Towards the Development of Biobetter Therapeutic Whole Antibodies

Towards The Development of Biobetter Therapeutic Whole Antibodies

A thesis submitted in fulfilment of the requirements for the award of Doctor of Philosophy

Vicki Sifniotis BSc (Hons) Biotechnology, UNSW

School of Pharmacy Faculty of Medicine and Health The University of Sydney

2019

Statement of Authentication

This thesis is submitted to the University of Sydney in fulfilment of the requirement for the degree of Doctor of Philosophy. The work presented in this thesis is, to the best of my knowledge and belief, original except as acknowledged in the text. I hereby declare that I have not submitted this material, either in full or in part, for a degree at this or any other institution.

Name: Vicki Sifniotis

Date: 08/02/2019

Acknowledgements

I am profoundly honoured and humbled to have been provided the opportunity to embark on this research journey in fulfilment of my PhD candidature. It has been a real privilege and pleasure to undertake my higher degree by research at the School of Pharmacy, Faculty of Medicine and

Health, University of Sydney, in A/Prof Kayser’s research group. The Usyd pharmacy school is full of wonderful and supporting colleagues and staff that hold a high and professional standard in research and teaching. In my time here, I have gained much confidence and professional development with being a skilled and valued researcher, and am very grateful to the many collaborations and friendships I have accrued.

I firstly want to thank my supervisor A/Prof Veysel Kayser and co-supervisor A/Prof Serdar

Kuyucak for the opportunity, their support, valuable insight and expertise, and their mentorship in guiding me through my candidature. The pharmacy school support staff have been impeccable,

Paddy Dhanvate, Suley Aktun, Laurence Rubacki and Tim Jones are all absolute champions and I am really grateful for your professionalism, integrity and kindness. Our school dean, Prof Andrew

McLachlan is just an absolute legend; I can’t gush over him enough, he is a supreme rock-star scientist. Dr Donna Lai and Dr Sheng Hua from the Bosch Institute (Usyd), Dr Mario Torrado del

Rey from Prof Mackay’s group (Usyd), Dr Tihomir Dodev from Dr Beavil’s group (King’s College

London) and David Mabon from GE Healthcare have been extremely helpful and I am very grateful for their kindness and expertise.

The direction and progression of antibody research in A/Prof Kayser’s group wouldn’t have eventuated if not for the inclusion of our most valuable post doc, Dr Zehra Elgundi. I am extremely grateful to Zehra joining our team despite her short term, as her expertise assisted us in moving our work forward; designing projects; teaching relevant skills and optimising techniques. Before her inclusion, there was absolutely no expertise in the group to be able to establish antibody projects, and all three antibody research students (me included) were struggling; unable to produce antibody let alone achieve tangible outcomes for more than 8 months. As soon as she joined our group, that situation immediately changed which markedly reflects how real engagement, expertise and leadership attributes to the success of a research team. I have Dr Zehra Elgundi to thank for the iii fruition of my work and professional development as a researcher, and am truly thankful to have worked with her.

I am very lucky to have such dear and valuable colleagues from the antibody research project,

Mouhamad Reslan and Esteban Cruz; I love you guys, I see you as my brothers and am so grateful to have worked with you *hugs*. Further students from our group Emma Li, Jacky Chan, Ziya

Yusuf Sahin, Senem Akkoç, Christina Limantoro and Kathy On, I’m so glad to have worked with you all. I have made some profound friendships; Vicky Lee, Wei-Ren Ke (William), Martin Wallin,

Dr Dipesh Khanal, Dr Sally Kim, Dr Thaigarajan Parumasivam (Thaigar), Dr Jiaqi Yu, Kamini

Divakarla, Daniela Eassey, Ariel Astudillo and Kevin Samnick. You guys, I love you all and it means the world to me to have crossed paths with you. And lastly, a deep and heartfelt thank you to my family and friends from IWCB for their generosity, love and support.

List of Publications and Conferences

Publications

Sarı, Y., Akkoç, S., Gök, Y., Sifniotis, V., Özdemir, İ., Günal, S., Kayser, V. Benzimidazolium-based novel silver N-heterocyclic carbene complexes: synthesis, characterisation and in vitro antimicrobial activity. J. Enzyme Inhib. Med. Chem 31, 1527- 1530 (2016).

Elgundi, Z., Sifniotis, V., Reslan, M., Cruz, E., Kayser, V. Laboratory Scale Production and Purification of a Therapeutic Antibody. J. Vis. Exp. (119), e55153 (2017).

Elgundi, Z., Reslan, M., Cruz, E., Sifniotis, V., Kayser, V. The state-of-play and future of antibody therapeutics. Adv. Drug Del. Rev. 122, 2-19 (2017).

Sifniotis, V., Cruz, E., Eroglu, B., Kayser, V., Current Advancements in Addressing Key Challenges of Therapeutic Antibody Design, Manufacture and Formulation. Antibodies (2019).

Poster presentations

Sifniotis, V., Reslan, M., Cruz, E., Kayser, V. (2018) Rational addition of glycosylation sites on trastuzumab; Glycoengineering a model antibody for enhanced stability.

Presented at:

The University of Sydney Annual Postgraduate Conference 2018 (won first prize). The 12th Annual BioProcessing Network Conference 2018 (won first prize). The 7th Crown Princess Mary Cancer Centre Symposium 2018.

Oral presentations

Sifniotis, V., Elgundi, Z., Reslan, M., Cruz, E., Kayser, V. (2016) Enhancing antibody affinity; towards a trastuzumab biobetter. The American Association of Pharmaceutical Sciences University of Sydney Student Chapter Annual Symposium 2016.

Sifniotis, V., Reslan, M., Cruz, E., Kayser, V. (2017) Rational addition of glycosylation sites on Trastuzumab; to reduce aggregation, increase solubility and improve formulation. The American Association of Pharmaceutical Sciences University of Sydney Student Chapter Annual Symposium 2017.

Contents

Statement of Authentication...... ii Acknowledgements ...... iii List of Publications and Conferences...... v Contents ...... vi List of Abbreviations ...... xii Thesis Abstract ...... 1 Thesis Overview ...... 3 Chapter 1 Therapeutic Antibodies ...... 6 1 Whole Therapeutic Monoclonal Antibodies; Introduction and Significance ...... 7 1.1 Trastuzumab; Model Antibody for Cancer Therapy ...... 9 1.2 Adalimumab; Model Antibody for Autoimmune Disorder Therapy ...... 10 2 Pharmaceutical Development Challenges and Considerations ...... 11 2.1 Discovery and Manufacture ...... 13 2.2 mAb Drug Design and Formulation...... 17 3 Computational Approaches for Prediction and Rational Design of mAbs ...... 20 4 Perspectives, Aims and Objectives ...... 23 Chapter 2 Vector Design and Cloning for Expression of Trastuzumab and Adalimumab ...... 24 Scope and Aims ...... 25 1 Introduction; Expression Vector Library ...... 25 1.1 pVITRO1 ...... 26 1.2 mAbXpress...... 28 1.3 gWiz ...... 29 2 Materials and Methods ...... 31 2.1 Instrumentation, Services and Software ...... 31 2.2 Molecular Biology Materials ...... 31 2.3 Culture and Purification Materials ...... 32 2.4 Cloning Strategies and Molecular Biology ...... 32 2.4.1 Cloning Tmab Variable Regions into mAbXpress Vectors ...... 33 2.4.2 Cloning Tmab Whole Genes into gWiz ...... 35 2.4.3 AdmAb Variable Region Gene Switching ...... 38

2.5 Isolating Positive Clones and Sequence Analysis...... 41 2.6 Transfection and Purification of Antibodies ...... 42 3 Results ...... 43 3.1 Cloning Tmab Variable Regions into mAbXpress Vectors ...... 43 3.2 gWiz Cloning of Whole Tmab Genes ...... 45 3.3 Variable Region Gene Switching in gWiz and pVITRO1 Vectors ...... 47 3.4 Sequence Analysis and Cloning Efficiency ...... 49 3.5 Preliminary Transient Expression Study...... 52 4 Discussion ...... 54 Chapter 3 Laboratory Scale Production and Purification of a Therapeutic Antibody ...... 57 Abstract ...... 58 1 Introduction ...... 58 2 Protocol ...... 60 2.1 Recovery and Scale-up of Vector DNA ...... 60 2.2 Stable Transfection of HEK-293 Cells ...... 61 2.3 Large Scale Antibody Production from Batch Overgrow Culture ...... 63 2.4 Antibody Purification by Affinity Chromatography Using an Automated Fast Protein Liquid Chromatography (FPLC) System...... 63 3 Representative Results ...... 66 4 Discussion ...... 71 Acknowledgements ...... 73 Disclosures...... 73 Chapter 4 Rationalised Residue Substitution of Trastuzumab for Enhanced Affinity to Target HER2 ...... 74 Abstract ...... 75 1 Introduction; Trastuzumab Affinity to HER2 for the Treatment of HER2 Positive Breast Cancer ...... 75 1.1 In silico Substitution Predictions and Analysis of Tmab ...... 77 1.2 Mutagenesis of Substitutions for Experimental Validation ...... 82 2 Methods ...... 83 2.1 Instrumentation, Services and Software ...... 83 2.2 Materials...... 83 2.3 Site Directed Mutagenesis ...... 84 2.4 Expression and Purification ...... 88 vii

2.5 HER2 Cell-binding Assay ...... 89

2.6 Binding Kinetics of Tmab Mutants to HER2 and FcγR1A ...... 89 3 Results ...... 90 3.1 Mutation Efficiency and Expression ...... 90

3.2 Affinity and Binding Kinetics to HER2 and FcγR1A ...... 93 4 Discussion ...... 97 5 Concluding Remarks ...... 101 Chapter 5 Rational Addition of Glycosylation Sites in Trastuzumab; Glycoengineering a Model Antibody for Enhanced Stability ...... 102 Abstract ...... 103 1 Introduction; Stability and Aggregation Propensity in Whole Antibody Therapeutics ..... 103 1.1 Strategic Addition of Glycosylation Sites ...... 105 1.2 Trastuzumab as a Model Antibody ...... 105 2 Methods ...... 109 2.1 Instrumentation, Services and Software ...... 109 2.2 Materials...... 110 2.3 Site Directed Mutagenesis ...... 111 2.4 Expression and Purification ...... 113 2.5 Mutations Confirmed through Mass Spectrometry ...... 114 2.6 Accelerated Thermal Stability and Monomer Loss Screen ...... 115 2.7 HER2 Cell-binding Assay ...... 116

2.8 Binding Kinetics of Tmab Mutants to HER2 and FcγR ...... 116 3 Results ...... 117 3.1 Mutation Efficiency and Expression ...... 117 3.2 Confirmation of Successful Mutation and N-glycan Addition ...... 121 3.3 Accelerated Thermal Stability Screening ...... 122

3.4 Retention of Biological Activity to HER2 and FcγRs ...... 127 4 Discussion ...... 132 5 Concluding Remarks ...... 135 Chapter 6 Enhancing the Stability of Adalimumab by Engineering Additional Glycosylation Motifs ...... 137 Abstract ...... 138 1 Introduction ...... 138 2 Materials and Methods ...... 141 viii

2.1 Materials...... 141 2.2 Cloning and Mutation of AdmAb WT and Variants...... 142 2.3 Expression and Purification of AdmAb WT and Variants ...... 142 2.4 LC-MS/MS Analysis of AdmAb Variants ...... 143 2.5 Accelerated Stability at Elevated Temperatures ...... 144 2.6 SE-HPLC Analysis of Monomer Loss ...... 145

2.7 AdmAb Melting Temperature (Tm) and Onset Temperature of Aggregation (Tagg) .. 145

2.8 Binding Kinetics (KD) to TNF-α and Fcγ receptors (FcγRs) ...... 146 3 Results ...... 147 3.1 Confirmation of Mutation and Glycan Attachment through LC-MS/MS...... 147 3.2 Monomer Loss following Accelerated Studies at Elevated Temperature ...... 149

3.3 Tm and Tagg ...... 149

3.4 Binding Kinetics of AdmAb to TNF- and FcγRs ...... 151 4 Discussion ...... 153 5 Conclusion ...... 155 Acknowledgements ...... 155 Chapter 7 Discussion; Outcomes, Reflections and Future Prospects ...... 157 1 Summary of Aims and Outcomes ...... 158 2 Establishing Lab-Scale mAb Expression Platforms ...... 159 3 In Pursuit of Modulating Target Affinity ...... 160 4 In Pursuit of Reducing Self-Association and Aggregation Propensity ...... 161 5 Concluding Remarks ...... 164 References ...... 165 Appendix ...... 184 1 E. coli Protocols ...... 184 1.1 Media and Culture Conditions ...... 184 1.2 Glycerol Stock Preparation and Resuscitation ...... 185 1.3 Calcium Chloride Preparation of Competent E. coli ...... 185 1.3.1 Reagents ...... 186 1.3.2 Calcium Chloride Method of Preparing In-house Competent JM109 ...... 186 1.4 Transformation of Competent Cells by Heat-Shock method ...... 186 2 DNA Protocols ...... 186 2.1 DNA Extraction Kits ...... 186 2.1.1 GenElute™ Plasmid Miniprep Kit ...... 187 ix

2.1.2 PureLink® HiPure Plasmid Midiprep Kit ...... 187 2.1.3 JetStar® LFU Plasmid Maxiprep Kit ...... 187 2.1.4 Genelute™ HP Plasmid Maxiprep Kit ...... 188 2.1.5 Nucleospin® DNA Purification kit ...... 188 2.2 Sample Preparation for Sequencing (AGRF) ...... 188 2.3 Agarose Gel Electrophoresis ...... 189 2.3.1 Agarose Gel Casting ...... 189 2.3.2 DNA Preparation for Loading into Gels ...... 190 3 Tissue Culture Protocols ...... 190 3.1 Maintenance Media, Reagents and Culture Conditions ...... 190 3.2 Cell Counting ...... 192 3.3 Suspension 293F / CHOS Culture Maintenance ...... 193 3.4 Suspension 293F / CHOS Cryopreservation and Resuscitation ...... 193 3.5 Adherent SK-BR3 / CHO-K1 WT Culture Maintenance ...... 194 3.6 Adherent SK-BR3/CHO-K1 WT Cryopreservation and Resuscitation ...... 194 3.7 Reagents for Protein Expression ...... 195 4 Antibody Purification and Characterisation Protocols ...... 196 4.1 FPLC Buffers and Protocol: ...... 196 4.2 Buffer Exchange, concentration and quantification ...... 197 4.3 Purity Determined by SDS PAGE Analysis ...... 197 4.4 Sequence and Glycosylation Confirmed by LC-MS/MS Analysis ...... 198 4.4.1 Reducing SDS PAGE Separation and Peptide Digestion ...... 198 4.4.2 LC-MS/MS Operation and Analysis ...... 199 4.5 Size Distribution of Antibody Samples Determined by SE-HPLC ...... 200 4.5.1 Reagents for HPLC ...... 200 5 Affinity and Binding Kinetic Assays ...... 201 5.1 ELISA ...... 201 5.1.1 ELISA Protocol ...... 202 5.1.2 ELISA Analysis ...... 202 5.2 HER2 Cell-binding Assay Reagents and Protocols ...... 203 5.2.1 SK-BR3 Cell Harvesting and Preparation ...... 204 5.2.2 Antibody Titrations and Cell Binding Assay ...... 204 5.2.3 Detection and Analysis with Flow Cytometry ...... 205 5.3 Binding Kinetic Assay Materials and Protocols ...... 206 6 Additional data for Chapter 4 and Chapter 5 ...... 209

6.1 LC-MS/MS Representative Data of Tmab Glycosylation Library ...... 209 6.2 SPR Sensorgrams of Tmab Glycosylation Library ...... 210 6.3 SPR Residual data of Tmab Affinity Library ...... 212 6.4 SPR Residual data of Tmab Glycosylation Library ...... 213 7 Additional data for Chapter 6 ...... 215 7.1 Primers; Mutation Efficiency of AdmAb Glycosylation Library ...... 215 7.2 LC-MS/MS Representative Data of AdmAb Glycosylation Library...... 215 7.3 SPR Sensorgrams of Monomer AdmAb Glycosylation Library ...... 222 7.4 SPR Results and Sensorgrams of AdmAb Glycosylation Library, pre-SEC Purification ...... 224 7.5 SPR Residual data of AdmAb Glycosylation Library ...... 227 7.6 Representative Thermal Melting and Aggregation curves of Monomer AdmAb Glycosylation Library ...... 231 Authorship Attribution Statement ...... 233 Poster and Publications ...... 234

List of Abbreviations aa amino acid

ADC antibody-drug conjugate

ADCC antibody-mediate antibody-dependant cell-mediated cytotoxicity

AdmAb adalimumab

AGRF the Australian Genome Research Facility

AH adalimumab heavy chain

AL adalimumab light chain

APR aggregation prone region

AUC area under the curve

BLAST Basic Local Alignment Search Tool bp base pairs

CDC complement-dependent cytotoxicity

CDR complementarity-determining region

CH1 constant heavy region 1

CH2 constant heavy region 2

CH3 constant heavy region 3

Chinge constant hinge region

Cjoin constant joining region

Cκ constant kappa region

DMARD disease-modifying antirheumatic drugs

DNA deoxyribonucleic acid

EDTA ethylenediaminetetraacetic acid

Fab antigen binding fragment

FBS foetal bovine serum

Fc fragment crystallisable

xii

FITC fluorescein isothiocyanate

FPLC fast performance liquid chromatography

GOI gene of interest

HER2 Human epidermal growth factor receptor 2 hr hour/s

IgG immunoglobulin isotype G

IgG1 immunoglobulin isotype G sub-class 1

IMGT the International ImMunoGeneTics Information System

IV intravenous kb kilobase pairs

LB Luria-Bertani mAb monoclonal antibody

MCS multiple cloning site

MD molecular dynamic min minute/s

MW molecular weight

NCBI the National Centre for Biotechnology Information

O/N overnight

PBS phosphate buffered saline

PCR polymerase chain reaction

PDB Protein Data Bank

PIPE polymerase incomplete primer extension

RE restriction enzyme rpm rotations per minute rSAP shrimp alkaline phosphatase

RT room temperature

SC subcutaneous

xiii scFv single-chain variable fragment

SDM site directed mutagenesis sec second/s

SE-HPLC size-exclusion high-performance liquid chromatography

SFM serum free media

SLS static light scattering

SP secretory signal peptide

TAE tris acetate EDTA

TH trastuzumab heavy chain

TL trastuzumab light chain

Tm melting temperature

Tmab trastuzumab

TNF tumour necrosis factor

Tris tris(hydroxymethyl)aminomethane

VH variable heavy region

Vjoin variable joining region

Vκ variable kappa region

VPA valproic Acid v/v volume for volume

WT wildtype w/v weight for volume

xiv

Thesis Abstract

The development of antibody based therapeutics has come of age, being a profoundly significant and dominant technology for the pharmaceutical industry. Antibody based therapies treat a plethora of indications including cancers, infections and autoimmune diseases, with many under development for new or improved treatments. The specificity of antibodies to biological targets and their ability to elicit an immune response are the primary functions behind the treatment strategy. Biotechnological advancements have led to the strategic design of fragment, drug conjugated and fusion antibody platforms along with modulated affinity to biological targets and immune effector functions, for the development of highly tailored and more efficacious treatment strategies.

To achieve improved antibody manufacturing outcomes and treatment strategies, considerations in drug design can address key challenges through the antibody discovery, production and formulation processes. Currently, the manufacture of antibody based therapeutics is costly, owing to production being limited to cell-based expression systems and loss of biologically active antibody in downstream processes that cause degradation and aggregation. Furthermore to achieve 100% bioavailability, antibody based therapeutics are currently limited to lyophilised and liquid based formulations for intravenous or subcutaneous delivery.

In considering the current technologies being applied to antibody based therapeutics, along with the challenges faced in manufacture, formulation and treatment strategies, this body of work was directed towards the strategic design of model antibodies trastuzumab and adalimumab, for enhancing target affinity and reducing self-association and aggregation propensity. Mutations were designed in trastuzumab’s Fab region for enhanced affinity to target HER2, and N-linked glycosylation sites introduced for the steric hindrance of colloidal self-association interactions. The hyper-glycosylated mutations were transferred to adalimumab to demonstrate proof-of-concept in this strategy of reducing self-association. Prior to mutational project design, appropriate expression systems for lab-scale production of trastuzumab and adalimumab were established. Mutant libraries were designed, produced, expressed and characterised prior to stability profiling and binding kinetic analysis against their relevant biological targets and Fcγ receptors.

The findings from this body of work demonstrated that in vitro assessment of affinity mutations corroborated with in silico predictions, highlighting the success of this approach for biobetter development through election of lead candidates by design. Furthermore, in vitro assessment of strategically hyper-glycosylated libraries produced several lead candidates with enhanced stability profiles. Moreover, the stability profiles were transferrable between the trastuzumab and

1 adalimumab mutants demonstrating the applicability and adaptability of this approach in engineering stability enhanced mAbs by design.

The implications from the outcomes of this body of work show great potential in the use of in silico analysis for the strategic and efficient development of biobetters, with the aim of improving manufacture and treatment outcomes by design.

Thesis Overview

A brief overview of the chapters in this thesis is as follows:

Chapter 1 is a comprehensive literature review that discusses the current knowledge and state of the art in the field of therapeutic antibodies, providing significant insight and grounding for the reasoning behind the projects pursued for this body of work. Chapter 2 reports on establishing an expression vector library for the lab scale production of trastuzumab and adalimumab, as model antibodies for engineering projects. Chapter 3 reports on the optimisation of expression and purification in the lead up to the design and implementation of antibody engineering projects for enhanced affinity (Chapter 4) and stability (Chapter 5 and Chapter 6). Chapter 7 is a comprehensive discussion of the results and outcomes from this body of work, further understanding and future perspectives that this work has contributed to current knowledge in the art.

A more detailed overview of the chapters in this thesis is as follows:

Chapter 1 discusses and highlights the clinical significance of antibody based immunotherapies, specific challenges, current platforms and future trajectories of the development of antibody based immunotherapies. Therapeutic antibodies trastuzumab and adalimumab are discussed specifically as model antibodies chosen for this body of work, representing current treatment strategies for cancer and autoimmune diseases. State of the art knowledge in antibody engineering is extensively and comprehensively discussed as a tool for addressing key challenges in the manufacture, formulation and administration of therapeutic antibodies. And finally, this chapter aims to highlight the rationale of the directions pursued in this body of work, of which trastuzumab and adalimumab are chosen as model antibodies for engineering to enhance affinity and stability for improving formulation and administration strategies.

Chapter 2 reports on the development of a library of expression vectors that carry the genes for expression of trastuzumab and adalimumab in mammalian cell culture. Upon successful cloning, antibody expression was compared between the vectors in a pilot transient expression system to ascertain the most appropriate expression systems to proceed with for the chapters following. This was the crucial starting point from which antibody expression conditions were optimised in the lead up to the design and production of antibody mutant libraries for in vitro testing.

Chapter 3 reports on establishing a high yielding stable trastuzumab expression system from which purification strategies were optimised, and further describes a simple format of adapting adherent cells to suspension in the lead up to expression. This protocol was developed to be simple, cost- effective and reasonably high-yielding to express, purify and characterise trastuzumab in 3 preparation of producing mutant libraries. Chapter 3 was published in the Journal of Visualized Experiments:

Elgundi, Z., Sifniotis, V., Reslan, M., Cruz, E., Kayser, V. (2017). Laboratory Scale Production and Purification of a Therapeutic Antibody. Journal of Visualized Experiments, 119: e55153.

Chapter 4 reports on trastuzumab being used as a model antibody to investigate rationally designed residue substitutions for enhanced affinity to target HER2. These substitutions were predicted to enhance affinity based on in silico results reported to us from our collaborators from A/Prof Serdar Kuyucak’s Computational Biophysics Group, School of Physics, The University of Sydney. Several QuikChange™ (Agilent Technologies) whole-plasmid mutagenesis strategies were compared and optimised to produce a trastuzumab affinity mutant library in the lead up to in vitro testing of relative binding affinity and kinetics. In vitro and in silico results were in agreement suggesting that computational tools are useful and powerful in predicting and rationally designing lead antibody candidates with a tailored specificity to target.

Chapter 5 reports on trastuzumab being used as a model antibody to investigate rationally designed residue substitutions that introduce N-linked glycans in the Fab region for enhancing stability and resistance to aggregation. Substitutions were identified as regions that had the potential to sterically hinder a reported aggregation-prone region. A trastuzumab glycosylation mutant library was produced through an optimised mutagenesis and transient expression system in the lead up to in vitro screening of stability enhancement and retention of biological functions. The library produced mutants of varying stabilities with 2 lead candidates elected for further investigation, demonstrating the effectiveness of this approach in designing antibodies with enhanced stability profiles.

Chapter 6 reports on demonstrating the proof-of-concept in stability enhancement from the glycosylation project reported in Chapter 5 by transferring the glycosylation mutations in adalimumab for comparison. As per the trastuzumab glycosylation study, an adalimumab glycosylation mutant library was produced through an optimised mutagenesis and transient expression system in the lead up to in vitro screening of stability enhancement and retention of biological functions. Chapter 6 employed additional purification strategies and more rigorous stability data to validate the interpretation of stability enhancement in the mutant library. Chapter 6 was prepared for publication, pending patent embargo, to the European Journal of Pharmaceutics and Biopharmaceutics:

Reslan, M., Sifniotis, V., Cruz, E., Sumer-Bayraktar, Z., Cordwell, S., Kayser, V. (2018). Enhancing the stability of adalimumab by engineering additional glycosylation motifs. European Journal of Pharmaceutics and Biopharmaceutics.

Chapter 7 summarises and discusses the findings of this body of work, in reflection of the aims and challenges addressed. The success of strategic engineering approaches applied in this fbody of work is highlighted, demonstrating the success of this approach in elucidating lead candidates by design for more efficient biobetter development. The achieved outcomes are reiterated with consideration and further understanding of what this body of work has contributed to current knowledge in the art. And finally, further investigation of the elected lead candidates are discussed for consideration of future perspectives in biobetter development and formulation strategies.

Chapter 1 Therapeutic Antibodies

Since thesis submission, Chapter 1 had been repurposed as a review article entitled

“Current Advancements in Addressing Key Challenges of Therapeutic Antibody Development, Manufacture and Formulation”

and is accepted for publication in Antibodies:

Sifniotis, V., Cruz, E., Eroglu, B., Kayser, V., Current Advancements in Addressing Key

Challenges of Therapeutic Antibody Design, Manufacture and Formulation. Antibodies (2019) 8,

36. DOI: 10.3390/antib8020036

1 Whole Therapeutic Monoclonal Antibodies; Introduction and Significance Since the first therapeutic monoclonal antibody (mAb) Orthoclone OKT3® was approved by the FDA in 1986, whole antibody therapeutics have persistently remained the most dominant and significant biologic therapeutic platform in the pharmaceutical industry [1, 2]. To date, therapeutic antibodies treat a plethora of indications including cancers, infections, autoimmune, cardiovascular and neurological diseases [3]. The development of therapeutic antibodies has been revolutionary due to the specificity of antibodies to a biological target and their ability to elicit a response that engages the immunological destruction of the target. The whole antibody therapeutics platform is regarded as the most promising class of pharmaceutical technology to date; it is continually being applied to newly identified biological targets and implemented in many formats to produce strategically engineered next generation antibody therapeutics, otherwise termed “biobetters” [4-9]. The international ImMunoGeneTics information system® (IMGT®) database reveals that as of December 2018, 65 whole antibodies and 19 next generation fragment or recombinant fusion antibody based therapies are approved for clinical use, with hundreds more in clinical trials expected to reach market.

Of the therapeutic whole antibodies approved for use, the majority (74%) are IgG1 isotype (Figure 1A) owing to its superior ability to activate the complement pathway and mediate antibody- dependant cell-mediated cytotoxicity (ADCC) and complement-dependent cytotoxicity (CDC) whilst maintaining a long half-life (approximately 23 days) in patient serum [4]. IgG2 and IgG4 therapeutic isotypes have also been developed despite their inferior immunological functions as compared to IgG1 and are thus applied as antagonists, blocking the activation of key disease bearing signalling pathways, and more recently as carriers for potent drugs that is ideal for their minimal engagement with immune effector cells [10-12].

Originally derived from animal origins such as mouse and rat, humanisation of therapeutic mAbs was the first revolutionary step to improving the treatment platform, as animal derived mAbs exhibit immunogenicity and are less efficient than human mAbs at activating the human immune system [13]. Humanisation has become standard practice for the development of new mAb therapies, which has led to the majority of whole therapeutic mAbs as either humanised (41%) or wholly human (44%) approved for clinical use (Figure 1B).

Whole therapeutic mAbs are presently the dominant antibody platform approved for clinical use (Figure 1C) although antibody engineering technologies have advanced in recent years to produce highly optimised, strategically engineered biobetter therapies, along with biosimilar mAbs reaching market to compete against their originator. Further whole mAb formats include antibody-drug conjugates (ADCs) that have exceptional potency, isotype switched and glyco-engineered whole 7

mAbs for modulated Fc effector function and bispecific mAbs with engagement to dual biological targets. Fragments of mAbs such as the crystallisable fragment (Fc), antigen binding fragment (Fab) and single-chain variable fragment (scFv) possess key functions such specificity to a biological target or immunological activation. The isolation of these fragments for fusion with other mAb

IgG1/κ

A IgG1/λ

IgG2/κ

IgG2/λ

IgG2-4/κ hybrid

IgG4/κ

B Wholly Human Humanised

Chimeric

Murine

Naked Whole mAbs C ADCs Whole mAb Bispecifics

Biosimilars

Fabs

Fab'

Fc Fusions

scFv Fusions

scFv Bispecific (BiTE)

Figure 1 Pie charts depicting the proportions of whole therapeutic mAb isotypes (A), whole therapeutic mAbs species (B) and therapeutic formats (C) approved for use as of December 2018, IMGT®. 8 fragments, biologically functional proteins, mediators, cytotoxic drugs or drug carriers has been the crux of ingenuity in developing the next generation of biobetter therapies [4-12, 14-16].

1.1 Trastuzumab; Model Antibody for Cancer Therapy Breast cancer is the most diagnosed and has the highest morbidity of all cancers in females worldwide, and is the second most commonly diagnosed of all cancers, posing a substantial burden on global disease [17]. HER2-positive breast cancer is diagnosed in approximately 20% of breast cancers and is a particularly aggressive cancer that can progress to metastasis [18]. The development of therapeutic whole antibody trastuzumab (Herceptin®) has been revolutionary in the evolution of HER2-positive breast cancer treatment whereby adjuvant treatment in combination with chemotherapy dramatically improves prognosis and overall survival [18-22]. HER2 is a cell receptor that is overexpressed in HER2-positive breast cancers and contains an intracellular and extracellular region. Trastuzumab is a humanised IgG1/κ mAb and acts by binding to the extracellular domain of HER2 to promote ADCC of tumour tissue for direct elimination [21] and downregulate the PI3K/AKT pathways that promote tumour cell survival and proliferation, thereby dampening the tumours aggression and making chemotherapy more potent [22]. However, the binding of trastuzumab to HER2 induces cellular uptake, internalisation and catabolism of trastuzumab into the breast cancer tumour cells, which limits the effect of ADCC [23]. Kadcyla®, a trastuzumab-drug conjugate takes advantage of this internalisation pathway to induce tumour cell death through the targeted delivery and release of cytotoxic drug mertansine into HER2 overexpressing cancerous tissue [24].

Despite the improved clinical outcomes of HER2-positive breast cancers through trastuzumab combination therapy, several challenges have arisen that require the need for additional treatment options. Resistance to trastuzumab therapy is observed in HER2-positive breast cancer patients either initially (de novo) or it becomes acquired over time which eventually leads to disease progression [25]. Furthermore, trastuzumab is reported to induce adverse side effects such as cardiotoxicity leukopenia, pulmonary toxicity and hypersensitivity [18, 20-22, 26-28]. Pertuzumab (Perjeta®) was developed as another whole mAb therapeutic that targets a different epitope of HER2 to produce a similar clinical effect and enhanced efficacy in combination with trastuzumab; however, cardiotoxicity was displayed in pertuzumab suggesting a correlation of this adverse effect with targeting HER2 specifically [21, 22, 29]. Small molecule drugs that inhibit HER2 have also been developed for use following resistance to trastuzumab, such as lapatinib (Tykerb®), afatinib (Gilotrif®) and neratinib (Nerlynx®); however, adverse side effects are more prevalent in these treatment strategies, making trastuzumab the preferred first line treatment in combination with chemotherapy [20, 21, 26, 30].

Trastuzumab was traditionally formulated as a lyophilised powder for intravenous (IV) administration, although more recently as a fixed dose in pre-filled syringe for subcutaneous (SC) injection, subject to FDA approval [31-33]. The IV formulation reconstitutes the lyophilised powder to a concentration of 21 mg/mL, to be further diluted for administration to 4 mg/kg as a loading dose, and 2 mg/kg for weekly maintenance doses [28]. In comparison, the fixed SC dose (600 mg, 5 mL) is administered once every three weeks, regardless of patient weight [33]. Between the formulation strategies, SC administration produces more adverse side effects such as infections that may be attributed to the mode of administration, although it does not require preparation and patients reported that they preferred SC administration [20, 33].

Trastuzumab is produced in a stable chinese hamster ovary (CHO) expression system, with gentamicin selection [28]. The entire trastuzumab peptide sequences for the heavy and light chains are publicly available on the IMGT® (2D ID: 7637) and DrugBank® (ID: DB00072) databases [34, 35]. Crystal structures of various trastuzumab fragments and mutant variants are publicly available on the Research Collaboratory for Structural Bioinformatics Protein Data Bank (RCSB PDB®) [36]. Of notable interest, crystal structure of the HER2 extracellular domain complexed to trastuzumab Fab (PDB ID: 1N8Z) is useful for in silico analysis of target binding.

1.2 Adalimumab; Model Antibody for Autoimmune Disorder Therapy Autoimmune disorders encompass a wide range of chronic diseases that are often associated with comorbidities; they pose a substantial global disability burden and are predominantly in the aging population [37]. Adalimumab (Humira®) is used to treat many types of autoimmune disorders and stands out globally as the single most lucrative drug, generating 18.4 billion US dollars in 2017 [1]. Many autoimmune disorders such as rheumatoid arthritis, osteoarthritis, psoriatic arthritis, Crohn’s disease and ulcerative colitis are hypersensitised to TNF-α levels making it a primary target for therapy [38-41]. TNF-α is a key cytokine produced and secreted in activated innate immune cells that induces systemic inflammation, in the context of autoimmune disorders inflammation is spontaneously activated and misdirected to healthy tissues leading to tissue degenerative diseases. Hypersensitisation to TNF-α could be due to a deregulation in TNF-α expression (TNF-α overexpression) or single-nucleotide polymorphisms in key downstream immune pathways causing a demodulated inflammatory response (increased sensitivity to TNF-α) [41, 42]. Adalimumab is one of five mAb based disease-modifying antirheumatic drugs (DMARDs) that specifically targets TNF-α and have been revolutionary in the treatment of autoimmune diseases as they elicit a rapid and sustained suppression of TNF-α induced inflammatory disease. Adalimumab acts by binding to both the secreted form of TNF-α (soluble homotrimeric) and to membrane-bound TNF-α to promote ADCC for direct elimination and block downstream inflammatory pathways [43]. 10

Adalimumab is a wholly human IgG1/κ therapeutic whole mAb, developed as a second generation antibody from its chimeric predecessor infliximab (Remicade®). Along with adalimumab and infliximab, additional mAb based anti-TNF-α DMARDs include wholly human mAb golimumab (Simponi®), recombinant Fc fusion protein to a TNF receptor (CD120b) etanercept (Enbrel®) and PEGylated humanized Fab certolizumab pegol (Cimzia®); however, adalimumab remains the dominating anti-TNF-α DMARD treatment option [40, 43, 44].

As discussed with trastuzumab, resistance to treatment is an ongoing problem relating to all mAb therapies, which is more pronounced with chronic autoimmune diseases that require ongoing treatment until remission, with the potential to relapse [45, 46]. Adverse effects in adalimumab therapy, as with all anti-TNF-α based therapies is the increased risk of infection and delayed wound healing due to the dampened immune system, and a higher incidence of lymphoma. To mitigate risk of these challenges, mAb therapeutics are not issued as a first-line treatment and are primarily considered in combination with current effective small molecule DMARD therapies for a superior recovery strategy. In the instance of resistance to treatment over prolonged exposure, switching between mAb therapies is a common practice to prolong efficacy [40, 41, 44].

Adalimumab is produced in a CHO expression system and is formulated as a fixed dose (40 mg, 0.8 mL) in pre-filled syringe for SC injection, administered once every other week, regardless of patient weight [47]. The entire adalimumab peptide sequences for the heavy and light chains are publicly available on the IMGT® (2D ID: 7860) and DrugBank® (ID: DB00051) databases [34, 35]. Crystal structures of various adalimumab fragments are publicly available on RCSB PDB® [36], notably TNF-α complexed to adalimumab Fab (PDB ID: 3WD5) for in silico analysis of target binding.

2 Pharmaceutical Development Challenges and Considerations Whole therapeutic mAbs require a mammalian expression system to produce the biologically functional product; however, mAb fragment and recombinant fusion mAb products with simplified (or lacking) glycosylation are suitable for lower organism expression platforms [48]. Unlike oligopeptides that can be chemically synthesised, whole therapeutic mAbs are considerably larger (monomer ranging from 140 – 160 kDa) and comprise of four peptide chains (two heavy and two light chains) bound together by disulphide bonds. Further to this, antibodies contain glycosylation in a conserved region of the Fc (N297) that contributes to its stability and immune effector functions [49]. Aside from peptide synthesis, cell machinery is required to glycosylate, fold, orient and covalently bind the antibody peptide chains in order to produce the complete, biologically functional antibody product for secretion and purification. Manufacturing biologically functional whole mAb product is therefore commercially unfeasible through chemical synthesis, and insufficient in lower organism expression platforms such as bacteria, yeast, insect and plant cells 11 that do not produce the equivalent tertiary structure and glycosylation profiles. In particular, bacteria are completely deficient in the machinery to add post-translational glycosylations, yeasts hyper-mannosylate glycans which causes immunogenicity, insect cells are deficient in sialylation machinery and produce immunogenic glycan structures [48]. Secretion of the mAb product for purification is suboptimal for several lower organism expression platforms such as E. coli due to poor productivity and harsh culture conditions promoting product degradation. Protein is therefore produced intracellularly as inclusion bodies and harvest involves further processing steps such as cell lysis, inclusion body recovery, protein solubilisation and renaturation prior to further downstream purification steps [50, 51]. Despite these pitfalls, development of lower organism expression systems is of high commercial interest due to the simplified culture conditions, cheaper media requirements, rapid organism growth and higher product yield as compared to mammalian expression systems.

In considering the requirements through the entire process of mAb discovery, manufacture, formulation and disease treatment, several key challenges arise which have sparked overwhelming interest in pursuit of achieving better mAb manufacturing outcomes and treatment strategies. As with all biotherapeutics, mAbs and mAb based therapeutics are limited to production in cell-based expression systems which is considerably costly and inefficient, yield can vary depending on the product and expression system, and downstream processing is necessary to remove biological contaminants introduced from the expression system. Despite affinity chromatography being a robust technology for the initial capture of mAb for purification, the capture process and further downstream processes such as viral inactivation applies the mAb product to harsh pH and salt conditions, which can chemically degrade the mAb leading to product instability and loss [5, 52, 53]. A common challenge as seen with all biotherapeutics, mAbs and mAb based therapeutics are currently restricted to lyophilised and liquid based formulations for IV or SC delivery to achieve 100% bioavailability. Protein self-association and intrinsic stability drive this limitation, in that viscosity and propensity to aggregate is dependent on mAb concentration. Formulations are optimised to achieve the highest dosing concentration (minimum achievable volume) without compromising the quality of the mAb in formulation. Viscosity remains a key limiting factor for formulating as a SC administration, certain mAb therapies are suitable and others not based on their solubility, self-association and aggregation profiles. Alternative non-invasive administration strategies such as pulmonary delivery causes additional mechanical stress that contribute further to mAb instability and loss, and oral delivery is unsuitable due to chemical degradation and poor absorption in the gastric and intestinal environments [5, 54-58].

