes in temperature. Among these, fluorescence anisotropy-based methods are particularly advantageous because they are less affected by changes in the probe concentration and irradiation conditions. Here we present intracellular temperature measurements in cancer cells and live organism using the green fluorescent protein and a method to add thermosensitivity to any protein thereby transforming them into nanothermoters . The method consists of covalently attaching a dye to the protein, which increases the rotational time of the dye-protein system compared to the free dye, and confers thermosensitivity to the resulting bioconjugates. With this method, we transformed bovine serum albumin, glucose oxidase and catalase into nanothermoters. This also allowed us to analyze the anisotropy signal changes occurring during the catalytic cycle of catalase, as well as their correlation with the reaction exothermicity. In addition, it will also be presented the theoretical model that predicts the optimal sensitivity for anisotropy-based thermometers based on protein size and dye fluorescence lifetime. Using this model, most of the proteins and dyes can be converted to nanothermometers. The utilization of these nanothermometers by a broad spectrum of disciplines within the scientific community will bring new knowledge and understanding that today remains unavailable with current techniques

BIOT 286

Systematic engineering of virus-like particles to identify optimal characteristics for nanoparticle delivery

Bon Ikwuagwu1, [email protected], Emily Hartman2, Danielle Ercek3. (1) Chemical and Biological Engineering, Northwestern University, Evanston, Illinois, United States (2) Chemistry, UC Berkeley, Berkeley, California, United States (3) Dept of Chemical and Biological Engineering, Northwestern University, Evanston, Illinois, United States

Self-assembling protein scaffolds known as virus-like particles (VLPs) are promising delivery vehicles for diagnostic and therapeutic applications. VLPs are easily produced in a bacterial expression system, are biocompatible, and DRAFT have defined chemical handles that can attach targeting ligands. However, there are some key challenges to be solved. In particular, VLPs are highly stable and must come apart upon entry to the target cell. Moreover, it is difficult to control their size relative to other delivery platforms. Here, we engineered VLPs based on the MS2 bacteriophage by combining comprehensive codon mutagenesis, one-pot selections, and high-throughput sequencing to create protein fitness maps in a strategy we termed Systematic Mutagenesis and Particle Selection (SyMAPS). Using SyMAPS, we identified several mutations that confer a desired acid instability, as well as a single point mutation that uniformly alters the size and geometry of the VLP. We were intrigued by the mutation that alters the size of the VLP because size is a known characteristic that affects accumulation in tumors. Therefore we applied similar mutations to MS2 homologs to understand if the mutation’s effects were generalizable, and went on to identify alternative mutations that would confer the same size shift. Here, we present the detailed results of this suite of studies, and discuss the role of our engineered VLPs both for theranostic applications and as tools for investigating the role of particle size on tumor penetration.

BIOT 287

DNA-RNA hybrid materials via dual enzyme polymerization for carrier- free therapeutic RNA delivery

Sangwoo Han1, [email protected], Yongkuk Park3, Hyejin Kim2, Jong Bum Lee1. (1) Chemical Engineering, University of Seoul, Seoul, Korea (the Republic of) (3) University of Massachuseets, Amherst, Massachusetts, United States

Development in nucleic acid nanotechnology has been attracted with their therapeutic functions and nontoxicity. As one of them, rolling circle replication (RCR) has been considered as a powerful tool to synthesize various nucleic acid-based architectures. We developed a novel enzymatic synthetic approach to produce cell-targetable DNA-RNA hybrid nanovectors. This material was generated by the dual enzyme polymerization, especially rolling circle amplification (RCA) for DNA and rolling circle transcription (RCT) for RNA with enhanced resistance to nuclease degradation compared to the natural DNA and RNA. DNA strands contains tumor-specific aptamer sequence and RNA strands includes the siRNA precursors resulting in the carrier-free tumor-targeted siRNA delivery. In further steps, DNA-supported RNA hydrogel embedding the therapeutic RNA particles was also fabricated. In previous studies, there are few reports in regard to RNA hydrogel due to DRAFT the structural instability. To overcome this limitation, we employed dual polymerization resulting in macroscale RNA-based structure. This hydrogel is produced by self-assembly of DNA and RNA and their direct hybridization. This system is considered as a smart double-sustained release of RNA particles from hydrogel then siRNA from the particles. As a proof of concept, the hydrogel was applied to tumor-targeted green fluorescence protein (GFP) silencing assay by the double-releasing system in physiological condition. Taken together, our strategy for the synthesis of DNA-RNA hybrid materials has high potential to capacitate further exploration in gene and cancer therapy.

BIOT 288

Development of a platform media and buffer control strategy leveraging scale-down media mixing models and at-scale data lakes

Christopher VanLang, [email protected], Amanda Laidler, Alyssa Fidanza, Elena Banegas Nunez, Elizabeth Valentine, George Boras, Kimberly M. Lamsal, Amy Nehring, Ben Dionne. Process Development, Amgen, Shirley, Massachusetts, United States

A heavy use of a platform media and buffer preparation process allows for the agile development of bioprocesses over a diversity of biologics processes. We have built upon that platform by carefully interrogating and standardizing the individual process steps. With small-scale mixing models, we have built a mechanistic understanding of how various process parameters impact media preparation. Combining these physical insights with a wealth of at-scale data, we can design a flexible media and buffer preparation platform that can better support complex manufacturing processes in an efficient manner.

To enhance our process understanding of the major contributors to performance, we developed mixing models at the 7-L scale using a power per unit volume geometric scale-up method to reflects media preparation up to the 20,000 L scale. Given the role of dissolved carbon dioxide in bicarbonate buffered solutions, it was expected that carbon dioxide entrainment would impact pH. However, process parameters that impact dissolved carbon dioxide levels have been unknown. We present a few case studies that tested hypotheses of media preparation consistency using this mixing model. In one scenario, the order of addition influenced the media pH; changes to the order of addition resulted in a 25% recovery in cell culture productivity. In another scenario, mix time was identified as a major contributor to pH. Increasing the mix time increased the pH by 0.015 units which improved the consistency of DRAFT at-scale batching. These small scale mixing models help to better translate the role of mixing dynamics on process consistency.

These small-scale findings can be combined with real-time process monitoring from data lakes to define a proactive control strategy. Building statistical models to relate process inputs with process outputs, we can leverage plant data to identify potential causes of process variability. Along with the small scale data, representative plant experience can be fed back into the media and buffer platform and aid early stage development. For instance, we have used plant information to compare the process capability of shared media across various scales and mixing geometries to better define the expected process performance. We outline our strategy to define and harmonize the media and buffer preparation process and showcase best practices using small scale models to build process understanding.

BIOT 289 kLa characterization of O2 and CO2 allows for improved prediction of process variables at pilot scale

Xiaoming Liu1, [email protected], Jeffrey C. Swanberg2, Chao Ma3, Jongchan Lee4. (1) Bristol-Myers Squibb, Devens, Massachusetts, United States (2) Process Development, Bristol-Myers Squibb, Maynard, Massachusetts, United States (3) Chemical and Biological Engineering / 870203, The University of Alabama, Tuscaloosa, Alabama, United States (4) Mailstop L-2, Bristol-Myers Squibb, East Syracuse, New York, United States

Mass transfer, including oxygen transfer and carbon dioxide stripping, is critical to cell culture performance in a bioreactor system. As a process scales up from bench scale, it is crucial to make sure oxygen demand is met and carbon dioxide accumulation remains below levels that reduce process productivity or product quality. In this study, complete profiles of oxygen transfer coefficient, kLa O2, and carbon dioxide stripping coefficient, kLa CO2, were established in 500L and 5L bioreactors. A range of scale dependent parameters were studied and different sparger types were compared. Sparging flow rate was found to be the dominant factor for both oxygen transfer and carbon dioxide stripping in both scales, while headspace was not a significant contributor. Van’t Riet equations were established for mass transfer coefficients and applied to real process data. Maximum oxygen demand in pilot scale bioreactor was predicted using kLa O2 and small scale process data. The average error of prediction was 6.3% for six different cell culture processes. Both kLa O2 and DRAFT

kLa CO2 data were used to explain CO2 stripping limitations at the pilot scale. An online pCO2 control strategy was then applied to small scale processes to successfully reproduce pilot scale pCO2 profile.

BIOT 290

Scalability of impurity precipitation via a heterogeneous mixing application of the population balance

Mohamed Agoub2, [email protected], Steven J. Traylor2, Jing Guo3, Weixin Jin2, Helen Hua2, Xuankuo Xu4, Russell Diemer5, Abraham M. Lenhoff6, Sanchayita Ghose1. (1) Bristol-Myers-Squibb, Westford, Massachusetts, United States (2) Biologics Process Development, Bristol- Myers Squibb, Hopkinton, Massachusetts, United States (3) Teva pharmaceuticals, West Chester, Pennsylvania, United States (4) Biologics Process Development, Bristol-Myers Squibb, Devens, Massachusetts, United States (5) University of Delaware, Newark, Delaware, United States (6) Univ of Delaware, Newark, Delaware, United States

The isoelectric precipitation of impurities during viral inactivation neutralization has proven to be difficult to adequately characterize across scales. Namely, efforts have been made to connect critical mixing factors, such as Reynolds number or specific energy, to the observed particle size distributions in order to develop a scalable model of impurity precipitation. However, experimental results have suggested that these factors alone are insufficient in describing the precipitating system across scales. The objective of this work is to further investigate variables contributing to the mechanism, facilitated by simulation of the mixing vessel using computational flow dynamics, and quantifying particle formations via a sectional population balance model. Experiments were conducted for one scaled-up system and two scaled-down systems of different geometries to quantify the trajectory of particle size distributions via flow microscopy. CFD analysis of each system displayed the prominence of heterogeneous mixing conditions in lab-scale experiments, lending credence to the possibility that non-uniform shear rates throughout the vessel impacts the shape of the particle distribution. This heterogeneity can be represented via a two-compartment mixing model, illustrated as individual regions of high- and low-energy agitation exchanging material. A parameter set can be generated to describe these compartments, and subsequently applied to the population balance equations to predict the particle distribution.