2.1 Discovery and Manufacture Traditional technology for whole therapeutic antibody discovery required the immunisation of animals, primarily mice with a target antigen for the generation of a mixed population of B- lymphocytes producing antibody against the target. The B-lymphocytes would be isolated for immortalisation to produce monoclonal hybridoma cell lines that secrete antibody candidates of interest, for further panning through display technologies to isolate potential leads [5]. This method led to the generation of highly specific mAbs of non-human origin that were potentially immunogenic and less efficacious at eliciting an immune response as compared to wholly human mAbs. Technologies arose to address immunogenicity by producing chimeric and humanised antibodies through grafting of the variable domains (chimeric) or complementarity-determining region (CDR) residues (humanised) isolated from lead non-human antibodies to human antibody framework. To further improve the humanisation strategy, transgenic mice were developed with their murine antibody heavy and light chain genes replaced with equivalent human genes, consequently expressing wholly human antibodies for discovery [4-6].

Phage display technology remains the most prominent selection technology for panning antibody gene pools for specificity to a target antigen, it effectively couples the expression of mAb proteins to the genes that encode them for panning of high affinity leads. mAb variable genes from B- lymphocyte pools are isolated through PCR, cloned into a phage expression vector that presents the genes on the surface of the phage, the library of vectors is rescued in phage and the phage library screened against a specific target antigen. Phage pools undergo several rounds of selection to isolate high affinity candidates, the variable genes of the lead mAb candidates can then be isolated and sequenced for the design of a mAb drug format and development of a suitable expression platform [5, 59].

Expression from hybridoma cell line to cell line can vary quite substantially, so to meet the commercial need for high expression efficiency, scalability, quality and reproducibility, high yielding mammalian expression systems have been developed. mAb genes of interest are introduced into suitable expression vectors and transfected into highly efficient mammalian cell lines for antibody expression and secretion; mAb can then be directly captured from the culture supernatant through affinity chromatography. Currently the majority of mammalian expression systems for commercial whole therapeutic antibody expression are based on stable chinese hamster ovary (CHO), mouse myeloma (NS0) and mouse hybridoma (Sp2/0) cell lines [4, 5]. However, in the context of research and development, transient expression in human cell lines such as embryonic kidney (HEK 293), amniotic (CAP), hybrid of HEK 293 and lymphoma (HKB-11) and embryonic retina (PER.C6) are favoured over stable CHO expression due to the ease and speed of production

13 of workable quantities of antibody for preliminary studies [48, 60-64]. Candidates that show potential for further investigation would then be developed for high yielding stable expression and more rigorous characterisation leading up to therapeutic development. However, a drawback from changing from human to hamster expression systems is the difference in glycosylation profile. CHO expressed mAbs produce immunogenic non-human glycoform Neu5Gc and have a higher composition of sialylation which results in reduced ADCC. Therefore in the early stages of development, leads need to be identified and thoroughly characterised in the expression system to be developed for commercial manufacture before committing to industrial scale up [48, 60, 61]. Amongst other biotherapeutics, approved mAb Fc fusion therapeutics dulglutide (Trulicity®), efmoroctocog alpha (Eloctate®) and eftrenonacog alpha (Alprolix®) are manufactured in a HEK 293 expression system, which is giving rise to the acceptance of human-based expression systems for the production of mAb based therapeutics [4]. However, HEK 293 expression systems are prone to inducing mAb aggregation in culture, which is detrimental to cell viability and loss of product during manufacture, hence HKB-11 and PER.C6 are the preferred commercial human cell line for mAb expression [48, 60, 62].

Established lower organism expression systems for fragment or recombinant fusion mAb therapeutics include bacteria such as E. coli, yeast such as S. cerevisiae and P. pastoris, plant such as tobacco, algae and insect such as silkworm [5, 48, 65-67]. Expression platforms that use E. coli in particular are considered high risk due to potential endotoxin contamination in the mAb product; however, several approved mAb fragment and recombinant based therapies are produced in an E. coli based expression system including pegol conjugated Fab’ certolizumab pegol (Cimzia®), Fab ranibizumab (Lucentis®) and recombinant Fc fusion romiplostim (Nplate®). An emerging in vitro cell-free synthesis technology is being developed with bacterial and CHO cell lysates which have the potential to alleviate formation of undesirable biological byproducts, as the machinery from the cell lysates purely express protein from the mAb genes that are introduced to the system. Although this technology is not currently applicable for industrial scale manufacture, it holds much promise as an alternative to live culture in that culture maintenance and highly defined media are no longer necessary, reactions can run continuously or lysates can be recycled with reproducible results, unmodified linear DNA is suitable for the system alleviating a need for multiple cloning steps and additional enzymes can be introduced to the system for engineering specific post-translational modifications [68-73].

Current bioprocess engineering and cell line development strategies have been crucial in enhancing the manufacturability of mAbs. Many current mAb therapeutic expression platforms make use of CHO cell lines that have been commercially developed with dihydrofolate reductase or glutamine

14 synthetase deficiency, for enhanced stability selection through resistance to methotrexate and methionine sulfoximine inhibition, respectively [64, 74]. Mammalian cell lines have been adapted to suspension culture as it supports higher cell densities and mAb titre in the absence of serum from the culture media, for large scale fed-batch, perfusion or continuous culture systems. Further to this, cell lines are continually being engineered for enhanced metabolic functions, introducing glycosylation pathways, superior secretion and resistance to apoptosis for prolonged survival, in efforts to generate super-producer cell lines that can sustain continuous culture [75-81]. mAb gene and expression vector design and development are specifically tailored to the expression system to maximise mAb expression and cell line stabilisation, through host cell codon optimisation and the addition of highly efficient transcription, secretion, selection and integration elements [82- 86]. Vectors can contain a single site for gene insertion, for subsequent co-transfection of vectors with that carry the antibody heavy and light chains, or both chains can be cloned into a dual expression vector. In more elaborately designed recombinant mAb drug formats that require expression of several genes, vector technologies have implemented multicistronic expression to enhance the efficiency of the vector system. Stable or transient transfection of vectors is performed on highly viable cell cultures with high cell density. Antibody heavy and light chain ratios may require adjustment in transfection for optimising expression and secretion, a reporter vector that expresses green fluorescent protein is typically included to observe transfection efficiency during optimisation [87, 88]. Liposome mediated transfection is preferential over all other mammalian transfection strategies, induced chemically through the complexing of DNA to a cationic lipid prior to or during addition to culture. Polyethylenimine is the most prevalently used transfection reagent despite the development of superior reagents such as Lipfectamine™ 2000, Freestyle™ MAX, LyoVec™, FuGENE 6™ and TransIT-PRO®, owing to its lower cost and relative efficiency [86, 88- 91].

The commercial manufacture of mAbs have transitioned to serum-free, chemically defined, animal- free media which has removed a source of expression variability and has been driven particularly from the advent of mad cow disease [64, 92]. Several media supplements such as plant and yeast digests (peptones / hydrolysates), surfactants (e.g. Pluronic F-68), DNA methyltransferase (azacytidine) and histone deacetylase (sodium butyrate, valproic acid) inhibitors have been found to support cell viability and enhance mAb expression [92-97]. Mild hypothermia of the culture has also demonstrated to reduce cell expansion, support prolonged cell viability and enhance mAb expression [86, 98-102]. Further to this, continuous supplementation in expression culture of uridine, manganese chloride and galactose demonstrated successful glycan engineering of expressed mAb, where mAb hyper-galactosylation was promoted to enhance CDC activity [103].

In the manufacturing process of fed-batch, perfusion and continuous culture systems, media is continuously fed in the culture system whilst parameters such as cell viability, cell density and metabolite levels are monitored in real time until the parameters indicate that it is optimal time to harvest the expressed mAb from the culture supernatant. In perfusion and continuous culture systems, a feed is removed from the culture simultaneously to the media addition, the difference being that the cell dilution rate in continuous culture is optimised to remain equal or higher than the cell growth rate, which allows the perpetuation mAb expression and requires continuous harvest of the supernatant [104-106]. Harvest of the supernatant requires clarification technologies such as the use of precipitants / flocculants (e.g. polyethylene glycol, diethylaminoethyl dextran, caprylic acid and polyethylenimine), high throughput centrifugation and filtration methods to remove cells and biological debris, for lowering the loading burden in the lead up to mAb capture through affinity chromatography [50, 107-112]. Centrifugation and filtration causes unavoidable mechanical stresses which induces mild mAb product loss through fragmentation and aggregation.

Protein A-ligand based affinity chromatography is the most robust, efficient and prevalently used mAb capture technology, owing to its selectivity and high affinity to various human Fc, and efficient dissociation of captured mAb at low pH for reuse. mAb fragment and recombinant formats based on Fabs and single chain variable fragments (scFv) are unable to take advantage of Protein A affinity capture and alternatives have been developed such as capture to Protein G, M and L which bind at different mAb epitopes, ion exchange chromatography and polyhistidine tagged capture through immobilised metal chromatography [112-114]. A further advantage of affinity chromatography in manufacture is the integration of a viral inactivation step by holding the low pH eluted mAb product prior to further purification steps; however, this applies the mAb product to low pH at high concentrations which can induce mild mAb instability, aggregation and formation of acidic variants [5, 52, 112, 115, 116].

Post mAb capture, further polishing steps are required to remove further contaminants such as host cell proteins and DNA, leached affinity chromatography ligand and mAb degradation products such as fragments, aggregates and ionic variants. The polishing steps are designed based on the specific properties of the mAb product such as the isoelectric point and molecular weight (MW), to implement appropriate chromatographic technologies such as anion and cation exchange, hydrophobic interaction, multimodal and size exclusion that will separate the mAb product from the manufacture-introduced impurities [52, 112, 115]. The additional chromatographic steps unavoidably apply the mAb product to buffers of varying ionic strength and high concentrations, which can again induce mild mAb instability and aggregation. Post polishing steps, the purified mAb product undergoes a further viral removal step such as filtration, then buffer exchange and

16 concentration through ultrafiltration or diafiltration methods to prepare the bulk mAb product for formulation [112, 117].

2.2 mAb Drug Design and Formulation Strategies to improve formulation of mAbs and mAb based therapies continues to be an ongoing challenge, as is faced with all biotherapeutics. Firstly, whole therapeutic mAbs are unsuitable for non-invasive oral, nasal or pulmonary routes of administration as they are susceptible to chemical and enzymatic degradation in the gastrointestinal tract and bioavailability through these routes is poor, owing to mAbs polar surface charge and relatively large MW limiting transport through mucosal membranes. Fragment mAb platforms together with excipient and PEGylation technologies have been developed to help circumvent transportation limitations for pulmonary delivery specifically, although physical stresses applied to mAb (e.g. shear stresses from aerosolisation of liquid formulations for pressurised meter dose and nebulisation delivery, or from production of dry powder for inhalation) pose further challenges of mAb instability leading to degradation and reduced efficacy. [8, 55, 57]. Few biologics have been successfully approved for pulmonary delivery, most notably insulin formulation Afrezza®, and currently an erythropoietin-Fc fusion and a nanobody targeting respiratory syncytial virus are in clinical trials, showing promise for further development of this administration strategy for both systemic and localised delivery of mAb based therapeutics [5, 57]. Several drug carrier technologies such as microencapsulation, liposome and nanoparticle formulations inherently enhance the stability and control the release of mAbs which can prolong their half-life. These nanocarrier formulation strategies are of intense interest as they hold promise for developing less invasive inhaled formulations of mAb based therapeutics with superior attributes to currently established formulations [118].

The majority of currently approved mAb therapeutics are formulated for IV although commercial interest has directed technologies to develop injectable mAb formulations which has seen success in many approved mAb therapies thus far, primarily SC for systemic delivery, along with few intravitreal and intramuscular formulations for tissue specific indications or larger dosing volume, respectively. Injectable administrations offers several advantages over IV administration, especially in the treatment of chronic disease in regards to reduced burden to allied health services, patient tolerance and adherence to treatment; however, the intrinsic physicochemical properties of mAbs may be undesirable for injectable formulation. In considering the low volume, injection pressure and the typically high (> 100 mg) effective dose of mAb for injectable delivery, viscosity and aggregation propensity of mAbs become key formulation challenges to address as they are dependent on mAb concentration. Certain mAb therapies are suitable and others not based on their solubility, viscosity, self-association, intrinsic stability, aggregation and precipitation profiles. mAb

17 injectable formulations can further be improved with the use of excipients to increase solubility, reduce viscosity and enhance the stability of mAbs, and co-formulated with recombinant human hyaluronidase specifically as a permeation enhancer for more efficient absorption into tissue, although the inclusion of this additional biologic adds further burden to the formulation’s viscosity and propensity to aggregate [5, 8, 33, 54-57, 119-121]. mAb therapies for IV administration are prepared as lyophilised powder for reconstitution and further dilution, and injectable administrations as liquid based formulations in pre-filled syringes. Liquid formulations of mAbs are more susceptible to physiochemical degradation, are less stable and have a reduced shelf-life as compared to lyophilised formulations. However, drying technologies to produce lyophilised formulations apply the mAb to physical stresses that again induce instability and degradation [122- 124]. mAb therapies are currently restricted to invasive parenteral routes of administration for 100% bioavailability and systemic distribution, although local delivery to the specific target tissue such as tumours for cancer treatments continues to be a challenge. Penetration to tissue from blood vessels is again poor, owing to mAbs polar surface charge and relatively large MW limiting transport through physiological barriers and efficiency of tissue penetration is further influenced by systemic and local mAb clearance rates. Enhancement to mAbs affinity to their biological target through maturation and strategic mutation technologies have led to faster diffusion rates of mAb to target for increased efficacy. However, for tumourous tissue specifically, penetration through tumours is restricted by the tumour’s binding site barrier, in which antigens expressed on the tumours periphery capture the majority of mAb released to surrounding tissue and this effect is exaggerated with overexpressed antigen. Enhancement of whole mAbs affinity beyond 1 nM has specifically shown that there is no further improvement of tumour diffusion rates and tissue penetration and accumulation [125-127]. Increased affinity of mAbs to cellular targets also leads to increased uptake, internalisation and catabolism of mAbs which reduces ADCC and increases mAb clearance rate, this mechanism is exploited through the development of high affinity ADCs to target the delivery and release of potent drugs, inducing targeted cell death [125, 128, 129].

Fragment mAb platforms have shown better tissue penetration and biodistribution than whole mAb therapeutics; however, a pitfall of peptides lacking an Fc region is the highly reduced in vivo half- life and poor retention times. Technologies to improve the half-life of mAb fragments have been primarily through PEGylation, along with strategic Fc mutation and glycosylation engineering to enhance neonatal Fc receptor (FcRn) recycling. Furthermore, hyper-glycosylation technology has been applied successfully in other biotherapeutics, such as with glycosylated erythropoietin Aranesp®. Conversely, nanocarrier platforms that are relatively larger as compared to whole mAbs

18 have demonstrated superior pharmacokinetics and tumour retention, as well as previously mentioned enhanced stability and sustained, controlled release of mAbs. These enhanced properties have generated considerable interest in developing systemic delivery of mAb-nanoparticle platforms for superior tumour penetration, targeted and sustained therapeutic drug delivery. Further to nanocarriers, additional formulation strategies developed for the controlled release of mAbs include hydrogels and crystalline antibody, which have shown success as stable, injectable formulations for development [5, 8, 16, 23, 118, 125, 127, 130, 131]. In addition, to further enhance the stability and half-life of other biotherapeutics, recombinant technologies allowed their fusion to mAb Fc as to introduce FcRn recycling as a protection mechanism, which has innovatively expanded the applicability of mAb based therapeutic platforms [15, 126, 132].

Modulation of mAbs effector functions through isotype switching, glycoengineering and strategic mutations have proved advantageous for the development of more effective mAb treatment strategies. mAb binding to Cq1 promotes the complement cascade, FcγR1/BA, FcγR2A and

FcγR3A/B receptors activate immune effector functions, FcγR2B counter-balances the effector response and FcRn prolongs mAb half-life, mAb binding to these receptors is primarily through sites in the Chinge, CH2 and the conserved glycosylation region in the Fc (N297). IgG2 and IgG4 mAbs have reduced effector function as compared to IgG1, and are thus used in instances where minimal engagement to the immune system is warranted to increase safety of the mAb therapy, for instance with ADCs reducing off-target cytotoxicity. Eculizumab (Soliris®) is the first (and thus far only) approved recombinant IgGκ 2(CH1/Chinge) – 4(CH2/CH3) hybrid whole mAb, and further cross-sub-class variant mAb therapies are in development with key amino acid substitutions from the IgG2 and IgG4 sub-classes introduced for intentionally suppressed effector functions [4, 12]. Removal of the conserved glycosylation site through amino acid substitution of N297 or T299 (aglycosylation) has also demonstrated reduced effector function; however, at the expense of mAb stability and is therefore not a suitable strategy for whole mAb therapeutics [10, 15, 16, 133-135]. On the other hand, enhancing ADCC and CDC is useful for engaging the immune system to tumour tissues in the absence of a conjugated cytotoxic drug. Certain glycoengineered modifications to the conserved glycosylation region in the Fc such as deficiency in fucose (afucosylation) and hyper- galactosylation have demonstrated enhanced mAb binding to FcγR3A specifically for an enhanced ADCC effect [12, 125]. Afucosylated mAbs benralizumab (Fasenra™) and mogamulizumab (Potleigeo®), and low fucose content mAb obinutuzumab (Gazyva®) are currently approved therapies, with several further in development which demonstrates the commercial interest and applicability of this technology in improving mAb therapeutics through enhancing ADCC. However, the significance of manipulating sialyation in mAb therapeutics is a controversial topic,

19 with several reports demonstrating a reduction in ADCC and CDC, whilst others reporting no observable difference, highlighting a clear need to characterise and define the in vivo efficacy of such manipulation [10, 12, 15, 125]. Although no currently approved mAb therapies contain amino acid substitutions for improved half-life through modulated binding properties to FcRn, improved CDC through binding to complement factor Cq1 or improved ADCC through enhanced binding affinity to FcγR1A and 3A, several mutations have been reported and are in development, demonstrating the commercial interest in this technology for improving mAb therapeutics [6, 12, 15, 16, 126, 133, 136-138].

And finally, the stability profiles of mAbs can be greatly affected by the amino acid composition, structure and potential glycosylation in their variable region CDRs that are isolated though the screening and maturation process. Elucidating the causes for reduced solubility or increased aggregation propensity have to be considered together with the molecular interactions between the mAb to biological target as to not compromise affinity. Computational tools have been developed to elucidate amino acids within the mAb CDR structure that have a high propensity to aggregate, and strategies for improving solubility and aggregation profiles have been through direct amino acid substitutions and strategic addition of glycosylation sites [139-143]. Characterisation of mAb stability and target interaction for optimisation is considered a fundamental step as part of preliminary drug discovery, as the applicability for development, manufacture and formulation of the mAb therapeutic is governed by the mAbs stability profile.

3 Computational Approaches for Prediction and Rational Design of mAbs The advancement of in silico analysis of mAb peptide sequences, structures, conformation and their associated biological interaction has been integral to the development of various computational tools for characterising, designing and optimising mAb based therapeutics. The generation and continued pursuit of mAb structural data for in silico analysis has led to the development of several integral databases, notably IMGT®, DrugBank® and PDB® serving as key resources for data mining [34-36, 144-146]. Molecular dynamic (MD) simulation analysis has driven the computational analysis of the discovery and characterisation of relevant molecular interactions within the mAb molecule, mAb binding to biological targets and mAb surface association with the surrounding environment. These interactions can infer intrinsic stability, target binding associations, solubility and aggregation propensity of the mAb. MD simulations and free energy calculations from crystal structures remain key for the highly specific elucidation of mAb-target intramolecular interactions that correlate to binding affinity, as significant interactions such as hydrogen-bond formation can be predicted and determine the strength of the molecular associations [147]. However, elucidation of mAb self-association, solubility and aggregation propensity have been driven by the development

20 of computational modelling and simulation tools that simultaneously analyse mAb topography and surface polarity. Notably, AGGRESCAN3D, TANGO and PASTA are the most prominent tools for predicting site specific aggregation propensity of mAbs [148, 149].

The preliminary elucidation of mAb structural data, that being amino acid sequences and higher order structure (HOS), is experimentally derived to produce crystal structures that model the solid- state 3D structure of the mAb. Mass spectrometry technologies have come of age, to produce high throughput and orthogonal analysis to elucidate mAb peptide sequences, oxidation, deamidation and glycosylation heterogeneity, and more recently for native, destabilised and aggregated HOS elucidation [150, 151]. X-ray crystallography technologies have been the underlying workhorse for elucidating the crystal structure of mAbs. Producing crystalline mAbs is challenging owing to the complexity of mAb HOS and the degree of mAb conformational heterogeneity. However, cryogenic-electron microscopy and nuclear magnetic resonance spectroscopy have evolved as a complementing technologies to ascertain structure and interactions in recent years [149, 152-155]. Thus far, only four whole IgG antibodies have a successfully determined crystal structure (PDB ID: 1HZH, 1IGT, 1IGY and 5DK3), notably 1HZH as a wholly human IgG1, and 5DK3 as a humanised IgG4/κ, which have been used as model structures for MD simulations [156]. mAb fragments yield much higher success with producing crystal structures, which has led to much pursuit and contribution in generating mAb fragment and target complexing crystal structure libraries for data mining and in silico analysis [36, 157].

Upon obtaining relevant crystal structures for a candidate mAb, detailed crystal structure and MD simulation analysis has been extensively used to elucidate the mAbs intermolecular interactions that confer intrinsic stability and intramolecular interactions to confer interactions to biological targets and self-association. The particular binding interface of interest is analysed, amino acids in the mAb structure are identified for substitution that can disrupt molecular interactions in the interface in the instance of diminishing target binding, or identify potentially unutilised bonding sites and polarity mismatches in the interface that can be improved in the instance of enhancing target binding. MD simulations are then carried out for the native and substituted mAb structures to elucidate any relevant bond formations and more favourable binding free energy produced by the substituted structure, of which promising candidates can be synthesised and validated through in vitro analysis. This predictive approach has been successfully applied to identify mutations in mAbs for modulating effector functions, target binding and optimise affinity capture for manufacturing, and more recently for designing antibodies de novo from targets of interest [15, 158-162]. mAb mutations have been specifically identified for enhanced binding to FcγR2A, FcγR3A and Cq1, and reduced binding to FcγR2B for a more pronounced ADCC and CDC effect, enhanced binding to

FcRn for extended half-life, and reduced binding to FcγRs and Cq1 for a diminished ADCC and CDC effect, all of which are transferrable to mAbs of the same isotype sub-class. Further to this, the molecular interactions of glycoengineered mAb variants to FcRs have been characterised through MD simulations to corroborate predictions with the modulated effector functions observed [12, 16, 133, 136, 137]. mAb self-association, solubility and aggregation propensity are further evaluated by computational modelling tools that specifically characterise the topography and surface polarity of mAbs, concurrently analysing the local spatial arrangement of amino acids in the mAb structure, solvent accessibility and local surface charge [5]. In particular, surface exposed hydrophobicity is sought as the lead mechanism for protein self-association driving aggregation propensity, in which aggregation-prone regions (APRs) are identified in the mAb structure. Substitution of identified APRs to gatekeeper (i.e. polar) amino acids has experimentally demonstrated resistance to aggregation and improved solubility, which validates this strategy for improving mAb stability [163-170]. The majority of APRs identified are located in biologically relevant mAb regions, that being the CDRs and the Fc. Specifically targeting these APRs through mutation may disrupt important biological functions and mAb structure. Analysis of the mAbs intermolecular interactions are necessary to validate the intrinsic stability of substituted mAb variants, if conformational fluctuations are elucidated then the biologically active interface is likely disrupted. Interestingly, mutations for enhancing biological functions have demonstrated a negative impact on mAb solubility and stability, further suggesting that the self-association profile of mAbs is linked to its biological activity [139, 165, 171].

Aside from direct residue substitution in mAb structures, other strategies reported to profoundly interfere with the effects of APRs include isotype switching and the strategic addition of N-linked glycans [139, 168, 172, 173]. Additionally, N-linked glycosylation in the CDR regions of mAbs although uncommon, have demonstrated improved solubility and stability profiles of mAbs without impacting their target affinity, as compared to the same mAb with removed CDR glycans [139, 174]. The rationale behind strategic glycan addition is based on the election of N-linked glycosylation sites that have apparent spacial proximity to APRs, the introduced glycan therefore sterically hindering the APR from self-association interactions. The benefits of extending mAb half- life with strategic glycan addition as seen with hyper-glycosylated biotherapeutics, as well as improving mAb solubility and stability for improved formulation strategies, has yet to be realised and suggests a very intriguing and highly relevant technology for perusal.

4 Perspectives, Aims and Objectives In considering the current technologies being applied to mAb based therapeutic platforms, along with the challenges faced in manufacture, formulation and treatment strategies, this body of work was directed towards the strategic design and development of antibodies for enhancing target affinity, and reducing self-association and aggregation propensity. The modulation of a mAbs affinity to its target is an imperative part of the drugs design, where mAb fragment and ADC platforms take advantage of high target affinity. For diseases such as cancers, tumour tissue penetration, retention and immunological engagement are necessary considerations for improved treatment design and take advantage of mAb platforms with reduced target affinity (although retained specificity), including nanocarrier delivery strategies. Further to this, mAb self-association and aggregation propensity seems an ongoing challenge throughout the manufacture process leading to product loss and reduced quality, and through limiting formulation strategies for efficacious treatment. Specifically designed modifications in a mAbs framework for considerable stability enhancement can be transferrable to other mAb platforms in the same isotype sub-class, and would be of great commercial benefit for the whole class of biotherapeutics.

In this body of work, trastuzumab and adalimumab were elected as model antibodies for the design, expression and in vitro analysis of mutational projects for enhanced affinity and reduced self- association. Specifically, mutations were designed in trastuzumab for enhanced affinity to target HER2 and N-linked glycosylation sites introduced for steric hindrance of APRs, the glycosylation mutations transferred to adalimumab to demonstrate proof-of-concept in reduced self-association. Prior to mutational project design, appropriate expression systems for lab-scale production of trastuzumab and adalimumab were established. Several cloning strategies were implemented as to ascertain the most appropriate expression vector platforms for optimising antibody expression and purification, prior to mutagenesis optimisation and the expression of mutant libraries. Mutant libraries were designed based on in silico analysis of crystal structures, the trastuzumab affinity mutations were based on results from our collaborators from A/Prof Serdar Kuyucak’s Computational Biophysics Group, School of Physics, The University of Sydney, and the glycosylation mutations based on literature search of reported APRs. Mutant libraries were then produced and characterised before in vitro assessment of modulated target affinity, comparison of retained biological functions and stability profiling, to determine lead candidates for further investigation.

Chapter 2 Vector Design and Cloning for Expression of Trastuzumab and Adalimumab

Scope and Aims This chapter reports on the development of a library of expression vectors that carry the genes for expression of trastuzumab (Tmab) and adalimumab (AdmAb) in mammalian cell culture. Upon successful cloning, antibody expression was compared between the vectors in a pilot transient expression system to ascertain the most appropriate expression systems to proceed with for the chapters following. This was the crucial starting point from which antibody expression conditions were optimised in the lead up to the design and production of antibody mutant libraries for in vitro testing.

Here we report on the cloning strategies applied to the assembly of Tmab and AdmAb expression vectors for expression in transient 293F culture. Tmab genes were cloned into a library of vectors and the expression yield was compared in a preliminary transient expression experiment. Variable region gene switching was then performed on the vector library to produce AdmAb expression vectors which was again expressed transiently for comparison. The cloning strategies for introducing genes into vectors was relatively simple in design, had low background and error rate to yield successfully cloned isolates. Between the vector systems, the dual expression vector pVITRO1_Trastuzumab_ IgG1/κ and single expression vectors gWiz TH/TL expressed Tmab readily and were selected for further cloning, expression optimisation and mutagenesis in the further chapters. The cloning strategy for variable gene switching involved more scrupulous design, had high background with the gWiz vector system and cloning into the dual gene pVITRO1 vector system was unsuccessful. The single gene expression vectors gWiz AH/AL expressed AdmAb readily therefore they were adopted for mutagenesis in the further chapters.

1 Introduction; Expression Vector Library The library of expression vectors sourced for this study were derived from three suites of vectors; that being pVITRO1 (InvivoGen, USA), mAbXpress (Acyte Biotech, Australia) and gWiz (Genlantis, USA).

The pVITRO1 suite of vectors are empty dual expression vectors with the capacity for stable transfection in mammalian culture [175-177]. Dodev et al from Dr Andrew Beavil’s research group, School of Medicine, King’s College London successfully developed a library of antibody expression vectors from an empty pVITRO1 vector that possessed Hygromycin selection [178]. Beavil’s library of vectors have the capacity for stable and transient expression and have been made available through the international plasmid repository AddGene.

The mAbXpress suite of vectors are single gene expression vectors containing the antibody constant region of either heavy or light chain [179]. This suite of vectors was designed for cloning in

25 antibody variable regions of interest through SacI digestion [180]. Successfully cloned mAbXpress heavy and light chain vector pairs would then be suitable for co-transfection in mammalian culture with the capacity for stable transfection.

The gWiz vector is an empty single gene expression vector, with no capacity for stable transfection. For antibody expression, cloning of the whole antibody heavy or light chain genes would be required and transient expression through co-transfection of successfully cloned gWiz heavy and light chain vector pairs.

The pVITRO1 and gWiz vectors were originally commercially sourced, whereas the mAbXpress vectors were developed by the National Biologics Facility at the Australian Institute for Bioengineering and Nanotechnology, for the specific applications of their partner, Acyte Biotech. The mAbXpress vectors are patented by Acyte Biotech and are not commercially available, although they have been made available for research purposes only.

1.1 pVITRO1 pVITRO1_Trastuzumab_IgG1/κ was the pVITRO1 vector sourced for this study, a kind donation from Dr Andrew Beavil’s research group. pVITRO1_Trastuzumab_IgG1/κ was developed from the commercial vector pVITRO1_hygro_mcs vector (pvitro1-mcs, InvivoGen) which is a dual expression vector with Hygromycin resistance [175]. Dodev et al produced pVITRO1_Trastuzumab_IgG1/κ as part of a library of different antibody expression vectors that contained paired antibody heavy and light chain genes in the one vector. This library was produced firstly by cloning gene cassettes of secretion signal peptide (SP) coupled with whole antibody genes into pVITRO1_hygro_mcs, and subsequently by replacing different combinations of antibody constant and variable regions in the expression vector by PIPE cloning method [178]. All vectors in the pVITRO1 suite were high yielding, capable of stable transfection and antibody was extracellularly secreted for ease of purification. pVITRO1_Trastuzumab_IgG1/κ was the subsequent vector produced for complete Tmab antibody expression (Table 1 and Figure 2).

For the purposes of this study, pVITRO1_Trastuzumab_IgG1/κ contained the Tmab genes used for the cloning strategies of the other vectors produced. Moreover, the cloning strategy to produce an AdmAb dual chain expression vector was to exchange the Tmab variable region genes with AdmAb variable region genes of pVITRO1_Trastuzumab_IgG1/κ through an In-Fusion® cloning method (Clontech). A version of this vector with omitted variable regions (pVITRO1_dV_IgG1/κ) was used as a variable gene insert negative control.

Table 1 Summary of features of pVITRO1_Trastuzumab_IgG1/κ vector.

Vector pVITRO1_Trastuzumab_IgG1/κ

Size 8583 bp

Bacterial Selection Hygromycin

Mammalian Selection Hygromycin

Gene Insert 1 SP coupled to Tmab IgG1 heavy chain

Gene Insert 2 SP coupled to Tmab kappa chain

Expression System Mammalian, secreted

Transfection Stable and transient

Figure 2 Vector map of pVITRO1-Trastuzumab-IgG1/κ.

1.2 mAbXpress The mAbXpress vectors sourced for this study were mAbX_hKappa and mAbX_hG1, a kind donation from Acyte Biotech [181]. mAbX_hKappa and mAbX_hG1 were produced as part of a library of expression vectors that contained a single expression cassette of a SP coupled to a SacI restriction enzyme (RE) site and an antibody constant region of either heavy or light chain. Specifically, mAbX_hKappa contains a SP coupled to SacI RE site, to the last 2 amino acids (aa) of the joining region to Cκ, and mAbX_hG1 contains a SP coupled to SacI RE site, to the last 4 aa of the joining region to the complete IgG1 constant regions CH1, hinge, CH2 and CH3 (Table 2 and Figure 3). This suite of vectors were designed to make antibody variable region insertion simple by SacI digestion of the vector and In-Fusion® cloning (Clontech) of the variable region of interest [179].

Vector pairs of mAbXpress heavy and light chain would be suitable for co-transfection in mammalian culture. This suite of vectors was specifically designed and optimised for efficient cloning, transfection and high yielding expression in suspension CHO culture [179], although they can be applied to other mammalian expression systems.

As the cloning strategy is SacI digestion based, the start and end codons for insertion of the variable region are restricted to glutamic acid (E) and leucine (L); i.e. E - variable insert - L. For the purposes of this study, the antibody variable regions were modified to suit the cloning strategy; however, they subsequently produced antibody with modified aa sequences to their reference product. The significance of modifying the aa sequences was not pursued.

Table 2 Summary of features of mAbXpress vectors.

Vector NBF304 mAbX_hKappa NBF305 mAbX_hG1

Size 5064 bp 5742 bp

Bacterial Selection Kanamycin

Mammalian Selection G418

RE site SacI

SP coupled to SacI RE site, to SP coupled to SacI RE site, to Gene Insert human Cκ, from last 2 aa of human IgG1, from last 4 aa of joining region joining region

Expression System Mammalian, co-expression (optimised for CHO)

Transfection Stable and transient

C Figure 3 Portions of the vector maps with cloning site for mAbX_hKappa (A) and mAbX_hG1 (B). Portion of the vector sequence for mAbX_hG1 (C) highlighting the SacI RE site for cloning.

1.3 gWiz The commercial gWiz blank vector sourced from this study was purchased from Genlantis, USA [182]. The gWiz expression vector contains a single multiple cloning site (MCS) for the cloning of either a heavy or light chain gene coupled to a SP, with subsequent expression in mammalian culture through co-transfection of vector heavy and light chain pairs (Table 3 and Figure 4). The gWiz vector does not possess any mammalian selection so is therefore only suitable for transient expression.

Table 3 Summary of features of gWiz vector.

Vector gWiz

Size 5063 bp

Bacterial Selection Kanamycin

Mammalian Selection None

RE site 1 MCS for gene insertion

Expression System Mammalian

Transfection Transient only

For the purposes of this study, the strategy for cloning Tmab heavy and light chain genes into the gWiz vector was RE based, the strategy designed from the gWiz MCS. Subsequently, AdmAb expression vectors were cloned from the Tmab heavy and light chain gWiz vectors by exchanging the Tmab variable regions with AdmAb variable regions through an In-Fusion® cloning method (Clontech).

Figure 4 Vector map of gWiz blank vector (A) and portion of the gWiz sequence displaying the multiple cloning site with unique RE sites (B). *RE sites in grey are blocked by Dam or Dcm methylation for consideration of the cloning strategy.

2 Materials and Methods 2.1 Instrumentation, Services and Software Cloning and colony PCR reactions were conducted on a Bio-Rad T100 thermal cycler, with heated lid. Agarose gels were prepared in Bio-Rad Mini-Gel casters and ran in a Bio-Rad Mini-Sub Cell GT system. Gel imaging was performed on a Bio-Rad Gel Doc XR system and analysed on Bio- Rad Image Lab 5.2.1 software. DNA quantification was performed on a NanoDrop ND-1000 spectrophotometer.

Cloned isolate vectors were prepared as purified DNA samples and sent to AGRF for Sanger Sequence analysis. Sequencing results were analysed on Applied Biosystems Sequence Scanner 2 software.

The complete aa sequence of Tmab and AdmAb light and heavy chains were obtained from the DrugBank database (https://www.drugbank.ca). The corresponding nucleic acid sequences were obtained from the IMGT database (http://www.imgt.org).

Visualisation and cloning plans of the vectors, genes and primers were performed on GSL Biotech

SnapGene Viewer 4.0.5 software. Tm calculations were performed on NEB Tm calculator software (http://tmcalculator.neb.com). PCR product molar ratios were derived from Clontech molar ratio calculator software (http://bioinfo.clontech.com/infusion/molarRatio.do). Alignment of clone isolate sequences to designed vector DNA sequences were performed on the NCBI BLAST software (https://blast.ncbi.nlm.nih.gov/Blast.cgi).

Antibody was purified from culture supernatant on a GE HealthCare ÄKTApurifier 100 FPLC system, using Unicorn 5.11 system and analysis software.

2.2 Molecular Biology Materials pVITRO1_Trastuzumab_IgG1/κ (61883) and pVITRO1_dV_IgG1/κ (52213) vectors were obtained from AddGene and mAbXpress NBF304 and NBF305 vectors were obtained from Acyte Biotech (Australia) through materials transfer agreements. gWiz blank expression vector (P000200) was purchased from Genlantis (USA).

The primers designed for cloning mAbX_TH and mAbX_TL vectors and sequencing mAbX inserts were synthesised from Life Technologies (Australia). The primers designed for sequencing pVITRO1 and gWiz inserts and cloning gWiz_TH and gWiz_TL vectors were synthesised from

Geneworks (Australia). The primers designed for cloning AdmAb heavy (AH) and light (AL) ® variable regions into all vectors, and the AH and AL variable region genes (gBlocks gene fragments) were synthesised from Integrated DNA Technologies (Singapore).

The SacI (1078A) RE and In-Fusion® cloning kit (38909) containing Stellar™ competent E. coli (636763), Nucleospin DNA purification kit (740609.1) and In-Fusion® HD Cloning mastermix (639648) were purchased from Scientifix (Australia). The Phusion high-fidelity PCR kit (E0553S), OneTaq® Quick-Load PCR mastermix (M0486L), high-fidelity SalI (R3138S) and BamHI (R3136S) RE, T4 DNA Ligase (M0202S), rSAP (M0371S), gel loading dye (B7025) and DNA ladders 1 kb (N3232S) and 100 bp (N3231S) were purchased from Genesearch (Australia).

Biotechnology grade agarose (BIOD0012), PBS tablets (09-2051-100) and TAE 25x stock solution (PARTA82820) were purchased from Astral Scientific (Australia). SYBR® Safe nucleotide stain (S33102) was purchased from Life Technologies (Australia). Plasmid maxiprep kit (G221020) was purchased from Astral Scientific (Australia), miniprep (PLN70) and maxiprep (NA0310) kits were purchased from Sigma-Aldrich (Australia).

2.3 Culture and Purification Materials Yeast extract (70161), Sodium chloride (S9888) and Kanamycin solution (K0254) were purchased from Sigma-Aldrich (Australia). Tryptone (LP0042B) and Geneticin® (10131-035) were purchased from Thermo Fisher Scientific Australia and agar (20767.232) was purchased from VWR International (Australia). Hygromycin B solution (ant-hg-1), Hygromycin agar (fas-hg-s) and Terrific Broth (fas-hg-l) sachets were purchased from Jomar Life Research (Australia).

FreeStyle 293F cells were a kind donation from Dr Mario Torrado del Rey of Prof Joel Mackay’s research group, School of Life and Environmental Sciences, The University of Sydney. FreeStyle™ 293 expression media (12338018) and OptiPRO™ Serum free media (12309050) were purchased from Life Technologies (Australia). Linear Polyethylenimine (PEI) 25 kDa (23966-2) was purchased from BioScientific (Australia). PBS tablets (09-2051-100) and 1 M Tris-HCl pH 9 (BIOSD814) were purchased from Astral Scientific (Australia). Disposable baffled culture flasks (CLS431405), centrifuge tubes (CLS430291, CLS430791), vacuum systems (CLS430769), glycine hydrochloride (G2879), HiTrap® Protein A affinity column (GE17-0403-01) and Amicon® Ultracel 50 kDa centrifugal filters (Z740177, Z740193) were purchased from Sigma-Aldrich (Australia).