BIOT 291 DRAFT

Streamlining process characterization efforts using the high-throughput ambr® crossflow system for ultrafiltration and diafiltration processing of monoclonal antibodies

Lara Fernandez-Cerezo1,2, [email protected]. (1) Downstream Process Development & Engineering, Merck & Co., Inc (USA), Kenilworth, New Jersey, United States (2) Drug Product & Downstream Technology, Merck Sharp & Dohme (MSD), Schachen, Switzerland

Final development stages for biopharmaceuticals often involve process characterization (PC) studies to gain process knowledge and understanding in preparation for process validation. One common approach to conduct PC activities is by using Design of Experiments (DoE), which can provide data for the product quality attributes, operating ranges and system parameters. Qualified scale-down systems are typically used to conduct these studies. For example, for an ultrafiltration/diafiltration (UF/DF) application, a traditional scale-down system is fitted with: a 88 cm2 or 0.11 m2 Pellicon flat-sheet tangential flow filtration (TFF) cassette; a diaphragm crossflow pump; a retentate valve to control mean transmembrane pressure (ΔPTMP); and a feed/retentate recirculating tank connected to a diafiltration buffer tank. This system, however, requires hundreds of milliliters of material per run and can only conduct one experiment at a time. This poses a challenge in resources, in terms of material, time and personnel, as there could be 20+ experiments required for a typical DoE UF study. One solution to circumvent this, is the use of high-throughput screening systems which enable parallel experimentation by only using a fraction of the material requirements. Recently, Sartorius Stedim Biotech has commercialized the ambr® crossflow system to meet this need by only using 10 cm2 proprietary cartridge and 5 mL channel hold-up volume. In this presentation, the performance of this system will be compared with a traditional pilot-scale TFF system for a monoclonal antibody process during a PC study. 2 Studies across a range of ΔΡTMP (12 to 18 psi); QF (3 to 9 L/min/m ); and diafiltration concentration (35 to 55 mg/mL) were conducted in both scales. The main comparative metrics used will be: the permeate flux as a measure of the membrane performance; and the aggregate content as a measure of the product quality. Recovery and turbidity values will also be shown.

BIOT 292

Pilot-scale bioreactor production of recombinant butyrylcholinesterase from transgenic rice cell suspension cultures DRAFT

Kantharakorn Macharoen1, [email protected], Min Du1, Seongwon Jung1, 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

Plant-based biopharmaceuticals have become more attractive due to several advantages of using whole plants (molecular farming) or plant cells as a production platform. For example, plants do not produce prions, and they are unable to propagate human pathogens like mammalian cells. Plants also provide proper protein folding and complex protein post-translational modifications. Unlike whole plants, plant cell cultures can be grown in low cost chemically-defined media in a contained sterile environment such as a bioreactor, which makes the system controllable, reproducible, and compatible with regulatory guidelines such as current good manufacturing practice (cGMP). However, there are limited studies on scale-up to pilot-scale bioreactors, especially in conventional stainless-steel stirred tank bioreactors (SSTBR), to produce recombinant proteins in plant cell suspension cultures. In this study, we successfully scaled-up the production of rice recombinant butyrylcholinesterase (rrBChE), a complex hydrolase enzyme that can be used to prophylactically and therapeutically treat against organophosphorus nerve agents and pesticides exposure, from metabolically regulated transgenic rice cell suspension cultures in a 40-L pilot-scale SSTBR. Employing cyclical operation together with the efficient simplified-process operation (controlling gas sparging rate rather than dissolved oxygen and allowing natural sugar depletion) identified in lab-scale bioreactor studies, we found consistent maximum total functional rrBChE production level of 46-58 µg/g fresh weight in four cycles over 80 days of operation. Here we demonstrate the potentially cost-effective and scalable platform for rrBChE production as a result of the simplified bioreactor operation and the low cost of plant cell culture media.

BIOT 293

Unexpected high levels of HHL formation observed during a tech transfer to a 2000-L single-use bioreactor: A case study

Chentian Zhang, [email protected], Eric Hodgman. Manufacturing Science and Technology, Bristol-Myers Squibb, Devens, Massachusetts, United States DRAFT

An unexpected high-level of impurity formation (heavy-heavy-light chain (HHL)) was observed in a legacy monoclonal antibody cell culture fed-batch process when transferred from a 5000-L stainless steel facility to a 2000-L single-use bioreactor (SUB). Root cause analysis identified that the formation likely occurred during the production bioreactor or harvest steps. To our surprise, the lab studies mimicking the reducing environment of a single-use harvest operation did not induce HHL formation for a normal batch. Further investigation using a 5-L scale-down model of the production step identified that the root cause of HHL formation was actually the byproduct of changing sparger types and the use of a microsparger in the single-use bioreactor (Figure 1). Additional scale-down model work and material hold studies identified that the HHL formation occurs extracellularly. Overall, the evidence suggests that the microsparger itself is not inducing HHL formation; rather its presence set off a cascade of events that ultimately led to excess free thiols in the culture broth present at the end of the run. Specifically, the presence of microsparging leads to poor growth, resulting in overfeeding the culture and unconsumed free-thiol containing components that induce HHL formation. Using this knowledge we were able to replicate HHL formation using orthogonal methods and ultimately close the process characterization gap. This talk will overview the tech transfer considerations used to design the process in the initial 2000-L SUB and supporting characterization studies to characterize HHL formation. Further, this talk will also review tech transfer considerations to a secondary 2000-L SUB facility with a drilled-hold sparger that did not produce any elevated levels of HHL.

BIOT 294

Industrial Biotechnology Award: Development and scale-up of the commercial fed batch refolding process for Lumoxiti, an anti-CD22 two chain immunotoxin

Alan Hunter, [email protected]. AstraZeneca, Gaithersburg, Maryland, United States DRAFT

Lumoxiti (moxetumomab pasudotox) is an orphan drug and the first approved recombinant immunotoxin relying on an antibody fragment for selective targeting of cancer cells. It is composed of the VH and VL portions of an anti- CD22 antibody genetically fused to a truncated form of Pseudomonas exotoxin (PE38) by a peptide linker. Immunotoxins are similar to the Antibody Drug Conjugate (ADC) concept where a toxic payload (in this case PE38) is directed against a target of interest (CD22) using an antibody or antibody fragment. The Lumoxiti CMC process development team, responsible for development of a commercializable manufacturing technology, overcame numerous obstacles and industrialized a truly innovative approach to protein refolding to bring this promising therapy to patients. Industrial protein refolding presents a number of unique challenges. In particular, low refold titers and high levels of misfolded protein and aggregates commonly encountered in refold reaction mixtures often prove exceedingly difficult to purify, leading to low yield and high cost of goods. In this work, we present the development, scale-up, and validation of a commercial immunotoxin refolding and purification process. Limitations of the existing clinical process are explained. The development strategy is summarized and data is presented describing improvements made to the process. Results from the 950L commercial scale refolding process are provided, demonstrating a greater than fivefold improvement in volumetric productivity compared to the clinical process. Lastly, several novel challenges that arose during late stage development are discussed including identification and control of low level expression of endogenous DE3 bacteriophage with a recA deficient E.coli host strain, as well as validation strategy for a process where the two polypeptide chains comprising Lumoxiti are expressed and recovered in separate fermentation unit operations.

BIOT 295

Supercharged protein cargo-templated assembly of Archaeoglobus fulgidus ferritin cages

Joshua A. Bulos, [email protected], Rui Guo, Zhiheng Wang, Maegan A. DeLessio, Jeffery G. Saven, Ivan J. Dmochowski. University of Pennsylvania, Philadelphia, Pennsylvania, United States

Extensive work has been conducted in the field of encapsulation of cargo by a supramolecular protein complex. Many examples exist of loading a variety of cargo molecules, from enzymes to inorganic nanoparticles, using protein cages such as viral capsids, carboxysomes and ferritins. These have been used for a variety of applications, ranging from nanoparticle synthesis to DRAFT developing novel methods for drug delivery. Recent studies have shown templated self-assembly of thermophilic ferritin (AfFtn) from Archaeoglobus fulgidus around a supercharged protein, GFP(+36), as well as around enzyme-GFP(+36) fusions. Here, the charge requirements of protein cargo to template self-assembly of AfFtn using GFPs is investigated with a spectrum of different surface charges (GFP(+28), GFP(+14), GFP(+9), GFP(-30)). Similarly, the surface of human carbonic anhydrase II (hCAII), a zinc - metalloenzyme that hydrates CO2 into HCO3 , was redesigned using computational protein design for encapsulation within AfFtn without the need to fuse it to GFP(+36). The designed hCAII variants retain catalytic function. The AfFtn-hCAII complexes are highly active and can potentially be used for CO2 fixation in extreme temperature conditions.

BIOT 296

Lasso peptides as hyperstable antimicrobials

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

Lasso peptides are miniature proteins (14-33 aa) that are defined by a unique macrocyclic, threaded structure. This structure endows lasso peptides with protease resistance as well as thermostability and resistance to denaturation by organic solvents. We have developed genome mining methods to predict novel lasso peptides from genome sequences and synthetic biology tools for the high level heterologous production of these peptides. This talk will focus on the unique biophysical properties of lasso peptides as well as how these biophysical properties translate to antimicrobial activity.

BIOT 297

Engineering antimicrobial proteins: Co-evolutionary models aid molecular discovery

Benjamin Hackel, [email protected]. Chemical Engineering and Materials Science, University of Minnesota - Twin Cities, Minneapolis, Minnesota, United States

Antibiotic-resistant infections are a pressing challenge in healthcare and agriculture. Antimicrobial proteins (AMPs) present a compelling opportunity for efficient discovery of a suite of potent, selective therapeutics with diverse modes of action. This seminar will discuss the development and use of protein DRAFT engineering platforms to enhance stability, selectivity, and potency in AMPs via a hybrid of bioinformatics-guided library design and high-throughput discovery assays. Co-evolutionary sequence models enhanced library strategies to engineer enhanced stability and activity in endolysins against Clostridium perfringens and vancomycin-resistant Enterococcus. A high-throughput sequence depletion assay elucidated the sequence-function relationship in the oncocin AMP against E. coli. In addition to identifying enhanced AMPs, these methods could be broadly applied to protein engineering.

BIOT 298

Kinetic analysis of cellular internalization and expulsion of unstructured D-chirality cell penetrating peptides

Manibarathi Vaithiyanathan1, Hannah Hymel1, Nora Safa1, Olivia Sanchez1, Jacob Pettigrew1, Cole Kirkpatrick1, Ted Gauthier2, Adam T. Melvin1, [email protected]. (1) Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana, United States (2) Biotechnology Lab, LSU AgCenter, Baton Rouge, Louisiana, United States

Cell penetrating peptides (CPPs) offer great potential as delivery agents for otherwise cell impermeable therapeutic molecules in addition to allowing for the introduction of novel biosensors to track intracellular processes. However, the majority of identified CPPs are unstructured which makes them susceptibility to proteolytic degradation and limits their overall utility. To address this, researchers have explored how altering the sequence identity or structure of the CPPs enhances both uptake efficiency and intracellular stability. Prior work from our lab found that utilizing a β-hairpin secondary structure for the CPP enhanced intracellular stability; however, it diminished overall uptake efficiency when compared to an unstructured, scrambled L- chirality sequences. Thus, a tradeoff was identified between uptake efficiency and intracellular peptide stability. The focus of this work is to expand upon these findings by incorporating D-chirality amino acids in unstructured CPPs to identify peptide sequences with superior uptake kinetics and long-term stability. Intracellular proteases and peptidases cannot recognize D-chirality peptides which allows them to exhibited extended lifetime in the cell. A small library of D-chirality CPPs were synthesized and their performance was evaluated using a series of time, concentration, temperature and energy dependent studies. A three-fold increase in cellular uptake was observed with the D-chirality peptides over their L-counterparts, with a slight alteration in the mode of entry. While majority of uptake occurred by direct penetration, CPP DRAFT internalization was partially controlled by endocytosis followed by endosomal escape and cytosolic access. The inversion of chirality from L- to D- provided increased protease resistance where the D-CPPs exhibited a half-life of ~600 min - a 20-fold increase in comparison to a half-life of ~ 30 min for its L- counterpart. Interestingly, it was observed that a percentage of the internalized D-CPPs were expelled from the cells in a time-dependent manner until the cell reached an effective steady state. Mechanistic studies identified that cells exported the D-CPPs via the exocytosis pathway and that it occurred in an energy-dependent manner. These findings highlight the potential and utility of D-chirality CPPs, but also identify an expulsion method suggesting a new trade-off exists between uptake kinetics, intracellular stability, and intracellular retention.

BIOT 299

Site-specific conjugation of antibodies to a protein carrier for intracellular antibody delivery

Anshul Dhankher, [email protected], Manuel Hernandez, Julie Champion. Chemical and Biomolecular Engineering, Georgia Institute of Technology, Atlanta, Georgia, United States

The application of antibodies as therapeutics is restricted to a small number of extracellular targets since antibodies cannot cross the cell membrane. To target intracellular proteins, we developed a self-assembling protein carrier (Hex carrier) that binds the conserved Fc region of antibodies and delivers them intracellularly. The carrier is comprised of a self-assembling hexamer barrel with an Fc binding domain (SPAB) fused to each monomer of the barrel. Characterization of serum stability of the Hex carrier fully loaded with Immunoglobulin G (IgG) showed that antibodies loaded on the carrier were exchanging with the excess IgG present in simulated serum conditions. In order to use Hex for in vivo drug delivery applications, antibodies needed to be conjugated to Hex to prevent exchange with serum IgG. The SPAB domain contains two phenylalanine residues directly facing the Fc region in the binding interface between the two proteins. To create a covalent linkage between the Fc region and SPAB domain, a non-canonical amino acid analog of phenylalanine, para-azido-phenylalanine (pAzF), was incorporated into the Hex carrier. Upon exposure to UV light, pAzF forms a reactive group that inserts into nearby C-H and N-H bonds of the protein backbone. Incorporation of pAzF was confirmed by ESI mass spectrometry and the Hex proteins containing pAzF (pHex) retained their binding function to the Fc region of antibodies. Upon crosslinking of IgG to pHex, antibody exchange with serum DRAFT

IgG was significantly reduced compared to the original Hex proteins and a non-crosslinked pHex control. Ongoing work is optimizing the reduction of antibody exchange with serum IgG and the minimizing the effects of UV exposure on the protein complex. Alongside the pAzF conjugation, other strategies are being explored to further improve the stability of Hex-IgG complexes in serum.

BIOT 300

Activity-based discovery and optimization of agonist antibodies

John S. Schardt, [email protected], Hark S. Jhajj, Ryen O'Meara, Matthew D. Smith, Peter M. Tessier. Chemical Engineering and Pharmaceutical Sciences, University of Michigan, Ann Arbor, Michigan, United States

Agonist antibodies that activate intracellular signaling via binding to extracellular receptors have emerged as a new paradigm in disease treatment (e.g., in cancer and neurodegenerative diseases). Despite this promise, the discovery of extremely rare antibodies capable of initiating receptor-mediated cellular activation remains a major biotechnological bottleneck. To address this issue, we have developed a high-throughput, direct function-based screening method to generate antibody variants with high agonist activity against the T- cell receptor Ox40. Our approach uses mammalian cells engineered to report intracellular, Ox40-mediated signaling via the NF-κB pathway in a manner that can be robustly detected using flow cytometry. By combining this cellular reporter system with the display of libraries of antibodies on the mammalian cell surface, we are able to directly screen for agonist activity in a high- throughput manner. We will discuss our validation of this approach using clinical-stage agonist and antagonist antibodies as well as our isolation and characterization of novel antibody variants that activate the Ox40 receptor. We expect that this type of activity-based screening method – which can be readily adapted to target other cellular receptors – will be valuable for improving the rapid and reliable isolation of agonist antibodies for diverse biological and therapeutic applications.

BIOT 301

Core annular flows to enhance the injectability of high concentration drug formulations DRAFT

Vishnu Jayaprakash, [email protected], Maxime Costalonga, Somayajulu Dhulipala, Kripa K. Varanasi. Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts, United States

Highly concentrated biologic drug formulations that can be delivered via subcutaneous injections offer tremendous benefits to global health, but they cannot be injected via commercial syringes and needles due to their high viscosities. Current approaches to solve this problem face several challenges ranging from cross contamination and high cost to needle clogging and protein inactivation. Here we report a simple method to enhance injectability using a core annular flow, where the transport of highly viscous drugs through a needle is made easier, thanks to co-axial lubrication by a less viscous fluid. We establish a phase diagram to ensure optimally lubricated core annular flow while minimizing the volume fraction of lubricant. This technique allows us to experimentally obtain up to a 7x reduction in injection pressure for the largest viscosity ratio that was tested, which is in agreement with theoretical predictions for eccentric core annular flows. Finally, we implement these findings into the design and fabrication of a double barreled syringe that significantly expands the range of injectable concentrations of several biologic formulations.

BIOT 302

Exploring linker composition to enhance biophysical properties of antibody fragments for cancer theranostics

Jeong Min Han, [email protected], Thomas J. Magliery. The Ohio State University, Columbus, Ohio, United States

Single-chain variable fragment (scFv) is the smallest functional unit of monoclonal antibody with a non-native peptide linker holding the variable domain together. Its small size allows bacterial expression making engineering easy, and further gains some pharmacokinetic advantages such as better tumor penetration and reduced immunogenicity. Besides, the engineered scFvs can be easily put into different formats (diabodies, bispecifics, conjugates) to tune the half-life and functions. However, despite the great potentials, scFvs and its derivatives have not made it into the clinic. Part of the problem is that nearly all scFvs have been engineered with flexible (Gly4Ser)n linkers, which are susceptible to proteolysis leading scFvs to poor stability. Yet there has been no comprehensive study on compositional effects of the linker. Here, we explored the linker composition to enhance biophysical properties of DRAFT

3E8; 3E8 is a scFv that has strong affinity to a disaccharide, sialyl-Tn on Tumor-Associated Glycoprotein-72, which is present in 80% of adenocarcinomas but in healthy human tissues making it a great candidate for cancer targeting. First, sets of linker libraries in varying lengths and compositions were constructed and characterized to determine favorable linker trends in scFvs. The results showed stabilizing effects for conformationally constrained linkers. Based on this, a much broader sequence space was explored via combinatorial linker library and phage display, a high-throughput screening method. After the fifth round of biopanning, the library was deep sequenced by next-generation sequencing, and the hits appearing in repeat were further characterized to confirm stability enhancements. The results showed a significant improvement in expression level, thermal stability and proteolytic susceptibility with high frequency of conformationally constrained amino acids throughout the linker. Moving forward, we are currently working on generating a high-resolution structure to understand the stability change at the molecular level. Also, we constructed complementarity-determining region libraries to alter its specificity towards a series of cancer-related glycans to generate novel binders for cancers with poor prognosis. Taken together, this study provides valuable insights into scFv engineering that is currently lacking in order to stabilize scFvs and expand its utility for cancer theranostics.

BIOT 303

Predicting antibody developability from molecular simulations and machine learning

PIN-KUANG LAI, [email protected], Amendra Fernando, Theresa K. Cloutier, Bernhardt L. Trout. Chemical engineering, Massachusetts Institute of Technology, Quincy, Massachusetts, United States

After the discovery of target binding antibodies, the characteristics of developability such as high levels of aggregation, escalated viscosity and other stability issues assume critical importance. However, the design space for improving antibody stability is enormous because of the large size of proteins and the dependence of physiochemical properties on the formulation conditions. In this work, we implemented a combination of experimental and computational approaches to elucidate the key determinants for aggregation of 21 therapeutic monoclonal antibodies (mAbs). We have applied accelerated aggregation rate assessment, molecular dynamics simulations and a machine DRAFT learning method for predicting antibody aggregation. The machine learning method combines features from sequence, amino acid properties, molecular descriptors (spatial aggregation propensity and spatial charge map) developed from our group. The contribution of different domains of antibodies is also included. With a rational feature selection, we extracted the most important features to classify and predict therapeutic mAbs with high and low aggregation rate. The cross-validation accuracy is 97% using only three features. The protocol proposed in this study has implications in training and predicting other stability properties with limited experimental data.

BIOT 304

Protein-protein interactions and self association in antibody solutions via static light scattering, small angle X-ray scattering and coarse grained molecular dynamics simulations

Amjad Chowdhury1, [email protected], Keith P. Johnston2. (1) The University of Texas at Austin, Austin, Texas, United States (2) Univ of Texas, Austin, Texas, United States

Highly concentrated therapeutic monoclonal antibody (mAb) solutions are desired for subcutaneous delivery in patients. However, at these high concentrations, protein-protein interactions (PPI) including local, short-range attractions may cause prohibitively high viscosities and poor storage stability. The understanding of this behavior has been limited by the lack of characterization techniques that are applicable at these extreme conditions. Here, we determine the effects of pH and co-solute on the strength and geometry of short-range anisotropic attractions for both a mAb and a polyclonal bovine IgG by comparing static light scattering (SLS) and small angle x-ray scattering (SAXS) data to those computed in molecular dynamics simulations. The simulations utilize 12-bead models exhibiting a van der Waals attraction and additional short-ranged attractions between specific beads and quantify the distribution of reversible clusters in solution. Since our model captures key features of the protein shape, it can describe the experimental SAXS scattering profiles for solutions of 10-200 mg/mL protein with only a small (<1 kBT) variation in the model’s attraction strength. The apparent molecular weights calculated from SLS show reduced net attractions with the addition of NaCl and arginine. Structure factors (S(q)) from SAXS for mAb solutions with added NaCl exhibit an upturn at low q, which our modeling reveals is due to the formation of large reversible oligomers resulting from strong anisotropic attractions in the Fab and Fc regions. For the bovine IgG, reducing the pH below the pI or adding NaCl weaken short-range anisotropic DRAFT attractions, but not enough to remove large reversible oligomers that raise viscosity. In contrast, for arginine added at pH 5.5, the scattering for both proteins can be approximated by a uniform attraction model exhibiting only monomers and dimers. The cluster size distributions are used to improve the understanding and prediction of experimental viscosities using an empirical viscosity model. The ability to discriminate between different interaction geometries and identify the presence of oligomers has great utility in designing and formulating low viscosity, stable mAb solutions.

BIOT 305

High throughput potency characterization of vaccine antigen by affinity chromatography

Qin Yan, [email protected], Mandy Alger, Li Ma, Matthew Brecher, Xianzhi Zhou, Jingning Li, Varnika Roy, Claudia Magagnoli. Analytical Research Development, Glaxosmithkline, Rockville, Maryland, United States

An antigen with multiple antigenic sites has been studied as a potential vaccine candidate. Antibodies against different epitopes were immobilized on columns to prepare affinity columns. Good stability of several such antibodies enabled the routine use of affinity chromatography on HPLC for characterization of the stabilities and binding percentages of different antigenic sites. The procedure is high throughput and samples are injected directly without any sample preparation. The method can be used for both content quantitation and binding percentage of the antigen for in-process samples, final drug substance and drug product.