2.4 Cloning Strategies and Molecular Biology The cloning strategy for producing Tmab heavy and light chain mAbX and gWiz vectors was through a RE digestion and ligation method. Briefly for the mAbX vectors, the Tmab variable heavy and light chain genes were amplified from the pVITRO1_Trastuzumab_IgG1/κ vector by high fidelity PCR, with primers incorporating 16 – 17 bp homology of the SP upstream and constant region (Vjoin / Cjoin) downstream of the mAbX vectors for seamless cloning. The mAbX

32 vectors were linearised with SacI, then the linearised mAbX vectors and the Tmab variable PCR fragments were combined for an In-Fusion® ligation reaction and rescued in E. coli.

Briefly for the gWiz vector, the complete Tmab heavy and light chain genes were amplified from pVITRO1_Trastuzumab_IgG1/κ by high fidelity PCR, with primers designed to flank the genes with unique RE sites for digestion. The PCR products were digested with the associated RE along with the gWiz vector. The digested PCR products and linearised vector were combined for a ligation reaction and rescued in competent E. coli.

The cloning strategy for producing AdmAb heavy and light chain pVITRO1 and gWiz vectors was through an In-Fusion® insert switching method. Briefly, the entirety of the pVITRO1_Trastuzumab_IgG/κ, gWiz TH and gWiz TL vectors were amplified by high fidelity PCR with the Tmab VH and VL genes omitted from the PCR products through primer design. AdmAb VH and VL genes were synthesised, incorporating 20 bp homology of the SP upstream and constant region (Vκ / CH1) downstream of the vector backbones for seamless cloning. The linear vector backbones and AdmAb gene fragments were combined for an In-Fusion® ligation reaction and rescued in E. coli.

In all cloning methods, single colony isolates were rescued and screened by colony PCR for the presence of gene insertion. Plasmid extractions were performed in isolates that screened insert positive, to then be prepared for sequence analysis. Isolates that were confirmed for successful cloning through sequence analysis were scaled up for a maxiprep plasmid extraction, in preparation of transfection.

2.4.1 Cloning Tmab Variable Regions into mAbXpress Vectors

The cloning strategy to produce mAbX Tmab light (mAbX TL) and heavy (mAbX TH) chain expression vectors was to isolate the genes for Tmab variable light (TVL) and variable heavy (TVH) ® chains and introduce them into the mAbX vectors through In-Fusion cloning. TVH and TVL were amplified from pVITRO1_Trastuzumab_IgG1/κ using a Phusion High Fidelity PCR kit (Table 4 and

Table 5) with the primers designed to introduce 16 – 17 bp homology up and downstream of the mAbX vectors for seamless cloning (Table 6). The PCR products were confirmed on a 0.8% agarose gel and purified using the Nucleospin DNA purification kit (Appendix 2.1.5).

The mAbX hG1 and hKappa vectors were linearised through SacI digestion, incubated at 37 °C for 1 hr. The linearised vectors were confirmed and separated on a 0.8% agarose gel and purified using the Nucleospin DNA purification kit (Appendix 2.1.5). The vector backbones were combined with

33 the Tmab variable region genes for In-Fusion® ligation reactions in a molar ratio of 2:1 gene to vector, incubated at 50 °C for 15 min. 2.5 µL of the ligation products were transformed into Stellar™ competent E. coli by heat-shock method, colonies were isolated in Luria-Bertani (LB) agar containing 50 µg/mL Kanamycin. Colony PCR was performed to rescue and screen for vector isolates containing intact single gene insertion. Positive isolates were purified and sequenced to confirm seamless integration of the TVH and TVκ genes, producing complete Tmab heavy (mAbX

TH) and light (mAbX TL) chain expression vectors.

Table 4 Reaction components for amplification of TVH. For amplification of TVL, the HC primer pair was replaced with the LC primer pair, all other components were equivalent.

Reaction Volume Component Concentration (µL)

5 x Phusion HF Buffer 1 x 10

10 mM dNTP 200 µM 1

10 µM mAbX_HC_Trast_Fwd 0.5 µM 2.5

10 µM mAbX_HC_Trast_Rev 0.5 µM 2.5

1 ng/mL pVITRO1_Trastuzumab_IgG1/κ 20 ng/mL 1

Nuclease free H2O - 32.5

Phusion DNA Polymerase 20 units/mL 0.5

Table 5 Cycling conditions for PCR amplification of TVH and TVL.

Temperature Cycle Step Cycles Time (sec) (°C)

Initial denaturation 1 98 120

Denaturation 98 40

Annealing 30 55 40

Extension 72 150

Hold 1 12 ∞

Table 6 List of designed primers for mAbX cloning and sequencing. Nucleotides with homology to mAbX vectors (underline), nucleotides with homology to Tmab genes in pVITRO1, including SP and join regions (bold). Tm calculated to pVITRO1_Trastuzumab_IgG1/κ with the Phusion PCR kit* and mAbX vectors with the OneTaq PCR kit#.

Homology Primer Sequence (bp) to Tm 5’ 3’ mAbX for (°C) In-Fusion®

mAbX_LC_Trast_Fwd CCGGCGTGCACTCCGACATCCAGATGACC 16 48* mAbX_LC_Trast_Rev GCCACGGTCCGCTTGATCTCCACCTTGGT 16 51*

mAbX_HC_Trast_Fwd CAGGTGTCCACTCCGAGGTGCAGCTGGTG 17 73* mAbX_HC_Trast_Rev GCGGAGGACACGGTGACAAGAGTACCT 16 71*

mAbX_SeqF TAACACCGCCCCGGTTTTCC 64# mAbX_Seq-hKapR GTTCAGCAGGCACACCACGG N/A 65# mAbX_Seq-hG1-hG2-R TCACCAGGCAGCCCAGAGCG 69#

2.4.2 Cloning Tmab Whole Genes into gWiz

The cloning strategy to produce gWiz Tmab light (gWiz TL) and heavy (gWiz TH) chain expression vectors was to isolate the whole genes for Tmab light and heavy chains from pVITRO1_Trastuzumab_IgG1/κ and introduce the genes into the gWiz vector through RE double digestion and ligation reactions.

The complete Tmab genes from the pVITRO1_Trastuzumab_IgG1/κ vector were amplified using a Phusion High Fidelity PCR kit using a 2-step PCR cycling method (Table 7 and

Table 8). Primers for amplifying the Tmab genes were designed to flank the genes with unique RE sites based on the gWiz MCS, SalI and BamHI were elected for the primers designed upstream and downstream, respectively (Table 9). It was found that a BamHI RE site was present at the beginning of the TH SP gene, hence the TH forward primer was further modified to remove that RE site, whilst preserving the aa sequence: R6 (arginine) AGG > AGA (Figure 5).

The linear PCR products were confirmed on a 0.8% agarose gel and purified using the Nucleospin DNA purification kit (Appendix 2.1.5). SalI-BamHI double digestion reactions were performed on the PCR products and gWiz vector, incubated at 37 °C for a total of 90 min (Table 10). To prevent 35 religation of the linearised gWiz vector, the digestion product was further dephosphorylated by the addition of rSAP 30 min into the digestion reaction.

The digested DNA products were again separately purified using the Nucleospin DNA purification kit (Appendix 2.1.5), then prepared for ligation with T4 DNA ligase in a molar ratio of 3:1 gene to vector, incubated O/N at room temperature (Table 11). 5 µL of the ligation products were transformed into Stellar™ competent E. coli by heat-shock method, cells containing intact vector were rescued in LB agar containing 50 µg/mL Kanamycin. Colony PCR was performed to rescue and screen for vector isolates containing intact, single gene insertion. Positive isolates were purified and sequenced to confirm seamless integration of the Tmab IgG1 gene (gWiz TH) or kappa gene

(gWiz TL).

Table 7 Reaction components for amplification of TH. For amplification of TL, the TH primer pair was replaced with the TL primer pair, all other components were equivalent.

Component Concentration Volume (µL)

5 x Phusion HF Buffer 1 x 10

10 mM dNTP 200 µM 1

10 µM Fwd pVitro TH into gWiz 0.5 µM 2.5

10 µM Rev pVitro TH into gWiz 0.5 µM 2.5

1 ng/ mL pVITRO1_Trastuzumab_IgG1/κ 20 ng/mL 1

Nuclease free H2O - 32.5

Phusion DNA Polymerase 20 units/mL 0.5

Table 8 Cycling conditions for 2-Step PCR amplification of TH and TL. 30 cycles were run for TH * + PCR with 45 sec extension time , 25 cycles were run of the TL PCR with 20 sec extension time .

Temperature Cycle Step Cycles Time (sec) (°C)

Initial denaturation 1 98 30

Denaturation 98 10 30*/25+ Annealing and Extension 72 45* / 20+

Final Extension 1 72 300

Table 9 List of primers designed for gWiz TH/TL cloning and sequencing. SalI/BamHI RE sites (underlined) and nucleotides with homology to Tmab genes (bold). Tm calculated to pVITRO1_Trastuzumab_IgG1/κ with the Phusion PCR kit* and gWiz vectors with the OneTaq PCR kit#.

Sequence Tm Primer 5’ 3’ (°C)

Fwd pVitro TH into gWiz GGTGGTGTCGACATGGACTGGACCTGGAGAATCCTCTTCTTG 76* Rev pVitro TH into gWiz GGTGGTGGATCCTCATTTACCCGGAGACAGGGAGAG 73*

Fwd pVitro TL into gWiz GGTGGTGTCGACATGTTGCCATCACAACTCATTGGG 71* Rev pVitro TL into gWiz GGTGGTGGATCCCCTAACACTCTCCCCTGTTGAAGC 70*

SeqF gWiz MCS AGTACTCGTTGCTGCCG 58# SeqR gWiz MCS ATGGCTGGCAACTAGAAGG 58#

Table 10 Reaction components for double-digestion of TH, TL and blank gWiz vector. The entire * quantity of TH and TL purified PCR products was digested. rSAP was added to the gWiz digestion reaction after 30 min incubation.

TH/TL gWiz

Component Concentration Volume (µL)

TH/TL PCR product (purified) - 28 -

122 ng/µL blank gWiz vector 40 µg/mL - 16.5

10 x CutSmart Buffer 1 x 5

Sal-HF 400 units/mL 1

BamHI-HF 400 units/mL 1

rSAP 20 units/mL - 1*

Nuclease free H2O - 15 25.5

Figure 5 Portion of the pVITRO1_Trastuzumab_IgG1/κ sequence highlighting the BamHI RE site in the SP of the TH gene. Primer Fwd pVitro TH into gWiz was designed to remove the BamHI site in the SP as to make BamHI a unique RE site to be introduced after the stop codon with the reverse primer.

Table 11 Ligation reaction components for cloning gWiz TH and gWiz TL vectors.

gWiz TH gWiz TL

Component Reaction Quantity

10 x T4 DNA Ligase Buffer 1 x

Linearised gWiz vector (purified) 84 ng

Digested TH product (purified) 72 ng -

Digested TL product (purified) - 36 ng

T4 DNA Ligase 20 units/µL

Nuclease free H2O Up to 20 µL

2.4.3 AdmAb Variable Region Gene Switching

The cloning strategy to produce AdmAb light (gWiz AL), heavy (gWiz AH) and dual (pVITRO1_Adalimumab_IgG1/κ) chain expression vectors was through variable region gene switching of pVITRO1_Trastuzumab_IgG1/κ, gWiz TL and gWiz TH. Each of the vectors was linearised using a Phusion High Fidelity PCR kit (Table 12 and Table 13), with primers designed to specifically omit the variable regions of the Tmab genes (Table 14). Two vector fragments were produced in separate PCRs for the pVITRO1_Trastuzumab_IgG1/κ that contained the heavy (pVITRO1 H/sL) or light (pVITRO1 L/sH) constant regions (Figure 6), whereas a single vector backbone fragment was produced from the gWiz TH (gWiz CH) and gWiz TL (gWiz CL) PCRs. The linear PCR products were confirmed on a 0.8% agarose gel and purified using the Nucleospin DNA purification kit (Appendix 2.1.5).

Table 12 Reaction components for PCR linearisation of gWiz TH, gWiz TL and pVITRO1_Trastuzumab_IgG1/κ fragments, omitting the Tmab variable regions.

pVITRO1 pVITRO1 gWiz CH gWiz CL H/sL L/sH Reaction Component Volume (µL) Concentration

5 x Phusion HF Buffer 1 x 10

10 mM dNTP 200 µM 1

10 µM LinCH_IgG1_Fwd 0.5 µM 2.5 - 2.5 -

10 µM LinCH_IgG1_Rev 0.5 µM 2.5 - - 2.5

10 µM LinCL_kappa_Fwd 0.5 µM - 2.5 - 2.5

10 µM LinCL_kappa_Rev 0.5 µM - 2.5 2.5 -

1 ng/mL gWiz TH 20 ng/mL 1 - - -

1 ng/mL gWiz TL 20 ng/mL - 1 - -

1 ng/mL pVITRO1_Trastuzumab_IgG1/κ 20 ng/mL - - 1 1

Nuclease free H2O - 32.5

Phusion DNA Polymerase 20 units/mL 0.5

Table 13 Cycling conditions for PCR linearisation of gWiz TH, gWiz TL and pVITRO1_Trastuzumab_IgG1/κ fragments, omitting the Tmab variable regions.

pVITRO1 pVITRO1 gWiz CH gWiz CL H/sL L/sH

Temperature Cycle Step Cycles Time (sec) (°C)

Initial denaturation 1 98 30

Denaturation 98 10

Annealing 25 65 10

Extension 72 100 90 60 70

Final Extension 1 72 300

Table 14 List of primers designed for pVITRO1_Adalimumab_IgG1/κ, gWiz AH and gWiz AL * cloning and sequencing. Tm calculated to the Tmab genes with the Phusion PCR kit and gWiz/pVITRO1 vectors with the OneTaq PCR kit#

Sequence Primer Tm (°C) 5’ 3’

LinCH_IgG1_Fwd GCTAGCACCAAGGGCCCATC 70* LinCH_IgG1_Rev GGAGTGCGCGCCTGTG 69*

LinCL_kappa_Fwd GTACGGTGGCGGCGC 68* LinCL_kappa_Rev ACCGCGGCTAGCTGGAAC 68*

SeqF gWiz MCS AGTACTCGTTGCTGCCG 58# SeqR gWiz MCS ATGGCTGGCAACTAGAAGG 58#

pVITRO1_hFwd TTTTGAGCGGAGCTAATTCTCGGG 62# pVITRO1_hRev AAAAAACCTCCCACACCTCC 58#

pVITRO1_kFwd GAGGCTAATTCTCAAGCCTC 55# pVITRO1_kRev TCTAGACCTGGAAAGACCAG 56#

Figure 6: PCR products of pVITRO1-Trastuzumab -IgG1/κ for In-Fusion® cloning pVITRO1 H/sL (A) and L/sH (B).

AL and AH genes were designed and custom synthesised to contain 20 bp homology of the SP upstream, and constant region (Vjoin / Cjoin) downstream of the vector backbones for seamless cloning. The vector backbones were combined with the AdmAb variable region genes for In- Fusion® ligation reactions in a molar ratio of 2:1 gene to vector, incubated at 50 °C for 15 min (Table 15).

2.5 µL of the ligation products were transformed into Stellar™ competent E. coli by heat-shock method, cells containing intact vector were rescued in LB agar containing 50 µg/mL Kanamycin for gWiz reactions, or 75 µg/mL Hygromycin for pVITRO1 reactions. Colony PCR was performed to rescue and screen for vector isolates containing intact, single gene insertion. Positive isolates were purified and sequenced to confirm seamless integration of the AdmAb variable regions producing complete AdmAb heavy and light chain expression vectors (gWiz AH, gWiz AL and pVITRO1_Adalimumab_IgG1/ κ).

® Table 15 In-Fusion ligation reaction components for cloning gWiz AH, gWiz AL and pVITRO1_Adalimumab_IgG1/κ vectors. *Fragments chosen for reference to derive molar equivalent of other fragment.

pVITRO1_ gWiz AH gWiz AL Adalimumab _IgG1/κ

Component Reaction Quantity

5 x In-Fusion HD Enzyme Premix 1 x

AH gene 26.5 ng - 29.4 ng

AL gene - 20.1 ng 26.4 ng*

gWiz CH 200 ng - -

gWiz CL - 150 ng -

gWiz CH 200 ng - -

pVITRO1 H/sL - - 150 ng*

Nuclease free H2O Up to total reaction volume based on DNA volumes used

2.5 Isolating Positive Clones and Sequence Analysis Ligation products were rescued in Stellar™ competent E. coli through a heat-shock method (Appendix 1.4), onto an LB agar plate containing selection antibiotic (50 µg/mL Kanamycin for mAbX and gWiz vector ligations, 75 µg/mL Hygromycin for pVITRO1 vector ligations). Single colonies were isolated for culture and screened for successful insertion through colony PCR (Table 16 and Table 17). 5 µL of colony PCR products were loaded on 0.8 % agarose gel stained with SYBR® Safe nucleotide stain, imaged and analysed using the Bio-Rad Gel Doc XR system.

Isolates that screened positive were prepared as glycerol stocks (Appendix 1.2) and plasmid miniprep extractions (Appendix 2.1.1) in preparation for sequencing (Appendix 2.2). Isolates that 41 screened positive for the clone insertion from sequence analysis were resuscitated and culture scaled up for plasmid maxiprep extraction (Appendix 2.1.3 - 2.1.4) in the lead up to preliminary expression studies.

Table 16 Reaction components for colony PCR, probing for complete TH, TL, AH and AL genes in clone isolates. Primer pairs for the reactions were associated with the sequence primers for mAbX, gWiz and pVITRO1 vectors.

Reaction Volume Component Concentration (µL)

OneTaq QuickLoad 2x Master Mix 1 x 12.5

Liquid o/n culture of isolate - 1

10 µM Fwd Seq primer 0.2 µM 0.5

10 µM Rev Seq primer 0.2 µM 0.5

Nuclease free H2O - 10.5

Table 17 Cycling conditions for colony PCR, probing for complete TH, TL, AH and AL genes in clone isolates.

TH/AH TL/AL

Temperature Cycle Step Cycles Time (sec) (°C)

Initial denaturation 1 94 120

Denaturation 94 15

Annealing 25 55 30

Extension 68 90 45

Final Extension 1 68 300

2.6 Transfection and Purification of Antibodies Successfully cloned vectors were prepared for transient transfection in suspension 293F culture. 24 hr prior to transfection, 293F cells were passaged to a seeding density of 5 x 105 cells/mL in Freestyle 293 expression media. Transfection proceeded if the live cell concentration reached 0.8 – 1.2 x 106 cells/mL and cell viability was ≥ 95% on the day of transfection. The total live cells of the culture was obtained and quantity of DNA and PEI were calculated as 2 µg PEI and 1 µg DNA per

106 cells in transfection culture. For co-transfection of light chain and heavy chain mabX or gWiz vectors, a 1:1 ratio of light chain to heavy chain vector was combined for transfection. Calculated quantity of DNA and PEI were separately diluted in OptiPRO SFM to total volumes of 600 µL and incubated at RT for 5 min. The diluted DNA and PEI solutions were combined, gently mixed and incubated at RT for 15 min prior to dropwise addition to transfection culture. Cultures were returned to incubate, 24 hrs post transfection the cultures were supplemented with 0.5% w/v tryptone to total culture volume, and again supplemented with 0.5% w/v tryptone on subsequent scale up at 48 hrs. Cell counts of the transfection cultures were taken daily to monitor progression and culture was harvested when the cell viability reached < 80%.

In the instance of expression with the mAbX vectors, transfection was conducted in 30 mL culture and harvested 5 days post transfection. In the instance of expression with the pVITRO1 vector, transfection was conducted in 30 mL culture which was scaled up to 60 mL at 24 hrs, then 100 mL at 48 hrs post transfection; harvested 7 days post transfection. In the instance of Tmab and AdmAb expression with the gWiz vectors, transfection was conducted in 60 mL culture which was scaled up to 120 mL at 24 hrs, then 240 mL at 48 hrs post transfection; harvested 8 days post transfection.

Culture supernatants were harvested by centrifugation (3,000 g, 15 mins), clarified and degassed by passing through a 0.22 μm vacuum filter. Secreted antibody was purified from the culture supernatant through Protein A affinity chromatography, using an ÄKTApurifier 100 FPLC system with a HiTrap® Protein A column (Appendix 4). Antibody was bound on the column with PBS as binding buffer, eluted from the column with 0.1 M glycine HCl pH 2.7 and elution fractions neutralised with 1 M Tris-HCl pH 9.

FPLC chromatograms detected the protein content passing through the column with UV absorbance at 280 nm. Elution peaks from the FPLC chromatogram indicated the total protein content purified from each expression system. The chromatograms were analysed to determine the area under the elution peak for comparison, as a preliminary indicator of total protein yield. Fractions containing eluted antibody as indicated from the FPLC chromatograms were pooled, buffer exchanged to PBS and concentrated using 50 kDa centrifugal filters. UV absorption of the purified antibody samples was measured at 280 nm to derive total antibody yield using the Beer Lambert law and the extinction coefficients of Tmab and AdmAb.

3 Results 3.1 Cloning Tmab Variable Regions into mAbXpress Vectors The amplified Tmab variable region genes and SacI linearised mAbX vectors yielded fragments of their predicted size – TVH 376 bp, TVL 346 bp, linearised mAbX hKappa 5064 bp and linearised

43 mAbX hG1 5742 bp (Figure 7); however, the PCR products ran fairly broadly which suggests incomplete extension in the PCRs. Purification of the DNA products post RE digestions was high yielding and therefore did not compromise the ligation reaction setup conditions.

Transformation of both ligation reactions successfully yielded many isolates for colony PCR screening, of which 4 mAbX TH and 2 mAbX TL tested isolates were positive (Figure 8). Blank mAbx hKappa and mAbX hG1 DNA were run for insert negative control. A consistent smear present in negative colony PCR samples (< 1000 bp) was attributed to the unreacted primers from the PCR mastermix. The 100 bp ladder was not resolved in the gels to quantify the PCR fragment sizes, although the difference between insert positive and negative was apparent. All colony PCR positive clones were prepared for sequencing. A B 1 2 L1 3 4 L2 bp - 1500 kb - 1200 - 10 - 1000 5.7 kb → - 8 5 kb → - 6 - 800 - 4 - 700 - 3 - 600 - 2 - 500 - 1.5 375 bp → - 400 350 bp → - 300 - 1 - 200 - 0.5 - 100

Figure 7 0.8% Agarose gels of Tmab variable region gene PCR products (A) and SacI digested mAbX vectors (B). 100 bp DNA Ladder (L1) 1 kb DNA Ladder (L2), TVL PCR product (1), TVH PCR product (2), linearised mAbX hKappa (3), linearised mAbX hG1 (4). 2 µL ladder, 1.5 µL PCR and digestion products were loaded.

A B L 1 2 3 4 5 6 7 8 9 10 -H 11 12 -L L

Figure 8 0.8% Agarose gels of colony PCR screening for positive insertion in mAbX TH (A) and mAbX TL (B) isolates. 100 bp DNA Ladder (L), 10 mAbX TH isolates (1 - 10), blank mAbX hG1 (- H), 2 mAbX TL isolates (11 - 12) and blank mAbX hKappa (-L). 2 µL ladder and 5 µL PCR samples were loaded.

3.2 gWiz Cloning of Whole Tmab Genes The amplification of Tmab genes and linearisation of gWiz blank vector yielded fragments of their predicted sizes - TH 1,434 bp, TL 726 bp and linearised gWiz 5,016 bp (Figure 9); however, the products ran fairly broadly which suggests incomplete extension in the PCRs. Purification of the DNA products post RE digestions was high yielding and therefore did not compromise the ligation reaction setup conditions (Table 18).

L 1 2 A L 3 4 5 6 7 B

kb kb 10 - 8 - 108 - 6 - 6 - 5 - 5 - ← 5 kb 3 - 3 - 2 - 2 - 1.5 - 1.5 - ← 1.5 kb 1 - 1 - 0.5 - ← 750 bp 0.5 -

Figure 9 0.8% Agarose gels of reaction products for cloning Tmab whole genes TL (A) and TH (B) into gWiz vector. 1 kb DNA Ladder (L), TL PCR product (1), TL purified RE digestion product (2), TH PCR product (3), gWiz RE / rSAP product (4) TH purified RE digestion product (5), gWiz purified RE / rSAP product (6), 40 ng circular gWiz blank (7). 2 µL ladder, PCR and digestion products, and 1.5 µL purified samples were loaded.

Transformation of both ligation reactions successfully yielded many isolates for colony PCR screening, of which 3 gWiz TH and 6 gWiz TL tested isolates were positive (Figure 10).

Additionally, pVITRO1_Trastuzumab_IgG1/κ DNA was run for a positive control, blank gWiz DNA for a negative control and PCR cocktail without DNA sample was run for a contamination control. A consistent band present in all colony PCR samples (< 100 bp) was attributed to the residual primers from the PCR mastermix. All colony PCR positive clones were prepared for sequencing.

Table 18 Purified DNA digestion products for cloning Tmab whole genes into gWiz for ligation reaction setup.

gWiz TH gWiz TL

Component Concentration Volume (µL)

10 x T4 DNA Ligase Buffer 10 x 2

Linearised gWiz vector (purified) 28.9 ng/µL 2.9

Digested TH product (purified) 121.0 ng/µL 0.6 -

Digested TL product (purified) 8.6 ng/µL - 4.2

T4 DNA Ligase 400 units/µL 1

Nuclease free H2O - 13.5 9.9

L 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 - + C A kb 10 - 6 - 3 - 1.5 - 1.6 kb → 1 - 0.5 -

L 23 24 25 26 27 28 29 30 31 32 33 - + C

kb B 10 - 6 - 3 - 1.5 - 1 - ← 0.8 kb 0.5 -

Figure 10 0.8% Agarose gels of colony PCR screening for positive insertion in gWiz TH (A) and gWiz TL (B) isolates. 1 kb DNA Ladder (L), 22 gWiz TH isolates (1 - 22), 11 gWiz TL isolates (23 - 33), blank gWiz DNA (-), pVITRO1_Trastuzumab_IgG1/κ DNA (+) and PCR mastermix omitting DNA (C). 2 µL ladder and 5 µL PCR samples were loaded. 46

3.3 Variable Region Gene Switching in gWiz and pVITRO1 Vectors

The PCR reactions for linearising gWiz TH and gWiz TL vectors, and backbone fragments of pVITRO1_Trastuzumab_IgG1/κ yielded fragments of their predicted sizes – gWiz CH 6,072 bp, gWiz CL 5,401 bp and pVITRO1 H/sL 3,782 bp; however, pVITRO1 L/sH product was a much smaller fragment than the expected 5,401 bp (Figure 11). The fragment size could not be elucidated for pVITRO1 L/sH product although the fragment band spanned 700 – 1500 bp. PCR cocktail without DNA sample was run as a contamination control for all linearising PCRs. All PCR products ran broadly and were low yielding suggesting incomplete extension and suboptimal cycling conditions. Despite the low product yields as indicated from the agarose gel, purification of the DNA products yielded manageable quantities and In-Fusion® ligation reactions were adjusted accordingly, in keeping with the 2:1 molar ratio of gene to vector (Table 19).

1 1C 2 2C 3 3C 4 4C L

6.2 5.5 3.8 kb kb kb - 10 kb → - 8 → - 6 → - 5 - 3 - 2 - 1.5 - 1 ← 0.7 - 1.5 kb - 0.5

Figure 11 0.8% Agarose gels of PCR products for linearised gWiz TH, gWiz TL and pVITRO1_Trastuzumab_IgG1/κ. 1 kb DNA Ladder (L), pVITRO1 H/sL product (1), pVITRO1 L/sH (2), gWiz CL (3), gWiz CH (4), PCR mastermix omitting DNA (1C - 4C). 2 µL ladder and 1.5 µL linearising PCR products were loaded.

Transformation of gWiz In-Fusion® reactions successfully yielded many isolates for colony PCR screening, of which 7 gWiz AH and 8 gWiz AL tested isolates were positive (Figure 12A and B). However, transformation of pVITRO1 In-Fusion® reaction was unsuccessful, yielding 1 isolate that screened negative for both AL and AH colony PCRs (Figure 12A and C). Additionally, gWiz TH, gWiz TL and pVITRO1_Trastuzumab_IgG1/κ DNA were run for insert positive controls, pVITRO1_dV_IgG1/κ DNA was run for insert negative control and PCR cocktail without DNA sample was run for a contamination control. All colony PCR positive clones were prepared for sequencing.

Table 19 Purified linearised vector products for AdmAb variable region switching In-Fusion® ligation reactions. Reaction volumes were adjusted to maintain a minimum of 20 ng gene in the reaction, keeping in the 2:1 molar ratio of gene to vector.

pVITRO1_ gWiz AH gWiz AL Adalimumab _IgG1/κ

Component Concentration Volume (µL)

5 x In-Fusion HD Enzyme Premix 5 x 3 2 4.5

AH gene 8.6 ng/µL 3.1 - 3.4

AL gene 8.6 ng/µL - 2.3 3.1

gWiz CH 23.8 ng/µL 8.4 - -

gWiz CL 29.4 ng/µL - 5.1 -

pVITRO1 H/sL 24.9 ng/µL - - 5.5

pVITRO1 H/sL 31.2 ng/µL - - 4.8

Nuclease free H2O - 0.5 0.6 1.2 A L 1 2 3 4 5 6 7 8 9 10 11 p+ gH+ dV

kb 10 - 8 - 6 - 4 - 3 - 2 - ← 1.5 kb 1.5 - ← 1.2 kb 1 - 0.5 - ← 0.7 kb

B L 12 13 14 15 16 17 18 19 20 21 gL+ p+ dV C C L 22 p+ gL+ dV C

kb 10 - kb 8 - 6 - 10 - 4 - 8 - 3 - 6 - 2 - 4 - 1.5 - 3 - 1 - ← 800 bp 2 - 0.5 - ← 550 bp 1.5 - 1 - ← 800 bp 0.5 - ← 550 bp

Figure 12 0.8% Agarose gels of colony PCR screening for positive insertion for full AH gene (A) and AL gene (B, C) in gWiz and pVITRO1 isolates. 1 kb DNA Ladder (L), 10 gWiz AH isolates (1 - 10), 1 pVITRO1 isolate (11, 22), 10 gWiz AL isolates (12 - 21), pVITRO1_Trastuzumab_IgG1/κ DNA (p+), gWiz TH DNA (gH+), gWiz TL DNA (gL+), pVITRO1_dV_IgG1/κ DNA (dV) and PCR mastermix omitting DNA (C). 2 µL ladder and 5 µL PCR samples were loaded. 48

3.4 Sequence Analysis and Cloning Efficiency All sequence results were analysed for quality and robustness of base calling prior to BLAST alignment analysis. Robust isolate nucleotide sequences were aligned against the nucleotide sequences of the predicted vector products to determine the success of seamless insertion. Cloned isolates were confirmed through sequence analysis in both (forward and reverse) directions.

An example of positive sequence analysis as shown in Figure 13, isolate gWiz TL1 forward sequence was aligned with the predicted nucleotide sequence for gWiz TL. Homology from 1874 –

2589 bp in the gWiz TL sequence confirmed successful integration of the TL gene. No further alignments (i.e. one single alignment) confirmed a single, seamless gene insertion and a single vector backbone. Continued homology beyond the gene insertion confirmed a portion of the vector backbone.

Through the sequencing analysis, it was realised that the primer design for the mAbX TL cloning produced a mismatch in the joining region that directly disrupted the SacI site and caused an aa substitution of Leucine 104 to Valine. This was observed in the sequencing of both mabX TL positive clones isolated. The significance of this bp and aa substitution was not investigated.

The efficiency and accuracy between the different cloning methods were compared (Table 20). Efficiency is described as the percentage of isolates that screened insert positive by colony PCR and accuracy is described as the percentage of insert positive clones that were confirmed as successful cloning through sequence analysis.

Query 1819 AGCTGACAGACTAACAGACTGTTCCTTTCCATGGGTCTTTTCTGCAGTCACCGTCGTCGACATGTTGCCATCACAACTCATTGGGTTTCTGCTGCTCTGGGTTCCAGCTAGCCGCGGTGACATCCAGATGACCCAGTCTCCTTCTTCTCT 1968 ||||||| |||| | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| Sbjct 8 AGCTGAC-GACT-ANAGACTGTTCCTTTCCATGGGTCTTTTCTGCAGTCACCGTCGTCGACATGTTGCCATCACAACTCATTGGGTTTCTGCTGCTCTGGGTTCCAGCTAGCCGCGGTGACATCCAGATGACCCAGTCTCCTTCTTCTCT 155

Query 1969 TTCTGCTTCTGTTGGTGATCGTGTTACTATTACTTGTCGTGCTTCTCAAGATGTTAATACTGCTGTTGCTTGGTATCAACAAAAACCTGGTAAAGCTCCTAAActtcttatttattctgcttcttttctttattctggtgttccttctcg 2118 |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| Sbjct 156 TTCTGCTTCTGTTGGTGATCGTGTTACTATTACTTGTCGTGCTTCTCAAGATGTTAATACTGCTGTTGCTTGGTATCAACAAAAACCTGGTAAAGCTCCTAAACTTCTTATTTATTCTGCTTCTTTTCTTTATTCTGGTGTTCCTTCTCG 305

Query 2119 tttttctggttctcgttctGGTACTGATTTTACTCTTACTATTTCTTCTCTTCAACCTGAAGATTTTGCTACTTATTATTGTCAACAACATTATACTACTCCTCCTACTTTTGGTCAAGGTACCAAGCTGGAGATCAAACGTACGGTGGC 2268 |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| Sbjct 306 TTTTTCTGGTTCTCGTTCTGGTACTGATTTTACTCTTACTATTTCTTCTCTTCAACCTGAAGATTTTGCTACTTATTATTGTCAACAACATTATACTACTCCTCCTACTTTTGGTCAAGGTACCAAGCTGGAGATCAAACGTACGGTGGC 455

Query 2419 GAGTGTCACAGAGCAGGACAGCAAGGACAGCACCTACAGCCTCAGCAGCACCCTGACGCTGAGCAAAGCAGACTACGAGAAACACAAAGTCTACGCCTGCGAAGTCACCCATCAGGGCCTGAGCTCGCCCGTCACAAAGAGCTTCAACAG 2568 |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| Sbjct 606 GAGTGTCACAGAGCAGGACAGCAAGGACAGCACCTACAGCCTCAGCAGCACCCTGACGCTGAGCAAAGCAGACTACGAGAAACACAAAGTCTACGCCTGCGAAGTCACCCATCAGGGCCTGAGCTCGCCCGTCACAAAGAGCTTCAACAG 755

Query 2569 GGGAGAGTGTTAGGGGATCCAGATCACTTCTGGCTAATAAAAGATCAGAGCTCTAGAGATCTGTGTGTTGGTTTTTTGTGGATCTGCTGTGCCTTCTAGTTGCCAGCCATCTGTTGTTTGCCCC-TCCCCCGTG-CCTTCCTTGACCCTG 2716 |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| ||||||||| ||||||||||||||| Sbjct 756 GGGAGAGTGTTAGGGGATCCAGATCACTTCTGGCTAATAAAAGATCAGAGCTCTAGAGATCTGTGTGTTGGTTTTTTGTGGATCTGCTGTGCCTTCTAGTTGCCAGCCATCTGTTGTTTGCCCCCTCCCCCGTGCCCTTCCTTGACCCTG 905

Query 2717 GAA-GGTGCC-ACTCCC-ACT-GTCCTTTCCTAATAAAATGA-GGAAA-TTGCATCGC-ATTGTC-TGAGT-AGGTGTCATT-CTATT-CTgggggg-tgggg 2807 ||| |||||| |||||| ||| ||||||||||||||||| || ||||| ||||||| | |||||| ||| | ||||||| || ||||| |||||||| ||||| Sbjct 906 GAAAGGTGCCCACTCCCCACTGGTCCTTTCCTAATAAAAAGAAGGAAAATTGCATCCCCATTGTCCTGAATNAGGTGTCCTTTCTATTTCTGGGGGGGTGGGG 1008

Figure 13 Example of a BLAST sequence alignment demonstrating a positive sequence analysis result. gWiz TL1 positive isolate (Subjct) aligned to the predicted nucleotide sequence of gWiz TL (Query).

Cloning efficiency of the light chain for both Tmab and AdmAb appeared higher than the heavy chain, although compared to other isolates, only 2 mAbX TL isolates were tested which likely does not represent the efficiency derived. Regarding gene insertion, cloning accuracy was 100% for all strategies, yet accuracy with the variable gene switching strategy was highly reduced due to a high degree of background vector containing the original Tmab variable gene insert.

Table 20 Comparison of cloning efficiencies between the different cloning strategies conducted in this study

Isolates Cloning Cloning Vector Yielded Efficiency Accuracy

mAbX TH 59 4/10 (40 %) 3/3 (100 %)

mAbX TL 47 2/2 (100 %) 2/2 (100 %)

gWiz TH 52 3/22 (13.6 %) 3/3 (100%)

gWiz TL > 300 6/11 (54.5 %) 6/6 (100 %)

gWiz AH > 300 7/10 (70 %) 1/7 (14.3 %)

gWiz AL 285 8/10 (80 %) 1/8 (12.5 %)

pVITRO1_Adalimumab_IgG1/κ 1 0/1 (0 %) N/A

The variable gene switching strategy was unfortunately not successful for the dual pVITRO1_Adalimumab_IgG1/κ expression vector. This can be attributed to the incomplete linearised vector backbone product pVITRO1 L/sH truncating the vector re-assembly in the In- Fusion® reaction. A truncated version of the pVITRO1_Adalimumab_IgG1/κ vector may have assembled in the In-Fusion® reaction although its rescue and isolation was not successful, suggesting that the assembly product was non-functional and therefore not useful.

3.5 Preliminary Transient Expression Study The expression of Tmab and AdmAb from the successfully cloned vector expression systems was compared through a pilot transient expression study in suspension 293F culture. pVITRO1_Trastuzumab_IgG1/κ dual expression vector was compared with the co-transfection of mAbX TH/TL, gWiz TH/TL and gWiz AH/AL single expression vectors. To further support the cell viability and protein expression prior to harvest, tryptone was supplemented to all transfection cultures and cultures were scaled up (aside from the mAbX expression system). The harvested culture supernatants were purified through an FPLC system using Protein A affinity chromatography and elution peaks from the FPLC chromatogram analysed to provide an indication of the total protein content purified from each expression systems.

FPLC chromatograms indicated that the gWiz expression vectors were the highest yielding of the three vector expression systems (Figure 14), although the purification strategy between the two antibodies required optimisation. Insufficient column washing post sample injection and broad elution were observed with the gWiz expressed Tmab chromatogram (Figure 14C), which suggests inefficient elution of Tmab off the column. However, the similar peak area between both gWiz expression systems suggests that the quantity of antibody yielded was relatively consistent and that despite inefficient elution, Tmab yield was not compromised. It is noted that a pressure fluctuation in the gWiz AdmAb purification occurred during the column wash post sample loading, although this did not affect the elution of AdmAb.

The pVITRO1_Trastuzumab_IgG1/κ and mAbX expression systems produced considerably less antibody (Figure 14A and B), although it is noted that seeding volume, total cells and quantity of DNA for these transfection reactions were exactly half of those prepared for the gWiz expression systems, and that the harvest time point slightly differed across the expression systems according to cell viability. However, in keeping the cells to DNA ratio consistent, the scale of transfection seeding correlates with protein expression, and that the majority of expression occurs within the first 96 hrs of transfection. It is also noted that the mAbX purification commenced on the FPLC system before the system had stabilised (as indicated by the negative baseline during elution), although this has no effect on the efficiency of the elution of Tmab, or determining the eluting peak area.

B Absorbance at 280 nm 280 at Absorbance

Volume (mL)

Figure 14 FPLC chromatograms of Tmab (blue) and AdmAb (red) purifications from pVITRO1 (A), mAbX (B) and gWiz (C) expression systems. Elution peaks of each chromatogram (inset).