The performance of the affinity chromatography method is demonstrated by specificity, precision, linearity and accuracy. The method has been used to evaluate lot-to-lot consistency of antigen manufacture process. The method demonstrates stability indication capability and has been used to support the stability study of both drug substance and drug product. The direct measurements of affinity chromatography are absolute %Bound and %Unbound, and they show strong correlation with relative %Potency by ELISA and %Unshiftable (the portion that doesn’t bind to the antibody) by Fab-shift, respectively. It has also been used to measure antigen concentration in in-process samples.

BIOT 306

Considerations in formulation development of a PEGylated protein DRAFT

Lori Burton, [email protected], William Ying, Rajesh Gandhi. Drug Product Development, Bristol Myers Squibb, New Brunswick, New Jersey, United States

PEGylation is a process in which a polyethylene glycol (PEG) chain is covalently conjugated to an active therapeutic (often a peptide, small protein, or an antibody fragment) in order to enhance solubility, reduce immunogenicity and/or prolong circulating half-life in the body. For biotherapeutics, most of which must be administered by injection due to poor oral absorption, the reduction in dose frequency afforded by an extended in- vivo half-life can provide significant advantage in terms of patient convenience, especially for indications requiring chronic treatment. However, PEGylated proteins can also present a number of special challenges during formulation development. These challenges may include potential for chemical instability of the conjugate and, in case of subcutaneous products where injection volumes are limited, increased intrinsic viscosity, which can pose challenges for processing of drug substance and make injection through a fine-gauge needle difficult. In this presentation, we will discuss formulation development of a small PEGylated protein and strategies used to address stability- and viscosity-related challenges to enable a viable drug product suitable for subcutaneous use.

BIOT 307

Understanding reaction mechanisms of monoterpene synthases using atomistic simulations

Hoshin Kim1, [email protected], Narayanan Srividya2, Bojana Ginovska3, B. Markus Lange2, Simone Raugei4. (1) Physical & Computational Science Directorate, Pacific Northwest National Laboratory, Richland, Washington, United States (2) Inst of Biological Chemistry, Washington State University, Pullman, Washington, United States (3) K1-83, PNNL, Richland, Washington, United States (4) Chemical Materials Sciences Division, Pacific Northwest National Laboratory, Richland, Washington, United States

Monoterpene synthases (MTSs) catalyze the production of a diverse array of terpenoid products, which are widely used as key ingredients for many industrial applications including pharmaceuticals and biofuels. MTSs control challenging reactions — such as hydrogen atom, hydride, and proton transfers — in a selective and stereospecific manner. However, despite their importance and versatility, a general understanding on how these enzymes achieve high specificities and efficiencies remain elusive. In this study, we DRAFT performed molecular dynamics (MD) simulations of three MTSs, (-)-limonene synthase, bornyl diphosphate synthase, and 1,8-cineole synthase, aimed at understanding how these enzymes control selectivity by exploiting steric confinement to promote specific configurations of catalytic intermediates. Mechanistic hypotheses based on MD simulations were validated against experimental mutagenesis studies. The potential impact of key residues on product specificity will be discussed.

BIOT 308

Two-step inactivation model of α-Amylase: How do non-aqueous solvents influence the enzyme activity?

Arshad Khan, [email protected]. Chemistry, Penn State, State College, Pennsylvania, United States

α-Amylase is an enzyme present in our saliva and pancreatic secretions and is responsible for the break-down of starch into sugar molecules. It undergoes reversible inactivation by first dissociation of calcium ions followed by irreversible inactivation of apoenzyme by heat. Since non-aqueous solvent suppresses the ionic dissociation step, the enzyme achieves a greater stability in non-aqueous solvents. The effect of heat and non-aqueous solvent will be discussed on the basis of the two-step model.

BIOT 309

Studying the transport of CPP-cargo protein fusions into Candida albicans using experiments and simulations

Sayanee Adhikari1, [email protected], Mahdi Ghorbani1, Jeffery B. Klauda1, Amy J. Karlsson1,2. (1) Chemical and Biomolecular Engineering, University of Maryland, College Park, College Park, Maryland, United States (2) Fischell Department of Bioengineering, University of Maryland College Park, College Park, Maryland, United States

Cell-penetrating peptides (CPPs) are a class of small peptides with the ability to cross membranes and carry biomolecules with them. We are exploring CPPs as vehicles for delivering biomolecular cargo into the opportunistic fungal pathogen, Candida albicans. We recombinantly produced the CPPs histatin 5 (Hst5) and MPG as fusions to green fluorescent protein (GFP) to explore the intracellular delivery of GFP into C. albicans using CPPs. MPG is a short amphipathic CPP that inserts readily into natural membranes. Hst5, DRAFT which is found in human saliva, is a histidine-rich antimicrobial peptide that must cross the cell membrane to exert its fungicidal activity. We studied CPP- mediated translocation of GFP using flow cytometry to detect GFP inside cells. Fusion to MPG led to translocation of GFP into 38% of C. albicans cells, which was significantly higher than the 13% cells that exhibit translocation of GFP without a CPP (p ≤ 0.001). In contrast, Hst5 did not translocate GFP into cells, with only 5% of cells showing Hst5-GFP translocation. Our results show that MPG can be used to deliver GFP, while Hst5 is not as promising. To better understand the translocation of MPG and Hst5, we used molecular dynamics simulations to study their interaction with C. albicans cell membranes. Our simulation studies showed that MPG enters the membrane model readily, with the N-terminal residues initiating the entry. In contrast and consistent with experimental results, Hst5 fails to enter the membrane even after 400 ns. Using knowledge gained from simulations, we designed MPG variants with amino acid substitutions to improve translocation into C. albicans cells and are currently testing the variants for enhanced translocation efficiency. Our results highlight the utility of combining experiments and simulations to study the translocation of CPPs and to engineer CPPs for improved delivery of cargo into C. albicans cells.

BIOT 310

Tissue-specific dystrophin isoforms have different stabilities and functional roles

Vaibhav Upadhyay1, [email protected], Sudipta Panja1, Swati Bandi1, Krishna Mallela2. (1) Dept of Pharmaceutical Sciences, University of Colrado, Aurora, Colorado, United States (2) Pharm Sci box C238-P15, Univ of Colorado Denver, Aurora, Colorado, United States

Mutations in Duchenne muscular dystrophy (DMD) gene that affect the expression of dystrophin protein lead to a number of disorders collectively called dystrophinopathies characterized by muscle degeneration and loss. Dystrophin is a large protein (427 KDa) made up of N-terminal calponin homology (CH) domains, 24 spectrin repeats, cystein rich (CR) domain and C- terminal (CT) domain. Numerous isoforms of dystrophin are known to exist due to alternative splicing and usage of alternate promoters. These variations occur in the functionally important N-terminal and the C-terminal domains. The N-terminal domain mediates interaction with actin, whereas the C-terminal domain interacts with cytosolic proteins like dystrobrevin and syntrophin. The impact of these isoform variations on the structure and function of dystrophin is unexplored. Dystrophin isoforms have differing tissue specificity and thus DRAFT the differences in their structure-function become important in understanding tissue specific symptoms of dystrophinopathies. We probed the impact of these variations on structure-function relationship of dystrophin. We found that minor N-terminal variations in the CH domain of dystrophin isoforms modulate their thermodynamic stability and actin-binding function, thus leading to specificity in dystrophin-actin interactions in various tissues. We also characterized the yet unexplored CT domain of dystrophin, and the variations in the CT domain of dystrophin lead to differences in the stability and binding specificity towards its binding partners. These results show that sequence variations in tissue specific isoforms do govern thermodynamic stabilities, thus affecting tissue-specific dystrophin expression, and dystrophin interaction networks. These differences should be considered before designing effective therapeutic strategies for dystrophinopathies that specifically affect different tissues with a myriad of symptoms.

BIOT 311

Engineering peptide insertions in virus-like particles using systematic apparent fitness landscapes

Stephanie A. Robinson2,1, [email protected], Emily Hartman1, Bon Ikwuagwu2, Matthew B. Francis1, Danielle Ercek2. (1) Chemistry, UC Berkeley, Berkeley, California, United States (2) Dept of Chemical and Biological Engineering, Northwestern University, Evanston, Illinois, United States

Peptide insertions in the primary sequence of proteins expand functionality by introducing new binding sequences, chemical handles, or membrane disrupting motifs. With these properties, proteins can be engineered as scaffolds for vaccines or targeted drug delivery vehicles. Virus-like particles (VLPs) are promising platforms for these applications since they are genetically simple, mimic viral structure for cell uptake, and can deliver multiple copies of a therapeutic agent to a given cell. Peptide insertions in the coat protein of VLPs can increase VLP uptake in cells by increasing cell binding, but it is difficult to predict how an insertion affects monomer folding and higher order assembly. To this end, we have engineered the MS2 VLP using a high-throughput technique, called Systematic Mutagenesis and Assembled Particle Selection (SyMAPS). In this work, we applied SyMAPS to identify and investigate a highly mutable loop in the MS2 coat protein to display 9,261 non-native tripeptide insertions. This library generates a discrete map of three amino acid insertions permitted at this location, validates the FG loop as a valuable position for peptide insertion, and illuminates how DRAFT properties such as charge, flexibility, and hydrogen bonding can interact to preserve or disrupt capsid assembly. Taken together, the results highlight the potential to engineer the MS2 VLP in systematic manner, paving the way to exploring the applications of peptide insertions in biomedically relevant settings.

BIOT 312

Design and directed evolution of conformational antibodies specific for amyloid aggregates

Alec A. Desai1, [email protected], Matthew D. Smith1, Yulei Zhang1, Peter M. Tessier1,2,3. (1) Chemical Engineering, University of Michigan, Ann Arbor, Michigan, United States (2) Pharmaceutical Sciences, University of Michigan, Ann Arbor, Michigan, United States (3) Biomedical Engineering, 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 and affinity of selected antibody variants against insoluble, complex antigens (amyloid fibrils of the Alzheimer’s Aβ42 peptide). First, we have developed a novel library design method that samples wild-type residues and a small number of the most frequently occurring residues at multiple positions in antibody binding loops based on natural antibody diversity. Second, we have developed methods for isolating antibodies with high conformational specificity and low off-target binding by performing selections against the antigen (Aβ fibril) and closely related non- antigens (Aβ monomer) and polyspecificity reagents. By evaluating the relative enrichment of mutations during antibody selections, we find that most mutations that enhance antibody affinity compromise conformational specificity and that deep sequencing methods are critical to identify extremely rare sets of mutations that increase affinity while maintaining strict conformational and sequence specificity. We are currently testing our library design, sorting and bioinformatics methods to generate novel antibodies against a number of other complex antigens (e.g., tau and α-synuclein fibrils and oligomers). We expect that these approaches will improve the in vitro isolation of diverse types of antibodies with high specificity and other drug-like properties.