The purified antibody samples were buffer exchanged, concentrated and the total antibody yield derived from UV absorbance at 280 nm. Antibody yields correlated to the elution peak areas (Table 21), further confirming that the gWiz expression system was the most efficient for transient expression of both Tmab and AdmAb. The Tmab yield from the pVITRO1 system was 18% of gWiz system after buffer exchange. The mAbX system yield was lost through the buffer exchange process and was therefore not determined. As the elution peak area for the mAbX system was 8- fold lower than the pVITRO1 system, the theoretical yield could be approximated to 30 µg, which is 2% of the gWiz system.

Table 21 Yield of Tmab and AdmAb generated from the expression vector library in the preliminary transient expression study.

Seed Harvest Peak Area Yield Vector Expression System Volume Volume (mL.mAU) (µg) (mL) (mL)

pVITRO1_Trastuzumab_IgG1/κ 100 31.04 235 30

mAbX TH/TL 30 3.95 ND

gWiz TH/TL 467.34 1300 60 240

gWiz AH/AL 459.27 1175

4 Discussion In developing the expression vector library for the expression of Tmab and AdmAb in mammalian culture, different cloning strategies were required for each vector suite. That is variable region gene switching in the pVITRO1 suite, variable region gene insertion in the mAbX suite and whole gene insertion then variable region gene switching in the gWiz suite. Of the cloning strategies conducted in this project, whole Tmab gene insertion into the blank gWiz vector using traditional restriction cloning and ligation was the least scrupulous and had the highest efficiency. AdmAb variable region gene switching into the gWiz suite was relatively unscrupulous also; however, the method requires optimisation to reduce the number of background isolates captured for sequencing. The gWiz Tmab and AdmAb vector suite produced the highest yield in the transient co-expression

54 system and was deemed the most suitable vector suite for transient expression optimisation leading up to mutagenesis projects.

The cloning efficiency in producing gWiz AH/AL through the variable region gene switching strategy was considerably low, in that many isolates had captured the background vector containing the original Tmab variable gene insert. Unlike with the gene insertion strategies, colony PCR for the variable gene switching strategy is not indicative of the cloning efficiency. In this instance, colony PCR does not differentiate between the complete original Tmab genes and cloned AdmAb genes in order to exclude background vector, but rather it excludes the rare event of isolating incomplete gene/vector constructs. The isolated background vector is attributed to the gWiz TH/TL vectors from the linearising PCRs being retained in the purification of the linearised products, and therefore carried over in residual quantities in the subsequent In-Fusion® reactions. In considering improvement to this cloning method, a separation step of the linearised PCR products on agarose gel prior to purification would have substantially reduced vector background, as was demonstrated with the higher cloning efficiency of the mAbX variable gene insertion cloning strategy. However, with this currently reported cloning method, gWiz AH/AL vectors were successfully isolated and they readily expressed AdmAb in transient co-transfection.

The mAbX variable gene insertion strategy was rather scrupulous; however, yielded high cloning efficiency and accuracy. Interestingly, as a dephosphorylation step was not included in this strategy as a method of preventing religation of linearised vector, a lower cloning efficiency was expected from what we reported. This further infers that agarose separation of linearised product is a more efficient way of removing non-linearised background vector in the final ligation product. Unfortunately primer design required revision before embarking on this strategy as an aa substitution was unintentionally introduced into the TL gene. Further to this, transient co-expression of Tmab using the mAbX system was poor yielding, hence it was determined that this system was not suitable for further investigation and optimisation. pVITRO1_Trastuzumab_IgG1/κ readily expressed Tmab in transient culture, and is a versatile vector in being able to produce stable cell lines. However, the variable gene switching strategy for producing pVITRO1_Adalimumab_IgG1/κ proved unsuccessful. Specifically, the amplification of pVITRO1 L/sH produced an incompletely amplified fragment that was smaller than predicted. Further investigation of the primer design did not give insight to the reason why the primers

55 produced a truncated product from pVITRO1_Trastuzumab_IgG1/κ, as no further binding sites were predicted when manipulating primer hybridization parameters such as primer length, CG content and minimum melting temperature. Considering that the primer set was designed with full homology in the Tmab genes, the primers were completely transferrable between the gWiz TH/TL and pVITRO1_Trastuzumab_IgG1/κ vectors. The incompletely amplified backbone product contained 20 bp homology to the synthetic AdmAb variable region genes as per the primer design, therefore the In-Fusion® reaction was expected to assemble a truncated version of pVITRO1_Adalimumab_IgG1/κ vector, with possible exclusion of the hygromycin resistance gene. It is likely that the outcome of the In-Fusion® reaction assembled a truncated version of pVITRO1_Adalimumab_IgG1/κ which may have lost function, replicability or hygromycin selectivity, rendering the ligation product incapable of being rescued and isolated.

Upon successful cloning, a preliminary transient expression study was conducted in 293F culture to compare yield and quality of antibody produced from the different vector systems. The vector suites had different expression requirements that impact the further considerations for producing libraries of antibody variants. Unlike the pVITRO1 and mAbX vector suites, the gWiz vectors were limited to transient expression. Furthermore, unlike the pVITRO1 suite, the mAbX and gWiz vector suites were single gene vectors that required co-expression of both heavy and light chain vectors for the expression of complete antibody. Therefore, a fair comparison of antibody expression between the vector suites required that all expression systems remained transient and that the ratio of heavy to light chain genes and DNA to cells in transfection culture remained consistent. To further support cell viability and enhance expression, tryptone was supplemented to transfection cultures.

Of the transient expressions conducted in this study, Tmab and AdmAb in the gWiz system was the highest yielding. Tmab expression in the pVITRO1 system was moderately yielding and in the mAbX system, poorly yielding. As the mAbX system was low yielding and the mAbX TL vector isolated for expression contained an aa mismatch due to primer design, we determined that this system was not suitable for further investigation and optimisation.

Investigations reported in Chapter 3 and Chapter 4 expressed Tmab libraries through pVITRO1_Trastuzumab_IgG1/κ before progressing to gWiz TH/TL (Chapter 4 and Chapter 5) and

AdmAb expression in gWiz AH/AL (Chapter 6) vector systems.

Chapter 3 Laboratory Scale Production and Purification of a Therapeutic Antibody

Chapter 3 was published in the Journal of Visualized Experiments:

Elgundi, Z., Sifniotis, V., Reslan, M., Cruz, E., Kayser, V. (2017). Laboratory Scale Production and Purification of a Therapeutic Antibody. Journal of Visualized Experiments, 119: e55153.

Abstract Ensuring the successful production of a therapeutic antibody begins early on in the development process. The first stage is vector expression of the antibody genes followed by stable transfection into a suitable cell line. The stable clones are subjected to screening in order to select those clones with desired production and growth characteristics. This is a critical albeit time-consuming step in the process. This protocol considers vector selection and sourcing of antibody sequences for the expression of a therapeutic antibody. The methods describe preparation of vector DNA for stable transfection of a suspension variant of human embryonic kidney 293 (HEK-293) cell line, using polyethylenimine (PEI). The cells are transfected as adherent cells in serum-containing media to maximize transfection efficiency, and afterwards adapted to serum-free conditions. Large scale production, setup as batch overgrow cultures is used to yield antibody protein that is purified by affinity chromatography using an automated fast protein liquid chromatography (FPLC) instrument. The antibody yields produced by this method can provide sufficient protein to begin initial characterization of the antibody. This may include in vitro assay development or physicochemical characterization to aid in the time-consuming task of clonal screening for lead candidates. This method can be transferable to the development of an expression system for the production of biosimilar antibodies.

1 Introduction The success of therapeutic antibodies continues to drive substantial investment into antibody development as a wave of next generation therapeutics begins. The antibody market is expected to be reshaped by antibody fragments [183], antibody-drug conjugates [184], bispecific antibodies [185] and engineered antibodies with favorable properties [186]. Another class gaining pharmaceutical interest are biosimilars. Biosimilar antibodies are ‘highly similar’ replicate products of a therapeutic antibody that has already received regulatory approval. A proposed biosimilar must be comparable with the originator antibody with respect to its structure, function, animal toxicity, clinical safety and effectiveness, human pharmacokinetics (PK), pharmacodynamics (PD) and immunogenicity [187, 188].

The approval rates of biosimilar antibodies have been slow due to the strict constraints on the final quality of the product. The exact manufacturing processes such as specific cell lines and culturing conditions through to the final processing steps can remain proprietary. What is more, the manufacturing of antibodies inherently involves a degree of variability which can add to the 58 challenge of producing a highly similar product. A comprehensive physiochemical and biophysical characterization and comparison is quite difficult, yet a number of studies demonstrating the characteristics of biosimilar antibodies are emerging in the literature [189-191].

Generating a therapeutic antibody begins with transfection of mammalian host cells with a vector carrying the genes for the respective antibody. Vector design, cell line and culture conditions are key considerations for setting up the expression system.

The DNA sequences of antibodies can be sourced from Drug Bank (www.drugbank.ca), IMGT (www.igmt.org) or research publications including patents. For example, the sequence of trastuzumab is available through Drug Bank (DB ID: DB00072). The amino acid sequence of the variable regions can undergo gene design and optimization for synthesis in the desired host species. It is important for a biosimilar antibody that no modification is made to the amino acid sequence. Once synthesized, antibody genes can be subcloned into the appropriate vector of choice.

Human IgG antibodies consist of two identical heavy chains and two identical light chains. Tightly regulated expression of both chains is essential for optimal production of heterologous IgG protein in mammalian cells [192]. Intra- as well as inter-chain disulfide bonds have to be formed and a number of post-translational modifications have to be introduced during protein biosynthesis. A number of vectors are available that have been designed specifically to express antibody genes (refer to Table S1 in supplementary information). These antibody-specific vectors usually express the constant regions for both heavy and light chains so only the variable regions of each chain require cloning.

Transfection of cells with two independent constructs (co-transfection) is the most common approach for delivering heavy and light chain-encoding genes. That is, each gene is driven by its own promoter and transcribed as separate antibody chains before being assembled in the endoplasmic reticulum. On the other hand, multi-cistronic vectors have internal ribosome entry site (IRES) elements incorporated that allow expression of multiple genes as a single mRNA transcript with translation permitted from internal regions of the mRNA [193]. In this instance, the heavy and light chain-encoding genes are coupled in an arrangement to achieve co-expression of both antibody chains [192, 194].

While transiently transfected cells yield sufficient protein to perform a limited number of experiments, stably transfected cell lines that have undergone selection for genome integration can

59 deliver higher yields. Higher protein amounts allow for assay development relating to in vitro characterization and can provide an indication of antibody quality in consideration for downstream applications such as clonal cell line and lead candidate selection.

The goal of this article is to describe the stable expression and purification of a therapeutic antibody produced in a mammalian expression system. Indeed, this method can be applied to the expression of a biosimilar antibody. The method can be used for the initial characterization of antibodies before proceeding on to the critical, albeit time-consuming steps of identifying a desirable clone for larger scale manufacturing. Moreover, this method can be used to express other proteins and not just antibodies.

The following detailed protocol describes the expression of therapeutic antibody trastuzumab. This consists of preparation of vector DNA followed by stable transfection in HEK-293 cell line and purification of antibody protein by an automated chromatographic method.

2 Protocol Note: A suitable mammalian expression vector must be used for this protocol. Here, a single construct containing two expression cassettes is used (i.e. heavy and light chain expression is driven by separate promoters). Trastuzumab heavy and light chains were previously cloned into the vector. This vector was a gift from Andrew Beavil, obtained through a not-for-profit plasmid repository [178].

2.1 Recovery and Scale-up of Vector DNA Note: Vector DNA was received as a soft agar stab culture in Escherichia coli XL-1 Blue strain; vector carries hygromycin resistance.

2.1.1 Insert a sterile inoculation stab or loop into the soft agar of the stab culture then streak a Luria Bertani (LB) agar plate prepared with 75 μg/mL hygromycin for isolated colonies. Incubate the plate at 37 °C for 18-24 h. 2.1.2 Inoculate a single colony into 5 mL Terrific broth (TB) containing 75 μg/mL hygromycin. Incubate the culture at 37 °C for 18-24 h with 225 rpm shaking. 2.1.3 Use the overnight culture to prepare a glycerol stock of vector DNA by gently mixing 800 μL of culture with 200 μL 80% glycerol in a cryovial and freeze at - 80 °C. 2.1.3.1 Add 100 μL of the overnight culture to 100 mL TB containing 75 μg/mL hygromycin in a

60 baffled shaker flask (i.e. 1/1000 dilution). Incubate the culture at 37 °C for 18 - 24 hr with 225 rpm shaking. 2.1.4 After overnight culture, extract and purify the DNA according to manufacturer’s instructions of midi/maxi preparation kit with the following exception; during the final step, elute or resuspend DNA with water (pH 7.0-8.5). 2.1.5 Check concentration and purity of DNA by absorbance readings at 260 and 280 nm then store DNA at -20 °C. Note: Optional (highly recommended): Sequence DNA using vector-specific primers to confirm identity.

2.2 Stable Transfection of HEK-293 Cells

2.2.1 Grow and maintain HEK-293 cells (suspension cells) according to standard protocols in serum-free media supplemented with 0.1% non-ionic surfactant in baffled flasks. Maintain at 2x105 cells/mL and subculture every fourth day. Culture cells at 37 °C with 5% CO2 and 120 rpm rotation. Note: Cells should have been in culture for no less than 4 days and no more than 4 weeks prior to transfection. HEK-293 cells grown as monolayers in serum-containing media can also be used for this procedure. 2.2.2 On the day prior to transfection, seed HEK-293 cells at 3x105 cells/mL in wells of a 12-well plate in 2 mL of Dulbecco's Modified Eagle Media (DMEM) supplemented with 10% heat- inactivated fetal bovine serum (FBS). On the day of transfection, check that cells have reached 80- 90% confluency 2.2.3 Dilute DNA and polyethylenimine (PEI) separately in transfection media then mix together. 2.2.3.1 Dilute 1.25 μg vector DNA per well (15 μg for 12 wells) in 300 μL transfection media. Incubate at room temperature for 5 min. 2.2.3.2 Dilute 2.5 μL of 1 mg/mL PEI solution (30 μL for 12 wells; i.e. 1:2 ratio of DNA to PEI) in 300 μL transfection media. Incubate at room temperature for 5 min. 2.2.3.3 Add diluted DNA to diluted PEI, gently mix and incubate at room temperature for 15 min. 2.2.4 Add 50 μL of DNA/PEI mixture dropwise to each well of the plate. Rock the plate gently to distribute the transfection mixture then place the plate at 37 °C, 5% CO2 for 24 h.

2.2.5 Add 50 μg/mL hygromycin B per well and return plate to 37 °C, 5% CO2 for 10 days.

Note: By Day 10, un-transfected cells will have died and detached from the surface of the plate, whereas stably-transfected cells will be viable and attached to the plate. 2.2.6 Replace the media in wells with ¼ volume serum-free media and ¾ volume DMEM supplemented with 10% FBS (i.e. 0.5 mL serum-free media + 1.5 ml DMEM = 7.5% final serum concentration) and incubate for 4 days. 2.2.7 Replace the media in wells with ½ volume serum-free media and ½ volume DMEM supplemented with 10% FBS (i.e. 1 mL serum-free media + 1 mL DMEM = 5% final serum concentration) and incubate for 4 days. 2.2.8 Replace the media in wells with ¾ volume serum-free media and ¼ volume DMEM supplemented with 10% FBS (i.e. 1.5 mL serum-free media + 0.5 mL DMEM = 2.5% final serum concentration) and incubate for 4 days. 2.2.9 Replace the media in wells with 2 mL of serum-free media supplemented with 0.1% non-ionic surfactant (i.e. 0% final serum concentration) and incubate for 4 days. Note: Maintain 50 μg/mL hygromycin B selective pressure on cells during serum-free adaptation. Cells adapted to serum-free media detach from the surface of wells and may cluster in suspension. Refer to step 2.2.1 for culture conditions with the additional supplement of 50 μg/mL hygromycin B. 2.2.10 Using a pipette, gently mix the suspension cultures in the 12-well plate and pool cells into a separate tube. 2.2.10.1 Using a hemocytometer, enumerate pooled cells and seed in 30 mL serum-free media in a baffled flask at 2x105 cells/ml. Culture cells at 37 °C with 5% CO2 and 120 rpm rotation. Subculture and expand every fourth day. 2.2.11 Continue to expand culture size to required cell density to obtain 10 vials at 1x107 cells/vial for cryopreservation in liquid nitrogen. Freeze the cells in serum-free media containing 10% dimethylsulfoxide (DMSO). Note: Conditioned media containing antibody can be harvested at each subculture to confirm protein production or used to optimize purification conditions. To harvest conditioned media and maintain cells for subculture, centrifuge the culture at 300 x g for 5 min then filter-sterilize supernatant through a 0.22 μm filter. Filtered supernatant may be kept at 4 °C for 1-2 weeks or -20 °C for longer term storage.

2.3 Large Scale Antibody Production from Batch Overgrow Culture Note: Antibody production can be followed on from step 2.2.11 where existing cells are expanded to the required cell density based on the batch overgrow culture volume to be setup. Otherwise, cells that have been thawed from cryopreservation begin at this stage once expanded to an appropriate cell density and volume. Various culture supplements may be used to optimize antibody production; the use of tryptone to increase antibody yield is demonstrated in this protocol.

2.3.1 Seed cells at 2x105 cells/ml in 100 mL serum-free media in two baffled flasks (to compare antibody yields from un-supplemented and nutrient-supplemented cultures). Culture at 37 °C, 5%

CO2 and 120 rpm. 2.3.2 After 24 h, add tryptone to a final concentration of 0.5% to nutrient-supplemented culture. Equalize volume of un-supplemented culture with serum-free media. 2.3.3 From day 8, perform cell counts of the cultures daily using a hemocytometer to monitor cell viability. Harvest the culture supernatants once cell viability is less than 80%. 2.3.3.1 Harvest supernatant by centrifugation of the cultures at 3000 x g for 15 min then filter- sterilize the supernatant (containing antibody) through a 0.22 μm filter. Store the supernatants at 4 °C (short-term) or freeze at -20 °C (long-term). 2.4 Antibody Purification by Affinity Chromatography Using an Automated Fast Protein Liquid Chromatography (FPLC) System Note: The following procedure can generally be applied to most automated systems. Purification can be performed at room temperature or at 4 °C (if FPLC system is kept in a cool room). A series of scouting tests can be performed to identify the optimal purification conditions including appropriate column matrix, binding buffer, elution buffer and pH to ensure maximum recovery of purified antibody from conditioned media (refer to Results section). The optimal conditions are dependent on the antibody or protein being purified. Purifications were performed on an automated FPLC system. Purifications were performed at room temperature using a 5 ml Protein A column.

2.4.1 Prepare the following buffers using ultrapure water and adjust to the recommended pH then filter through a 0.22 μm filter. 2.4.1.1 Prepare 1 L of phosphate buffered saline (PBS) pH 7.4 (binding buffer) by mixing the following: 0.14 M NaCl, 0.0027 M KCl, 0.01 M Na2HPO4 and 0.001 M KH2PO4. Adjust pH if necessary before filtering the buffer.

2.4.1.2 Prepare 500 mL of 0.1 M glycine-HCl pH 2.7 (elution buffer). Adjust pH before filtering the buffer. 2.4.1.3 Prepare 50 mL of 1 M Tris pH 9.0 (neutralization buffer). 2.4.2 Ensure all system power and communication connections with computer are made. Devices should be visible in the software ‘System Control’ module. Ensure UV cell is set at 280 nm wavelength. Optional: Calibrate pH meter if connected and to be used. 2.4.3 Immerse inlet tubes of A, B and sample pump in ultrapure water to wash the system. Purge the pumps with a syringe if there is air in the tubing or a suspicion of air in the system. Wash the system with water by manual operation via the system controller or via an automated method. Observe that pressure, conductivity, UV280 tracing and pH remain consistent and that the pressure limit for the column is not exceeded as per manufacturer’s specification. Note: Abnormalities may indicate air or blockage in the system that should be addressed before proceeding. 2.4.4 In the system controller module, start the flowrate at 1 mL/min manually via pump A in either load (bypassing sample loop) or inject (through the sample loop) position to begin connecting the column. Disconnect system tubing at the column position inlet and remove the stopper connected to the column inlet. 2.4.4.1 Allow water to flow from the system tubing dropwise on top of the column then attach the system tubing to the column. Having the column inlet and inlet fitting overflowing with drops of water ensures a connection free of air bubbles. 2.4.5 Attach the column outlet at the downstream outlet and verify all fittings are tightly fastened. With the column now attached to the system, do not exceed limits for maximum pressure limit and flowrate as outlined in column specifications. 2.4.6 Wash column with 5 column volumes (CV) of water. If necessary, continue column wash until UV280 tracing has stabilized. 2.4.7 Immerse inlet tubes of A in Binding buffer, B in Elution buffer and sample pump in Binding buffer to equilibrate the system in correct buffers. Fill the inlet tubes with buffer using PumpWash by manual operation. 2.4.8 In the ‘Method Editor’ module of the software, use the method wizard to setup an affinity chromatography method for the column that is intended to be used.

Note: Automated FPLC systems come with methods pre-filled with the recommended settings (i.e. flowrate and pressure limit) and run steps based on the column (manufacturer, matrix and size) selected for purification. 2.4.8.1 Use the following run steps for Protein A affinity chromatography: 2.4.8.1.1 Equilibrate system with 5 CV of Binding buffer and collect flowthrough into waste container. 2.4.8.1.2 Load sample onto the column (volume to be loaded is specified manually in the method) and collect flowthrough into a separate container. 2.4.8.1.3 Wash system with 5 CV of Binding buffer and collect flowthrough into a separate container. 2.4.8.1.4 Elute column with 5 CV isocratic fractionation using Elution buffer and collect purified antibody sample as fractions by fraction collector. 2.4.8.1.5 Equilibrate system with 5 CV of Binding buffer and collect flowthrough into waste container. 2.4.9 Once the method has been setup, specify the volume of conditioned media to be applied to the column and save the method. Note: The specified volume in the method should be 5-10 mL less than the actual volume to avoid introduction of air during the sample run. The specified volume used in this protocol is 90-120 mL. 2.4.10 Submerge the sample pump tubing into the vessel containing the conditioned media. 2.4.11 Prepare a separate container to collect sample flowthrough via the specified outlet tubing. Note: It is important to collect the flowthrough in the case an error occurs during the run or the binding capacity of the column is exceeded requiring the flowthrough to be reapplied to the column or the purification repeated. 2.4.12 Prepare collection tubes in the fraction collector. Add 100 μL of neutralization buffer per 1 mL fraction volume. The FPLC system will automatically elute the fractions into the collection tubes. 2.4.13 In the ‘System Control’ module of the software, open the method to be run. Note: The method run is initiated in a series of pages that include checking the variables of the method, fraction collector setup and defining result file name and storage location. 2.4.14 Click START to initiate the run. The run can be monitored in the ‘System Control’ module.

2.4.15 At the completion of the purification run, check the resulting chromatogram in the ‘Evaluation’ module in the system software. 2.4.16 Combine all protein-containing fractions into a separate tube, buffer exchange and concentrate in PBS using a centrifugal filter device with 30 kDa molecular weight cut off. Perform centrifugation steps according to manufacturer’s instructions. 2.4.17 Measure antibody concentration using bicinchoninic acid assay (BCA) according to manufacturer’s instructions. 2.4.18 If another purification run is to be performed, prime the sample pump tubing in Binding buffer and repeat steps 2.4.9-2.4.14. 2.4.19 At the completion of purification runs, immerse inlet tubes of A, B and sample pump in ultrapure water and wash the system and column as performed in step 2.4.3. 2.4.20 Submerge A, B and sample pump tubing in 20% ethanol and repeat wash procedure for storage of the system and column. 2.4.21 Disconnect the column from the downstream outlet and replace column stopper, then disconnect column at the inlet and replace stopper, reconnect tubing to the system. Store the column at 4 °C. 3 Representative Results Stable production of trastuzumab by transfected HEK-293 cells was confirmed using bio-layer interferometry (BLI) as presented in Figure 15. An IgG standard curve was generated by measuring the binding rate between an IgG antibody standard and protein A biosensor (Figure 15A). The crude supernatant sample was similarly measured, then its concentration interpolated from the standard curve (Figure 15B). The supernatant concentration was measured to be approximately 25 μg/mL (sampled from 50 mL collected at subculture) two weeks after serum-free adaptation and prior to setting up cells for overgrow batch cultures. Supernatant collected at each subculture after serum-free adaptation was pooled and used to optimize purification conditions; scouting of binding and elution buffers was performed as shown in Figure 16. The steps of purification are depicted in Figure 16A based on UV280 signal. Once the Protein A column equilibrates, an increase in UV280 signal represents sample loading; this increase is due to supernatant flowthrough passing through the column (containing unbound proteins which absorb UV light at this wavelength), while antibodies have been captured by the column. Once the sample has finished loading, the UV280 signal returns to baseline as any remaining unbound proteins are washed away. Elution occurs as

66 pH decreases to the optimal pH to release the antibodies captured by the column, depicted by an increase in UV280 signal with eluted antibodies fractionated and collected by the fraction collector. Throughout the sample run, conductivity changes based on the salt concentration in the buffers whereas system pressure should remain relatively constant.

Optimal elution, pH and buffer were scouted first (Figure 16B); it was observed that antibody eluted more efficiently off the column at lower pH using either 0.1 M glycine HCl or citric acid. However, below pH 2.7 there was no further improvement in the elution profile. In addition, elution peaks with 0.1 M glycine HCl appeared less broadened when compared with 0.1 M citric acid at similar pH (Figure 16B). Subsequently, 0.1 M glycine-HCl pH 2.7 buffer was used to optimize binding buffer conditions (Figure 16A). The effect of different binding buffers on elution was also compared (Figure 16A). It was observed that PBS pH 7.4 and 20 mM sodium phosphate pH 7 had comparable elution profiles, while the addition of 3 M NaCl to sodium phosphate buffer (to enhance antibody binding to the column) was not favorable. Due to the ease of PBS buffer preparation, PBS pH 7.4 was selected as the binding buffer for future purifications.

The purification chromatograms of batch overgrow cultures are shown in Figure 17; tryptone- supplemented culture is compared to un-supplemented culture. It was noted that the addition of tryptone to the supplemented culture resulted in an increased UV280 signal during the sample loading step, compared to the un-supplemented culture. The tryptone-supplemented culture yielded 3.8 mg and the un-supplemented culture 1.7 mg of trastuzumab, based on protein recovered after buffer exchange and concentration. Quality control of the purified antibody was confirmed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) (Figure 18). Trastuzumab grown in un-supplemented and tryptone-supplemented conditions results in similar antibody profiles under non-reducing and reducing conditions. As expected, a prominent band at approximately 150 kDa confirms correctly folded antibody; other bands represent fragmented forms of the antibody, a result of disulfide bond breakage and a method-induced artefact. Likewise, two

67 bands at 50 and approximately25 kDa confirm the presence of antibody heavy and light chains, respectively.

Figure 15 Confirmation of protein production by stably transfected HEK293-F cells using BLItz (bio- layer interferometry). (A) BLItz sensorgrams of antibody standards analysed in a 1:1 binding model to determine binding rates at each concentration. Binding curves of standards (grey) and crude supernatant sample of PEI-transfected trastuzumab (PEI-Tmab; black) to protein A biosensor was measured over 120 seconds. (B) Antibody standard curve of concentration versus binding rate, used to interpolate concentration of antibody protein in crude supernatant.

Figure 16 Purification chromatograms of elution and binding buffer scouting experiments. (A) The steps of purification are depicted according to UV280 signal (blue, green or black lines); sample loading onto the column followed by column wash to wash away unbound proteins then elution of bound protein from the column. Measurement of pH (red line), conductivity (brown line) and system pressure (grey line) is monitored throughout the run. Three buffers were tested for optimal binding of trastuzumab to the column and washing away of unbound protein to maximize elution yield. (B) 0.1 M glycine-HCl (pH 2.5-3.3) and 0.1 M citric acid (pH 2.5-3.5) buffers were tested for optimal elution of trastuzumab from protein A column.

Figure 17 Purification chromatograms of trastuzumab (TmAb) batch overgrow cultures grown with no supplement or supplemented with tryptone. Stably transfected HEK-293 cells were setup at 2x105 cells/ml and cultured for 8 days either un-supplemented (120 mL; black line) or supplemented with 0.4% tryptone (90 mL; green line). After harvesting, supernatants were purified using PBS pH 7.4 (binding buffer) and 0.1 M glycine-HCl pH 2.7 (elution buffer).

Figure 18 Analysis of purified trastuzumab by SDS-PAGE. Approximately 2 µg of trastuzumab purified from un-supplemented (-) or tryptone-supplemented (+) cultures was sampled by 10% SDS-PAGE under non-reduced or reduced conditions. The gel is stained with coomassie R-250.

4 Discussion This protocol details the transfection, stable expression and purification of a therapeutic antibody in HEK-293 cells. Stable expression of antibody genes is the first step in generating an antibody- producing cell line for the development and manufacture of a therapeutic antibody. While Chinese hamster ovary (CHO) cells remain the expression platform of choice for therapeutic proteins, the HEK-293 cell line is gaining prominence with the realization that proteins produced in these cells are a closer match to naturally occurring human proteins, in terms of post-translational modifications and function [195, 196].

Mammalian cell lines including CHO (for example, CHO-DG44 and CHO-K1) as well as non- immunoglobulin secreting murine B cell lines NS0 and SP2/0 are predominantly used in biopharmaceutical production. Expression systems based on these cell lines are based on selection processes whereby high producing clones are induced and selected through incremental addition of a specific selective drug [197, 198]. The selection process for a stable clone can therefore be arduous and time-consuming. In comparison to these existing methods, the HEK-293 cell line robustly transfects with high efficiency, the selection process is simplified and very easily adapts to serum-free suspension culture, making it an ideal expression system for laboratory scale production of proteins [199].

A limitation of stable transfection is time in which a practical amount of protein can be produced and utilized in experiments. To try and overcome this, the current protocol describes a culturing step that can achieve substantial amounts of protein within 3-6 weeks of transfection. The batch overgrow cultures (large scale production) provide the opportunity to produce substantial amounts of antibody for the setup of in vitro characterization experiments in the lead up to selection of a clonal cell line and lead candidates. This has clear advantages over transient transfection which can require higher amounts of DNA and reagents, and protein yield is not reproducible from batch to batch since expression is only temporary and determined by transfection efficiency.

71 on comparison of the antibody bands under non-reducing and reducing conditions (Figure 18). The addition of tryptone to the cultures is optional and is particularly used to achieve higher protein yields. Other supplements have been considered to enhance protein expression such as sodium butyrate and valproic acid [97, 202, 203].

The first checkpoint of the protocol is the confirmation that protein is being produced by the transfected cell line. This step may be performed any time after the two-week period of selective pressure used to select stable transformants. A number of methods may be used to confirm protein production including bio-layer interferometry (Figure 15) or the similar approach of an IgG ELISA. Alternatively, a western blot may be performed using an anti-IgG detection antibody to identify antibody protein in the crude supernatant or purified sample.

Other researchers have shown that higher transfection efficiency is achieved using adherent cells in serum-containing media [204]. The presence of supplements from animal origin is undesirable for biopharmaceutical production therefore, the sequential adaptation to serum-free media is a critical step in the protocol requiring patience and careful monitoring of cell viability. The 4-day turnover period suggested in this protocol is a guide and so if problems are encountered at a certain serum concentration, it is recommended that the cells be subcultured 2-3 times in the previous ratio of serum-containing to serum-free media before continuing with the next ratio. Prior to setting up batch cultures and cryopreservation of cells, consider that most cell lines are deemed fully adapted after three subcultures in 100% serum-free media.

One of the most crucial steps at the purification stage is the scouting for optimal conditions and buffers to ensure efficient binding of the antibody to the column and then complete elution off the column (Figure 16). The conditioned media collected at each subculture is useful for this purpose. In this case, the elution buffer of citric acid pH 3.5 generally recommended for protein A affinity chromatography was unsuitable for elution of trastuzumab from the column. The scouting experiment using two different elution buffers at various pH clearly showed that even within the narrow range of pH tested, the degree of elution was affected (Figure 16B).

In terms of downstream processing of the antibody fractions after purification, the antibody can be buffer-exchanged using a desalting column either by manual or automated operation. Alternatively, dialysis membrane can also be used. Final protein concentration can be determined by other protein quantification methods including measurement of the absorbance signal at 280 nm with

72 concentration deduced from Beer Lambert’s Law based on the extinction coefficient of the antibody.

This protocol has successfully produced antibody protein of trastuzumab by stable transfection of HEK-293 cells. The antibody was purified and characterized to confirm integrity. The steps in this protocol detailing DNA preparation, stable transfection followed by serum-free adaptation, large scale production and automated purification can be transferred to producing other proteins.

Acknowledgements The research was supported by the University of Sydney. pVITRO1-Trastuzumab-IgG1/κ was a gift from Andrew Beavil (Addgene plasmid # 61883). We thank Tihomir S. Dodev for useful discussions regarding pVITRO1-Trastuzumab-IgG1/κ.

Disclosures The authors declare that they have no competing financial interests.

Chapter 4 Rationalised Residue Substitution of Trastuzumab for Enhanced Affinity to Target HER2

Abstract This chapter reports on the development of a library of trastuzumab mutants that had predicted enhanced affinity toward target HER2 using in silico methods. Three of the reported five mutants were successfully produced in optimised transient 293F expression systems and tested in vitro for comparison of relative affinity and binding kinetics.

Residue substitution predictions, molecular dynamic (MD) simulations, free energy perturbation and thermodynamic integration calculations were conducted by our collaborators from A/Prof Serdar Kuyucak’s Computational Biophysics Group, School of Physics, The University of Sydney. Their group reported highly promising residue substitutions in the trastuzumab CDRs and Fab region for experimental validation. We produced three of their reported candidates for screening; compared and optimised site-directed mutagenesis protocols in the lead up to expression of the trastuzumab affinity mutant library for in vitro testing. Trastuzumab affinity mutants were compared for binding to extracellularly expressed HER2 in a cell-binding titration assay and real- time binding kinetics to immobilized HER2 through surface plasmon resonance.

Interestingly, as in silico results were preliminary at the time of reporting lead candidates to our group for investigation, experiments gave varying results which were later validated with the completion of the in silico analysis. One mutation disrupted trastuzumabs affinity to HER2 quite considerably and the other two showed negligible affinity enhancement experimentally which confirmed the in silico results in that the mutations did not show conclusive affinity enhancement to HER2.

The findings of this study suggest that in silico prediction correlates with in vitro analysis of an affinity disrupting mutation and that native inter and intramolecular interaction disruptions are a prerequisite part of MD simulation analysis for electing lead affinity mutation candidates. Two substitution mutations for affinity enhancement were unsuccessful after several rounds of mutagenesis and were therefore unable to corroborate a correlation between in silico prediction and in vitro analysis for affinity enhancement.

1 Introduction; Trastuzumab Affinity to HER2 for the Treatment of HER2 Positive Breast Cancer As outlined in Chapter 1, Trastuzumab (Herceptin®) is a humanized IgG1/κ monoclonal antibody used therapeutically for the treatment of HER2 positive breast cancers [21, 22]. HER2 is a cell 75 receptor that is overexpressed in approximately 20% of breast cancers and contains an intracellular and extracellular region. HER2 can dimerize with several activated receptors and this dimerization activates downstream pathways that promote cell proliferation, motility and survival [18, 21, 22]. Trastuzumab (Tmab) acts by binding to extracellular subdomain IV of HER2, which specifically inhibits HER2-HER3 dimerization, independent of HER3 ligand induction [22]. The consequence of this is the downregulation of the PI3K/AKT pathways that promote cell survival and proliferation [22]. Tmab being an IgG1/κ class antibody, its binding to the extracellular region of HER2 also specifically targets HER2 positive cancer cells for antibody-dependant cell-mediated cytotoxicity (ADCC) [21]. Tmabs relative binding affinity to the extracellular domain of HER2 has been screened through many assays, including cell-binding assays, enzyme-linked immunosorbent assays (ELISA), bio-layer interferometry (BLI) and surface plasmon resonance (SPR) [205-209]. The equilibrium binding constant (KD) for Tmab to HER2 has been experimentally determined to be very strong in the subnanomolar range, spanning from 19 pM – 1 nM [19, 205, 209].

Interestingly, in the context of HER2 breast cancer treatment, in vivo studies have shown that biodistribution and tumour retention was poorer in anti-HER2 biotherapeutics of higher affinity [23, 210, 211]. This association was confirmed with biotherapeutics that bound to the same HER2 epitope, where affinity to HER2 was manipulated through mutation of the biotherapeutic [211-213]. Biodistribution was highly dependent on the molecular size of the biotherapeutic and cellular uptake was preferential in the tumour periphery for all biotherapeutics of high affinity [210, 211, 214]. Specifically, biodistribution and tumour retention did not increase significantly in anti-HER single-chain Fv antibody fragments with an affinity stronger that 1 nM KD [213]. And similarly, the biodistribution of whole anti-HER2 antibodies of high affinity was confined to the tumour perivascular space as compared to lower affinity antibodies that had a broader distribution pattern and were achieving further penetration in the tumour [211].

Manipulation of Tmabs relative affinity to HER2 is of ongoing interest, as different therapeutic platforms can be developed for improved treatment of HER2 positive breast cancers. Fragment Tmab and Tmab-drug conjugate platforms would favour enhanced relative affinity for efficient catabolism and delivery of cytotoxic drug to target cancer cells [23]. Tmab with enhanced ADCC functions, extended half-life, and Tmab decorated nanoparticles would favour a reduced relative affinity for better biodistribution and tumour retention leading to a more insistent ADCC effect.

This study was designed to produce lead Tmab mutant candidates with rationally designed residue substitutions to have enhanced relative affinity to HER2. The overall aim was to produce a Tmab affinity mutant library based on in silico analysis and to corroborate these results with in vitro experimental validation.

1.1 In silico Substitution Predictions and Analysis of Tmab Preliminary in silico results produced from A/Prof Serdar Kuyucak’s Computational Biophysics Group (unpublished) were reported to us as warranted for experimental investigation. Firstly, their team analysed the binding interface of the crystal structure of Tmab Fab - HER2 fragment complex (PDB ID: 1N8Z) (Figure 19). Residues in the Tmab sequence that mismatch HER2 residues were identified as well as potential unutilised bonding sites of HER2 residues in close proximity to Tmab residues (Table 22). Suggested residue substitutions in the trastuzumab sequences were then trialled in MD simulations to detect the likelihood of newly formed bonds between the substitutions and HER2 residues. MD simulations were also conducted on the substitutions to confirm that the important intermolecular interactions from the complex were not disrupted and that intramolecular interactions within the Tmab Fab were not disrupted. For MD simulation results on substitutions that showed that newly formed bonds were likely, further free energy perturbation and thermodynamic integration calculations were conducted to determine the change in the binding free energy.

Residues in the binding interface between Tmab Fab and HER2 that contribute to the binding between trastuzumab and HER2 is outlined in Figure 19. Several potential substitutions that were identified for investigation is outlined in Table 22. In one case (D28), a coulomb repulsion interaction was converted to an attraction interaction by replacing the Tmab amino acid (aa) residue to the opposite polarity; i.e. from acidic aspartic acid (D) to basic arginine (R) or lysine (K).

In another case (S50), an introduction of electrostatic interaction or creation of a hydrophobic pocket was possible for the same site. In the instance of producing a hydrophobic pocket, surrounding aa in the Tmab sequence contained aromatic side chains which were predicted to cause steric hindrance at the binding interface with the addition of a further aromatic. Therefore the group proposed aa switching of tyrosine (Y) at position 49 and serine (S) at position 50 as opposed to the addition of a further aromatic in that region. In the instance of introducing an electrostatic interaction, the Tmab aa residue was replaced with an acidic aa residue; i.e. aspartic acid (D) or

77 glutamic acid (E); however, E was preferred as the longer sidechain would have closer proximity to HER2 aa residue to establish a hydrogen bond. In one further case (D102), this site being surrounded by aromatic residues from HER2 would create a hydrophobic pocket by replacing the Tmab aa reside to an aromatic i.e. phenylalanine (F). Unlike the case of S50, the Tmab aa sequence at the D102 site was not immediately surrounded by aromatics to be affected by steric hindrance, instead the residues at that site would directly interact and bury within the aromatics from a nearby loop of the HER2.