BIOT 313 DRAFT

Monoclonal antibody glycan analysis using mass spectrometry

Sumit Singh, [email protected], Kelvin H. Lee. Chemical and biomolecular engineering, University of Delaware, Newark, Delaware, United States

With the rapidly increasing use of monoclonal antibodies (mAbs) for clinical use, there is growing demand for fast, efficient and reliable analytical techniques for comprehensive glycosylation analysis. Existing evidence shows that glycosylation in mAbs play important roles in their biological activity, physicochemical properties and effector functions. Even small changes in linkage, position or site occupancy of glycans can adversely influence the effectivity of the products. Yet, the hallmark of mAb N- glycosylation is extensive heterogeneity associated with each glycosylation site. The current mass spectrometric approaches involve relatively complex workflows often performed off-line. Using a case example of NISTmAb, we attempted to investigate if this bottleneck could be overcome by acquiring the data using Waters Bioaccord LC-MS system which has features of automated calibration and method optimization. The performance of the system was benchmarked by comparing the data with the results obtained on Waters Synapt G2- si LC-MS system.

BIOT 314

Evaluation of non-immunoaffinity chromatographic methods for adeno- associated virus capture in Sf9-Baculovirus system

Matthew Luther, [email protected], Aditya Ansodaria, James Forster III, Abhiram Arunkumar, Nripen Singh. Technical Operations, Voyager Therapeutics Inc., Cambridge, Massachusetts, United States

As recombinant Adeno-Associated Virus (rAAV) based therapies continue to show increasing clinical progress, the business and scientific case for robust and efficient manufacturing processes grows proportionally. As such, traditional rAAV downstream processes should be evaluated for robustness and efficiency at each unit operation of the manufacturing process. In this study, multimodal and ion-exchange resins were evaluated as alternative chromatographic capture options for purification of rAAV produced in an Sf9-Baculovirus system. rAAV recovery and the clearance of process related impurities were evaluated along with resin reusability. The result of this evaluation provided valuable insights in developing an rAAV purification process that ensures production of pure and high-quality drug products. DRAFT

BIOT 315

Scalable adeno associated virus production process, from cell culture to purified bulk

Asa H. Hagner McWhirter, [email protected], Ann-Christin Magnusson, Robert Magnusson, Hans Blom, Daphne Areskoug, Markus Pitkänen, Mats Lundgren. Bioprocess application, Cytiva, Uppsala, Uppsala, Sweden

Adenovirus associated virus (AAV) vectors are increasingly used in clinical trials, evaluated for gene therapy applications. Manufacturing of safe and efficacious clinical-grade virus relies on a scalable and cost-effective production process. In this study, we have combined experimental work and process economy calculations, from AAV production in suspension cell culture to purified bulk product up to 10L scale. An efficient and scalable process for AAV production was developed by evaluation and optimization of each process step. We used AAV2 for process development, since this is the most used AAV vector in addition to serotype 5, 8, 9 and various synthetic capsid variants. AAV2 expressing the green fluorescent protein (GFP) was used for process development to facilitate analysis. First, suspension culture of HEK 293T cells adapted to serum-free medium and triple plasmid transfection were optimized using a DoE strategy for transfection efficiency and AAV2 production. The conditions were further developed for production in a single use stirred tank bioreactor system. Novel affinity and ion exchange based purification step alternatives were compared, including different chromatography resins and membrane formats. Finally, ion exchange chromatography and core bead technology were evaluated as an alternative to size exclusion chromatography for the polishing step before the formulation into storage buffer. Analytical methods for virus titer are challenging and depend on purity and quality of the sample. For viral genome titer qPCR was used and for total virus titer ELISA and HPLC methods were used. For infectious virus titer, we have used a flow cytometry-based assay counting GFP producing cells. Based on analytical data different downstream process alternatives were compared regarding load capacity, recovery, reduction of empty capsids and purity and we propose a robust and scalable process with a favorable process economy.

BIOT 316

Leveraging high-throughput strategies to accelerate AAV purification process development DRAFT

Xiaotong Fu, [email protected], Asher Williams, Meisam Bakhshayeshi. Gene Therapy, Biogen, Cambridge , Massachusetts, United States

The adeno-associated viral vector (AAV) has been broadly used as a delivery tool for gene therapy applications. Development of a robust purification process is essential in delivering high purity and potency AAV products to the clinic. The short clinical timelines and limitation of materials to support process development pose unique challenges when developing robust and scalable downstream purification processes. Our approach in overcoming this challenge is to leverage high throughput (HTP) technology for purification process development. In this study, we demonstrated the capability of the TECAN automated liquid handling system to develop purification processes using mini-columns and self-packed resin plates. The purification performance and product quality achieved in the high throughput format was compared with that obtained in the bench scale column chromatography. The established high throughput strategies demonstrated good feasibility, significantly reduced material needs, and reduced development timelines. With further expansion and improvement, we envision that this HTP platform can be used for optimization and characterization of downstream purification processes used for manufacturing of gene therapy products.

BIOT 317

Robust harvest clarification for adeno-associated virus vectors via depth filtration

Thomas Parker1, [email protected], Elina Gousseinov1, Claire Scanlan1, Youness Cherradi2. (1) Technology Management, MilliporeSigma, Burlington, Massachusetts, United States (2) Technology Management, Merck Chemicals N.V./S.A., Overijse, Belgium

The bioprocessing industry widely embraces depth filtration for cell culture clarification of recombinant proteins produced in mammalian cell lines; however, Adeno-Associated Virus (AAV) processes present a unique harvest challenge due to a potential lysis step and the surface characteristics of the target product. While efficient at removing cell debris and other insoluble impurities, many depth filters are optimized for adsorption of soluble impurities for monoclonal antibody process streams, where target products are relatively high concentration and have a slight net positive surface charge. Since AAV have slightly acidic isoelectric points, these viruses may interact with positively DRAFT charged adsorptive depth filters at typical harvest pH ranges. Process development, therefore, requires thorough screening and investigation of depth filters prior to selection for AAV clarification. In light of these concerns, MilliporeSigma reviewed internal data to understand trends of depth filter performance for clarification of AAV from SF9 and HEK293 cell lines; additionally, a controlled screening experiment confirmed findings from data mining activities. This body of work presents the results from these two investigations, which suggest Millistak+® depth filters to be a robust solution for clarification of multiple AAV serotypes from SF9 and HEK293 cell cultures. Additionally, the study demonstrates scalability from bench scale to process scale, giving confidence for implementation in cGMP. Robust performance, ease of use in manufacturing, and scalability make depth filtration an ideal selection for clarification of AAV.

BIOT 318

Two-step purification process for H1N1 virus using ion exchange chromatography resins

Wei Zhang2, Duy Tien Ta2, Kai Ling Chu2, Mark A. Snyder1, [email protected], William H. Rushton1, william_rushton@bio- rad.com. (1) Bio-Rad Laboratories, Hercules, California, United States (2) Downstream Processing Group (DSP), Bioprocessing Technology Institute (BTI), Singapore, Singapore

Yearly outbreak of influenza causes millions of cases of infections worldwide, with about ten percent being fatal. The risk of the disease and the high mutation rate of the viruses have urged scientists to continuously and rapidly develop new kinds of vaccines with higher qualities. In this work, we demonstrate the development of vaccine production process against the Influenza A H1N1 subtype, using recently developed ion exchange resins. The aim is to develop a low cost process while maintaining high quality and safety of the product.

H1N1 virus was propagated in serum free medium using Vero cells as a host and released into the culture supernatant. The supernatant was separated by centrifugation, heat-treated for virus inactivation, followed by Benzonase treatment and concentration by TFF. H1N1 virus concentrate was then captured on a wide pore anion exchanger and subsequently polished by wide pore cation-exchanger. High purity and good overall yields were achieved with this purification. Digital-droplet PCR (ddPCR) with high sensitivity was used as the main quantitative assay throughout the whole process. All purification DRAFT steps used here are scalable and can be adapted for large-scale production, making the process is an efficient alternative for Influenza A H1N1 vaccine production.

BIOT 319

Development of downstream purification process for lentiviral gene therapy vector

Ronit Ghosh, [email protected], Sushmita Koley, Sneha Gopal, Jonathan S. Dordick, Steven M. Cramer. Department of Chemical and Biological Engineering, Rensselaer Polytechnic Institute, Troy, New York, United States

Lentiviral vectors have been increasingly employed as gene delivery vehicles for the treatment of a variety of genetic, neurodegenerative and acquired diseases. As gene therapy treatments progress from pre-clinical to clinical phases, there is a growing demand for high purity and concentrated lentiviral vectors. However, the current manufacturing process presents major challenges such as low virus production titers and viral vector stability leading to low downstream recoveries. This work presents a workflow for robust, fast and scalable process development for LVV purification. Viral vector stability was first evaluated in different mobile phase conditions spanning a range of pH, ionic strength, and buffer species to identify the optimal conditions for the screening of chromatographic media. The stability results suggested that the vector was stable in a pH range of 6.5-7.5 and the stability was observed to decrease above or below this range. The effect of salt concentration on vector stability was also investigated and it was observed to reduce at higher ionic strength likely due to increased osmotic pressure. The results of this study helped to devise a robust purification strategy with the use of chromatographic resins, membranes and/or monoliths which could provide stable lentiviral vector. Process development stage included a high-throughput screening of resins in a bind-elute mode using a 96-well slurry plate format to minimize time and material requirements. Following this, the top candidates were evaluated in a miniature column setting for obtaining resolution/selectivity under different gradient and loading conditions and ranked based on capacity, yield, and impurity clearance criteria. A lab-scale column was performed to ensure the validity of these findings along with tracking the levels of process- related impurities like HCP and hcDNA. Finally, the employed screening strategies and resulting purification schemes aided to guide the development of a GMP process for the manufacturing of clinical-grade lentiviral vectors for gene therapy applications. DRAFT

BIOT 320

Purification of Cas9/sgRNA complexes from residual sgRNA by ultrafiltration

Ivan Manzano1, [email protected], Grace Vezeau2, Neil Taylor1, Howard Salis1,2, Andrew L. Zydney1. (1) Chemical Engineering, Penn State University, University Park, Pennsylvania, United States (2) Agricultural and Biological Engineering, Penn State University, University Park, Pennsylvania, United States

CRISPR technologies have created exciting opportunities for gene editing and novel biotherapeutics, using an engineered single guide RNA (sgRNA) that binds to specific DNA sites enabling modification / regulation of targeted genes. The development of CRISPR-based therapeutics requires robust methods for dowsntream processing, including the purification ribonucleoprotein complex from residual free RNA. We present the first study on the potential use of ultrafiltration membranes to separate Cas9/sgRNA complex from free unbound single guide RNA. The Cas9 protein was produced in E. coli NiCo21 (DE3) cells transformed with pET-28a-Cas9-Cys plasmid containing the S. pyogenes Cas9 with a 6XHis tag. Cells were grown in a 5L bioreactor, lysed, and clarified with the lysate purified by Nickel Affinity Chromatography using a 5 mL HisPure Ni- NTA column. A 48 kDa sgRNA was synthesized from a DNA template consisting of the T7 promoter sequence, a guanosine nucleoside to promote efficient initiation of transcription, and the appropiate DNA sequence. The DNA was prepared from a gBlock fragment and amplified by PCR, T7 RNA polymerase was used for in vitro RNA synthesis using the T7 HiScribe Kit. Synthesized RNA was purified using phenol:chloroform extraction and then coupled to the Cas9 protein. We performed stirred cell filtration experiments with the individual Cas9 and sgRNA to identify appropiate conditions for separation. Key variables were the membrane molecular weight cutoff (50 - 300 kDa), the membrane chemistries (PES and RC), and the filtration velocity. We separate the Cas9/sgRNA complex from free sgRNA via diafiltration, with samples analyzed by gel electrophoresis using SYBR Gold and SyPRO Ruby dyes for visualtization of the RNA and Ca9, respectively. Preliminary data showed effective removal of the sgRNA using a Biomax 100 kDa membrane at filtrate flux around 30 μm/s (100 LMH). These results demonstrate the potential of using ultrafiltration for the purification of ribonucleoproteins from free unbound RNA, providing a low- cost scalable system suitable for large-scale production of these new biotherapeutics. DRAFT