Preliminary MD simulations indicated that all suggested mutations did not disrupt the important intermolecular interactions between the Tmab and HER2 in the complex (not shown). However, aside from S50>E substitution, all other substitutions did not disrupt the intramolecular interactions within the Fab (not shown). Unfortunately S50 had an intramolecular interaction with V33 that destabilised with the substitution to E (not shown).

Figure 19 Portion of the crystal structure of the Tmab Fab – HER2 complex (1N8Z) zoomed in to the binding interface. TL Fab (yellow), TH Fab (orange), HER2 (green) and pairwise interactions between the contact residues (red dashed lines).

Table 22 Summary of reported Tmab aa substitutions identified for experimental validation. *Preferred aa due to the length of the sidechain. ^Additional electrostatic interaction identified.

Substitutions HER2 Tmab Proposed Interaction and Comments identified Remove coulomb repulsion interaction; replace E598 D28 R, K acidic to basic aa Introduce electrostatic interaction; replace to K593 D, E* basic aa Create hydrophobic pocket (recruiting Y49, I591 S50 F53, G103 and F104); replace S50 to aromatic P571 Y aa; consider steric hindrance in the region; K593^ switch Y49/S50 F573 Create a hydrophobic pocket; replace to V575 D102 F aromatic aa Y588

MD simulations of Tmab electrostatic substitution residues to the aa residues in HER2 (Figure 20) showed that of all electrostatic interactions, only D28>R introduced new interactions, that being stable hydrogen bonds presented in the MD simulations as molecular distance stabilisation at 3Å (Figure 20C and D). However, molecular distance instability as presented in Figure 20A and B suggested no newly introduced interactions through the D28>K and S50>E substitutions, respectively. D28>R was a lead candidate for binding free energy calculations; D28>R was predicted to enhance the affinity of Tmab to HER2 by - 5.86 kcal/mol, which is a substantial 43% increase as compared to Tmab WT to HER2 of - 13.6 kcal/mol.

MD simulations of Tmab aromatic substitution residues to the aa residues in HER2 showed that only Y49/S50>S49/Y50 introduced a stable hydrophobic and electrostatic interaction, along with key aa recruited for the hydrophobic pocket (Figure 21 and Table 23). HER2 I591 produced stable hydrophobic interactions with Tmab S50>Y and F104 (Figure 21A) and HER2 K593 produced stable hydrogen bonds with S50>Y and G103 (Figure 21B), presented in the MD simulations as molecular distance stabilisation at 4Å and 3Å, respectively. However, molecular distance instability as presented in Figure 21C suggested no newly introduced interactions through the D102>F substitution. The changes in molecular distances from key interactions observed for the mutations corroborate the MD simulation findings, in that Y49/S50>S49/Y50 introduced new interactions,

whereas D102>F did not (Table 23). Y49/S50>S49/Y50 is a lead candidate for binding free energy calculations that will be pursued as a future project with A/Prof Serdar Kuyucak’s Computational Biophysics Group.

A B 20 7 6 15 5

4 10 3

5 2

)

Å 0 0 5 10 15 20 25 0 5 10 15 20 25 30 35 C D 15 15

Molecular ( Distance 10 10

5 5 0

0 0 10 20 30 40 50 0 10 20 30 40 50 Time (ns)

Figure 20 MD simulations of Tmab electrostatic substitutions Tmab D28>K-NZ to HER2 E598- OE2 (A), Tmab S50>E-O2 to HER2 K593-NZ (B), Tmab D28>R-N1 to HER2 E598-O1/2 (C) and Tmab D28>R-N2 to HER2 E598-O1/2 (D). Stability at 3Å for residues at R28 (C and D) suggests successful hydrogen bond formation.

A B 15 10

8 10 6

4 5

Molecular Distance 2

0 0 0 10 20 30 40 50 0 10 20 30 40 50 Time (ns) 20 C 18 16 14 12 10 8

6 Molecular Distance 4 2 0 0 10 20 30 40 50 60 70 Time (ns)

Figure 21 MD simulations of Tmab aromatic substitutions S50>Y-Cε1 (and F104-Cδ1) to HER2 E591- Cδ (A), S50>Y-OH (and G103-O) to HER2 K593-NZ (B), and D102>F-Cδ1/ε1/δ2 to HER2 V575-Cγ1, F575-Cδ1 and Y588-Cε2 (C). Stability at 4Å for residue Y50 (A) suggests stable hydrophobic interaction and 3Å for residue at Y50 (B) suggests successful hydrogen bond formation.

Table 23 Summary of molecular distances between the residues of interest. Average molecular distances calculated from MD simulations, *introduced interactions through mutation, no data (ND).

Crystal MD Structure Tmab MD Average HER2 Tmab WT Average Molecular Mutants (Å) (Å) Distance (Å)

* I591-Cδ Y49-Cε1 7.5 4.7 ± 1.1 Y50-Cε1 4.7 ± 0.9 P571-Cγ Y49-Cε2 11.0 13.3 ± 0.7 Y50-Cε2 6.0 ± 0.6 * K593-NZ Y49-OH 8.6 9.4 ± 0.5 Y50-OH 3.0 ± 0.2

F573-Cδ1 D102-O2 7.0 9.7 ± 1.4 F102- Cε1 7.5 ± 1.2 V575-Cγ1 D102-O2 8.7 8.6 ± 2.1 F102- Cδ1 6.5 ± 1.2 Y588-Cε2 D102-O2 ND 12.5 ± 1.9 F102- Cδ2 9.3 ± 1.5

1.2 Mutagenesis of Substitutions for Experimental Validation Based on the preliminary in silico results presented in 1.1, substitutions elected for experimental validation are listed in Table 24. Codons from the Tmab WT genes for mutagenesis were identified from the pVITRO1_Trastuzumab_IgG1/κ expression vector and mutations were designed based on minimum bp mismatching to Tmab WT codons. Mutagenesis reactions were designed and performed based on the QuikChange™ method (Agilent Technologies), that is whole plasmid, single-round PCR [215]. Overlapping primer pairs with introduced mutations were designed and the entire vector was amplified through high fidelity inverse PCR through a variety of cycling methods, including conventional 3-step PCR, a 2-step PCR and single-primer reactions in parallel [216]. Vectors containing successful mutations were expressed in transient 293F culture and purified through protein A affinity chromatography in the lead up to affinity and binding kinetic assays. The library was screened for extracellular binding of HER2 in a cell-based assay, and binding kinetics to HER2 through SPR. Furthermore, the library was screened for binding kinetics to model Fc receptor FcγR1A through SPR to demonstrate retention of other biological functions. For the purposes of this study, pVITRO1_Trastuzumab_IgG1/κ expression vector was used for experimental validation of D28>K and S50>E substitutions and the gWiz TH/TL expression vectors for all other substitutions.

Table 24 Summary of Tmab aa substitutions elected for experimental validation. Codon bp that mismatch the WT gene codon for primer design consideration (bold). *Two primer pairs were designed with different mismatch codons for several mutagenesis attempts, ^simultaneous double substitution.

Site WT Codon Substitution Mutation Codon

K AAG D28 (V ) GAT κ R CGT/CGC*

^ Y49 (Vκ) TAT S TCT

E GAG S50 (V ) TCT κ Y^ TAT

D102 (VH) GAT F TTT

2 Methods 2.1 Instrumentation, Services and Software Site-directed mutagenesis (SDM) reactions were conducted on a Bio-Rad T100 thermal cycler, with heated lid. DNA quantification was performed on a NanoDrop ND-1000 spectrophotometer. Mutant isolate vectors were prepared as purified DNA samples and sent to AGRF for Sanger Sequence analysis. Sequencing results were analysed on Applied Biosystems Sequence Scanner 2 software.

Overlapping primer pairs for mutagenesis were designed using GSL Biotech SnapGene Viewer

4.0.5 software. Tm calculations were performed on NEB Tm calculator software with mismatched bp omitted (http://tmcalculator.neb.com). Alignment of mutant isolate sequences to vector DNA sequences were performed on the NCBI BLAST software (https://blast.ncbi.nlm.nih.gov/Blast.cgi).

Mutant antibodies were purified from culture supernatant on a GE HealthCare ÄKTApurifier 100 FLPC system, using Unicorn 5.11 system and analysis software.

Flow cytometry was performed on a Merck Guava® easyCyte flow cytometer, using InCyte 3.1 system and analysis software. SPR was performed on a GE HealthCare Biacore™ T200, using T200 Control and Evaluation software.

2.2 Materials pVITRO1-Trastuzumab-IgG1/κ vector (61883) was obtained from AddGene, gWiz TH and gWiz TL vectors were cloned from gWiz blank vector as described in Chapter 2. The primers for D28>K and S50>E mutations were synthesised from Life Technologies (Australia). The primers designed for D28>R, Y49/S50>S49/Y50 and D102>F mutations were synthesised from Integrated DNA Technologies (Singapore). The primers designed for sequencing pVITRO and gWiz inserts were synthesised from Geneworks (Australia).

The Phusion high-fidelity PCR kit (E0553S) and DpnI enzyme (R0176S) were purchased from Genesearch (Australia). Plasmid maxiprep kit (G221020) was purchased from Astral Scientific (Australia), miniprep (PLN70) and maxiprep (NA0310) kits were purchased from Sigma-Aldrich (Australia).

83 was purchased from Thermo Fisher Scientific Australia and agar (20767.232) was purchased from VWR International (Australia). Hygromycin B solution (ant-hg-1), Hygromycin agar (fas-hg-s) and Terrific Broth (fas-hg-l) sachets were purchased from Jomar Life Research (Australia).

FreeStyle 293F cells were a kind donation from Dr Mario Torrado del Rey of Prof Joel Mackay’s research group, School of Life and Environmental Sciences, The University of Sydney. SK-BR3 cells were a kind donation from A/Prof Thomas Grewal’s research group, Pharmacy School, The University of Sydney. FreeStyle™ 293 expression media (12338018), OptiPRO™ Serum free media (12309050), UltraPure™ 1M Tris-HCl pH 8.0 (15568-025), RPMI media (21870-076) and additional tissue culture reagents (Appendix 3.1) were purchased from Life Technologies (Australia). Linear Polyethylenimine (PEI) 25 kDa (23966-2 (POL)) was purchased from BioScientific (Australia), bovine serum albumin (BSA) (BSAS 0.1) was purchased from Bovogen (Australia) and T75 TC flasks (CORN430641U) were purchased from Bio-Strategy (Australia). PBS tablets (09-2051-100) and 1 M Tris-HCl pH 9 (BIOSD814) were purchased from Astral Scientific (Australia). Disposable baffled culture flasks (CLS431405), EDTA (E6758), FITC secondary antibody (F9512), isotype control antibody (I5154), centrifuge tubes (CLS430291, CLS430791), vacuum systems (CLS430769), glycine hydrochloride (G2879), HiTrap® Protein A affinity column (GE17-0403-01), HiTrap® MabSelect Sure column (29-0491-04) and Amicon® Ultracel 50 kDa centrifugal filters (Z740177, Z740193) were purchased from Sigma-Aldrich (Australia).

Herceptin® was a kind donation from Genentech (USA), HER2 HIS-tagged receptor (SRP6405) was purchased from Sigma-Aldrich (Australia) and FcγR1A HIS tagged receptor (10256-H08H-5) was purchased from Life Technologies (Australia). CM5 chips (29-1049-88), Amine coupling kit (BR-1000-50) and Anti HIS capture kit (28-9950-56) were purchased from GE Healthcare (Australia). HEPES (54457), NaCl (S9888) and Tween 20 (P9416) were purchased from Sigma- Aldrich (Australia).

2.3 Site Directed Mutagenesis Overlapping mutagenesis primers were designed to incorporate mutagenic bp substitutions with a minimum of 8 bp homology upstream and downstream of the mutation site (Table 25). The primer lengths were varied due to GC content at the mutation site, and were adjusted to meet similar

predicted Tm values, a GC clamp of 2 C/G bp in the 3’ end of the primers and 40 – 60 % GC content in all primers for optimal annealing.

Several PCR techniques were employed to optimise mutation efficiency (Table 26); inverse PCR (INV PCR) as 3-step and 2-step cycling conditions, differing number of cycles (18 and 25) and as single-primer reactions in parallel (SPRINP)[216]. The D28>K and S50>E mutations were performed on pVITRO1-Trastuzumab-IgG1/κ vector using the SPRINP and INV PCR 18 reaction

conditions. The D102F mutation was performed on the gWiz TH vector and the D28>R mutation

was performed on the gWiz TL vector using the INV PCR 25 and 2-step cycling conditions. The

Y49/50>S49/Y50 mutation was performed on the gWiz TL vector using the 2-step cycling condition only. All reactions were performed using a Phusion High Fidelity PCR kit (Table 27). Unlike the INV PCR reactions, the SPRINP reactions were performed with a single primer in each reaction to produce single stranded vector DNA product. The SPRINP complementary single stranded DNA products were then combined and annealed in gradually cooling reaction to form the complete double stranded vector DNA (Table 28).

Table 25 List of primers designed for TH and TL gene affinity mutations. Nucleotides that mismatch the Tmab genes aa sequence, introducing codon mutation (bold). Primer Tm was calculated to pVITRO1_Trastuzumab_IgG1/κ with the Phusion PCR kit, with mismatches omitted from the primer sequences. *1 G/C bp on 3’ end due to high GC content of sequence in the mutation site.

Sequence Tm Primer 5’ 3’ (°C)

SDMA TrasLC D28K Fwd CGTGCTTCTCAAAAGGTTAATACTGCTGTTGCTTGG SDMA TrasLC D28K Rev CCAAGCAACAGCAGTATTAACCTTTTGAGAAGCACG 75

SDMA1 TrasLC D28R Fwd CGTGCTTCTCAACGTGTTAATACTGCTGTTGCTTGG SDMA1 TrasLC D28R Rev CCAAGCAACAGCAGTATTAACACGTTGAGAAGCACG 76

SDMA2 TrasLC D28R Fwd GTCGTGCTTCTCAACGCGTTAATACTGCTGTTGC SDMA2 TrasLC D28R Rev GCAACAGCAGTATTAACGCGTTGAGAAGCACGAC* 72

SDMA TrasLC S50E Fwd GCTCCTAAACTTCTTATTTATGAGGCTTCTTTTCTTTATTCTGGTGTTCCTTCTCG SDMA TrasLC S50E Rev CGAGAAGGAACACCAGAATAAAGAAAAGAAGCCTCATAAATAAGAAGTTTAGGAGC 78

SDM TrasLC YS49/50SY Fwd GCTCCTAAACTTCTTATTCTTATGCTTCTTTTCTTTATTCTGGTGTTCC SDM TrasLC YS49/50SY Rev GGAACACCAGAATAAAGAAAAGAAGCATAAGAATAAGAAGTTTAGGAGC 76

SDMA TrasHC D102F Fwd CGTTGGGGTGGTTTTGGTTTTTATGCTATGG SDMA TrasHC D102F Rev CCATAGCATAAAAACCAAAACCACCCCAACG 75

All PCR products were digested with DpnI at 37 °C for 1 hr to remove native vector from the PCR product. 5 µL of the digestion products were transformed in competent E. coli via heat-shock method (Appendix 1.4), cells containing intact vector were rescued in LB agar containing selection antibiotic (75 µg/mL Hygromycin for pVITRO1 reactions and 50 µg/mL Kanamycin for gWiz reactions). All digestion products aside from the Y49/50S>S49/Y50 and D28>R with SDMA1 primer pair 2-step digestion products, were transformed into Stellar™ competent E.coli. The Y49/50S>S49/Y50 and D28>R with SDMA1 primer pair 2-step digestion products were transformed into in-house prepared JM109 competent E. coli (Appendix 1.4).

Table 26 Cycling conditions for the different PCR reaction methods for TH and TL gene affinity mutations. *2-step reaction only, ^extension time only pVITRO1-Trastuzumab-IgG1/κ; D28>K, S50>E SPRINP INV PCR 18

Temp Time Time Cycle Step Cycles Cycles (°C) (sec) (sec)

Initial denaturation 98 1 120 1 30

Denaturation 98 40 30

Annealing 55 30 40 18 60

Extension 72 150 150

* gWiz TH, gWiz TL; D102>F, D28>R, Y49/S50>S49/Y50 INV PCR 25 2-Step

Cycle Step Cycles Time (sec) Temp (°C)

Initial denaturation 1 30 98 98

Denaturation 10 98 98

Annealing 25 10 65 72^ Extension 120 72

Final Extension 1 300 72 72

Table 27 Reaction components for TH and TL gene affinity mutations. Two SPRINP reactions were performed for each mutation; each reaction containing either the forward or reverse primer.

Reaction Concentration Volume (µL)

Component INV PCR SPRINP INV PCR SPRINP

5 x Phusion HF Buffer 1 x 1 x 5 5

10 mM dNTP 200 µM 200 µM 0.5 0.5

10 µM Forward primer 1.5 pM 0.75 40 pM 4 10 µM Reverse primer 1.5 pM 0.75

Vector DNA 0.6 ng/µL 20 ng/µL as needed

Nuclease free H2O - to 25 µL

Phusion DNA Polymerase 20 units/mL 50 units/mL 0.25 0.38

Table 28 Cycling conditions for the annealing of complementary SPRINP products of TH and TL gene affinity mutations.

Cycle step Temp (°C) Time (sec)

Denaturation 95 300

90 60 80 60 70 30 Gradual anneal 60 30 50 30 40 30

Hold 37 ∞

87 resuscitated and culture scaled up for plasmid maxiprep extraction (Appendix 2.1.3 - 2.1.4) in the lead up to production of the Tmab affinity mutant library through expression in transient 293F culture.

2.4 Expression and Purification

Successful pVITRO1-Trastuzumab-IgG1/κ and gWiz TH/TL mutant vectors were prepared for transient expression in suspension 293F culture. 24 hr prior to transfection, 293F cells were passaged to a seeding density of 5 x 105 cells/mL in Freestyle 293 expression media in 30 mL culture for pVITRO1-Trastuzumab-IgG1/κ transfections and 60 mL culture for gWiz TH/TL transfections. Transfection proceeded if the live cell concentration reached 0.8 – 1.2 x 106 cells/mL and cell viability was ≥ 95% on the day of transfection. Quantity of DNA and PEI were calculated as 2 µg PEI and 1 µg DNA per mL of transfection culture. For co-transfection of light chain and heavy chain gWiz vectors, a 1:1 ratio of light chain to heavy chain vector was combined for transfection. Calculated quantity of DNA and PEI were separately diluted in OptiPRO SFM to total volumes of 2.25 mL for pVITRO1-Trastuzumab-IgG1/κ or 4 mL gWiz TH/TL transfections and incubated at RT for 5 min. The diluted DNA and PEI solutions were combined, gently mixed and incubated at RT for 15 min prior to dropwise addition to transfection culture. Cultures were returned to incubate and 24 hrs post transfection the cultures were supplemented with 0.5% w/v tryptone to total culture volume. 2 days post transfection, cultures were scaled up to double volume

(i.e. 60 mL for pVITRO1-Trastuzumab-IgG1/κ and 120 mL for gWiz TH/TL transfections) and supplemented with 0.5% w/v tryptone in the gWiz cultures only. 3 days post transfection, cultures were further scaled up to double volume (i.e. 120 mL for pVITRO1-Trastuzumab-IgG1/κ and 240 mL for gWiz TH/TL transfections) and again supplemented with 0.5% w/v tryptone in the gWiz cultures only.

All transfection culture supernatants were harvested 8 days post transfection by centrifugation (3,000 g, 15 min), clarified and degassed by passing through a 0.22 μm vacuum filter. Secreted antibody was purified from the culture supernatant through Protein A affinity chromatography, using an ÄKTApurifier 100 FPLC system with a HiTrap® Protein A column for pVITRO1- Trastuzumab-IgG1/κ expressed Tmab and mutants, and a HiTrap® MabSelect Sure column for gWiz TH/TL expressed Tmab and mutants (Appendix 4). Antibody was bound on the column with PBS as binding buffer, eluted from the column with 0.1 M glycine HCl (pH 2.7 with the Protein A

88 column or pH 3.0 with the MabSelect Sure column) and elution fractions neutralised with 1 M Tris- HCl (pH 9 with the Protein A column or pH 8 with the MabSelect Sure column). FPLC chromatograms detected the elution peaks with UV absorbance at 280 nm and elution fractions containing antibody were pooled, buffer exchanged to PBS and concentrated using 50 kDa centrifugal filters. UV absorption of the purified antibody samples was measured at 280 nm to derive total antibody yield using the Beer Lambert law and the extinction coefficient of trastuzumab (Appendix 4.2).

2.5 HER2 Cell-binding Assay The Tmab affinity mutant library was assessed for affinity of extracellularly expressed HER2 in a cell-binding end point HER2 affinity assay as detailed in Appendix 5.2. Tmab affinity mutant titrations were compared to Herceptin® buffer exchanged to PBS (PBS-Herc), Tmab WT expressed from the gWiz TH/TL vector system (Tmab gWiz) and expressed from the pVITRO1_Trastuzumab_IgG1/κ vector system (Tmab pVITRO1).

Briefly, overexpressing HER2 cell line SK-BR3 was harvested non-enzymatically with Versene Solution, washed and incubated with Tmab sample titrations for 20 min, 4 °C. After incubation the cells were washed, resuspended with a FITC-secondary anti-human IgG1/κ antibody and further incubated for 20 min, 4 °C. The cells were again washed, resuspended and passed through flow cytometry to detect the amount of green fluorescence per cell in the single cell population. For each titration batch PBS-Herc was run as a positive control, a polyclonal IgG1/κ isotype was run as a negative control and dilution buffer was run as a wash control. The controls were used to define the cell populations detected on flow cytometry as positive or negative cell binding. All titration samples were measured through the control defined parameters in flow cytometry, with 5,000 events collected in duplicate. For each sample, the positive fraction was calculated as the percentage of cells defined as positive (bound) amongst the entire count. The positive fraction of each sample was plotted against antibody concentration in a sigmoidal binding curve to extrapolate KD at equilibrium, that being antibody concentration at 50% positive fraction (bound).

2.6 Binding Kinetics of Tmab Mutants to HER2 and FcγR1A ® The binding kinetics of Herceptin buffer exchanged to PBS and Tmab affinity mutant library was assessed to HER2 and FcγR1A through surface plasmon resonance (SPR) as detailed in Appendix 5.3. HER2 kinetics was measured to provide more insight to the results from the cell binding assay

89 and FcγR1A kinetics was measured to ensure that the mutations did not contribute to a loss of immunological functions.

Briefly, a CM5 chip was prepared with anti-HIS antibody through amine coupling for the capture of

HIS-tagged HER2 for FcγR1A receptor (4 nM, 5 µL/min for 5 min). Measurements were performed at 25 °C in HBS-T running buffer as single cycle kinetic assays. 2-fold serial dilutions of Tmab samples were prepared from 0.5 - 8 nM and 2.5 - 40 nM for HER2 and FcγR1A respectively. Increasing concentrations of Tmab sample was applied over the receptor captured surface

(20 µL/min, 2 min) with dissociation times of 60 min and 30 min for HER2 and FcγR1A binding respectively.

The sensorgrams were analysed globally using a Langmuir 1:1 binding model to determine Ka, Kd and KD. All samples were conducted in duplicate to obtain average values and standard deviation. Kinetic rate and binding constants determined from sensorgrams were accepted if the goodness of 2 fit value (χ ) was within 5% of the maximum response level (Rmax) of the sensorgrams and KD was calculated as Kd/Ka.

3 Results 3.1 Mutation Efficiency and Expression Of the mutagenesis strategies implemented, three Tmab affinity mutant vectors were successfully isolated for expression. That being D28>K and S50>E mutations in the pVITRO1_Trastuzumab_IgG1/κ expression vector and D102>F mutation in the gWiz TH expression vector (Table 29). Of the mutagenesis strategies, the SPRINP and inverse PCR 2-step cycling methods were not successful. Both inverse PCR 3-step cycling methods were highly successful, with 25 cycles showing apparently higher efficiency for D102>F mutation specifically. Despite repeated attempts (of which several are not reported in this chapter), including re-design of primers for the D28>R mutation, isolation of a confirmed D28>R mutant was not achieved. One attempt was made at isolating mutation Y49/S50>S49/Y50 which was also unsuccessful.

An example of positive mutation identified through sequence analysis as shown in Figure 22 isolate pVITRO1_Trast_S50E_INVPCR(3) forward sequence was aligned with the nucleotide sequence for pVITRO1_Trastuzumab_IgG1/κ. Homology from 5,710 – 6,411 bp of the for pVITRO1_Trastuzumab_IgG1/κ sequence confirms the complete TL gene with bp mismatch from

5914 – 5916 TCT > GAG confirmed for mutation S50>E. No further alignments and continued homology beyond site of mutation confirmed no further artefacts produced by primer ligation products.

Transient expression and purification of the Tmab affinity mutant library was successful for all mutant vector expression systems. In keeping with previously reported results from Chapter 2 and

Chapter 3, the gWiz TH/TL vector system produced significantly higher yield than the pVITRO1- Trastuzumab-IgG1/κ vector system (not shown). All purified Tmab and affinity mutant samples were of workable yield and quality to proceed with subsequent characterisation and binding kinetics studies; that being Tmab pVITRO1 (700 µg), D28>K (430 µg), S50>E (150 µg), Tmab gWiz (1.65 mg) and D102>F (500 µg).

Table 29 Comparison of efficiencies between the different mutagenesis strategies conducted in this study. Successfully isolated mutants and associated SDM method (bold). Inverse PCR SDM method performed using *SDMA1 and #SMDA2 primer pairs, +rescued from JM109 competent cells.

Isolates Mutation Vector SDM method Yielded Efficiency

SPRINP > 300 0/4 (0 %) pVITRO1_Tras_D28K INV PCR 18 > 300 1/5 (20 %)

SPRINP > 300 0/5 (0 %) pVITRO1_Tras_S50E INV PCR 18 26 3/5 (60 %)

2-step 0 N/A gWiz TH D102F INV PCR 25 > 300 2/3 (66.7 %)

1*+ 0/1 (0 %) 2-step 4# 0/3 (0%) gWiz TL D28R 37* 0/10 (0 %) INV PCR 25 22# 0/10 (0 %)

gWiz TL Y49/S50>S49/Y50 2-step 1+ 0/1 (0 %)

Query 5653 TTTCTAAACCCGTTTCCAGGTGTTGTGAAAGCCACCGCTAATTCAAAGCAATCCGGAATGTTGCCATCACAACTCATTGGGTTTCTGCTGCTCTGGGTTCCAGCTAGCCGCGGTGACATCCAGATGACCCAGTCTCCTTCTTCTCTTTCT 5802 ||||| | ||||||| ||||||||||| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| Sbjct 8 TTTCT-AGCCCGTTT-CAGGTGTTGTG-AAGCCACCGCTAATTCAAAGCAATCCGGAATGTTGCCATCACAACTCATTGGGTTTCTGCTGCTCTGGGTTCCAGCTAGCCGCGGTGACATCCAGATGACCCAGTCTCCTTCTTCTCTTTCT 154

Query 5803 GCTTCTGTTGGTGATCGTGTTACTATTACTTGTCGTGCTTCTCAAGATGTTAATACTGCTGTTGCTTGGTATCAACAAAAACCTGGTAAAGCTCCTAAActtcttatttattctgcttcttttctttattctggtgttccttctcgtttt 5952 ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||||||||||||||||||||||||| Sbjct 155 GCTTCTGTTGGTGATCGTGTTACTATTACTTGTCGTGCTTCTCAAGATGTTAATACTGCTGTTGCTTGGTATCAACAAAAACCTGGTAAAGCTCCTAAACTTCTTATTTATGAGGCTTCTTTTCTTTATTCTGGTGTTCCTTCTCGTTTT 304

Query 5953 tctggttctcgttctGGTACTGATTTTACTCTTACTATTTCTTCTCTTCAACCTGAAGATTTTGCTACTTATTATTGTCAACAACATTATACTACTCCTCCTACTTTTGGTCAAGGTACCAAGCTGGAGATCAAACGTACGGTGGCGGCG 6102 |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| ||||||||||||||||||||| Sbjct 305 TCTGGTTCTCGTTCTGGTACTGATTTTACTCTTACTATTTCTTCTCTTCAACCTGAAGATTTTGCTACTTATTATTGTCAACAACATTATACTACTCCTCCTACTTTTGGTCAAGGTACCAAGCTGGARATCAAACGTACGGTGGCGGCG 454

Query 6103 CCATCTGTCTTCATCTTCCCGCCATCTGATGAGCAGTTGAAATCTGGAACTGCCTCTGTTGTGTGCCTGCTGAATAACTTCTATCCCAGAGAGGCCAAAGTACAGTGGAAGGTGGATAACGCCCTCCAATCGGGTAACTCCCAGGAGAGT 6252 |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| Sbjct 455 CCATCTGTCTTCATCTTCCCGCCATCTGATGAGCAGTTGAAATCTGGAACTGCCTCTGTTGTGTGCCTGCTGAATAACTTCTATCCCAGAGAGGCCAAAGTACAGTGGAAGGTGGATAACGCCCTCCAATCGGGTAACTCCCAGGAGAGT 604

Query 6253 GTCACAGAGCAGGACAGCAAGGACAGCACCTACAGCCTCAGCAGCACCCTGACGCTGAGCAAAGCAGACTACGAGAAACACAAAGTCTACGCCTGCGAAGTCACCCATCAGGGCCTGAGCTCGCCCGTCACAAAGAGCTTCAACAGGGGA 6402 ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||||||| Sbjct 605 GTCACAGAGCAGGACAGCAAGGACAGCACCTACAGCCTCAGCAGCACCCTGACGCTGAGCAAAGCAGACTACGAGAAACACAAAGTCTACGCCTGCGAAGTCACCCATCAGGGCCTGAGCTCGCCCGTCAC-AAGAGCTTCAACAGGGGA 753

Query 6403 GAGTGTTAGGGATCCCGTACGCCTAGGAGCAGGTTTCCCCAATGACACAAAACGTGCAACTTGAAACTCCGCCTGGTCTTTCCAGGTCTAGAGGGGTAACACTTTGTACTGCGTTTGGCTCCACGCTCGATCCACTGGCGAGTGTTAGTA 6552 ||||||||||||||||||||||||||||||| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| Sbjct 754 GAGTGTTAGGGATCCCGTACGCCTAGGAGCA-GTTTCCCCAATGACACAAAACGTGCAACTTGAAACTCCGCCTGGTCTTTCCAGGTCTAGAGGGGTAACACTTTGTACTGCGTTTGGCTCCACGCTCGATCCACTGGCGAGTGTTAGTA 902

Query 6553 ACAGCACTGTTGCTTCGTAGCGGAGCATGACGGCCGTGGGAACTCCTCCTTGGTAACAAGGACCCACGGGGCCAAAAGCCACGCCCACACGGGCCCGTCATGTGTGCAACCCCAGCACGGCGACTTTACTGCGAAACCCACTTTAAAGTG 6702 |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| |||||||||||||||||| ||||||||||||||||||||||||| |||||||||| ||| Sbjct 903 ACAGCACTGTTGCTTCGTAGCGGAGCATGACGGCCGTGGGAACTCCTCCTTGGTAACAAGGACCCACGGGGCCAAAAGCCACGCCCACAC-GGCCCGTCATGTGTGCAA-CCCAGCACGGCGACTTTACTGCGAA-CCCACTTTAA-GTG 1048

Query 6703 ACATTGAAACTGGTACCCACACACTGGTGACAGGCTAAGGATGCCCTTCAGGTACCCCGAGGTAACACGCGACACTCGGGATCTGAGAAGGGGACTGGGGCTTCTATAAAAGCGCTCGGTTTAAAAAGCTTCTATGCCTGAATAGGTGAC 6852 | |||||| || ||||||| ||||| |||| | ||||| ||||| | ||| ||||| ||| |||| |||| |||| ||| | ||||| ||||||||||| |||| | |||| | |||| ||| | |||| ||| ||| | Sbjct 1049 AMATTGAA-CT-GTACCCAMMCACTG-TGACTG-CTAAG-ATGCCGT-CAG-TACCC-GAG-TAACM-GCGAACCTCG--ATC-G-GAAGG--ACTGGGGCTTC-ATAA--GS-CTCGAT--AAAA-GCT-C-ATGC-TGA-TAGKA-A- 1169

Query 6853 CGGAGGTCGGCACCTTTCCTTTGCAATTACTGACCC 6888 |||| |||||||| |||| || |||| || ||||| Sbjct 1170 CGGAAGTCGGCAC-TTTCGTTGCCAATAACKGACCC 1204

Figure 22 Example of a BLAST sequence alignment demonstrating a positive mutation sequence analysis result. Positive isolate pVITRO1_Trast_S50E_INVPCR(3) (Subjct) aligned to the pVITRO1_Trastuzumab_IgG1/κ complete nucleotide sequence (Query), mismatches at mutation site (circled).

3.2 Affinity and Binding Kinetics to HER2 and FcγR1A An example of Tmab sample titration detected through flow cytometry for the cell-binding assay as shown in Figure 23. A 10-fold increase in fluorescent intensity was observed in Herceptin® treated cells (positive control) versus IgG1/κ isotype treated cells (negative control) to clearly define positive and negative binding regions (Figure 23A). The titration of Tmab gWiz produced fractions from the cell population that migrated from the positive (bound) to the negative (unbound) region with reducing concentration of Tmab gWiz (Figure 23B). The cell-binding titrations from the Tmab affinity mutants as shown in Figure 24 prior to analysis. D28>K and D102>F titrations demonstrated retained affinity to extracellular HER2, whereas the S50>E titration spanned almost

entirely in the negative (unbound) region.

100

A B

Count

40 50 60 70 80 90 80 70 60 50 40

0 10 20 30 40 50 60 70 80 90 100 90 80 70 60 50 40 30 20 10 0 0 10 20 20 0 10

Green Fluorescence (Log) Figure 23 Tmab gWiz titration detected through flow cytometry. Controls for defining positive and negative binding (A). Herceptin® treated (green), IgG1/κ isotype treated (aqua), dilution buffer treated (brown) and untreated cells (yellow). Tmab gWiz titration in 2-fold serial dilutions from 10 µg/mL (B).

300 A B C

Count 0 50 100 150 200 250 250 200 150 100 50 0

Green Fluorescence (Log) Figure 24 Cell-binding assay as detected on flow cytometry for titrations of Tmab affinity mutants D28>K (A), S50>E (B) and D102>F (C).

Binding curves and KD values derived from the flow cytometry data give a much clearer comparison of the binding between the Tmab affinity mutant library (Figure 25 and Table 30). The S50>E mutation was confirmed to have significantly diminished binding affinity to extracellularly expressed HER2, where the maximum concentration (10 µg/mL) reached 27% positive fraction and a KD value could not be determined. Between the control titrations, Tmab pVITRO1 binding was comparable to PBS-Herc; however, Tmab gWiz had an apparent 6-fold reduction in binding as compared to PBS-Herc. The D28>K mutation had an apparent 2-fold reduction in binding as compared to Tmab pVITRO1 and D102>F had an apparent 4-fold increase in binding as compared to Tmab gWiz, its binding was also comparable to Tmab pVITRO1.

In comparison, SPR derived binding kinetic values from the single cycle kinetic assay showed

1000-fold stronger KD of the Tmab affinity mutant library binding to surface captured HER2 (pM) as opposed to the cell-based affinity assay (nM), as well as reporting KD values in the nM range for surface captured FcγR1A (Figure 26 and Table 31). As the Tmab gWiz performed poorly in the cell-binding assay, it was excluded from SPR analysis.

Figure 25 Tmab affinity mutant library binding curves to SK-BR3 cells from the cell-binding HER2 affinity assay, detected on flow cytometry.

Table 30 KD values extrapolated from the cell-binding assay of the Tmab affinity mutant library. KD values were calculated as antibody concentration (µg/mL) at equilibrium (50% bound) and converted to nM for comparison with SPR data.

KD (µg/mL) KD (nM) PBS-Herc 0.2068 1.42

Tmab pVITRO1 0.3039 2.09

D28>K 0.5982 4.11

S50>E ND ND

Tmab gWiz 1.319 9.06

D102>F 0.3442 2.37

Tmab pVITRO1 has an apparent 4-fold reduction in binding (increase in KD) as compared to PBS- Herc to surface captured HER2. Mutants D28>K and S50>E had a slight increase (1.3 – 1.5-fold) in binding (decrease in KD) and D102>F had a 2-fold decrease in binding (increase in KD) as compared to Tmab pVITRO1 to surface captured HER2. However, inspection of the SPR sensorgrams for Tmab affinity mutant library showed satisfactory curve fitting for affinity and dissociation calculations aside from mutations S50>E and D102>F (Figure 26A). Inspection of the residuals to the curve fitting further confirmed the case, with S50>E and D102>F data points distinctly exceeding the acceptable range during transitions through the association phase (Appendix 6.3).

Mutation S50>E had the lowest maximum relative binding to the HER2 captured surface (Rmax =

11.6) which infers a lower association constant (Ka) than calculated. Mutation D102>F had a high maximum relative binding to the HER2 captured surface (Rmax = 27.68) which was comparable to the PBS-Herc (Rmax = 29.08) and Tmab pVITRO1 (Rmax = 24.64) although it was the lowest Ka calculated. The dissociation constant (Kd) of both S50>E and D102>F didn’t adequately match the data against the curve fitting. The rapid decay in S50>E transitioning to a steady state in the dissociation phase more likely resembles a biphasic interaction. The dissociation of D102>F did not reach steady state during the cycling in the association phase, which again indicates a more complex interaction. In both mutations, the monophasic 1:1 binding model was an inadequate model to describe the interaction therefore the derived rate constants for these 2 mutants were not considered for reliable comparison.

Tmab pVITRO1 was comparable to PBS-Herc to surface captured FcγR1A. All mutants had an

apparent increase (3 – 7-fold) in binding (decrease in KD) as compared to Tmab pVITRO1 to

surface captured FcγR1A. Inspection of the SPR sensorgrams for Tmab affinity mutant library showed satisfactory curve fitting for affinity and dissociation calculations aside from mutations D28>K and S>50E (Figure 26B). Notably, D102>F had a comparable sensorgram curvature to

PBS-Herc and Tmab pVITRO1 and had the highest maximum relative binding to the FcγR1A

captured surface (Rmax = 88.35), which was approximately 2-fold of the PBS-Herc (Rmax = 51.36)

and Tmab pVITRO1 (Rmax = 39.72). This is in agreement with the increased Ka and reduced KD

calculated values for D102>F in comparison to PBS-Herc and Tmab pVITRO1. The Ka of both D28>K and S>50E didn’t adequately match the data against the curve fitting. In both mutations, association was overestimated in the association phase due to incorrect compensation of the bulk signal refractive index in the analysis software. This is further indicated as maximum relative

binding to the FcγR1A captured surface for both mutants was 3 – 4.7-fold lower than Tmab

pVITRO1 (D28>K Rmax = 13.07, S50>E Rmax = 8.47), which infers a lower Ka than calculated. In this instance, to adequately describe the interaction the monophasic 1:1 binding model requires refinement, therefore the derived rate constants for these two mutants were not considered for reliable comparison.

Table 31 Binding kinetics of the Tmab affinity mutant library to surface captured HER2 and FcγR1A as measured through SPR.

HER2 FcγR1A

Ka Kd KD Ka Kd KD (x 106) M-1.s-1 (x 10-5) s-1 pM (x 105) M-1.s-1 (x 10-4) s-1 nM

PBS-Herc 3.85 ± 0.09 1.10 ± 0.60 2.85 ± 1.49 2.75 ± 0.06 11.31 ± 0.10 4.11 ± 0.05

Tmab pVITRO1 2.50 ± 0.03 3.03 ± 0.72 12.11 ± 2.72 2.11 ± 0.03 10.64 ± 0.56 5.05 ± 0.33

D28>K 3.21 ± 0.19 2.51 ± 0.18 7.81 ± 1.02 16.62 ± 8.69 11.65 ± 2.79 0.70 ± 0.16

S50>E 2.21 ± 0.02 1.95 ± 0.21 8.84 ± 0.87 4.88 ± 2.53 7.85 ± 0.14 1.61 ± 0.64

D102>F 1.41 ± 0.21 3.33 ± 0.99 23.69 ± 3.52 3.09 ± 0.11 2.28 ± 0.04 0.74 ± 0.01

RU 40 RU 35 40 Herceptin 30 25 30 in formulation 20 H e r c e p tin 15 20 10 in fo r m u la tio n

Response 10 5

0 Response 0 -5 -500 0 500 1000 1500 2000 2500 3000 3500 4000 4500 -10 Time s -500 0 500 1000 1500 2000 2500 3000 Time s RU RU 40 A 45 B 35 30 1 35 1 25 Herceptin 25 20 H e r c e p tin 15 in PBS 15 10

in P B S Response

Response 5 5 0 -5 -5 -500 0 500 1000 1500 2000 2500 3000 3500 4000 -500 0 500 1000 1500 2000 2500 3000 Time s Time s RU RU 35 30 30 25 25 2 20 2 p V IT R O 20 pVITRO 15 T m a b 15 Tmab 10 10 Response 5

Response 5

0 0

-5 -5 -500 0 500 1000 1500 2000 2500 3000 3500 4000 -500 0 500 1000 1500 2000 2500 3000 Time s Time s RU RU 14 25 RU RU 12 20 8 7 10 3 D 2 8 > K 7 D28>K 6 15 3 8 6 5 6 gWiz 4 10 g W iz 5 3 4 4 5

2 Response Tmab Response T m a b 2 3 1 0 2

Response 0 0 Response 1 -1 -5 -2 0 -2 -500 0 500 1000 1500 2000 2500 3000 3500 4000 4500 -500 0 500 1000 1500 2000 2500 3000 -500 0 500 1000 1500 2000 2500 3000 3500 4000 4500 -1 Time s Time s -500 0 500 1000 1500 2000 2500 3000 RU Time s RU Time s 16 9 14 7 12 4 4 10 S50>E S 5 0 > E 5 8 6 3 4

Response Response 2 1 0 -2 -1 -500 0 500 1000 1500 2000 2500 3000 3500 4000 4500 -500 0 500 1000 1500 2000 2500 3000 Time s Time s RU RU 90 30 25 70 5 20 5 50 D102>F 15 D 1 0 2 > F 10 30

Response

Response 5 10 0

-5 -10 -500 0 500 1000 1500 2000 2500 3000 3500 4000 4500 -500 0 500 1000 1500 2000 2500 3000 Time s Time s Figure 26 Analysed SPR sensorgrams from single cycle kinetic assays of the Tmab affinity mutation library to surface captured HER2 (A) and FcγR1A (B). Tmab samples PBS-Herc (1), Tmab pVITRO1 (2), D28>K (3), S50>E (4) and D102>F (5). Sensorgram data (red/green), curve fitting (black).

4 Discussion This study successfully produced three residue substituted Tmab mutants that were predicted from our collaborators as having enhanced affinity toward target HER2 through preliminary in silico analysis. The residue substitutions were confirmed for all mutants prior to in vitro biological assays

97 for determining relative affinity and binding kinetics. Results from the Tmab mutant library in vitro tests are in agreement with the completed in silico findings, in that mutation S50>E was disruptive to its affinity to HER2, and that D28>K and D102>F did not present conclusive affinity enhancement.

The two further mutations that were not successfully produced for in vitro testing (i.e. D28>R and Y49/S50>S49/Y50) were reported as candidates that were very likely to have enhanced affinity to HER2 from in silico analysis. Although these mutations were not successfully isolated, they are of ongoing interest to be produced and tested for a future study. Tmab mutant D28>R had previously been tested and reported as 1.86-fold enhanced relative affinity compared to Tmab [217] through a cell-binding assay and an ELISA. This further highlights that D28>R is of interest for characterisation and profiling its binding kinetics to HER2, as the completed in silico findings (1.1) reported an increase of 43% in free binding energy of Tmab mutant D28>R as compared to native Tmab binding to HER2. Tmab mutant Y49/S50>S49/Y50 has not been reported experimentally thus far, and is of ongoing interest to calculate the free binding energies and compare these results with in vitro results.

In comparing the SDM reactions for producing high yielding affinity mutants, the INV PCR 3-step, 25 cycle reaction was the most efficient. INV PCR 25 yielded many isolatable colonies with a 67% mutation efficiency for the D102>F mutation; however, not yielding any successful D28>R isolate from the two sets of primer pairs designed (Table 29). The INV PCR 3-step, 18 cycle reaction was also very successful, although yielding 10-fold less isolates with D28>K mutation, and reduced (20%) mutation efficiency for the S50>E mutation. This was expected, as an increase in the number of cycles in the reaction produces a higher concentration of PCR product for isolation [218]. However, the SPRINP and INV PCR 2-step methods were not successful in yielding mutation positive vector isolates. A further attempt to isolate mutants D28>R and Y49/S50>S49/Y50 using INV PCR 3-step at 18 or 25 cycles, at 55 °C annealing temperature will most likely yield higher mutation efficiency.

Strategies for preventing potential artefact primer ligation products was warranted for investigation as non-specific binding of elongated primers to self or template can occur at suboptimal annealing temperature [219]. Both the 2-step cycling method [218, 219] and the SPRINP method [216] are reported to minimise primer ligation artefacts and were trialled against the conventional 3-step cycling methods for improving mutation efficiency. Fortunately in this study, of all the isolates screened, no primer ligation artefacts were isolated.

The primer design for D28>K and S50>E mutants may be the attributing factor to the failure of the

SPRINP method in producing mutant isolates (Table 25). The D28>K primer pair Tm was at the lowest recommended range (75 - 85 °C) and the S50>E primer pair exceeded the recommended flanking nucleotide length (10 – 20 bp) [216]. The primer design with the S50>E mutation specifically could not be helped due to the nucleotide sequence in that region having low GC content. The higher primer annealing temperature (72 °C) in the 2-step cycling method may be the attributing factor for its failure to produce mutant isolates, as the Tm of mutagenic primers containing mismatches can only be approximated against the template [220]. Use of the Phusion High fidelity PCR kit must also be considered, where components in the mastermix induce a higher primer Tm than predicted with conventional mastermixes [221]. It is likely that the predicted mutagenic primer Tms were overestimated and that a reduced annealing temperature of 10 – 20 °C was more suitable in designing the cycling reactions [220], rendering the 2-step cycling method inappropriate with the primers used in this study.

Of the SMD reactions performed, successful mutants produced for in vitro testing included D28>K,

S50>E in the pVITRO1_Trastuzumab_IgG1/κ expression system and D102>F in the gWiz TH/TL expression system. Along with the Tmab affinity mutants, native Tmab was produced from both expression systems for comparison against the reference product Herceptin®. All Tmab samples, including Herceptin® were buffer exchanged to PBS prior to in vitro assessment. Removal of the excipients in the Herceptin® formulation was necessary as to normalise Tmab sample biological activity and stability for comparison.

SPR determined KD values of Tmab samples to HER2 were in the subnanomolar range which agrees with previous literature [19, 205, 209], as opposed to the nanomolar range derived from the cell-based affinity assay. The inconsistency of KD values between the cell-binding assay and SPR suggests that the cell-binding assay requires optimisation by extending the dissociation time to reach equilibrium [217, 222]. Due to the dilution series chosen to perform the cell-binding assay, the lowest Tmab dilution was 0.54 nM which was not sensitive enough to reach the subnanomolar range required for comparison with SPR. Another consideration limiting this technique is of Tmab internalisation and catabolism from binding to extracellular HER2 [23]. Discrepancies owing to internalised Tmab through the incubation time can be estimated with the use of an appropriate secondary antibody containing a fluorophore that can be quenched [223]. Estimation of non- equilibrium system and internalisation discrepancies is necessary in order to ascertain appropriate optimisation parameters. In further consideration of optimising the cell-binding assay, minimization

99 of Tmab internalisation though a pre-treatment of the cells with 0.2% NaN3 [212] and a 1 hr dissociation incubation time [205, 206, 217] would likely yield results that are comparable to SPR.

Interestingly, between the Tmab expression systems, Tmab pVITRO1 had higher biological similarity to PBS-Herc than the Tmab gWiz (Table 30 and Figure 25). The difference in relative affinity is likely due to varied content of acidic variants of the Tmab samples, caused by deamidation in the CDR [224] which mainly occurs in the production steps of expression and purification [225]. It was reported that Kd reduced 5-fold in Tmab sample containing 35.2% acidic variant compared to Tmab sample containing 14.9% acidic variant [224]. In further considering purification of the Tmab affinity mutant library for our study, assessing the Tmab samples for heterogeneity of charge variants through isoelectric focusing would have indicated whether a further ion exchange purification step was appropriate prior to commencing biological assays [205, 224].

Of the Tmab affinity mutant library produced for experimental investigation, S50>E clearly demonstrated a significant diminished relative affinity to extracellularly expressed HER2 through the cell-binding assay (Figure 25). SPR data for S50>E although unreliably calculated, did indicate a biphasic interaction in the dissociation phase which further confirmed a more complex dissociation than is described through a 1:1 binding model [226]. At the time of reporting preliminary in silico results of S50>E for experimental investigation, the previously described consideration of residue steric hindrance was not realised. The change in the crystal structure of the Tmab Fab-HER2 complex and the overall free energy of S50>E substitution was not further pursued by our collaborators. MD results demonstrated that the predicted electrostatic interaction did not produce a stable hydrogen bond with HER2 residue (Figure 20B) and an intramolecular interaction with V33 was destabilised which suggests conformational instability in the region. The experimental results for S50>E are in agreement with the in silico results, in that this mutation destabilised Tmabs conformation which inherently disrupted Tmabs relative affinity to HER2.

In regards to the D28>K and D102>F mutants, in silico analysis did not confirm enhanced affinity to HER2 as predicted hydrogen bonds and electrostatic interactions were not produced (Figure 20A and Figure 21C). Relative affinity to extracellularly expressed HER2 of both mutants was similar to Tmab as demonstrated through the cell-binding assay (Figure 25). D28>K had a 2-fold reduced relative affinity to HER2 through the cell-based assay, and a 1.5-fold enhanced relative affinity through SPR as compared to Tmab-pVITRO1. D102>F had a highly similar relative affinity to HER2 through the cell-based assay, and a 2-fold reduced relative affinity through SPR as compared

100 to Tmab-pVITRO1 (Table 30 and Table 31). The deviations between these results suggest negligible differences in relative affinity that are in keeping with the results from in silico analysis in that the mutations did not demonstrate conclusive affinity enhancement to HER2. And, as previously discussed, assessment of charge variant content in all Tmab samples to determine if an ion exchange purification step was appropriate would have likely produced more robust relative affinity data for refined comparison.

To validate that the mutations did not compromise further immunological activity, the Tmab affinity mutant library was further investigated for biological activity to high affinity IgG-receptor

FcγR1A through SPR. Unexpectedly, analysis revealed that D102>F showed significant (6.8-fold) enhanced relative affinity to FcγR1A as compared to Tmab pVITRO1 (Table 31 and Figure 26B). As previously discussed, the curve fitting and values calculated for D28>K and S50>E were not reliable in which Ka and KD values were overestimated, although binding was apparent with all mutants which suggests a retention in biological activity. Binding curves to FcγR1A were similar as to previous investigations reporting on various other human IgG1 antibodies [227-231] which suggests that the mutations did not compromise further immunological activity. The enhanced relative affinity of D102>F also suggests a potentially enhanced ADCC effect that warrants further investigation.

5 Concluding Remarks The findings of this study have shown that in silico analysis of predicted residue substitutions for Tmab affinity to HER2 correlates with in vitro investigation of an affinity disrupting mutation and of two mutations that did not show conclusive affinity enhancement. One of the mutations exhibited considerable enhanced affinity to FcγR1A which would likely result in a more pronounced ADCC effect, and is of interest for further investigation. Of the other predicted substitutions that were unsuccessfully isolated after several rounds of mutagenesis, the in silico results strongly suggest that they are lead candidates to pursue for enhanced relative affinity and are of further interest for investigation.

101

Chapter 5 Rational Addition of Glycosylation Sites in Trastuzumab; Glycoengineering a Model Antibody for Enhanced Stability

102

Abstract This chapter reports on the design and development of a library of trastuzumab mutants with hypothesised enhanced stability, through the strategic addition of glycosylation sites within close proximity to a previously reported aggregation prone region. The trastuzumab glycosylation mutant library was designed to introduce N-linked glycosylation sites through residue substitution and the mutant library was expressed in an optimised transient 293F expression system. The mutant library was confirmed through mass spectrometry for successful residue substitution and glycosylation addition before proceeding with accelerated thermal stability, aggregation and monomer loss studies. Along with the stability screen, retention of biological function was confirmed for the mutant library to its target HER2 and effector immune receptors FcγR1A, FcγR2B and FcγR3A through a cell-based binding assay and surface plasmon resonance. The mutant library displayed varying stabilities, of which 2 mutants were elected as lead candidates, possessing enhanced thermal intrinsic stability and monomer retention. This demonstrates the success of this antibody engineering approach for the design of improved whole antibody and antibody-based therapeutics.

1 Introduction; Stability and Aggregation Propensity in Whole Antibody Therapeutics Whole antibody therapeutics have persistently remained the most dominant and significant biologic therapeutic platform [1], for treatment of a plethora of indications including cancers, infections, auto-immune, cardiovascular and neurological diseases [3]. Aggregation remains one of the leading challenges in the development, manufacture and formulation of whole antibody therapeutics as it negatively affects the quality, efficacy and safety of the dosage form [5, 232, 233]. Aggregation is a highly complex process driven by self-association, in which interactions from surface exposed hydrophobic sites in antibodies lead to irreversible agglomeration, loss of higher order structure and biological activity [234]. As a consequence, aggregation leads to loss of the active antibody therapeutic and the presence of aggregates in the dosage form may elicit an adverse immunogenic response and resistance to treatment over time [233-236].

Aggregation can occur either through native protein-protein interactions of the antibody, or through chemical or physical degradation of the antibody that leads to conformational instability, unfolding (reversible) and denaturation (irreversible). In either pathway, colloidal interactions of the protein species induce irreversible aggregation, the propensity of an antibody to aggregate is therefore concentration dependant [151, 232, 234, 237]. Antibodies follow the degradation pathway to aggregation when thermally or chemically stressed, where the intrinsic conformational stability of the protein is challenged with high ionic strength or energy [232]. Despite antibodies possessing relatively high intrinsic stability, much of their manufacturing process involves thermal and

103 chemical stress that is variable and cannot be compromised, such as cell culture conditions, affinity chromatography purification and viral inactivation steps [5, 64, 149]. In formulating the final therapeutic product for administration, the intrinsic stability of antibodies can be modestly enhanced with stabilising excipients and lyophilisation, to prevent aggregation from long term storage and handling [55, 120, 122, 123]. However, enhancement of whole therapeutic antibody intrinsic stability is of ongoing commercial interest to improve recovery and quality of antibody in the manufacture process, increase shelf life and accommodate for more versatile formulation options such as dry powder for inhalation [122, 238] and highly concentrated subcutaneous for injection [54, 55, 120].

Correlations between intrinsic stability, self-association, aggregation propensity, solubility and viscosity have been extensively researched through an array of experimental techniques and molecular dynamic simulations [166-170, 239-244]. Experimental findings have directed the development of an extensive array of computational tools to predict native protein-protein interactions of antibodies through amino acid (aa) sequence and native protein structure [5, 148, 165, 245]. The current consensus for computational analysis of predicting sites of aggregation propensity in antibodies is in considering the native tertiary structure, the local spacial arrangement of aa residues, solvent accessibility and the density of surface exposed polarity versus hydrophobicity [5].

For instance, the spacial-aggregation-propensity (SAP) computational tool has been reported to determine aggregation-prone regions (APR) in antibodies, experimentally confirmed by substitution of predicted surface exposed hydrophobic residues to polar aa residues [168, 169, 246]. This strategy has markedly demonstrated that removing APRs through directed single-point modifications to the antibody peptide sequence significantly improves its intrinsic stability and resistance to aggregation. This strategy has drawn significant attention to the continued discovery and characterisation of antibody APRs for enhancing stability through modification of the antibody peptide sequence. Stability enhancement would be translatable to the whole antibody class if modifications reside in any of the homologous antibody constant regions, although further considerations such as effector function may be compromised through such modifications. Pursuit of APR-removed antibodies is of ongoing commercial interest to derive strategically designed biobetter therapeutic antibodies with enhanced intrinsic stability and reduced aggregation propensity.

104

1.1 Strategic Addition of Glycosylation Sites In consideration of designing mutational projects for enhanced intrinsic stability and resistance to aggregation of a model antibody, literature had previously identified and experimentally confirmed APRs through direct residue substitution in Rituximab [169] and Bevacizumab [168]. Interestingly, it was previously found that the conserved glycosylation site in the Fc contributes to the intrinsic stability of antibodies, proposing the hypothesis that the glycans sterically hinder the interactions of APRs identified in close proximity to glycosylation site [247]. Additionally, residue substitution was applied to successfully introduce N-linked glycans to a poorly soluble antibody for increased solubility [173]. The direction of this mutational project design was therefore to identify surface exposed residues for substitution, to introduce N-linked glycans within close proximity (10 Å) to reported APRs as to sterically hinder APR colloidal associations.

Further considerations with introduction of N-linked glycans to a model antibody was to not interfere with native antibody structure that: binds to target (specificity) in the CDRs, binds to Fc receptors (effector functions) and Protein A (purification) in the Fc region. Therefore identification of APRs and introduction of glycosylation sites was restricted to the Fab antibody regions excluding the CDRs, making the substitutions compatible with whole antibody and Fab based formats.

Hyper-glycosylation has also been described with biotherapeutics as increasing solubility and extending half-life [118, 173], which is of further benefit to improving formulation and treatment strategies. Strategic glycosylation addition has the added advantage of orienting a specific functional group on the antibody for directed drug, dye, linker and nanoparticle conjugations. Beyond stability enhancement, the benefits for strategic addition of glycosylation sites in therapeutic antibodies are highly apparent for antibody engineering and is therefore of immense interest for investigation to uncover the full potential that can be exploited from this technology.

1.2 Trastuzumab as a Model Antibody Trastuzumab (Tmab) was used as a model antibody to identify potential APR candidates from literature to design mutational projects. Each project was designed around a lead APR, with an additional project designed from a report of improved solubility, as summarised in Table 32. Visual inspection of substitutions listed in Table 32 to the crystal structure of Tmab Fab fragment complexed to extracellular domain of HER2 receptor (1N8Z) was performed to identify and confirm surface exposed APRs and residues for introducing glycosylation (Figure 27). Glycosylation site candidates were excluded from further consideration if the residue for substitution was a glycine or proline, as these residues are integral in the native antibody structure for flexibility and rigidity, respectively. Further exclusions to glycosylation site candidates were if

105

the solvent accessible area (SAA) of the substituted residue was predicted to be less than 15 Å, as it was hypothesised to be inaccessible to glycosylases in native structure for post-translational glycan addition [168, 169].

Table 32 Surface exposed aa residues identified in Tmab Fab region for substitution. Mutations and project elected (bold), APR (underlined), *SAA of substitution predicted as less than 15 Å.

aa Residue aa Project Region Proposed Function (Position) Substitution K/D Remove APR by introducing polar aa [168, 169] L (154) S/T Remove APR and introduce glycosylation site [168] APR 1 L (175)* P (204) N Sterically hinder APR by introducing glycosylation site G (157)* [168] APR 1/2 E (195) Q (160) S/T

Cκ K Remove APR by introducing polar aa [168] L (201) APR 2 N Remove APR and introduce glycosylation site [168]

* F (139) S/T Sterically hinder APR by introducing glycosylation site APR 2/3 G (200)* N [168] V (110) K Remove APR by introducing polar aa [168] Y (140) S/T

APR 3 K (169) Sterically hinder APR by introducing glycosylation site D (170) N [168]

K (107) Vκ K/S Remove APR by introducing polar aa [168, 169] L (177) CH1 Remove APR and introduce glycosylation site [168, 172] Q (160)

* S (176) Cκ APR 4 * T (178) Sterically hinder APR by introducing glycosylation site L (115) [168] Cjoin T (117) N L (182) Solubility/ Sterically hinder APR by introducing glycosylation site Q (178)* APR 4 [168, 248] CH1 Improve solubility and sterically hinder self-association by T (198) Solubility introducing glycosylation site [166, 248] A (121) Improve solubility by introducing glycosylation site [248]

106

Interestingly, an excluded substitution identified from this research (Q178>N) had previously been produced successfully and demonstrated enhanced solubility [173] which indicated a potentially misguided rationale for exclusion. Of the mutational projects designed, the solubility and APR

A B

D C

Figure 27 Crystal structure 1N8Z of Tmab Fab fragment complexed to extracellular domain of HER2 receptor. Tmab VH/CH1 (blue), Vκ/Cκ (green), HER2 fragment (orange), APR (purple), surface exposed aa identified for substitution to an N-linked glycosylation site reported to increase solubility (red) or sterically hinder APR (pink). (A) Solubility Project, (B) APR Project 1, (C) APR Project 2, (D) APR Project 3, (E) APR Project 4 107 project 4 were pursued for this body of work with introduced glycosylations directing steric hindrance of APR L177; polar substitution L177>K was included for comparison (Figure 28).

Figure 28 Crystal structure 1N8Z, highlighting elected mutations pursued in this study based on Solubility Project and APR Project 4. Tmab VH/CH1 (blue), Vκ/Cκ (green), HER2 fragment (orange), APR (red) and surface exposed aa identified for substitution to an N-linked glycosylation site reported to sterically hinder APR (purple).

Residue substitutions elected for the Tmab glycosylation mutation library are listed in Table 33.

Codons from the Tmab WT genes for mutagenesis were identified from the gWiz TH/TL expression vectors and mutations were designed based on minimum bp mismatching to Tmab WT codons. As previously described in Chapter 4, mutagenesis reactions were designed and performed based on the QuikChange™ method (Agilent Technologies). Overlapping primer pairs with introduced mutations were designed and the entire vector was amplified through high fidelity inverse PCR. Vectors containing successful mutations were expressed in transient 293F culture and purified through protein A affinity chromatography in the lead up to stability screening. For the purposes of this study, the gWiz TH/TL expression vectors were used for producing the Tmab glycosylation mutant library.

108

Further to screening for stability enhancement, the Tmab glycosylation mutation library binding kinetics to HER2 and a panel of FcγRs was screened to ensure that introduced mutations and glycosylation do not contribute to a loss of immunological functions. Tmab targets extracellular HER2 for the treatment of HER2 positive breast cancers [21, 22]. As with all IgG1 antibodies,

Tmab binding to FcγR1A and FcγR3A induces immune response activation to the target to elicit an

ADCC response, and binding to FcγR2B inhibits this immune response [12, 16, 133, 249].

Table 33 Summary of Tmab aa substitutions elected for experimental validation. Codon bp that mismatch the WT gene codon for primer design consideration (bold)

Site WT Codon Substitution Mutation Codon

K AAA L177 (C 1) CTA H N AAT

Q160 (Cκ) CAG AAT

L115 ( Cjoin ) CTT AAT

T117 ( Cjoin ) ACC AAC

A121 ( CH1) GCT N AAT

Q178 ( CH1) CAG AAT

L182 ( CH1) CTC AAC

T198 ( CH1) ACC AAC

2 Methods 2.1 Instrumentation, Services and Software Crystal structure analysis for identification and confirmation of surface exposed N-linked glycosylation sites was performed on Dassault Systèmes Biovia Discovery Studio Visualizer v17.2.0.16349 software. Site-directed mutagenesis (SDM) reactions and accelerated thermal monomer loss studies were conducted on a Bio-Rad T100 thermal cycler, with heated lid. DNA quantification was performed on a NanoDrop ND-1000 spectrophotometer. Mutant isolate vectors were prepared as purified DNA samples and sent to AGRF for Sanger Sequence analysis. Sequencing results were analysed on Applied Biosystems Sequence Scanner 2 software.

Overlapping primer pairs for mutagenesis were designed using GSL Biotech SnapGene Viewer

4.0.5 software. Tm calculations were performed on NEB Tm calculator software with mismatched bp

109 omitted (http://tmcalculator.neb.com). Alignment of mutant isolate sequences to vector DNA sequences were performed on NCBI BLAST software (https://blast.ncbi.nlm.nih.gov/Blast.cgi).

Mutant antibodies were purified from culture supernatant on a GE HealthCare ÄKTApurifier 100 FLPC system, using Unicorn 5.11 system and analysis software. Liquid chromatography was performed on a Thermo Fisher Dionex™ 3500RS HPLC and mass spectrometry was performed in a Thermo Fisher Orbitrap Fusion™ Tribid™ MS using the Protein Metrics Byonic™ search engine.

Accelerated thermal stability screening was performed and analysed on an Unchained Labs UNcle™ system and analysis V2.0 software. Size-exclusion high-performance liquid chromatography (SE- HPLC) was performed on an Agilent Technologies Agilent™ 1200 Liquid Chromatography system, using ChemStation system and analysis software.

2.2 Materials gWiz TH and gWiz TL vectors were cloned from gWiz blank vector as per described in Chapter 2. The primers for designed for mutagenesis and sequencing gWiz inserts were synthesised from Geneworks (Australia).

Stellar™ competent E. coli (636763) was purchased from Scientifix (Australia), JM109 competent E. coli was prepared as per Appendix 1.3. Yeast extract (70161), Sodium chloride (S9888) and Kanamycin solution (K0254) was purchased from Sigma-Aldrich (Australia). Tryptone (LP0042B) was purchased from Thermo Fisher Scientific Australia and agar (20767.232) was purchased from VWR International (Australia).

(POL)) was purchased from BioScientific (Australia), bovine serum albumin (BSA) (BSAS 0.1) was purchased from Bovogen (Australia) and T75 TC flasks (CORN430641U) were purchased from Bio-Strategy (Australia). PBS tablets (09-2051-100) and 1 M Tris-HCl pH 9 (BIOSD814) were purchased from Astral Scientific (Australia). Disposable baffled culture flasks (CLS431405), EDTA (E6758), FITC secondary antibody (F9512), isotype control antibody (I5154), centrifuge tubes (CLS430291, CLS430791), vacuum systems (CLS430769), glycine hydrochloride (G2879), HiTrap® Protein A affinity column (GE17-0403-01) and Amicon® Ultracel 50 kDa centrifugal filters (Z740177, Z740193) were purchased from Sigma-Aldrich (Australia).

250 kDa Precision Plus™ molecular weight ladder (1610374) was purchased from Bio-Rad Laboratories (Australia) and all chemicals for reducing SDS-PAGE separation were purchased from

Astral Scientific (Australia). Trypsin (T6567) chymotrypsin (11418467001) and C18 Ziptips (Z720070) were purchased from Sigma-Aldrich (Australia). PNGase F (P0708) was purchased from Genesearch (Australia) and all chemicals for LC-MS/MS were purchased from Sigma-Aldrich (Australia).

The 20 cm × 75 μm Reprosil-Pur C18-AQ, 3 μm particle size 120 A in-house packed HPLC separating column (r13.aq) was purchased from Dr. Maisch GmbH (Germany). The 7.8 x 300 mm, ® ® 5 µm particle size TSKgel G3000SWXL HPLC column (0008541) and 6 x 40 mm TSKgel SWXL guard column (0008543) were purchased from Tosoh Bioscience GmbH (Japan).

® Herceptin was a kind donation from Genentech (USA) and FcγR1A HIS-tagged receptor (10256- H08H-5) was purchased from Life Technologies (Australia). HER2 HIS-tagged receptor

(SRP6405), FcγR2B HIS-tagged receptor (SRP6396), FcγR3A HIS-tagged receptor (SRP6436), HEPES (54457), NaCl (S9888) and Tween 20 (P9416) were purchased from Sigma-Aldrich (Australia). CM5 chips (29-1049-88), Amine coupling kit (BR-1000-50) and Anti HIS capture kit (28-9950-56) were purchased from GE Healthcare (Australia).

2.3 Site Directed Mutagenesis Overlapping mutagenesis primers were designed as described in Chapter 4, to incorporate mutagenic bp substitutions with a minimum of 8 bp homology upstream and downstream of the mutation site (Table 34). The primer lengths were varied due to GC content at the mutation site, and were adjusted to meet similar predicted Tm values, a GC clamp of 2 C/G bp in the 3’ end of the primers and 40 – 60 % GC content in all primers for optimal annealing. Mutagenesis was performed as an inverse PCR 2-step cycling condition to optimise mutation efficiency (Table 35) and minimise

111 primer ligation artefacts. All mutagenesis reactions were performed using a Phusion High Fidelity

PCR kit (Table 36) on the gWiz TH/TL expression vector system.

Table 34 List of primers designed for TH and TL glycosylation mutations. Nucleotides that mismatch the Tmab gene sequences, introducing codon mutation (bold). Primer Tm was calculated * to gWiz TH/TL with the Phusion PCR kit, with mismatches omitted from the primer sequences. 1 G/C bp on 3’ end due to high GC content of sequence in the mutation site.

Sequence Tm Primer 5’ 3’ (°C)

SDMC_TH1_L177K_Fwd CTTCCCGGCTGTCAAACAGTCCTCAGG 76 SDMC_TH1_L177K_Rev CCTGAGGACTGTTTGACAGCCGGGAAG*

SDM2_TH1_L177N_Fwd CTTCCCGGCTGTCAATCAGTCCTCAGG 74 SDM2_TH1_L177N_Rev CCTGAGGACTGATTGACAGCCGGGAAG*

SDM2_TkC_Q160N_Fwd CAATCGGGTAACTCCAATGAGAGTGTCACAGAGC 77 SDM2_TkC_Q160N_Rev GCTCTGTGACACTCTCATTGGAGTTACCCGATTG*

SDM2_THV_L115N_Fwd GGTCAAGGTACTAATGTCACCGTCTCCTCAGC 76 SDM2_THV_L115N_Rev GCTGAGGAGACGGTGACATTAGTACCTTGACC

SDM2_THV_T117N_Fwd GGTACTCTTGTCAACGTCTCCTCAGCTAGC 71 SDM2_THV_T117N_Rev GCTAGCTGAGGAGACGTTGACAAGAGTACC

SDM0_TH1_A121N_Fwd CCGTCTCCTCAAATAGCACCAAGGGC 74 SDM0_TH1_A121N_Rev GCCCTTGGTGCTATTTGAGGAGACGG

SDM0/2_TH1_Q178N_Fwd CCGGCTGTCCTAAATTCCTCAGGACTCTACTCC 75 SDM0/2_TH1_Q178N_Rev GGAGTAGAGTCCTGAGGAATTTAGGACAGCCGG

SDM2_TH1_L182N_Fwd CCTACAGTCCTCAGGAAACTACTCCCTCAGCAGCG 78 SDM2_TH1_L182N_Rev CGCTGCTGAGGGAGTAGTTTCCTGAGGACTGTAGG

SDM0_TH1_T198N_Fwd GCAGCTTGGGCAACCAGACCTACATC* 74 SDM0_TH1_T198N_Rev GATGTAGGTCTGGTTGCCCAAGCTGC

All PCR products were digested with DpnI at 37 °C for 2 hr to remove native vector from the PCR product. 5 µL of the digestion products were transformed in competent E. coli via heat-shock method (Appendix 1.4), cells containing intact vector were rescued in LB agar containing 50 µg/mL Kanamycin as selection antibiotic. All digestion products were transformed into in-house prepared JM109 competent E. coli (Appendix 1.3). If mutants were not isolated from that round of transformation then digestion products were transformed again into commercial Stellar™ competent E. coli.

112

Single colonies from transformations were isolated and cultured for glycerol stock preparation (Appendix 1.2) and plasmid miniprep extraction (Appendix 2.1.1) in preparation for sequencing (Appendix 2.2). Isolates that screened positive for the mutation from sequence analysis were resuscitated and culture scaled up for plasmid maxiprep extraction (Appendix 2.1.3 - 2.1.4) in the lead up to production of the Tmab affinity mutant library through expression in transient 293F culture.

Table 35 Cycling conditions for the mutagenesis inverse PCR 2-step reaction for TH and TL glycosylation mutations.

Cycle Step Cycles Time (sec) Temp (°C)

Initial denaturation 1 30 98

Denaturation 10 98 25 Annealing and Extension 120 72

Final Extension 1 300 72

Table 36 Reaction components for TH and TL glycosylation mutations.

Reaction Volume Component Concentration (µL)

5 x Phusion HF Buffer 1 x 5

10 mM dNTP 200 µM 0.5

10 µM Forward primer 1.5 pM 0.75

10 µM Reverse primer 1.5 pM 0.75

Vector DNA 0.6 ng/µL as needed

Nuclease free H2O - to 25 µL

Phusion DNA Polymerase 20 units/mL 0.25

2.4 Expression and Purification

Successful gWiz TH/TL mutant vectors were prepared for transient expression in suspension 293F culture. 24 hr prior to transfection, 293F cells were passaged to a seeding density of 5 x 105 cells/mL in Freestyle 293 expression media in 60 mL culture. Transfection proceeded if the live cell

113 concentration reached 0.8 – 1.2 x 106 cells/mL and cell viability was ≥ 95% on the day of transfection. 120 µg PEI and 60 µg DNA (1:1 ratio of gWiz TH : gWiz TL) were separately diluted in OptiPRO SFM to total volume of 4 mL and incubated at RT for 5 min. The diluted DNA and PEI solutions were combined, gently mixed and incubated at RT for 15 min prior to dropwise addition to transfection culture. Cultures were returned to incubate, 24 hr post transfection the cultures were supplemented with 0.5% w/v tryptone to total culture volume. 2 days post transfection, cultures were scaled up to double volume (120 mL) and supplemented with 0.5% w/v tryptone. 3 days post transfection, cultures were further scaled up to double volume (240 mL) and again supplemented with 0.5% w/v tryptone. All transfection culture supernatants were harvested 8 days post transfection by centrifugation (3,000 g, 15 min), clarified and degassed by passing through a 0.22 μm vacuum filter. Secreted antibody was purified from the culture supernatant through Protein A affinity chromatography, using an ÄKTApurifier 100 FPLC system with a HiTrap® Protein A column (Appendix 4). Antibody was bound on the column with PBS as binding buffer, eluted from the column with 0.1 M glycine HCl (pH 2.7) and elution fractions neutralised with 1 M Tris-HCl (pH 9). FPLC chromatograms detected the elution peaks with UV absorbance at 280 nm and elution fractions containing antibody were pooled, buffer exchanged to PBS and concentrated using 50 kDa centrifugal filters. UV absorption of the purified antibody samples was measured at 280 nm to derive total antibody yield using the Beer Lambert law and the extinction coefficient of trastuzumab (Appendix 4.2).

2.5 Mutations Confirmed through Mass Spectrometry Purified Tmab and glycosylation mutants were confirmed of the predicted aa sequence and successful introduction of a glycosylation site through site-specific liquid chromatography tandem mass spectrometry (LC-MS/MS) as detailed in Appendix 4.4. Briefly, purified Tmab samples were separated on 10% reducing SDS-PAGE, the separated heavy and light chains were excised from the gel, washed and digested with trypsin and chymotrypsin. A fraction of the digestions were further treated with PNGase F for glycan release studies. Peptide digestions and glycan released peptide digestions were desalted in preparation of liquid chromatography separation. The digestion products were separated through HPLC using an in-house packed Reprosil-Pur C18-AQ column over a 90 min 0 - 40% solvent gradient from 0.1% (v/v) formic acid to 80% (v/v) acetonitrile, 0.1% (v/v) formic acid, running at 300 nL/min, 60 °C. Tandem mass spectrometry was performed on the separated digested peptides using higher-energy collisional dissociation (HCD), collision induced dissociation (CID) and electron transfer dissociation (ETD) methods. The spectra were analysed by performing peptide spectrum matches (PSM) against the corresponding native or predicted mutation

114 aa sequence fasta file. Spectral annotations were manually validated and matches were compared between the HCD, CID and ETD methods to confirm and generate the complete aa sequences of the Tmab samples light and heavy chains. Successful introduction of a glycosylation site was confirmed through comparison of the glycan released sample fractions to their corresponding glycan intact samples.

2.6 Accelerated Thermal Stability and Monomer Loss Screen The Tmab glycosylation mutant library was assessed through an accelerated thermal stability screen ™ to derive melting temperature (Tm) and onset temperature of aggregation (Tagg) using the UNcle system. The Tmab library was further assessed in an accelerated thermal relative monomer loss study as previously described [250], quantified by SE-HPLC. Prior to stability screening, Tmab sample concentrations were normalised to 1 mg/mL in PBS.

Briefly for the stability screen, 9 µL of each Tmab samples was loaded on the UNcle™ system and a linear temperature ramp (15 – 95 °C) was applied at a ramp rate of 1 °C/min. A continuous scan for static light scattering (SLS) and intrinsic tryptophan fluorescence was measured through sample excitation at 266 nm, SLS intensity was further measured through sample excitation at 473 nm. The barycentric mean (BCM) of the Tmab samples were calculated through the UNcle™ analysis software (i.e. the mean fluorescence wavelength) and were plotted against temperature to generate thermal melting curves of the Tmab samples. Protein unfolding was observed as an increase in

BCM, sudden or gradual increase reflected the rate of unfolding. Tms were calculated as the midpoint of the transition states (inflection points), identified through first differentials of the melting curves. Analysis and manual inspection of differentials was conducted using the UNcle™ analysis software.

SLS intensity of the Tmab samples were plotted against temperature to generate thermal aggregation curves. Generation and accumulation of soluble aggregates in the sample was observed as an increase in SLS intensity with sensitivity to small particle formation at 266 nm and large particle formation at 473 nm. Sudden or gradual increase in intensity reflected the rate of aggregation. The midpoint of the transition state (inflection point) was identified through first differentials of the thermal aggregation curves. The temperature where signal intensity was 10% of the intensity at the inflection point was derived as the Tagg of the sample. Analysis and manual inspection of differentials was conducted using the UNcle™ analysis software.

Briefly for the monomer loss study, three sets of Tmab samples were prepared to 30 µL. Tmab sample was either incubated in a thermal cycler with heated lid at 68 °C for 10 min or 30 min, or

115 prepared without incubation. Samples were immediately transferred to an ice bath after incubation, until injection into the HPLC system (25 µL) for characterisation and quantification. HPLC was performed on all samples as detailed in Appendix 4.5 using an Agilent 1200 liquid chromatography ® system with a TSKgel G3000SWXL HPLC column coupled to a guard column. The HPLC system was running at 22 °C at a flow rate of 1 mL/min, with mobile phase of 150 mM potassium phosphate, pH 6.5. Total protein content of the sample was detected through UV absorption at 280 nm to produce chromatograms that profiled the protein size distribution in the Tmab samples. Monomer peaks of the sample chromatograms were specifically quantified for the area under the curve (AUC) to derive a percentage of relative monomer loss against the AUC of its corresponding non-incubated sample.

2.7 HER2 Cell-binding Assay The Tmab glycosylation mutant library was assessed for affinity of extracellularly expressed HER2 in a cell-binding end point HER2 affinity assay as detailed in Appendix 5.2. The Tmab glycosylation mutant library titrations were compared to Tmab WT expressed from the gWiz TH/TL vector system (Tmab gWiz). As previously described in Chapter 4, overexpressing HER2 cell line SK-BR3 was harvested non-enzymatically with Versene Solution, washed and incubated with Tmab sample titrations for 20 min, 4 °C. After incubation the cells were washed, resuspended with a FITC-secondary anti-human IgG1/κ antibody and further incubated for 20 min, 4 °C. The cells were again washed, resuspended and passed through flow cytometry to detect the amount of green fluorescence per cell in the single cell population. For each titration batch Herceptin® buffer exchanged to PBS (PBS-Herc) was run as a positive control, a polyclonal IgG1/κ isotype was run as a negative control and dilution buffer was run as a wash control. The controls were used to define the cell populations detected on flow cytometry as positive or negative cell binding. All titration samples were measured through the control defined parameters in flow cytometry, with 5,000 events collected in duplicate. The fraction of cells defined as positive (bound) in each sample was plotted against antibody concentration in a sigmoidal binding curve to extrapolate KD at equilibrium, that being antibody concentration at 50% positive fraction (bound).