BIOT 322

Developing therapeutic monoclonal antibodies for COVID-19 at pandemic speed

Brian Kelley, [email protected]. Vir Biotechnology, San Francisco, California, United States

Outbreaks of infectious diseases have become increasingly common in recent decades. A few have become pandemics, including HIV, H1N1 Flu and most recently COVID-19. Novel approaches may enable rapid evaluation of antibodies for passive immunization or treatment. The fastest timeline from discovery to clinical evaluation of novel antibodies has been a focus of the biopharmaceutical industry for decades. For potentially life-saving therapies, the benefits of the earliest clinical testing may translate to accelerated pivotal trials and maximal patient benefit. Process and product development groups at major biopharmaceutical companies have reduced phase 1 timelines for antibody production through a universal convergence on similar technologies and strategies. Yet there may be opportunities for substantially faster timelines arising from a combination of the latest technological advances with acceptance of higher business risk or costs without an increased risk profile to patients in the first clinical trials.

BIOT 323

Massively parallel COVID-19 diagnostic assay for simultaneous testing of 19200 patient samples

Alexander Reis1, Ayaan Hossain1, Lauren Cooper2, Howard Salis1, [email protected]. (1) Pennsylvania State University, University Park, Pennsylvania, United States (2) IU Health Pathology Laboratory, Indiana University, Indianapolis, Indiana, United States

The COVID-19 pandemic requires advanced diagnostic assays that can be scaled to many thousands of patient samples. Here, we developed a massively parallel diagnostic assay (MPDA) that combines Reverse Transcriptase (RT), Polymerase Chain Reaction (PCR), and Next-generation Sequencing (NGS) for diagnostic COVID-19 testing on up to 19200 patient samples per workflow. To do this, we applied our Non-Repetitive Parts Calculator to computationally design high-performance RT primers and PCR primers with highly unique barcode sequences that link patient samples to cDNA products across 3 amplicons (N1 and N2 SARS-CoV-2, RNAseP DRAFT human control). After the RT reaction, synthesized cDNA is combined together and amplified in pooled PCRs, followed by next-generation sequencing of the barcoded amplicon library. Within a highly optimized computational pipeline, barcodes and amplicon sequences are then mapped and counted, yielding verified read counts indicative of viral RNA levels. The approach also includes internal spike-in controls for improved quantitation and internal validation. Compared to current approaches, each workflow requires much less equipment, including only one PCR thermocycler and one next- generation sequencer, yielding positive/negative indications on up to 19200 patient samples. We present validation results using both mock samples as well as clinical samples, measuring the limit of detection and range of the assay. We also describe our efforts to automate the workflows using Opentrons robotic workstations. Preliminary efforts to deploy this workflow and seek FDA EUA approval are currently underway at several testing labs.

BIOT 324

Detection of SARS-CoV-2 via Microbubbling Digital Immunoassay

Hui Chen, [email protected], Sheng Feng, Zhao Li, Ping Wang. Pathology and Laboratory Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, United States

The global pandemic of COVID-19, the disease associated with betacoronavirus severe acute respiratory syndrome (SARS-CoV-2) infection, has widely spread the world, causing over 150,000 deaths, millions of confirmed infection cases and billions of influenced people. Rapid, accessible and sensitive diagnosis of SARS-CoV-2 infection is critical for preventing the transmission of the disease. However, rtPCR, the currently widely used testing technology for screening and diagnosis of patients with suspected COVID-19 syndromes, has a typical turnaround time over 24 h, given the need of sample shipping. Although serology tests are more rapid and require much less equipment, their sensitivity are limited, postponing detectable signal several days after symptom onset. Herein, we report the development of a rapid (<1 h), RNA-extraction-free, smartphone accessible and ultrasensitive (with rtPCR matchable sensitivity) microbubbling digital assay for the early diagnosis of COVID-19 by detecting SARS-CoV-2 nucleocapsid protein (N-protein) from respiratory swab. In the microbubbling digital assay picolitre-sized microwells together with platinum nanoparticle labels enable the discrete “visualization” of SARS-CoV-2 N-protein molecules via immobilized-microbubbling with smartphone camera. We also use computer vision and machine learning to develop an automated image analysis DRAFT smartphone application to facilitate accurate and robust analysis of the assay results.

BIOT 325

Oligomerization of COVID19 accessory proteins: Structure, specificity and signal transduction

James Tang1, Nathanael Sallada3, Madison Mann4, Jeffery B. Klauda2, Bryan Berger1, [email protected]. (1) Chemical Engineering, University of Virginia, Charlottesville, Virginia, United States (2) Chemical and Biomolecular Engineering, University of Maryland, Silver Spring, Maryland, United States (3) Biomedical Engineering, University of Virginia, Charlottesville, Virginia, United States (4) Chemical Engineering, Univeristy of Virginia, Charlottesville, Virginia, United States

COVID19 is an emerging, critical threat to human health. Unlike other envelope viruses, coronaviruses such as COVID19 produce unique accessory proteins that do not contribute to viral envelope structure or propagation in vitro, but have been strongly implicated in suppressing host immune response to allow rapid propagation. Prior studies have demonstrated that ORF7a is a single-pass transmembrane proteins, is expressed in vivo and in vitro, and localizes to the endoplasmic reticulum-Golgi intermediate compartment (ERGIC). Furthermore, expression of ORF7a is correlated with virulence in vivo; for ORF7a, a direct association with host bone marrow stromal antigen 2 (BST-2) suppresses its anti-viral activity. However, a specific model for how BST-2 and ORF7a interacts is unknown. In this presentation, we will discuss our collaborative efforts to understand the specific interface responsible for BST-2-ORF7a interaction. Using a bacterial-based oligomerization assay (DN- AraTM), specific interaction domains responsible for heterooligomerization are identified, and confirmed using mammalian cell-based FRET, immunofluorescence and functional assays. Given the known structures of BST-2 and ORF7a, an ensemble of putative heterodimeric structures have been identified using a multiscale molecular dynamics approach. These models are informed by experimental results as well as providing predictive insight into the basis for heterooligomerization to be tested experimentally. Additional work investigating peptide-based inhibitors of identified heterooligomeric interfaces will also be discussed.

BIOT 326 DRAFT

Rapid process development for production of SARS-CoV-2 spike proteins with the baculovirus-insect expression system using a multi well microscale bioreactor system

Jort Altenburg1, [email protected], Carolina Santos Fernandes2, Dirk Martens1. (1) Bioprocess Engineering, Wageningen University & Research, Wageningen, Netherlands (2) Applikon Biotechnology, Delft, Netherlands

To control the pandemic SARS-CoV-2 outbreak, rapid development of a COVID-19 vaccine is needed as well as serological tests to determine whether or not someone has contracted the virus and has built up immunity. The baculovirus-insect cell expression system can be used to produce the SARS-CoV-2 spike protein for vaccine and serological test applications. To produce sufficient amounts of spike protein in a relatively short period of time, rapid screening for optimal process conditions is of utmost importance. The Laboratory of Virology and the Bioprocess Engineering Group of Wageningen University will start with screening different conditions for optimal cell growth and protein production in Sf9 cells using a multi well microscale bioreactor system (Micromatrix, Applikon Biotechnology, Delft, NL). This microscale stirred reactor system permits simultaneously performing 24 individual cultivations in a deep well plate format with control of DO, temperature, and pH. Optimal conditions will be checked in 500 ml stirred tank bioreactor systems (Applikon Biotechnology, Delft, NL) and then scaled up to a 20 liter bioreactor to produce sufficient material for pre-clinical animal trials and serological testing.

BIOT 327

COVID-19 Virus-like particle vaccines for diagnostic and neutralizing mAbs

Asheley Chapman1, [email protected], Liangjun Zhao2, Robert Hincapie1, Alexandre Marques1, M.G. Finn1. (1) Chemistry and Biochemistry, Georgia Institute of Technology, Decatur, Georgia, United States (2) Chemistry and Biochemistry, Georgia Institute of Technology, Atlanta, Georgia, United States

COVID-19, the disease borne of novel coronavirus SARS-CoV-2, has resulted in a global pandemic responsible for over 2 million infections and 160,000 deaths worldwide (as of 4/19/2020), and caused social and economic disruption on an unprecedented scale in 2020. The lack of preexisting DRAFT immunity combined with a shortage of reliable diagnostics resulted in widespread infection. The need for rapid, accurate, ease-of-use antibody- based diagnostics has never been more apparent. Previous coronavirus outbreaks have provided knowledge about the structure, viral entry, and immunogenicity of betacoronaviruses that can inform on the rational design of potential vaccines against SARS-CoV-2. Neutralizing antibodies were developed against S1 of the older SARS-CoV and MERS-CoV pathogens, highlighting the Spike protein as a viable vaccine and therapeutic target. Here, I discuss the identification of potential SARS-CoV-2 peptide epitopes by structural analysis and immunological prediction, the genetic expression and chemical attachment of such candidates on virus-like particle scaffolds, and the results of their immunization in mice. Recombinant SARS-CoV-2 Spike 1 proteins were also arrayed on virus-like particles via interactions between fused Fc tags and Fc-binding domains on VLPs, to present these antigens in a different way. Using hybridoma technology, these vaccines were used to develop SARS-CoV-2 monoclonal antibodies (mAbs) at the Centers for Disease Control and Prevention. MAbs generated using this strategy were screened for viral specificity by ELISA, and binding affinities were characterized by biolayer interferometry. The results of neutralization assays may be available at the time of this presentation. Depending on their performance, these mAbs may be incorporated into rapid serological tests, and tested for prophylactic or therapeutic efficacy.

BIOT 328

Is the environmental sustainability of upcoming continuous manufacturing processes better than the traditional processes?

Sri Madabhushi, [email protected], Jack Huang, Matthew Manahan, Nuno Pinto, Sianny Christanti, Henry Lin. Merck & Co., Inc., Kenilworth, New Jersey, United States

Biologics encompass a wide variety of therapeutics including monoclonal antibodies, fusion proteins, nanobodies, vaccines, and enzymes, among others. With the biologics market volume increasing at a rapid rate, there is a need to assess the efficiency of manufacturing processes to better understand their environmental impact. Based on the recommendation from ACS GCIPR Biopharma Focus group, Process Mass Intensity (PMI) can be used as the benchmarking metric to evaluate the material efficiencies of biologics manufacturing processes. With the rise of continuous biologics manufacturing processes, it is important to understand how the material efficiencies of these processes compare against the standard fed batch processes. Here, we DRAFT calculated the PMI of a continuous manufacturing process and compared it to the PMI of traditional fed-batch processes for monoclonal antibodies (mAbs). Sensitivity analysis was performed to understand the differences and identify potential strategies to develop more efficient processes in terms of their environmental footprint. Furthermore, we compared the PMI of mAbs, vaccines and small molecules from a perspective of drug demand and dosage to gain insights into the practical implications of PMI.

BIOT 329

Development of a novel electrofuels platform for biochemical production from CO2 using synthetic designed proteins and non-growing cells

Scott Banta1, [email protected], Nadim Massad1, Jonathan Preston2, Eskil Andersen2, Aleksandr Uvaydov2, Ronald L. Koder2. (1) Chemical Engineering, Columbia University, Fairfield, Connecticut, United States (2) Marshak Building Room 419, The City College of New York, New York, New York, United States

We are developing a new bioprocess for the production of biofuels from CO2 using renewable energy (Electrofuels). Autotrophic organisms frequently grow slowly, rely on the Calvin cycle for carbon fixation, and can be challenging to engineer for high yields and productivities. We are working to address these challenges by replacing the Calvin cycle with artificial photosynthetic proteins and designed metabolic pathways that will operate in non-growing bacteria cells suspended in photobioreactors. Renewable electricity will be used to power LEDs that specifically energize the artificial photosynthetic pathway in the stationary cells. This will allow for the efficient conversion of CO2 to biofuels and biochemicals such as glycerol We have developed a novel designed light harvesting complex that can split water and regenerate the ubiquitous biological cofactor NADH. At the same time, we have developed a synthetic metabolic pathway that utilizes NADH to reduce CO2 to formaldehyde and then the formaldehyde is ligated to create glycerol using another designed enzyme (formolase). Once installed in E. coli, the pathways will allow for light-powered NAD(H) regeneration which will power CO2 reduction to glycerol that is independent of the cellular metabolism. Thus the system will be operated in photobioreactors which will specifically energize the designed photosynthetic pathway in cells that are not growing. The most recent results of this promising new Electrofuels platform will be presented, including performance metrics for the newly designed photoprotein, pathway fluxes through the carbon fixation pathway, and protein DRAFT engineering results on two key enzymes in the carbon fixation pathway (formaldehyde dehydrogenase and formolase).

BIOT 330

Sustainable bioprocessing for recovery of critical metal Indium

Astha Upadhyay, [email protected], Rouzbeh Tehrani. Temple University, Philadelphia, Pennsylvania, United States

The environmental impact of indium mining from the mine tailings of zinc ore is colossal, and the usage of excessive chemicals has ecological footprints. Bioprocesses and green chemistry can be used to develop more sustainable and cleaner practices. Phytoextraction is an innovative technology to reduce the use of chemicals in the recovery of critical metals. The potential of hyperaccumulators as possible renewable feedstock to mediate the treatment of waste feedstock is evaluated in this study. Electronic waste is exploited as a source for energy-efficient and sustainable secondary production of these metals. Among the other critical metals that can be extracted from e-waste, Indium is the most commonly recovered one from Indium tin oxide (ITO) scrap. The initial analysis of the hyperaccumulator through Inductively coupled plasma mass spectrometry (ICP-MS) demonstrated 98% recovery of indium from the ITO feed solution, which was further validated by Energy Dispersive X-Ray Spectroscopy (EDX) analysis. Although the secondary production of critical metal using the chemical leaching process is well researched, there remains considerable uncertainty over the environmental footprint. Phytoextraction, on the other hand, eliminates the usage of extensive compounds for chemical leaching and promotes greener technology for the recovery of metals from e-waste. Further genetic research on the plant has the potential to intensify the hyperaccumulation. Life cycle assessment (LCA) reinforced the hypothesis that phytoextraction can be more energy-efficient and sustainable compared to the chemical leaching process. Further development of demonstrated process to a large-scale production will assist in achieving the eventual goal of a sustainable circular economy.

BIOT 331

Sustainable and functional tandem repeat protein fibers and films

Melik C. Demirel1, Oguzhan Colak2, [email protected]. (1) ESM Dept, Penn State University, University Park, Pennsylvania, United States (2) Penn State University, State College, Pennsylvania, United States DRAFT

Production of repetitive polypeptides that comprise one or more tandem copies of a single unit with distinct amorphous and ordered regions have been an interest for the last couple of decades. Their molecular structure provides a rich architecture that can micro-phase-separate to form periodic nanostructures (e.g., lamellar and cylindrical repeating phases) with enhanced physicochemical properties via directed or natural evolution that often exceed those of conventional synthetic polymers. In this talk, we review programmable design, structure, and properties of functional fibers and films from squid-inspired tandem repeat proteins, with applications in advanced textiles.

BIOT 332

Engineering Escherichia coli for methanol-dependent growth and production

Michael Dillon1,2, [email protected], R. K. Bennett1,2, Gerald Har1,2, Maciek Antoniewicz3, Eleftherios T. Papoutsakis1,2, Kelley Hestmark4, Stephanie Jones4, Nyarie Dzvova4, Noah Helman4, Derek Greenfield4, Elizabeth Clarke4. (1) Chemical and Biomolecular Engineering, University of Delaware, Newark, Delaware, United States (2) Delaware Biotechnology Institute, Newark, Delaware, United States (3) Chemical Engineering, University of Michigan, Ann Arbor, Michigan, United States (4) Industrial Microbes, Alameda, California, United States

An abundance of methane from natural gas reserves has raised interest in engineering synthetic methylotrophs that can utilize methane or methanol, the oxidized form of methane, for growth and chemical production. Synthetic methylotrophs are nonmethylotrophic platform organisms that have been engineered to utilize reduced C1 compounds. Since native methylotrophs lack many of the well-developed genetic engineering tools that platform organisms possess, the development of robust synthetic methylotrophs is of considerable interest. Our lab previously developed a synthetic Escherichia coli methylotroph that consumes 1 mol of methanol with 1 mol of glucose concurrently, making this strain “methanol-dependent,” as published in our work Engineering Escherichia coli for methanol-dependent growth on glucose for metabolite production in Metabolic Engineering. This allows the sugar feedstock cost to be reduced by supplying a portion of the carbon substrate as methanol, which is less expensive. Using adaptive laboratory evolution and process optimization, we report improvements to growth and chemical production. One such evolved strain exhibited robust methanol-dependent growth in glucose minimal medium, whereas negligible growth was observed DRAFT using the ancestral strain. Furthermore, methanol-dependent acetone and 1- butanol production was achieved via incorporation of heterologous enzymes from Clostridium. As a result of dependence, high average carbon labeling (up to 33%) in acetone and 1-butanol was observed when using a 13C-methanol tracer during fermentation. Future work will involve scale-up to a recently developed cellrecycle bioreactor system for high-titer production at bench- scale. Additionally, this work has been extended to methane utilization via incorporation of methanotrophic enzymes that allow the biological conversion of methane to methanol in recombinant E. coli. By using a 13C-methane tracer, we demonstrate incorporation of methane-derived carbon into central metabolites during gas fermentations in engineered E. coli. This work greatly expands the knowledge of the synthetic methylotrophy field by highlighting the importance of using reduced C1 compounds, alone or together with sugar- based feedstocks, for the production of liquid fuels and chemicals.

BIOT 333

Noncovalent derivatization: Sustainable interfaces between molecules and between disciplines

John Warner, [email protected]. Warner Babcock Institute, Willmington, Massachusetts, United States

The defining text "Green Chem.: Theory and Practice" and the 12 principles of Green Chem. were published twenty years ago in 1998 by John Warner and Paul Anastas. There are now dozens of Green Chem. textbooks, journals, conferences, university courses and degree programs around the world documenting mol. level mechanistic approaches to reducing or eliminating neg. impacts of materials and products on human health and the environment. In the early days of Green Chem., Warner, as an industrial chemist, develop the concepts of Noncovalent Derivatization, a first principles approach to designing multimol. complexes to control chem. and phys. properties. In 2014 he received the Perkin Medal (consider the highest honor in US Industrial Chem.) for this work. This presentation will discuss Warner's basic concepts of Green Chem. and Noncovalent Derivatization. Examples of technologies developed and commercialized at the Warner Babcock Institute for Green Chem. will be described.

BIOT 334

Characterizing the DNA-cleaving and gene-editing abilities of NgAgo and other programmable endonucleases DRAFT

Kok Zhi Lee1, [email protected], Michael Mechikoff1, Archana Kikla4, Arren Liu4, Paula Pandolfi4, Kevin S. Fitzgerald1, Zachary Hartley5, Tyler J. Rankin4, Frederick Gimble3, Kevin Solomon2. (1) Agricultural and Biological Engineering, Purdue University, West Lafayette, Indiana, United States (2) Agricultural & Biological Engineering, Purdue University, West Lafayette, Indiana, United States (3) Biochemistry, Purdue University, West Lafayette, Indiana, United States (4) Biological Sciences, Purdue University, West Lafayette, Indiana, United States (5) Plant Genetics, Breeding, and Biotechnology, Purdue University, West Lafayette, Indiana, United States

Prokaryotic Argonautes (pAgos) have been proposed as more flexible gene- editing tools as they do not require a sequence motif adjacent to their targets for activity, unlike popular CRISPR/Cas systems. One promising candidate from the halophilic archaeon Natronobacterium gregoryi (NgAgo), however, has been subject to intense debate regarding its endonuclease activity and gene-editing potential in eukaryotes. Here, we develop methods and characterize NgAgo in vitro and in vivo, which revealed several unrecognized features that may explain the inconsistent results in the literature. First, NgAgo expresses poorly in non-halophilic hosts with the majority of protein being insoluble and inactive even after refolding, similar to other halophilic proteins. While our data suggest that soluble NgAgo does indeed act as a DNA endonuclease, it does not display significant levels of targeted activity in vitro. Second, structure homology modeling revealed that NgAgo not only shares canonical domains with other catalytically active pAgos but also contains a previously unrecognized single-stranded DNA binding domain (repA). Phylogenetic analysis showed that this repA domain is a hallmark of halophilic pAgos, which form a unique clade that is distinct from other characterized pAgos. We showed that both repA and conserved PIWI domains are required for DNA cleavage in vitro and in vivo. In vivo, these domains can be programmed with guides to induce targeted double-stranded DNA breaks that reduce survival ~51% when targeted to loci. In the absence of such loci, survival remains unaffected relative to unguided controls. Similarly, this targeted cleavage enhances homologous recombination, or gene-editing, in E. coli by 107%. In summary, our results provide new insight into poorly characterized NgAgo for subsequent gene-editing tool development and shed new light on seemingly contradictory reports. Meanwhile, the methods we develop can be applied to other endonucleases, providing new tools for screening programmable endonucleases.