2.8 Binding Kinetics of Tmab Mutants to HER2 and FcγR

The binding kinetics of PBS-Herc and Tmab glycosylation mutant library was assessed to HER2,

FcγR1A, FcγR2B and FcγR3A through surface plasmon resonance (SPR) as detailed in Appendix

5.3. HER2 kinetics was measured to complement results from the cell binding assay and FcγR kinetics was measured to ensure that the mutations did not contribute to a loss of immunological functions.

116

Briefly, a CM5 chip was prepared with anti-HIS antibody through amine coupling for the capture of

HIS-tagged HER2 or FcγR (4 nM, 5 µL/min for 5 min). Measurements were performed at 25 °C in HBS-T running buffer as single cycle kinetic assays. 2-fold serial dilutions of Tmab samples were prepared from 0.5 - 8 nM, 2.5 - 40 nM and 25 – 400 nM for HER2, FcγR1A and FcγR2B/3A respectively. Increasing concentrations of Tmab sample was applied over the receptor captured surface (20 µL/min, 2 min) with dissociation times of 60 min and 30 min for HER2 and FcγR binding respectively. The sensorgrams were analysed globally using a Langmuir 1:1 binding model to determine Ka, Kd and KD. All samples were conducted in duplicate to obtain average values and standard deviation. Kinetic rate and binding constants determined from sensorgrams were accepted 2 if the goodness of fit value (χ ) was within 5% of the maximum response level (Rmax) of the sensorgrams and KD was calculated as Kd/Ka.

3 Results 3.1 Mutation Efficiency and Expression

Mutagenesis was performed on gWiz TH/TL expression vectors through an inverse PCR 2-step cycling condition to optimise mutation efficiency and minimise primer ligation artefacts. Mutagenesis was successful for the entire Tmab glycosylation mutant library; however, several mutants produced isolatable primer ligation artefacts (Table 37). The majority of mutants were isolated from commercial Stellar™ competent E. coli, several were due to an unsuccessful round of capture from in-house JM109 competent E. coli. Reactions L177>K, L177>N and Q160>N were rescued directly into Stellar™ cells, along with mutants L115>N, T117>N and Q178>N that were not isolated from the first round of transformation into JM109 cells. Of the reactions rescued in JM109, no isolates were rescued from the T117>N and Q178>N reactions, and only 1 isolate was rescued from the L115>N reaction that screened negative for the mutation and positive for a primer ligation artefact through sequence analysis (not shown).

An example of positive mutation identified through sequence analysis is shown in Figure 29, and of a primer ligation artefact is shown in Figure 30. Nucleotide forward sequence of mutation positive isolate gWiz TL Q160>N(1) was aligned with the nucleotide sequence for gWiz TL (Figure 29).

Homology from 1880 – 2581 bp in the gWiz TL sequence confirms the complete TL gene with bp mismatch from 2414 – 2416 CAG > AAT confirmed for mutation Q160>N. No further alignments and continued homology beyond site of mutation confirmed no further artefacts produced by primer ligation products. Nucleotide forward sequence of primer ligation artefact isolate gWiz TL

Q160>N(2) was aligned with the nucleotide sequence for gWiz TL (Figure 30). Homology from 1880 – 2413 bp in the

117 gWiz TL sequence confirms the TL gene until the mutation site, bp mismatch from 2414 – 2416 CAG > AAT confirmed for mutation Q160>N; however, the primer sequence was repeated in the isolate nucleotide sequence before continued homology from 2417 bp.

Transient expression and purification of the Tmab glycosylation mutant library was successful for all mutants, with varied yield amongst the mutants (not shown). Aside from gWiz Tmab (1.55 mg), the highest yielding Tmab mutants from the library were L177>K (1.95 mg), A121>N (1.32 mg) and L177>N (1.00 mg). The poorest yielding were L182>N (370 µg) and T117>N (190 µg). All purified Tmab glycosylation mutants were of good yield and quality to proceed with mutation and glycosylation confirmation, stability and binding kinetics studies.

Table 37 Comparison of efficiencies between the different mutations conducted in this study, +rescued from in-house JM109 competent E. coli.

Isolates Mutation Primer Ligation Mutation Yielded Efficiency Artefacts

L177 > K > 300 1/5 (20 %) 3/5 (60%)

L177 > N 31 5/5 (100%) 0/5 (0%)

Q160 > N > 300 2/5 (40%) 3/5 (60%)

L115 > N 222 1/1 (100%) 0/1 (0%)

T117 > N 13 3/3 (100%) 0/3 (0%)

A121 > N 13+ 2/3 (67%) 0/3 (0%)

Q178 > N 126 2/3 (67%) 1/3 (33%)

L182 > N 4+ 2/3 (67%) 0/3 (0%)

T198 > N 9+ 1/3 (33%) 2/3 (67%)

118

Query 1819 AGCTGACAGACTAACAGACTGTTCCTTTCCATGGGTCTTTTCTGCAGTCACCGTCGTCGACATGTTGCCATCACAACTCATTGGGTTTCTGCTGCTCTGGGTTCCAGCTAGCCGCGGTGACATCCAGATGACCCAGTCTCCTTCTTCTCT 1968 ||||||| |||| | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| Sbjct 9 AGCTGAC-GACT-ANAGACTGTTCCTTTCCATGGGTCTTTTCTGCAGTCACCGTCGTCGACATGTTGCCATCACAACTCATTGGGTTTCTGCTGCTCTGGGTTCCAGCTAGCCGCGGTGACATCCAGATGACCCAGTCTCCTTCTTCTCT 156

Query 1969 TTCTGCTTCTGTTGGTGATCGTGTTACTATTACTTGTCGTGCTTCTCAAGATGTTAATACTGCTGTTGCTTGGTATCAACAAAAACCTGGTAAAGCTCCTAAActtcttatttattctgcttcttttctttattctggtgttccttctcg 2118 |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| Sbjct 157 TTCTGCTTCTGTTGGTGATCGTGTTACTATTACTTGTCGTGCTTCTCAAGATGTTAATACTGCTGTTGCTTGGTATCAACAAAAACCTGGTAAAGCTCCTAAACTTCTTATTTATTCTGCTTCTTTTCTTTATTCTGGTGTTCCTTCTCG 306

Query 2119 tttttctggttctcgttctGGTACTGATTTTACTCTTACTATTTCTTCTCTTCAACCTGAAGATTTTGCTACTTATTATTGTCAACAACATTATACTACTCCTCCTACTTTTGGTCAAGGTACCAAGCTGGAGATCAAACGTACGGTGGC 2268 |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| Sbjct 307 TTTTTCTGGTTCTCGTTCTGGTACTGATTTTACTCTTACTATTTCTTCTCTTCAACCTGAAGATTTTGCTACTTATTATTGTCAACAACATTATACTACTCCTCCTACTTTTGGTCAAGGTACCAAGCTGGAGATCAAACGTACGGTGGC 456

Query 2419 GAGTGTCACAGAGCAGGACAGCAAGGACAGCACCTACAGCCTCAGCAGCACCCTGACGCTGAGCAAAGCAGACTACGAGAAACACAAAGTCTACGCCTGCGAAGTCACCCATCAGGGCCTGAGCTCGCCCGTCACAAAGAGCTTCAACAG 2568 |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| Sbjct 607 GAGTGTCACAGAGCAGGACAGCAAGGACAGCACCTACAGCCTCAGCAGCACCCTGACGCTGAGCAAAGCAGACTACGAGAAACACAAAGTCTACGCCTGCGAAGTCACCCATCAGGGCCTGAGCTCGCCCGTCACAAAGAGCTTCAACAG 756

Query 2569 GGGAGAGTGTTAGGGGATCCAGATCACTTCTGGCTAATAAAAGATCAGAGCTCTAGAGATCTGTGTGTTGGTTTTTTGTGGATCTGCTGTGCCTTCTAGTTGCCAGCCATCTGTTGTTTGCCCCTCCCCCGTGCCTTCCTTGACCCTGG- 2717 ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| ||| Sbjct 757 GGGAGAGTGTTAGGGGATCCAGATCACTTCTGGCTAATAAAAGATCAGAGCTCTAGAGATCTGTGTGTTGGTTTTTTGTGGATCTGCTGTGCCTTCTAGTTGCCAGCCATCTGTTGTTTGCCCCTCCCCCGTGCCTTCCTTGACCNTGGA 906

Query 2718 AAGGTGCCACTCCCACTGTCCTTTCCTAATAAAATGAGGAAATTGCATCGCATTGTCT-GAGTAGGTGTCATTCTATTCTgggggg--tgggg-t-gggg-C-AGCACA-GCAAGGGGG-A-GGATTGGG-AAGA-CAATAGCAGG-CAT 2854 ||||||||||||||||||||||||||||||||||||||||||||||||||||||||| ||||||||||||||||||||||||||| ||||| | |||| | || | ||||||||| | |||||||| || | ||||||| || ||| Sbjct 907 AAGGTGCCACTCCCACTGTCCTTTCCTAATAAAATGAGGAAATTGCATCGCATTGTCCGGAGTAGGTGTCATTCTATTCTGGGGGGGGTGGGGGTTGGGGNCNAGGAANNGCAAGGGGGGAAGGATTGGGGAANAACAATAGCCGGGCAT 1056

Query 2855 GCTGGGG-A-TGCGGTGGG-CTCTATGGG-TACctctctctctctctctctctctctctctctctctctc 2920 ||||||| | ||||| ||| ||| ||||| |||||| ||||| |||| |||| | ||||||| |||||| Sbjct 1057 GCTGGGGGAATGCGGGGGGGCTCCATGGGGTACCTC-CTCTC-CTCTNTCTCCC-CTCTCTCCNTCTCTC 1123

Figure 29 Example of a BLAST sequence alignment demonstrating a positive mutation sequence analysis result. gWiz TL Q160>N(1) positive isolate (Subjct) aligned to the complete nucleotide sequence of gWiz TL (Query), mismatch at mutation site (green, circled).

119

Query 1819 AGCTGACAGACTAACAGACTGTTCCTTTCCATGGGTCTTTTCTGCAGTCACCGTCGTCGACATGTTGCCATCACAACTCATTGGGTTTCTGCTGCTCTGGGTTCCAGCTAGCCGCGGTGACATCCAGATGACCCAGTCTCCTTCTTCTCT 1968 ||||||| |||| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| Sbjct 9 AGCTGAC-GACT-ACAGACTGTTCCTTTCCATGGGTCTTTTCTGCAGTCACCGTCGTCGACATGTTGCCATCACAACTCATTGGGTTTCTGCTGCTCTGGGTTCCAGCTAGCCGCGGTGACATCCAGATGACCCAGTCTCCTTCTTCTCT 156

Query 1969 TTCTGCTTCTGTTGGTGATCGTGTTACTATTACTTGTCGTGCTTCTCAAGATGTTAATACTGCTGTTGCTTGGTATCAACAAAAACCTGGTAAAGCTCCTAAActtcttatttattctgcttcttttctttattctggtgttccttctcg 2118 |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| Sbjct 157 TTCTGCTTCTGTTGGTGATCGTGTTACTATTACTTGTCGTGCTTCTCAAGATGTTAATACTGCTGTTGCTTGGTATCAACAAAAACCTGGTAAAGCTCCTAAACTTCTTATTTATTCTGCTTCTTTTCTTTATTCTGGTGTTCCTTCTCG 306

Query 2119 tttttctggttctcgttctGGTACTGATTTTACTCTTACTATTTCTTCTCTTCAACCTGAAGATTTTGCTACTTATTATTGTCAACAACATTATACTACTCCTCCTACTTTTGGTCAAGGTACCAAGCTGGAGATCAAACGTACGGTGGC 2268 |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| Sbjct 307 TTTTTCTGGTTCTCGTTCTGGTACTGATTTTACTCTTACTATTTCTTCTCTTCAACCTGAAGATTTTGCTACTTATTATTGTCAACAACATTATACTACTCCTCCTACTTTTGGTCAAGGTACCAAGCTGGAGATCAAACGTACGGTGGC 456

Query 2269 GGCGCCATCTGTCTTCATCTTCCCGCCATCTGATGAGCAGTTGAAATCTGGAACTGCCTCTGTTGTGTGCCTGCTGAATAACTTCTATCCCAGAGAGGCCAAAGTACAGTGGAAGGTGGATAACGCCCTCCAATCGGGTAACTCCCAG-- 2416 ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| | Sbjct 457 GGCGCCATCTGTCTTCATCTTCCCGCCATCTGATGAGCAGTTGAAATCTGGAACTGCCTCTGTTGTGTGCCTGCTGAATAACTTCTATCCCAGAGAGGCCAAAGTACAGTGGAAGGTGGATAACGCCCTCCAATCGGGTAACTCCAATGA 606 * Query 2417 ------GAGAGTGTCACAGAGCAGGACAGCAAGGACAGCACCTACAGCCTCAGCAGCACCCTGACGCTGAGCAAAGCAGACTACGAGAAACACAAAGTCTACGCCTGCGAAGTCACCCATCAGGGCCTG 2539 ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| Sbjct 607 GAGTGTCACAGAGCGGGTAACTCCAATGAGAGTGTCACAGAGCAGGACAGCAAGGACAGCACCTACAGCCTCAGCAGCACCCTGACGCTGAGCAAAGCAGACTACGAGAAACACAAAGTCTACGCCTGCGAAGTCACCCATCAGGGCCTG 756 * Query 2540 AGCTCGCCCGTCACAAAGAGCTTCAACAGGGGAGAGTGTTAGGGGATCCAGATCACTTCTGGCTAATAAAAGATCAGAGCTCTAGAGATCTGTGTGTTGGTTTTTT-GT-GGATCTGCTGTG-CCTTCTAGTTGCCAGCCATCTGTTG-T 2685 |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| || |||||||||||| ||||||||||||||||||||||||| | Sbjct 757 AGCTCGCCCGTCACAAAGAGCTTCAACAGGGGAGAGTGTTAGGGGATCCAGATCACTTCTGGCTAATAAAAGATCAGAGCTCTAGAGATCTGTGTGTTGGTTTTTTTGTGGGATCTGCTGTGCCCTTCTAGTTGCCAGCCATCTGTTGGT 906

Query 2686 TTGCCCC-TCCCCC-GTG-CCTTCCTTGACCCTGGAA-GGTGCC-ACTCCC-ACTG-TCCTTTCCTAATAAAA-TG-AGGAAATTGCATCGC-ATTGT-CTGAGTA-GGTGTC-ATTCTATT-CTggggggtggggtgggg 2812 ||||||| |||||| ||| |||||||||||||||||| |||||| |||||| |||| |||||||||||||||| || ||||||||||||| | ||||| | ||||| |||||| |||||||| |||||||| ||| ||||| Sbjct 907 TTGCCCCCTCCCCCCGTGCCCTTCCTTGACCCTGGAAAGGTGCCCACTCCCCACTGTTCCTTTCCTAATAAAAATGAAGGAAATTGCATCCCCATTGTTCGGAGTAAGGTGTCCATTCTATTTCTGGGGGG-GGG-TGGGG 1045

Figure 30 Example of a BLAST sequence alignment demonstrating a primer ligation artefact sequence analysis result. gWiz TL Q160>N(2) primer ligation artefact (Subjct) aligned to the complete nucleotide sequence of gWiz TL (Query). Complete forward primer sequence (blue), mutation site in primer (red), repeated primer sequence (underlined) with * marking the first bp of the repeat sequence.

120

3.2 Confirmation of Successful Mutation and N-glycan Addition The Tmab glycosylation mutant library was separated on a 10% reducing SDS-PAGE for separation of the heavy and light chains in preparation for mass spectrometry (Figure 31). A primary indication of successful glycosylation was apparent from the gel analysis, with glycosylation mutants of the heavy chain (L115>N not shown) all migrating to 53 kDa, above the Tmab WT and non-glycosylated mutant L177>K heavy chain which migrated at 50 kDa. Glycosylation mutant of the light chain Q160>N (not shown) did not have an apparent difference in migration of the light chain through the gel as compared to the native Tmab light chain at 26 kDa, which suggested that glycosylation was not apparent in the light chain of this mutant. Gel analysis also showed that a proportion of non-glycosylated heavy chain species existed in T198>N (12%) and L182>N (7%). Residual fragment impurities of common size were observed in many of the purified Tmab samples (35 and 70 kDa) which match similarly to expected residual Protein A (42.0 kDa) fragment and HER2 (72.4 kDa) that had been captured together with the Tmab samples. Both the heavy and light chains of Tmab gWiz were excised from the gel in preparation for mass spectrometry. For all other Tmab mutants, the chain corresponding to its mutation was excised from the gel in preparation for mass spectrometry, i.e. the light chain for Q160>N and heavy chain for all other mutants.

L 1 2 3 4 5 6 7 8 kDa

250 - 150 - 100 - 75 -

← 53 kDa 50 - ← 50 kDa

37 -

← 26 kDa 25 -

Figure 31 10% Reducing SDS-PAGE of Tmab glycosylation mutant library. Bio-Rad Precision plus Ladder (L), Tmab gWiz (1), L177>K (2), L177>N (3), Q178>N (4), T198>N (5), T117>N (6), L182>N (7) and A121>N (8).

121

LC-MS/MS analysis confirmed that the entire Tmab glycosylation mutant library had successful aa substitution (Appendix 6.1). Fragmentation techniques were further applied to HCD, ETD and CID to provide compositional glycan information of the mutation-introduced N-linked glycosylation sites. All Tmab glycosylation mutants aside for Q160>N and L177>K were found to contain an addition N-glycan compared to the corresponding native Tmab chain, demonstrating successful glycosylation of the mutation-introduced glycosylation sites. This result agrees with the previous SDS-PAGE analysis observing no difference in molecular weight between the light chain of Q160>N, or the heavy chain of L177>K to the native corresponding Tmab chains despite successful substitution.

The MS glycan results for L177>K was as expected, as this substitution was not designed to introduce an N-linked glycosylation site. However, the lack of glycan addition to Q160>N was not expected as this substitution was designed to introduce an N-linked glycosylation site. This disagrees with previous literature that claims success in producing this particular glycosylated mutant in bevacizumab and that its performance indeed enhanced bevacizumab’s stability [168]. All further results obtained for this mutant is therefore due to the residue substitution alone.

3.3 Accelerated Thermal Stability Screening The Tmab glycosylation mutation library was assessed on the UNcle™ system through an accelerated thermal stability screen to derive Tm and Tagg. Intrinsic tryptophan fluorescence was monitored at 266 nm excitation through the temperature ramp to observe protein unfolding events in the antibody samples, generating melting curves (Figure 32) from which Tms were calculated (Table 38). SLS was monitored to observe aggregate generation and accumulation in the samples through the temperature ramp at 266 nm for small particle formation (Figure 33) and 437 nm for large particle formation (Figure 34), generating thermal aggregation curves from which Tagg was calculated (Table 38).

As glycan addition was not present with mutant Q160>N, it was excluded from this assay. The

Tmab WT had highly similar Tms to the reference PBS-Herc, although interestingly the Tagg was measured as 3 °C higher in the Tmab WT than PBS-Herc (Table 38). Mutant T117>N produced a low signal through the stability screen, measurements reflected background noise of the system and therefore Tm and Tagg were not derived for this mutant. The aggregation events for all other samples in the Tmab glycosylation mutant library occurred concurrently with the second melting event. Of the thermal melting curve results, L177>N and L115>N was the most stabilising for the Tm1,

L115>N and T198>N for the Tm2 as compared to the WT. The melting curve of L182>N showed

122 apparent stabilisation, with a delayed and reduced rate of melting from the Tm1 as compared to the

WT; however, the Tm values calculated did not reflect this observation. Of the thermal aggregation curve data, precipitation of aggregates was observed for PBS-Herc and mutant A121>N in both the 266 nm and 473 nm curves. Mutant L177>K had an apparent earlier onset of aggregation in both curves that was reflected in the calculated Tagg values. The aggregation curves of L115>N, L182>N and T198>N had an apparent resistance to aggregation as compared to WT.

However, the Tagg values of L115>N did not adequately reflect what was observed from the curves, in which it presented as the most resistant to aggregation and having a reduced rate of aggregation as compared to WT. A deviation in the data set was observed for L182>N at 65 °C which caused an error in the calculation of Tagg and was therefore excluded from the results. The aggregation curves of L177>N, A121>N and Q178>N had apparent similar onset of aggregation to WT, although the calculated Tagg values reflect an improvement.

The Tmab glycosylation mutant library was further assessed in an accelerated thermal relative monomer loss study. The Tmab mutants were incubated at 68 °C at 10 and 30 min time points. The amount of monomer remaining in the samples after incubation was quantified through SE-HPLC and compared against Tmab samples that were not incubated to produce a relative percentage of monomer loss.

Table 38 Tm and Tagg values calculated from the thermal melting and aggregation curves of the Tmab glycosylation mutant library.

Tm1 (°C) Tm2 (°C) Tagg (°C) 266 nm Tagg (°C) 437 nm

PBS-Herc 68.94 82.50 78.02 78.60

Tmab WT 68.50 81.31 81.54 81.70

L177 > K 69.00 79.27 78.51 80.59

L177 > N 70.50 78.78 82.13 82.65

L115 > N 71.00 81.92 74.15 75.90

T117 > N ND ND ND ND

A121 > N 69.60 80.50 82.41 82.78

Q178 > N 69.40 80.50 82.22 82.50

L182 > N 68.55 81.31 ND ND

T198 > N 69.80 82.50 80.61 83.73

123

Figure 32 Melting curves of the Tmab glycosylation mutant library generated from the UNcle™ system. (A) Full temperature ramp of all samples highlighting the unfolding events in PBS-Herc that correspond to the calculated Tm1 and Tm2, (B) curve magnified over the melting regions for comparison of mutant Tms against Tmab WT (PBS-Herc excluded).

124

Figure 33 Thermal aggregation curves of the Tmab glycosylation mutant library generated from the UNcle™ system, measured at 266 nm. (A) Full temperature ramp of all samples highlighting the initiated aggregation event in PBS-Herc that correspond to the calculated Tagg, (B) curve magnified over the region where aggregation initiates for comparison of Tmab mutants Tagg against WT (PBS- Herc and T117>N excluded).

125

Figure 34 Thermal aggregation curves of the Tmab glycosylation mutant library generated from the UNcle™ system, measured at 473 nm. (A) Full temperature ramp of all samples highlighting the initiated aggregation event in PBS-Herc that corresponds to the calculated Tagg, (B) curve magnified over the region where aggregation initiates for comparison of Tmab mutants Tagg against WT (PBS-Herc and T117>N excluded).

126

Figure 35 Relative monomer loss of the Tmab glycosylation mutation library at 68 °C, 10 min (solid) and 30 min (pattern) incubation time as quantified by SE-HPLC.

Due to sample limitations and poor performance from the stability screen, mutant T117>N was excluded from this assay. The accelerated monomer loss study performed at 68 °C demonstrated substantial differences in the Tmab glycosylation mutation library (Figure 35). Mutants L177>N and L182>N showed superior retention of monomer content as compared to WT. Mutants Q160>N, Q178>N and T198>N showed considerably (2 – 3 fold) reduced monomer content as compared to WT. Mutants L177>K, L115>N and A121>N also displayed reduced monomer content as compared to WT, but to a lesser extent. Mutants L177>N and L182>N possess apparent thermal stability as reflected in their melting curves, which appears effective at enhancing monomer retention under thermal stress.

3.4 Retention of Biological Activity to HER2 and FcγRs The Tmab glycosylation library was screened for retention of biological activity to extracellularly expressed HER2 target in a cell-binding assay, detected through flow cytometry. Binding curves and KD values derived from the cell-binding assay revealed a retention of affinity to HER2 amongst the entire Tmab glycosylation mutant library (Figure 36 and Table 39), with most mutants performing similarly, if not demonstrating a modest increase in binding affinity compared to Tmab WT. Notably, mutant L177>K had the most significant increase (2.5-fold) in binding as compared

127 to WT, followed by mutant Q178>N (2-fold) and L115>N (1.7-fold). Mutants T117>N and A121>N exhibited a slight loss in binding affinity (1.1 – 1.3-fold) although the significance of this reduction is negligible.

Figure 36 Tmab glycosylation mutant library binding curves to SK-BR3 cells from the cell-binding HER2 affinity assay, detected on flow cytometry.

SPR single-cycle kinetic assays of the Tmab glycosylation mutant library (Appendix 6.2) produced sensorgrams of typical curvature for the panel of HER2 and FcγR receptors, as shown with Tmab WT in Figure 37. Inspection of the Tmab glycosylation library sensorgrams confirmed satisfactory curve fitting (Appendix 6.4) for calculated values as presented in Table 40 and Table 41. The SPR assay screening Tmab library against surface captured FcγR2B was not robust despite multiple attempts. T117>N mutation performed poorly in all SPR assays despite repeated attempts. As the kinetic values derived from T117>N SPR were not robust, all calculated values were excluded from the results presented.

Tmab WT performed similarly to PBS-Herc on all receptors, aside from a 4-fold reduction in binding (increase in KD) to HER2. In comparing KD values to HER2 specifically, SPR derived kinetic values showed 1000-fold stronger KD of the Tmab glycosylation library binding to surface captured HER2 (pM) (Table 40) as opposed to the cell-based affinity assay (nM) (Table 39). Unlike

128 the results presented from the cell-binding assay, all Tmab glycosylation mutants exhibited a modest decrease in KD as compared to Tmab WT, notably that the dissociation constant (Kd) was the most drastically affected with mutants A121>N and T198>N. The entire Tmab glycosylation library KD to HER2 ranged from 30 – 84 pM, which was a 2.5 – 7-fold reduction in binding as compared to Tmab WT, with mutants T198N and L177>K performing poorest.

Table 39 KD values extrapolated from the cell-binding assay of the Tmab glycosylation mutant library. KD values were calculated as antibody concentration (µg/mL) at equilibrium (50% bound) and converted to nM for comparison with SPR data.

KD (µg/mL) KD (nM)

Tmab WT 1.63 11.22

L177 > K 0.661 4.54

L177 > N 1.131 7.77

Q160 > N 1.028 7.06

L115 > N 0.947 6.51

T117 > N 2.169 14.90

A121 > N 1.849 12.71

Q178 > N 0.783 5.38

L182 > N 1.220 8.38

T198 > N 1.557 10.70

SPR assays of the Tmab glycosylation mutant library binding to surface captured FcγR1A (Table

40), FcγR2B and FcγR3A (Table 41) report on KD values in the nM range. All mutants possessed modest 1.5 – 4-fold increase in KD to FcγR1A as compared to Tmab WT, with mutants Q178>N and T198>N presenting as the most superior. Despite adequate curve fitting for several of the mutants, the kinetic values calculated to FcγR2B are not reliable and cannot be interpreted. For instance, mutants A121>N and Q178>N required longer dissociation times to achieve steady state, they had high relative binding to the FcγR2B captured surface (Rmax) values and curvature that matched similarly to Q160>N and L115>N (Appendix 6.2). This supports the sensorgram interpretation of similarity in association binding with a stronger dissociation. Sensograms also indicated that

L182>N showed a drastic disruption to the binding kinetics with FcγR2B. Of the SPR screen to surface captured FcγR3A, mutant L177>K showed similarity in KD to Tmab WT, whereas all other 129 glycosylation mutants had a modest decrease (3.3 – 9.5-fold). However, inspection of the curve fittings suggested stronger Ka values than calculated for L177>N, L115>N and especially L182>N and T198>N due to their Rmax of 15.88, 25.15, 82.44 and 92.56 respectively. L177>N specifically had a slightly higher Rmax than L177>K (13.08), despite the 10-fold reduction in Ka calculated, therefore kinetic values calculated to FcγR3A are not entirely reliable for interpretation.

Overall, SPR analysis demonstrated that relative binding to HER2 slightly decreased and FcγR1A slightly increased for the entire Tmab glycosylation library, aside from mutant T117>N. For

FcγR2B and FcγR3A, results were varied and kinetic data was not robust for reliable interpretation. However, sensorgram interpretation showed that binding was similarly retained amongst the Tmab glycosylation library, aside from L182>N displaying drastic disruption in binding to FcγR2B and markedly enhanced affinity to FcγR3A, along with T198>N.

Table 40 Binding kinetics of the Tmab glycosylation mutant library to surface captured HER2 and * FcγR1A (± SD) as measured through SPR, 1 value only.

HER2 FcγR1A

Ka Kd KD Ka Kd KD (x 106) M-1.s-1 (x 10-5) s-1 pM (x 105) M-1.s-1 (x 10-4) s-1 nM

PBS-Herc 3.85 ± 0.09 1.10 ± 0.60 2.85 ± 1.49 2.75 ± 0.06 11.31 ± 0.10 4.11 ± 0.05

Tmab WT 2.50 ± 0.03 3.03 ± 0.72 12.11 ± 2.72 2.11 ± 0.03 10.64 ± 0.56 5.05 ± 0.33

L177 > K 1.00 ± 0.25 7.28 ± 1.42 72.97 ± 3.64 2.73 ± 0.12 7.94 ± 0.36 2.91 ± 0.01

L177 > N 1.84 ± 0.13 6.32 ± 0.08 34.28 ± 2.01 5.01 ± 0.18 15.04 ± 1.27 3.00 ± 0.15

Q160 > N 0.83 ± 0.02 2.78 ± 1.26 33.53 ± 15.86 2.55 ± 0.02 6.62 ± 0.11 2.59 ± 0.02

L115 > N 3.82 ± 3.80 18.16 ± 9.14 47.56 ± 25.89 1.79 ± 0.20 4.20 ± 0.06 2.34 ± 0.26

A121 > N 4.40 ± 1.71 23.84 ± 4.35 54.24 ± 10.24 3.83 ± 0.36 8.77 ± 1.14 2.29 ± 0.08

Q178 > N 1.35 ± 0.27 4.29 ± 0.67 31.88 ± 2.11 7.38 ± 0.11 9.23 ± 0.30 1.25 ± 0.02

L182 > N 1.1 * 3.2 * 29.1 * 3.67 ± 0.55 12.37 ± 2.64 3.37 ± 0.22

T198 > N 6.11 ± 1.17 50.97 ± 3.23 83.44 ± 77.77 4.71 ± 0.02 8.09 ± 0.10 1.72 ± 0.03

130

RU 30 A 25

Response 5

-5 -500 0 500 1000 1500 2000 2500 3000 3500 4000 Time s

RU 35 B 30 25 20 15 10

Response 5 0 -5 -500 0 500 1000 1500 2000 2500 3000 Time s

RU 16 14 C 12 10 8 6 4 Response 2 0 -2 -500 0 500 1000 1500 2000 2500 3000 Time s

RU 25 D 20

Response 0

-5 -500 0 500 1000 1500 2000 2500 3000 Time s

Figure 37 RepresentativeA analysed) H E R 2 BSPR) F c sensorgramsR1A C ) F c  R 2 Bof D single) F c  R 3 cycleA kinetic assays of Tmab WT to surface captured HER2 (A), FcγR1A (B), FcγR2B (C) and FcγR3A (D). Sensorgram data (red/green), curve fitting (black).

131

Table 41 Binding kinetics of the Tmab glycosylation mutant library to surface captured FcγR2B and FcγR3A (± SD) as measured through SPR.

FcγR2B FcγR3A

Ka Kd KD Ka Kd KD (x 104) M-1.s-1 (x 10-4) s-1 nM (x 104) M-1.s-1 (x 10-4) s-1 nM

PBS-Herc 3.05 ± 0.04 1.99 ± 0.13 6.55 ± 0.35 10.51 ± 0.29 1.90 ± 0.54 1.81 ± 0.47

Tmab WT 5.76 ± 1.01 4.75 ± 0.41 8.25 ± 0.69 11.41 ± 2.06 1.29 ± 0.64 1.13 ± 0.32

L177 > K 7.61 ± 1.66 5.05 ± 0.91 6.63 ± 2.19 23.94 ± 2.69 2.39 ± 2.20 1.00 ± 0.25

L177 > N 2.95 ± 1.82 6.88 ± 4.26 23.28 ± 44.62 2.83 ± 0.22 3.02 ± 0.05 10.67 ± 1.03

Q160 > N 4.01 ± 0.32 1.85 ± 0.27 4.62 ± 0.32 21.17 ± 4.95 8.01 ± 2.95 3.78 ± 0.46

L115 > N 0.65 ± 0.26 0.78 ± 0.44 11.92 ± 3.57 3.35 ± 0.85 2.74 ± 0.23 8.17 ± 1.60

A121 > N 2.78 ± 1.99 1.95 ± 0.43 7.01 ± 3.84

Q178 > N 1.95 ± 0.20 0.98 ± 0.20 5.04 ± 1.47 ND ND ND L182 > N 3.12 ± 0.25 1.95 ± 0.20 6.24 ± 1.30

T198 > N 2.25 ± 0.14 0.86 ± 0.15 3.81 ± 0.41

4 Discussion This chapter reports on the rational design of strategically introduced N-linked glycans to Tmab as a model antibody, to enhance intrinsic stability and reduce aggregation propensity through the steric hindrance of the APR identified at L177 in the CH1. Mutagenesis and expression was successful for the entire library designed; residue substitution and glycan addition was confirmed for the entire library aside from Q160>N. Stability screening revealed L177>N and L182>N as promising candidates for more rigorous investigation, possessing enhanced intrinsic stability and retention of monomer as comparted to Tmab WT. Biological screening generally demonstrated retained biological functions of the entire library aside from T117>N. Binding to target HER2 was modestly compromised and binding to FcγR1A and FcγR3A was modestly enhanced for the entire library, which warrants more rigorous investigation for the potential of modulating a more pronounced ADCC effect to HER2 positive tumorous tissue.

In characterising the Tmab glycosylation mutant library, SDS-PAGE and LC-MS/MS analysis confirmed glycan addition for the entire library aside from Q160>N. The equivalent Q160>N

132 mutation had previously been produced in bevacizumab [168] reporting successful glycosylation addition through SDS-PAGE, which contradicts our findings in transferring this mutation to Tmab. Therefore the further stability and affinity data presented in this chapter from Tmab Q160>N is only in relation to the aa substitution, and not of an additional glycosylation site. This also highlights that N-glycan introducing mutations between antibodies of identical constant region sequences are not necessarily transferable.

Due to the varied yield amongst the Tmab glycosylation mutation library, sample limitations restricted the stability screen to single measurements; however, the results presented show promise for further investigations with L177>N and L182>N as lead candidates. Mutant T117>N was specifically the poorest yielding of the library; the quality of T117>N produced was likely the contributing factor to its poor performance through the stability screen and biological assays and warrants reproduction.

Of the stability screen, several mutants exhibited improved intrinsic stability as reflected through their melting curves (Figure 32), particularly mutants L177>N, L115>N, Q178>N, L182>N and T198>N. Several exhibited resistance to aggregation as reflected through their aggregation curves (Figure 33 and Figure 34), particularly mutants L177>N, A121>N, Q178>N, L182>N and T198>N. All mutations aside from A121>N and Q160>N (not tested) exhibited improved solubility as indicated from the precipitation of insoluble aggregates generated through the thermal ramp of the aggregation curves (Figure 33 and Figure 34). And finally mutants L177>N and L182>N exhibited superior retention of monomer content as reflected through the monomer loss study (Figure 35). Equivalent mutation of L177>K was previously reported to exhibit resistance to aggregation in bevacizumab [168] and promotes aggregation in rituximab [169]. Our findings demonstrate that in Tmab, L177>K modestly decreases Fab stability, promotes aggregation and monomer loss. L177>N and L182>N were elected as lead candidates from the stability screen as they possessed enhanced thermal stability and monomer retention.

The stability screen profiling the glycosylation mutant library does not correlate resistance to aggregation with the retention of monomer. The correlation between thermal intrinsic stability, depletion of monomer and resistance to aggregation appears different with each Tmab glycosylation mutant, which highlights that profiling the stability for any individual protein is complex and requires orthogonal screening methods for thorough characterisation [251, 252]. As a minimum, repeating stability analysis in triplicate and longer term monomer loss studies would give more insight and robust information to the stability profiling of the Tmab glycosylation mutant library.

133

The transfer of these mutations to another antibody for equivalent stability profiling will demonstrate proof-of-concept that indeed, the strategic addition of N-linked glycans is a robust platform for enhancing intrinsic stability and resistance to aggregation of antibodies.

In optimising mutation efficiency, despite the reduced amount of isolates yielded from the in-house JM109 competent cells, mutation efficiency was relatively high for those that were rescued. A first round of rescue with in-house JM109 competent E. coli was more cost-effective, in that the few isolates rescued have a high mutation efficiency and a reduced probability of capturing miscellaneous reaction construct / artefact than with the commercial competent E. coli. The 2-step PCR method was implemented to reduce the frequency of capturing primer ligation artefacts; however, mutations L177>K, Q160>N, L182>N and T198>N presented artefacts in the screening process demonstrating that primer design required refinement. The mutation strategy implemented for the Tmab glycosylation mutation library was the most successful of all strategies reported in this thesis, and all further mutational work was carried out with this method. In optimising production and purification, SDS-PAGE analysis revealed residual fragment impurities in many of the purified Tmab samples; predominantly at 35 and 70 kDa which match similarly to the expected impurities of leached Protein A (42.0 kDa) fragment from the affinity purification process [253] and host cell protein HER2 (72.4 kDa) captured from antibody during expression [254]. Although the proportion of impurities was residual, their presence was of concern for the stability studies as uncharacterised unfolding from the impurities may accelerate aggregation and monomer loss in the sample. Specifically, this is demonstrated with mutant T198>N that possessed the highest proportion of impurities (Figure 31) and appeared to have enhanced stability (Table 38), although detrimental monomer loss (Figure 35). This indicates a need for a polishing step such as ion exchange or size exclusion chromatography that will exclude these additional species from the purified antibodies. The presence of protein impurities leads to the overestimation of protein content in the purified antibody samples, which makes normalisation of samples for assays inadequate for characterisation. For the purposes of this study, the antibody yield was inadequate for optimising additional purification steps prior to stability and biological screening and remains to be addressed for additional investigation.

Further to screening for stability enhancement, the Tmab glycosylation mutation library binding kinetics to HER2 and a panel of FcγRs was screened to ensure that introduced mutations and glycosylation do not contribute to a loss of immunological functions. Tmabs KD to HER2 is reported in the subnanomolar range, spanning 19 pM – 1 nM [19, 205, 209], IgG1 is reported to have a KD to FcγR1A of 1 – 10 nM, FcγR2B of 100 nM and FcγR3A of 10 µM [16, 133]. As per the

134 affinity and binding kinetics results presented in Chapter 4, KD values of the Tmab glycosylation mutant library derived from the cell-based assay resided in the nM range, unlike the SPR values that determined KD in the subnanomolar range. The SPR values of Tmabs binding kinetics to HER2 are in agreement with previous literature, which indicates that the cell-binding assay requires optimisation to achieve results that are in agreement with SPR, as previously discussed in Chapter

4. The SPR values of Tmabs binding kinetics to HER2 and FcγR1A are in agreement with literature, unlike the values to FcγR2B and FcγR3A which are not robust for reliable interpretation. KD to

HER2 was modestly reduced and KD to FcγR1A was moderately enhanced with the entire Tmab glycosylation mutant library in comparison to Tmab WT. Varied and unreliable kinetic rate constants were derived for the Tmab library binding to FcγR2B and FcγR3A. However, sensorgram interpretation demonstrated binding similarity amongst the Tmab glycosylation library. Significant deviations of sensorgram similarity of mutants to Tmab WT include L182>N and T198>N. L182>N displayed drastic disruption in binding to FcγR2B, both L182>N and T198>N displayed markedly enhanced affinity to FcγR3A (Appendix 6.2).