DRAFT

BIOT 335

Scaling up synthetic biology: Engineering synthetic genetic systems with 100s to 1000s of non-repetitive genetic parts

Howard Salis, [email protected]. Pennsylvania State University, University Park, Pennsylvania, United States

Engineered genetic systems can fail when designers are forced to re-use similar parts, due to the introduction of repetitive DNA sequences. To solve this challenge, we developed a novel algorithm, the Non-Repetitive Parts Calculator, that rapidly designs thousands of highly non-repetitive genetic parts from specified constraints, utilizing a newly developed linear-time, memory-efficient approach. As demonstrations, we designed and characterized 4350 highly non-repetitive bacterial σ70 promoters with transcription rates that varied across a 820,000-fold range and 1722 highly non-repetitive yeast promoters with transcription rates that varied across a 25,000-fold range, specified by rational design constraints with desired functionalities. We show how utilizing non-repetitive genetic parts greatly reduced homologous recombination as a source of genetic instability.

We then applied non-repetitive genetic part toolboxes and rational design to stably co-express up to 22 single-guide RNAs within highly compact extra- long sgRNA arrays (ELSAs). ELSAs use CRISPR interference to simultaneously knock down the expression of up to 22 targeted genes by up to 3500-fold. As demonstrations, we introduced designed ELSAs into the E. coli genome to create and stably maintain highly selective phenotypes, characterized by long-term growth, RT-qPCR, RNA-Seq, and LC-MS measurements. A 15-sgRNA ELSA targeting hisD, proC, lysA, tyrA, aroF, pheA, leuA, ilvD, and argH resulted in multiple amino acid auxotrophy. A 20-sgRNA ELSA targeting poxB, sdhC, sdhD, ackA, pta, and iclR increased succinic acid production by 150-fold. A 22-sgRNA ELSA targeting yncG, plsB, dkgA, yncE, ansP, narQ, yncH, adiA, iclR, ycfS, marR, and wzb disrupted stress response, inhibited membrane biosynthesis, and reduced persister cell formation by 21-fold after antibiotic treatment. We developed a design algorithm, the ELSA Calculator, that facilitates the automated design of ELSAs to regulate the expression of 20+ desired genes.

Overall, by combining rationally designed toolboxes of non-repetitive genetic parts with system-level design algorithms, we show how 100s to 1000s of DRAFT genetic parts can be combined together into stable genetic systems with evolutionary robust, programmed phenotypes.

BIOT 336

Expanding the capabilities of the Cas6 based dynamic high affinity scaffold for spatial and temporal co-localization of proteins

Alexander A. Mitkas1, [email protected], Wilfred Chen2. (1) Chemical and Biomolecular Engineering, University of Delaware, Newark, Delaware, United States (2) Chemical Engineering, University of Delaware, Newark, Delaware, United States

In modern biotechnology, platform organisms such as E. coli and S. cerevisiae are used produce small chemicals of interest through non-native enzymatic pathways. Traditional approaches for optimizing nonnative pathway productivity involve directed evolution of the key enzymes of the pathway and manipulation of the cell’s native metabolism through deletions or overexpression of enzymes to drive product formation. These approaches require multiple iterations and do not always yield the desired results. An alternative approach involves clustering nonnative enzymes together in the cell in order to maximize metabolic flux through the nonnative pathway. If the clustering is not dynamic, metabolite imbalances can occur which can prove detrimental to maximizing product titer. To address this issue a dynamic high affinity scaffold to colocalize enzymes dynamically was developed. The scaffold utilizes small RNA sequences and orthogonal Cas6 proteins taken from the CRISPR/Cas Type I systems. The Cas6 proteins bind with high affinity and sequence specificity to short RNA hairpin sequences. By designing the remaining RNA sequences appropriately, assembly of the scaffold through base pair hybridization has been demonstrated. Furthermore, the scaffold can be disassembled through the addition of a short 6-18 nt toehold sequence on one of the hybridized strands. Upon expression of a third RNA strand, the trigger, the scaffold quickly disassociates via toehold mediated strand displacement (TMSD). In previous work, successful scaffold assembly and disassembly via trigger strand expression has been demonstrated using the split luciferase reporter system by correlating increased and decreased luminescence to scaffold assembly and disassembly. Scaffold assembly has been demonstrated to occur only when all correct scaffold components are expressed. So far, scaffold disassembly was only demonstrated using a synthetic trigger. Recently scaffold disassembly using native sRNA sequences as the trigger (RyhB)has been successfully demonstrated. Furthermore, a turn ON system which relies on DRAFT the expression of the trigger to assemble the scaffold has also been developed using the split luciferase system. Finally, the toolkit has been applied to increase the product yield of model metabolic pathways. The expanded capabilities of this scaffold toolkit increase the scope of its application and allows users more freedom to design a scaffold that directly fits their needs.

BIOT 337

Desktop digital genome engineering: CRISPR editing of microbes on a massively parallel scale

James J. Lalonde, [email protected], Richard Fox, Michael Clay, Nandini Krishnamurthy, Eric Abbatte. Inscripta, Inc., Boulder, Colorado, United States

To fuel the pace of discovery through CRISPR editing, the research community requires tools to deliver genome-scale edits at the gene, pathway, and genome level. This would enable testing thousands, perhaps hundreds of thousands, of individual ideas rapidly and efficiently. The most urgent need in current CRISPR pipelines is a significant improvement in the ability to generate diversity in a scalable manner, which includes introducing both large numbers of edits as well as a broad variety of edit types. Achieving high edit efficiencies is a serious challenge in high throughput editing workflows. We have developed a platform which dramatically increases the scale of Digital Genome Engineering, making it possible to easily perform CRISPR-based forward engineering experiments through high-throughput diversity generation in E. coli and yeast. The platform simplifies the complex editing workflow solution from design to diversity generation – including software for oligo design, reagents, benchtop instrument and analytics. The platform enabled high-throughput diversity generation with 10’s of thousands of edits across the genome in both coding and non-coding regions to improve lysine production in E. coli. Similar experiments have been conducted in yeast to improve tyrosine production. This technology has also fueled large-scale discovery of gene- phenotype relationships in antibiotic resistance and abiotic stress responses. Examples of unprecedented genome wide studies performed on tens of thousands of different microbial strains containing a spectrum of gain and loss of function edits spanning a variety of environmental conditions.

DRAFT

BIOT 338

Low-pass filtering effect of degradation-driven protein turnover

Bahareh Mahrou2, Azady Pirhanov2, Yong-Jun Shin2, Yongku Cho1,2, [email protected]. (1) Chemical and Biomolecular Engineering, University of Connecticut, Storrs, Connecticut, United States (2) Biomedical Engineering, University of Connecticut, Storrs, Connecticut, United States

A low-pass filter passes low-frequency signals while attenuating high- frequency ones. To remove high-frequency noise, low-pass filtering is widely used in science and engineering. Low-pass filtering behavior is also found in various biological systems, which allow robust adaptation to continuously changing environment. Since the intracellular environment is highly stochastic, reduction of high frequency noise is critical in maintaining homeostasis and on the other hand allowing efficient adaptation. Previous studies have reported low-pass filtering in the cellular transcription machinery, using fluctuating levels of galactose inducer with defined frequency. The other side of protein turnover is protein degradation, and continual balance between production and degradation results in observed protein levels in cells. Here, we hypothesized that protein degradation driven protein turnover also filters out high-frequency fluctuations. To experimentally validate this, we applied a light- activated protein degradation machinery, which enables fluctuations in protein degradation rate by delivering temporally controlled light stimulation. This optogenetic approach allowed us to observe degradation-driven fluctuations in fluorescent reporter protein levels in the yeast Saccharomyces cerevisiae. We show that under continuous transcription (as assessed by quantitative PCR of the transcript), the protein level fluctuates according to the light stimulation at frequencies (1/f) ranging from 100 min to 20 min. As stimulation frequency increases, the delay between input (light stimulation) and output (protein level) increases and the amplitude of fluctuation decreases. These characteristic behaviors of low-pass filtering indicate that the degradation machinery also filters out high frequency noise, but respond robustly to slow fluctuations with response times faster than that of the transcription machinery. Our results combined with previous studies conclude that both protein production and degradation enable protein turnover to exhibit low-pass filtering.

BIOT 339 DRAFT

Data mining for mechanistic models using enhanced symbolic regression

Nicole Beauregard1, Steven Harris2, Mark Marten3, Ranjan Srivastava1, [email protected]. (1) Univ of Connecticut, Storrs Mansfield, Connecticut, United States (2) University of Manitoba, Winnipeg, Manitoba, Canada (3) Engineering Bldg 314, UMBC, Chem., Biochem. Environ. Engr., Baltimore, Maryland, United States

Biological systems can often be represented as complex mechanistic networks illustrating the intricate relationships among biological species. Large quantities of data are often generated from experiments investigating the existence or function of different pathways or mechanisms. The ability to predict biological networks from experimental data would be invaluable for biotech or pharmaceutical industries as a means for rapid identification of network bottlenecks, determination of potential drug targets, or optimization of bioprocesses. Even with advances in machine learning and computing power, the ability to deconvolute networks to identify the mechanistic network from data is still a challenge today.

This work examines the use of enhanced symbolic regression via genetic programming as a technique for inferring biochemical networks from data. Biochemical networks can be decomposed into systems of differential equations using assumptions from mass action kinetics. Genetic programming, a type of evolutionary algorithms, uses recombination and mutation principles to evolve equations over generations. This algorithm is notorious for its computational intensity and long simulation times. We implemented heuristics to this genetic program based on domain knowledge of our system to constrain both the complexity of the equations generated and the parameter search space. To assess the feasibility of the enhanced algorithm, we produced a synthetic dataset using a known system of equations describing intracellular viral kinetics. The known system consisted of three ordinary differential equations and six parameters. After 100 generations, we were able to evolve an ensemble of models that bound the dataset. To validate the model, we perturbed the initial conditions and still saw good performance from the ensemble of models. The enhanced symbolic regression technique shows good promise as a technique to infer information about biochemical networks from data. However, it is clear that to handle the challenges posed by larger system, heuristics grounded in domain specific knowledge will be needed to reduce the search space to a manageable size.

BIOT 340 DRAFT

Optogenetic circuits for dynamic control of Escherichia coli metabolism

Makoto A. Lalwani, [email protected], Jose L. Avalos. Chemical and Biological Engineering, Princeton University, Princeton, New Jersey, United States

The ability to induce gene expression in Escherichia coli is a cornerstone of basic and applied research in the life sciences and biotechnology. Optogenetic regulation of gene expression offers additional capabilities such as easy tunability, reversibility, dynamic regulation of induction strength, and the ability to exert control without the expense and potential side effects of chemical inducers. Here, we develop optogenetic circuits for the bacterium Escherichia coli that induce gene expression in darkness and repress it under blue light. Applying them to metabolic engineering improves mevalonate and isobutanol production compared to chemically induced controls in light-controlled fermentations that are scalable to at least lab-scale bioreactors. Furthermore, we show that our circuits can be used to control protein production with light. Our results show that light is a suitable alternative to chemical induction for microbial production of chemicals and proteins.