Retention of the Tmab glycosylation mutation library’s affinity to HER2 is ideal; however, as comprehensively discussed in Chapter 1 and Chapter 4, manipulation of affinity to HER2 can become useful for different therapeutic platforms. Fragment Tmab and Tmab-drug conjugates favour enhanced affinity to promote efficient catabolism and delivery of cytotoxic drug to target cancer cells [23]. Tmab with enhanced ADCC functions, extended half-life, and Tmab decorated nanoparticles favour reduced affinity for better biodistribution and tumour retention leading to a more insistent ADCC effect. Also, the consequence of enhancing Tmab affinity to FcγR1A and

FcγR3A or reducing Tmab affinity to FcγR2B would be an enhanced ADCC effect [16, 133]. The entire Tmab glycosylation mutant library appears to possess marginally adjusted receptor binding kinetics for an enhanced ADCC effect, however this needs to be extensively investigated to reach a satisfactory conclusion. However, the biological assays demonstrated that the addition of an N- linked glycan to the Fab region of Tmab does not contribute to a significant disruption in binding of

Tmab to HER2 or FcγR1A. Mutant L182>N appears a lead candidate due to the binding disruption to FcγR2B and enhancement to FcγR3A, warranting further investigation for the hypothesised enhanced ADCC effect.

5 Concluding Remarks The findings of this study have shown that the strategic addition of N-linked glycans in the Fab region of Tmab can enhance intrinsic stability and resistance to aggregation through the steric hindrance of an aggregation prone region. Several of the glycosylation mutants displayed varying

135 stabilities and aggregation propensities, of which two were elected as lead candidates for further investigation. Addition of an N-linked glycan to the Tmab Fab was further demonstrated to retain antigen binding specificity and immunological functions. Enhanced ADCC effect, increase in solubility and longer half-life are further therapeutic benefits of glycan addition warranted for further investigation, alongside directed conjugations for alternative therapeutic formats. This study demonstrates the effectiveness of this approach in enhancing biotherapeutic stability for improved formulation and treatment options, warranting a proof-of concept study, as reported in Chapter 6.

136

Chapter 6 Enhancing the Stability of Adalimumab by Engineering Additional Glycosylation Motifs

Chapter 6 was prepared for publication in mAbs;

manuscript submitted 14/05/2019, pending peer review

Reslan, M., Sifniotis, V., Cruz, E., Sumer-Bayraktar, Z., Cordwell, S., Kayser, V. (2019). mAbs

137

Abstract Monoclonal antibodies (mAbs) are of high value in the diagnostic and treatment of many debilitating diseases such as cancers, auto-immune disorders and infections. Unfortunately, protein aggregation is one of the ongoing challenges, limiting the development and application of mAbs as therapeutic products by decreasing half-life, increasing immunogenicity and reducing activity. We engineered an aggregation-prone region of adalimumab, the top selling mAb product worldwide - with additional glycosylations to enhance its resistance to aggregation by steric hindrance as a next generation biologic. We found that the addition of N-glycans in the Fab domain significantly enhanced its conformational stability, with some variants increasing the melting temperature of the Fab domain by more than 6 °C. The mutations tested had minimal impact on antigen binding affinity, or affinity to Fcγ receptors responsible for effector function. Our findings highlight the significant utility of this rational engineering approach for enhancing the conformational stability of therapeutic mAbs and other next-generation antibody formats.

1 Introduction Monoclonal antibodies (mAbs) are amongst the top-selling therapeutic prescription products worldwide. In 2017, Humira® (adalimumab) alone, generated over 18 billion US dollars, maintaining its position as the top-selling drug worldwide [1, 255]. mAbs are revolutionary biopharmaceuticals used to treat cancers, auto-immune and inflammatory conditions, and even some types of asthma. Approximately 80 antibody products have been approved and marketed by the FDA and EMA with hundreds more in early and advanced clinical trials [9]. Advances in antibody engineering technology have led to the development of a multitude of novel, next- generation antibody-derived products often termed biobetters, which include bispecifics, nanobodies, antibody-drug conjugates, and nanoparticles functionalised with antibodies [9, 256].

Despite the four-decade-long progress and ever-growing interest in antibody research, one of the major challenges in the development of antibody therapeutics remains unsolved: protein aggregation [250, 256-258]. Protein aggregation is one of the most common form of antibody degradation, having significant implications for a therapeutic product under development; aggregation can result in reduced therapeutic efficacy, and may trigger severe immunogenic side effects when administered [259, 260]. Formulations of biologics including mAbs and other protein products are stabilized with the use of additives such as salts, sugars, amino acids and surfactants [261-263]. Some novel formulation additives have also been investigated to inhibit protein aggregation, including deuterated solvents [264], hydrogels [265, 266], peptide dendrons [267] and ionic liquids [268]. However, the type and amount of additives allowed in any given formulation is

138 limited, especially in parenteral formulations – which are often used to administer mAbs. Thus, other approaches to enhance the aggregation-resistance of mAbs are needed.

Another approach to significantly enhance the stability of a mAb is to modify the antibody structure itself. A single amino acid mutation in an aggregation-prone region (APR) can be sufficient to significantly improve the stability of an antibody against aggregation and/or increase its melting temperature. Predictions of aggregation-prone regions are often aided by computational tools – these include TANGO, AGGRESCAN3D, SAP [269-271], docking studies and/or molecular dynamics simulations [272]. For instance, the spatial aggregation propensity (SAP) tool has previously been used to guide mutations of aggregation-prone motifs of rituximab. These motifs predicted by SAP were ‘neutralised’ by single-point mutations, and multiple mutations were then combined to increase the overall antibody stability [169].

A variation of this approach, involves engineering an additional N-glycosylation site in a spatially proximate region to an APR, to sterically hinder aggregation in that APR [168, 273]. Naturally occurring N-linked glycans in the Fc domain are known to significantly influence antibody stability and effector function and to increase half-life in patient serum. The introduction of extra N-linked glycosylation sites in the Fab domain may not only improve resistance to aggregation, but also enhance solubility and conformational stability of the antibody, offering additional benefits to a single-point mutation in a hydrophobic region.

Courtois, Agrawal, Lauer and Trout [168] recently tested a series of Fab-glycosylated bevacizumab variants and reported significant improvements in overall stability. Nakamura, Oda-Ueda, Ueda and Ohkuri [273] also tested a glycan-introducing mutation in the Fab of adalimumab expressed in Pichia pastoris and found that it inhibited protease digestion and precipitation under pH stress. However, its impact on soluble aggregation of the full antibody is not clear.

In this study, we report on and discuss the stability of Fab-glycosylated adalimumab variants that have spatial proximity to a previously determined APR in the IgG1 CH1 region of rituximab [169] and bevacizumab [168]. Specifically, the amino acid of the APR transposed to the adalimumab IgG1 peptide sequence is leucine 178. The substitution of this hydrophobic amino acid to lysine in bevacizumab [168], or to serine in rituximab [169] in the equivalent position, had experimentally demonstrated increased stability. This amino acid can also be substituted with asparagine (e.g. L178N) to introduce an N-linked glycosylation site. For this study, surface exposed amino acids were identified that can be substituted to asparagine to introduce N-linked glycosylation sites i.e. Asn-X-Ser/Thr, where X is any amino acid except proline (Figure 38). All substitutions except

139

T199N, were previously identified to have 10 Angstrom spatial proximity to the APR of interest [168, 169]. T199N was specifically chosen as it had been reported to dramatically increase the solubility of a poorly soluble antibody [248], however it does not have spatial proximity to the APR. This particular substitution also has the potential to sterically hinder reversible self- association [166] which warranted further investigation through this study.

The substitutions that were identified and pursued for this study are listed in Table 42. In addition to studying the effect of these Fab N-linked glycosylations on the aggregation propensity of adalimumab, we also investigated the conformational stability of the variants, and characterised their binding affinity to TNF-, and Fcγ receptors (FcγRs) responsible for effector function, to understand the broader impact of these additional glycans.

Table 42 Surface exposed amino acid (AA) residues identified in adalimumab Fab region for substitution

AA Position Region Mutation Proposed Function Residue

Remove APR by introducing polar amino acid K [168] L 178 CH1 Remove APR and introduce glycosylation site [168]

Q 160 Ckappa Sterically hinder APR by introducing L 116 glycosylation site [168] Cjoin T 118

N Improve solubility by introducing glycosylation A 122 site [248]

Q 179 Sterically hinder APR by introducing CH1 glycosylation site [168, 248] L 183

Improve solubility and sterically hinder self- T 199 association by introducing glycosylation site [166, 248]

140

T199

L178 L116

T118

A122 Q179 L183 Q160

Figure 38 Crystal structure 3WD5 of adalimumab Fab fragment complexed to partial TNF- fragment. Adalimumab VH/CH1 (green), VL/CL (aqua), TNF- fragment (orange), APR site (purple) and surface exposed amino acids identified for substitution to an N-linked glycosylation site (pink).

2 Materials and Methods 2.1 Materials Humira® subcutaneous injections (40 mg/0.8 mL) were purchased from Symbion (VIC, Australia) and used as a reference. Adalimumab (AdmAb) wild-type (WT) and mutant variants were expressed transiently in FreeStyle™ human embryonic kidney cells adapted for serum-free media (HEK293F). The expression vector gWiz (blank) was obtained from Genlantis (CA, USA) and vector pVITRO1_Trastuzumab_IgG1/κ was a gift from Andrew Beavil (Addgene plasmid # 61883). Primers and synthetic AdmAb variable region gene constructs were purchased from Integrated DNA Technologies (Singapore). In-Fusion cloning kit was obtained from Scientifix (Australia). FreeStyle™ 293 Expression Medium, OptiPRO™ Serum Free Medium (SFM) and HIS-tagged FcRγ1A receptor were purchased from Life Technologies Pty Ltd (VIC, Australia). Polyethylenimine (PEI), Linear, MW 25 kDa was purchased from BioScientific (Australia). 0.22 µm PVDF filters), 50 kDa centrifugal filters, trypsin and chymotrypsin were purchased from Sigma

141

Aldrich (Australia). PCR kits and all other enzymes were purchased from GeneSearch (Australia). Phosphate Buffered Saline (PBS) was purchased from Astral Scientific Pty Ltd (NSW, Australia). HiTrap MabSelect SuRe column, Biacore CM5 and ProA chips, amine coupling kit and HIS capture kit were purchased from GE Healthcare (Australia). All other receptors, chemicals, consumables and reagents were obtained from Sigma Aldrich (Australia).

2.2 Cloning and Mutation of AdmAb WT and Variants The blank gWiz expression vector was used to construct separate AdmAb IgG1 and kappa chain expression vectors. The secretion signalling genes, IgG1 and kappa chain genes were sourced from genes contained in pVITRO1_trastuzumab_IgG1/κ vector; they were cloned separately into the empty gWiz vector through conventional restriction cloning techniques. The variable regions of AdmAb genes were designed and synthetically produced; they were exchanged with the trastuzumab variable regions in the gWiz vectors through In-Fusion cloning to produce gWiz AdmAb IgG1 and gWiz AdmAb kappa vectors.

Mutations were then introduced to the gWiz AdmAb IgG1 and kappa vectors through high fidelity inverse PCR to produce a library of AdmAb mutant expression vectors. AdmAb WT and mutant variants were confirmed through sequencing and expressed through transfection of vector pair combinations.

2.3 Expression and Purification of AdmAb WT and Variants AdmAb WT and mutants were transiently expressed in suspension HEK293F culture. HEK293F culture was maintained and transfected in Freestyle 293F media, 120 rpm, 37 °C, 5% CO2, humidified incubation conditions. 24 hr prior to transfection, HEK293F culture was seeded to 6 - 7 x 105 cells/mL in a 60 mL culture and the culture was adjusted to 1 x 106 cells/mL on the day of transfection.

The complexing DNA for transfection was prepared as 1 µg total DNA per 1 x 106 cells. The gWiz AdmAb IgG1 and kappa vectors were combined in a w/w ratio of 1:1, and the complexing w/w ratio of PEI to total DNA was 2:1. The vector DNA and PEI were separately diluted in OptiPRO SFM. The dilutions were combined and left to complex for 15 min before adding dropwise to the transfection culture.

After 24 hr incubation of the transfection culture, the culture was scaled up to 120 mL with Freestyle 293F media and supplemented to a total concentration of 0.5% Tryptone w/v. After a further 24 hr of incubation, the culture was scaled up to 240 mL with Freestyle 293F media and supplemented to a total concentration of 0.5% Tryptone w/v. The transfection culture remained in

142 incubation for a total of 8 days before the media containing secreted antibody was harvested by centrifugation, and the culture supernatant was clarified by filtration.

AdmAb WT and variants were purified from the culture supernatants using protein A affinity with a 1mL HiTrap MabSelect SuRe column. PBS pH 7.4 was used as the binding and wash buffer, and glycine HCL pH 3 was used to elute bound antibody. Eluted antibody samples were neutralised in 1 M Tris-HCl pH 8 and concentrated to <1 mL using 50 kDa centrifugal filters. To purify the samples from any aggregates and obtain pure monomer samples, concentrated samples were fractioned using size-exclusion chromatography (as per 2.6) and the monomer fractions were collected and concentrated in 150 mM potassium phosphate buffer, pH 6.5, for analysis.

2.4 LC-MS/MS Analysis of AdmAb Variants Successful mutation and introduction of glycosylation motifs of the AdmAb variants was confirmed through site-specific liquid chromatography tandem mass spectrometry (LC-MS/MS). Briefly, AdmAb variants were separated on reducing SDS-PAGE and digested with trypsin and chymotrypsin. A fraction of the digestions was further treated with PNGase F for glycan release. The digestion products were analysed by LC-MS/MS using higher-energy collisional dissociation (HCD), collision induced dissociation (CID) and electron transfer dissociation (ETD) methods.

The AdmAb variants (10 μg) were run on a reducing SDS-PAGE, in-gel trypsin digestion was performed on the heavy chain band (50 - 60 kDa) of variants L116N, T118N, A122N, L178K, L178N, Q179N, L183N, T199N and the light chain band (25 - 30 kDa) of Q160N. The bands were excised, diced and washed with 50% (v/v) acetonitrile (MeCN) in 100 mM NH4HCO3. The gel pieces were then reduced with 10 mM DTT at 56 °C for 45 min, followed by alkylation by addition of 55 mM iodoacetamide (IAA) in 100 mM NH4HCO3, and incubation at 25 °C for 30 min. The samples were subsequently washed with 50% (v/v) MeCN in 100 mM NH4HCO3.

The alkylated polypeptides were incubated with trypsin (T6567) (1 μg trypsin per 50 μg antibody) in 50 mM NH4HCO3, pH 6.8 at 37 °C overnight. The resulting tryptic peptides were further digested with chymotrypsin (11418467001) in 100 mM Tris-HCl, 10 mM CaCl2, pH 7.8 at room temperature overnight.

Glycan release was performed to confirm amino acid substitution; a fraction of the chymotryptic peptides (excluding L178K) was incubated with recombinant PNGase F (P0708S) (400 units per 1 μg of peptide) in 50 mM sodium phosphate buffer pH 7.5 at 37 °C overnight. The chymotryptic peptides and the de-glycosylated fractions were desalted with C18 ZipTips (Millipore), eluted in

143

80% (v/v) MeCN, dried by vacuum centrifugation then redissolved in 0.1% formic acid (FA) for LC-MS/MS analysis.

Glycopeptides and de-glycosylated peptides were separated on an in-house packed 20 cm × 75 μm Reprosil-Pur C18AQ (3 μm, 120 A; Dr. Maisch GmbH) column using a Dionex 3500RS HPLC over a 90-min 0-40% solvent B gradient at a flow rate of 300 nL/min, at 60°C (solvent A was 0.1% (v/v) formic acid; solvent B was 80% (v/v) acetonitrile and 0.1% formic acid). MS analysis was performed on an Orbitrap Fusion MS (Thermo Scientific) in positive mode. Instrument parameters were set up as follows: source voltage = 2.3 kV, S-lens RF level = 68%, and capillary temperature = 275 °C. The initial MS scan was acquired in the Orbitrap mass analyzer (350–2000 m/z; MS AGC = 6 × 105) with a resolution of 60,000 at 400 m/z. MS1 was followed by data-dependent HCD using top speed, where as many dependant scans as possible are acquired in a specified time. HCD fragmentation was followed by re-isolation of the top two most intense precursors, and fragmentation with CID and ETD MS/MS. HCD parameters were as follows: activation time = 0.1 ms, resolution = 30,000, maximum injection time = 200 ms, dynamic exclusion = enabled with repeat count 1, normalized energy = 40, exclusion duration = 20 s, default charge state = 2, and MSn AGC 2.0e5. CID parameters were as follows: 30,000 resolution in orbitrap, activation time = 10 ms, max injection time = 300 ms, dynamic exclusion = enabled with repeat count 1, normalized energy = 35, exclusion duration = 20 s, default charge state = 2, and MSn AGC 5.0e4. ETD parameters were as follows: 30,000 resolution in orbitrap, 5.0e4 AGC, ETD reaction time = 50 ms, Max injection time = 300 ms.

The peptide spectrum matches (PSM) were generated through a search against the corresponding fasta file (wild-type or variant Adalimumab amino acid sequence) utilising the search engine Byonic (Protein Metrics Inc.). Annotation of the HCD and ETD spectra were automated through Byonic and the PSM assignments were validated by manually annotating the aforementioned spectra and the complementary CID spectrum. Further confirmation of the identity assignment was achieved through analysis of the de-glycosylated chymotryptic peptides utilising the same LC- MS/MS approach as for the glycosylated fraction.

2.5 Accelerated Stability at Elevated Temperatures We performed accelerated studies at an elevated temperature as previously described [250]. Briefly, 20 µL of each AdmAb sample (1 mg/mL) was pipetted into a 0.2 mL PCR vial and incubated at 65 °C for 1, 2 or 3 hr in a Thermal Cycler (Applied Biosystems, CA, USA) to induce aggregation. Each experiment was repeated three times. A thermal cycler was used to minimize sample amount and prevent evaporation of the sample and consequent changes in antibody concentration during

144 incubation (lid/cover of thermal cycler is heated above incubation temperature). Following incubation, incubated samples were removed and immediately stored on ice to halt aggregation and then characterized by size-exclusion high-performance liquid chromatography (SE-HPLC).

2.6 SE-HPLC Analysis of Monomer Loss Size-exclusion high-performance liquid chromatography (SE-HPLC) was used to quantify AdmAb monomer loss following incubation at elevated temperature as previously described [250]. Analysis was performed using an Agilent 1200 Liquid Chromatography system (Agilent Technologies, California, USA) with a Zorbax GF-250 column coupled to a guard column at 22 °C, using a 150 mM potassium phosphate pH 6.5 mobile phase, and a flow rate of 0.5 mL/min. 5 μL of each incubated sample was then immediately injected and each injection was repeated three times. Monomer peaks were detected using an in-line UV signal detector set at 280 nm. The area under the curve (AUC) of the monomer peak was averaged over the three runs and the mean relative monomer % was calculated for each sample, by setting the monomer AUC of the non-incubated samples as 100% and calculating the change in monomer AUC accordingly. The standard deviations (SD) were plotted as error bars in the figures.

2.7 AdmAb Melting Temperature (Tm) and Onset Temperature of Aggregation (Tagg) ® The Tm and Tagg of Humira , AdmAb WT, and variants were simultaneously measured using intrinsic tryptophan fluorescence and static light scattering (SLS), respectively, on the UNcle system (Unchained Labs, CA, USA). A linear temperature ramp from 15 to 95 °C at a 1 °C/min scan rate was performed whilst measuring tryptophan fluorescence and SLS simultaneously through laser excitation at 266 nm. 8.5 µL of each AdmAb sample (1 mg/mL) was pipetted undiluted into the UNcle UNI (a sample holding unit containing 16 quartz cells) in triplicates. A holding time was not used to maximise the frequency of measurements. Tm and Tagg were determined by the UNcle

Analysis software. The barycentric mean (BCM) was used to plot the Tm curves, which is defined by the following equation:

∑휆 휆퐼(휆) 휆퐵퐶푀 = ∑휆 퐼(휆)

The equation is defined over the range 300-450 nm; each wavelength value (λ) is multiplied by the tryptophan fluorescence intensity (I) at that wavelength, and the sum of that value for all wavelengths between 300-450 is divided by the sum of the intensities at those wavelengths. This results in an ‘averaged’ peak wavelength (λBCM) for a given spectrum.

145

2.8 Binding Kinetics (KD) to TNF-α and Fcγ receptors (FcγRs) ® The binding kinetics of Humira , AdmAb WT and variants were assessed to TNF- and FcγR1A, 2B and 3A through surface plasmon resonance (SPR) to confirm that the mutations did not contribute to a loss of biological function. Briefly, a Biacore T200 system (GE Healthcare, Australia) was used, running single cycle kinetic assays. Monomeric AdmAb samples were captured on a ProA chip and increasing dilutions of TNF- were injected, with a 60 min dissociation. HIS-tagged FcγR1A, 2B and 3A were captured on an anti-HIS chip and increasing dilutions of monomeric AdmAb samples were injected, with a 30 min dissociation. The sensograms were analysed to determine association (Ka), dissociation (Kd) and equilibrium binding constants

(KD). All runs were performed at 25 °C with running buffer HBS-T (10 mM HEPES, 150 mM NaCl, 0.05% v/v Tween 20, pH adjusted to 7.4).

Soluble homotrimeric TNF- (H8916) was assayed against monomeric AdmAb samples on a ProA chip under continuous flow rate of 20 µL/min. Monomeric AdmAb samples were prepared to 10 nM and captured on the sensor chip surface with a contact time of 10 sec. TNF- was injected in 2- fold increasing concentrations (2.5 – 40 nM) with a contact time of 120 sec at each concentration, then a dissociation time of 60 min. The chip surface was regenerated with 50 mM NaOH for 30 sec at a flow rate of 30 µL/min.

Monomeric AdmAb samples were assayed against HIS tagged FcRγ receptors. Anti-HIS antibody was covalently attached to a CM5 biosensor chip using an amine coupling kit and HIS capture kit.

HIS tagged FcγR1A (10256H08H), FcγR2B (SRP6396) and FcγR3A (SRP6436) were prepared to 4 nM and captured to the sensor chip surface with a contact time of 300 sec at 5 µL/min flow rate.

AdmAb samples were injected on FcγR1A coated surface in 2-fold increasing concentrations (2.5 – 40 nM) at a flow rate of 20 µL/min with a contact time of 120 sec at each concentration, followed by 30-min dissociation time. AdmAb samples were injected on FcγR2B and FcγR3A-coated surfaces in 2-fold increasing concentrations (25 – 400 nM) at a flow rate of 10 µL/min with a contact time of 120 sec at each concentration, followed by 1800 sec dissociation time. The chip surface was regenerated with 10 mM Glycine HCl, pH 1.5 for 60 sec at a flow rate of 30 µL/min.

The sensograms were analysed globally using a Langmuir 1:1 binding model to determine Ka, Kd and KD. All samples were conducted in duplicate to obtain average values and standard deviation. Kinetic rate and binding constants determined from sensograms were accepted if the goodness of fit 2 value (χ ) was within 5% of the maximum response level (Rmax) of the sensograms and KD was calculated as Kd/Ka.

146

3 Results 3.1 Confirmation of Mutation and Glycan Attachment through LC-MS/MS

Detailed analysis of LC-MSMS data confirmed that all AdmAb variants had successful amino acid substitution. Incorporation of HCD, ETD and CID fragmentation techniques further provided compositional glycan information of the new N-glycosylation sites. Except for Q160N mutant, all other AdmAb glycosylation-site variants were found to contain an N-glycan (data not shown).

Here, we show the characterization of glycan site-engineered AdmAb variant T199N as a representation of the analysis performed for all AdmAb variants. HCD spectrum of SLSSVVTVPSSSLGNQTY peptide observed at m/z 909.65(4+) showed b- and y-ion series confirming amino acid sequence and the asparagine substitution was shown specifically by y4-y11 fragment ions (Figure 39A). HCD of T199N mutant also contained diagnostic HexNAc (m/z 204.09) and HexNAcHex (m/z 366.14) oxonium ions, strongly indicating glycosylation on the mutant peptide (Figure 39A). ETD further confirmed the structure of T199N glycopeptide by multiple fragment ions corresponding to dissociated peptide backbone with intact glycan (Figure 39B, c-15, c-16, c-17 and z-16 ions). Theoretical monosaccharide composition of the N-glycan was derived using Byonic (Protein Metrics) search results and GlycoMod (https://web.expasy.org). CID data in Figure 39C shows sequential loss of monosaccharides from glycopeptide fragments of T199N mutant and contains complementary oxonium ions m/z 366.14, m/z 528.19, and m/z 569.22 of the observed glycan in the low mass region. Extensive glycosidic bond cleavages (Figure 39C, y-ions, m/z 1616.2(2+), m/z 1535.19(2+), m/z 1461.67(2+)) observed in CID confirmed that SLSSVVTVPSSSLGNQTY peptide is occupied by a complex type N-glycan with

HexNAc5Hex4Fuc1 composition. CID spectra of deglycosylated T199N peptide m/z 609.64(3+) showed enhanced peptide backbone fragmentation and provided additional information for the expected mass shift from threonine to asparagine (y4-y18 ions and b17 ion, Figure 40). Same ion series also displayed deamidation (+1Da) of the new asparagine residue which is a characteristic result of PNGaseF treatment of a formerly N-glycosylated peptide.

147

Figure 39 (A) HCD, (B) ETD and (C) CID spectra of chymotryptic T199N glycopeptide observed at m/z 909.65 (4+). Peptide sequence SLSSVVTVPSSSLGNQTY and attached complex type N-glycan HexNAc5Hex4Fuc1 are shown. Blue square; N-acetylglucosamine, red triangle; fucose, green circle; mannose, yellow circle; galactose, Hex; hexose, HexNAc; N-acetylhexosamine, M; precursor ion.

Figure 40 CID spectrum of de-glycosylated chymotryptic T199N mutant peptide SLSSVVTVPSSSLGNQTY observed at m/z 609.64(3+). b- and y- ions show CID fragments of the peptide.

148

3.2 Monomer Loss following Accelerated Studies at Elevated Temperature Incubation at 65 °C revealed modest differences in aggregation propensity between the reference Humira®, WT and single-mutant variants (Figure 41). Variants L116N and T118N were completely degraded after one hr of incubation and were therefore not included in the figure (Figure 41). Variant Q160N had very similar monomer loss data at all incubation temperatures to the W T, while L183N had a noticeably higher monomer loss. Variants A122N, Q179N and T199N retained higher monomer content than the control at most incubation conditions, with A122N having the least monomer loss (Figure 41). While Q179N had comparable monomer loss after one hr, monomer loss was lower at two and three hr, when compared to the WT. Interestingly, variants L178N and L178K had comparable monomer loss profiles which were not significantly different to the WT (Figure 41). Humira® reference had slightly lower monomer loss than the WT.

Figure 41 Monomer loss of Humira®, AdmAb wild-type (WT) and variants following incubation at 65 °C for one, two and three hours. Error bars represent the SD.

3.3 Tm and Tagg ® The WT AdmAb had a very similar Tm1 and Tm2 to the Humira reference product; a slightly lower

Tm1 was measured at 69.6 °C (WT) compared with 70.8 °C (Table 43). Aside for Tm2, all SD observed between the AdmAb WT and Humira were relatively low and comparable, demonstrating insignificance of the sample variation and robustness of derived values. The SD observed for Tm2 for Admab WT (1.4 °C) versus Humira (0.8 °C) showed more substantial sample variation,

149

® however; the derived mean Tm2 was comparable at 82 °C (WT) compared with 82.5 °C (Humira ).

Variant Q160N had a similar melting profile to the WT, with an identical Tm1 and comparable Tm2 of 69.6 and 82 °C, respectively. Variants T118N and L116N had a significantly lower Tm1 at 59.7 and 63.6 °C, respectively; the Tm2 was also lower at 68 and 78.6 °C, respectively. Interestingly, a third Tm was recorded for variants T118N and L116N at 87.8 and 88.3 °C, respectively. Variants

L178K, A122N and Q179N had similar Tms and were comparable to the WT. Variants T199N and

L178N had similar Tm1 values to the WT (70.6 and 69.9 °C, respectively), but significantly higher

Tm2 values (88.2 and 89.3 °C, respectively). Variant L183N had a similar Tm1 to the WT, but had a consistent minor unfolding event at ~65 °C (not shown); Tm2 was higher than the WT (88.3 °C,

Table 2). The Tagg of the WT and variants ranged from 70.6 to 72.5 °C, except for variants T118N and L116N, which had significantly lower Tagg values (61.7 and 64.7 °C, respectively; Table 43).

Most variants had a slightly higher Tagg than the WT (70.7 °C) by approximately 1 – 2 °C, while the

Tagg of Q160N and L178K were unchanged. Compared with L178K, the Tagg of glycosylated variant L178N was 1.7 °C higher. Results presented in Table 43 are derived from analysed sample melting and aggregation curves as presented in Appendix 7.6.

Table 43 The melting (Tm) and onset of aggregation temperatures (Tagg) of the AdmAb variants compared with the wild-type (WT) and reference Humira® product (± SD); ^see discussion

AdmAb Tm1 (°C) Tm2 (°C) Tm3 (°C) Tagg @266 nm

Humira® 70.8 ± 0.2 82.5 ± 0.8 - 71.3 ± 0.4

WT 69.6 ± 0.5 82 ± 1.4 - 70.7 ± 0.1

Q160N 69.6 ± 0.3 81.1 ± 1.4 - 71 ± 0.4

T199N 70.6 ± 0.2 88.2 ± 2.2 - 71.7 ± 0.1

T118N 59.7 ± 0.5 68 ± 0.7 87.8 ± 2.5 61.7 ± 0.6

L178K 68.4 ± 0.6 83.5 ± 0.5 - 70.6 ± 0.3

L178N 69.9 ± 0.5 89.3 ± 0.2 - 72.3 ± 0.3

A122N 69.7 ± 0.5 83.4 ± 0.7 - 71.8 ± 0.4

Q179N 69.4 ± 0.2 82.7 ± 0.9 - 71.7 ± 0.5

L116N 63.6 ± 0.5 78.6 ± 0.5 88.3 ± 1.2 64.7 ± 0.3

L183N 70^ ± 0.5 88.3 ± 1.1 - 72.5 ± 0.7

150

3.4 Binding Kinetics of AdmAb to TNF- and FcγRs The binding kinetics of reference Humira®, monomeric AdmAb WT and variants to TNF- and

FcγR1A, 2B and 3A were assessed using SPR single cycle kinetic analysis. Samples were run in duplicate and sensograms were fitted to a Langmuir 1:1 binding model to obtain Ka, Kd and KD values for comparison as per previous studies [228, 231]. It was determined that in general, the majority of AdmAb variants did not experience a significant loss of biological activity.

The KD range determined for monomeric AdmAb variants binding to soluble homotrimeric TNF- ® spanned 36.69 – 110.27 pM, while Humira had a mean KD of 30.77 pM (Table 44). Monomeric

AdmAb WT had a KD of 66.43 ± 2.37 pM which lies within the range, with no significant outlier

(Table 44). Of all variants tested, T199N had the strongest KD to TNF- (36.69 pM) - approximately 1.8-fold stronger than the WT - while Q179N had the weakest KD (110.27 pM).

Table 44 Kinetics and binding affinity of reference (Humira®) and monomeric AdmAb mutants to soluble homotrimeric TNF- (± SD).

TNF- AdmAb Ka Kd KD (x 105) M-1.s-1 (x 10-5) s-1 (pM)

Humira® 11.7 ± 0.01 3.59 ± 0.80 30.77 ± 6.83

WT 6.08 ± 0.07 4.04 ± 0.10 66.43 ± 2.37

L178K 4.84 ± 0.01 4.61 ± 0.64 95.33 ± 13.3

L178N 5.33 ± 0.03 5.42 ± 0.67 101.7 ± 12.1

Q160N 6.2 ± 0.07 2.86 ± 0.05 46.05 ± 1.31

L116N 5.19 ± 0.09 2.84 ± 0.1 54.77 ± 2.87

T118N 5.04 ± 0.02 3.59 ± 0.21 71.19 ± 4.50

A122N 6.47 ± 0.03 2.87 ± 0.05 44.32 ± 0.57

Q179N 5.02 ± 0.40 5.53 ± 0.1 110.3 ± 6.56

L183N 5.58 ± 0.14 3.05 ± 0.03 55.37 ± 0.86

T199N 7.15 ± 0.69 2.62 ± 0.01 36.69 ± 3.54

151

The KD range determined for FcγR1A binding to monomeric AdmAb variants spanned 2.1 – 6.81 ® nM, while Humira had a mean KD of 1.84 nM (Table 45). Monomeric AdmAb WT had a KD of 4.4 ± 0.01 nM which lies within the range with no significant outlier (Table 45). Of all variants tested, the strongest KD was recorded for mutant T118N (2.1 nM), while the weakest KD was recorded for L178N (6.81 nM).

Due to sample limitations, monomeric AdmAb Q160N was not tested for the cycle assays against

FcγR2B (Table 45). The KD range determined for FcγR2B binding to monomeric AdmAb variants ® spanned 1.81 – 9.61 nM, while Humira had a mean KD of 2.55 nM. Monomeric AdmAb WT had a

KD of 4.47 ± 0.1 nM which lies within the range with no significant outlier (Table 45). T199N had the strongest KD to FcγR2B (1.81 nM), while T118N had the weakest KD (9.61 nM). However, due to the large variation in KD values for the WT and several mutants, most differences were not deemed significant.

Table 45 Binding affinity of reference Humira® and monomeric AdmAb mutants to FcγR1A, FcγR2B and FcγR3A (± SD), no data (ND).

KD (nM) AdmAb FcγR1A FcγR2B FcγR3A

Humira® 1.84 ± 0.18 2.55 ± 1.67 1.84 ± 0.68

WT* 4.4 ± 0.01 4.47 ± 1.00 5.96 ± 1.62

L178K 4.14 ± 0.06 9.16 ± 6.86 2.37 ± 1.19

L178N 6.81 ± 0.45 3.67 ± 0.93 2.15 ± 0.73

Q160N 5.3 ± 0.01 ND 16.2 ± 3.05

L116N 3.87 ± 0.14 6.15 ± 1.10 6.71 ± 1.91

T118N 2.1 ± 0.01 9.61 ± 1.38 13.3 ± 0.45

A122N 5.58 ± 0.18 7.53 ± 2.74 18.01 ± 7.4

Q179N 2.74 ± 0.01 5.27 ± 0.11 2.92 ± 0.16

L183N 2.6 ± 0.04 2.52 ± 2.05 1.48 ± 0.84

T199N 6.49 ± 0.06 1.81 ± 1.25 11.80 ± 0.2

152

The KD range determined for FcγR3A binding to monomeric AdmAb variants spanned 1.48 – 18.01 ® nM, while Humira had a mean KD of 1.84 nM. Monomeric AdmAb WT had a KD of 5.96 ± 1.62 nM which lies within the range with no significant outlier (Table 45). The strongest KD was recorded for L183N (1.48 nM), while the weakest KD was recorded for A122N (18.01 nM) with large ranges. Overall, all differences in KD recorded compared to the WT, were no larger or smaller than 3-fold for TNF- and all receptors tested. All kinetic results presented in Table 44 and Table 45 are derived from analysed sample sensograms as presented in Appendix 7.3 - 7.5.

4 Discussion We rationally designed and produced a library of glycan-engineered AdmAb variants to enhance stability through steric hindrance of an APR. Successful mutation and glycan addition was confirmed for all mutants aside from Q160N, through LC-MS/MS. Consequently, the stability and binding kinetics results for the Q160N variant are directly related to the amino acid substitution alone. Since Q160N had identical stability and antigen binding affinity to the WT, it was concluded that the amino acid substitution did not have any significant impact on the antibody properties tested. The KD values for the AdmAb WT produced in our study (Table 44 and Table 45) are

® comparable to Humira KD values observed, and as reported in the literature; TNF- (30 - 110 pM), FcγR1A (15 - 23 nM), FcγR2B (4.3 - 14.2 nM) and FcγR3A (5.8 - 7.9 nM) [274, 275].

Likewise, most measured KD values for the mutants tested lay within a close range to reported literature values for the reference AdmAb, showing that these mutations had minimal impact on binding properties.

The highest improvement in resistance to aggregation was observed in mutant A122N with a slight increase in Tagg and reduction in monomer loss (Table 43 and Figure 41). The A122N mutation resulted in 1.5-fold increased binding affinity to TNF- with slightly reduced binding to FcγRs tested and with no impact on conformational stability. Pepinsky, Laura, A., Graham, Alexey, Lee, John, Allan, Sha, Christilyn and Ellen [248] reported that a mutation equivalent in position to AdmAb variant A122N significantly improved the solubility of an anti-LINGO antibody, with no impact on binding affinity or conformational stability. Improvements in solubility often correlate with reduced precipitation and insoluble aggregate formation.

This may explain the minor improvement in Tagg we observed with mutant A122N which reflects a higher resistance to precipitation at higher temperatures. Although only a minor improvement in monomer loss, the improvement is consistent at higher concentrations tested (15 mg/mL, data not

153 shown) and is significant in a clinical setting, where typically no more than 5% of aggregates are considered acceptable by regulatory standards.

No significant improvements in monomer loss were observed with other mutants (e.g. T199N and Q179N). Courtois, Schneider, Agrawal and Trout [169] hypothesised that a mutant equivalent in position to Q179N in bevacizumab could not be glycosylated as it was inaccessible to glycosylases. However, we confirmed that Q179N was glycosylated; likewise, Pepinsky, Laura, A., Graham, Alexey, Lee, John, Allan, Sha, Christilyn and Ellen [248] were able to express a glycosylated variant with a mutation in an equivalent position to Q179N which mildly improved antibody solubility. With the additional N-glycan, Q179N was expected to sterically hinder Fab-mediated aggregation in the targeted APR; while T199N was reported to significantly improve solubility [248] and sterically hinder self-association [166], however this was not observed for AdmAb. The Q179N mutation also mildly reduced antigen binding affinity by ~1.6-fold. Interestingly, Pepinsky, Laura, A., Graham, Alexey, Lee, John, Allan, Sha, Christilyn and Ellen [248] reported no change in the Tm of the mutation equivalent in position to AdmAb variant T199N, while we observed a significant improvement in Fab Tm with variant T199N (~6 degrees higher). Additionally, T199N had 1.8-fold improved KD to TNF- with minimal differences in KD to FcγRs. The lack of improvement in resistance to aggregation, despite the improvement in Fab stability, may be related to the mechanism of aggregation in AdmAb. Since the CH2 has the lowest Tm in most IgG1s, it is expected that most aggregation steps are at least initiated with unfolding of the CH2; thus, an improvement in Fab stability may not significantly impact overall resistance to soluble aggregation.

Similarly, we observed an improvement in Fab Tm for mutants L178N and L183N. L178N is an interesting example, because when compared to the non-glycosylated variant L178K, there is an improvement in Tagg and Fab Tm through glycosylation. This emphasises the impact of the N- glycans added via mutation on the conformational stability of the Fab and its resistance to precipitation, and highlights that these changes are not due to the amino acid substitution. The mutation significantly improved the conformational stability of the Fab domain. Furthermore, despite suppressing pH-induced precipitation when tested on the Fab alone [273], the L178N mutation had minimal impact on soluble aggregation when tested on the full AdmAb. Surprisingly, the non-glycosylated variant L178K which was reported to significantly suppress aggregation in bevacizumab [168], but enhance aggregation in rituximab [169], also had minimal impact on the stability of AdmAb. This finding highlights that framework regions differ significantly between IgG1 antibodies despite having identical sequences.

154

Variants L116N and T118N had multiple Tms, with a lower Tm1, higher Tm3 and a significantly reduced Tagg; additionally, they were completely degraded at 65 °C. The Tms of these mutants are harder to interpret, and do not necessarily represent the CH2, Fab and CH3 in that order. As previously reported [143], disturbances in the joining region between the Fab domains can create additional Tms for each separate domain. Thus, we expect that the Tm1 value of these mutants is representative of only part of the Fab (for e.g. the CH1 or VL); since the Tm of this part of the Fab is now less stable than the CH2, aggregation proceeded more rapidly at 65 °C. Again, our results do not correspond with findings reported for the mutation equivalent to AdmAb L116N in bevacizumab, which showed improved resistance to aggregation with no destabilization of the Fab [168]. Some perturbation in Fab stability was also observed with mutant L183N, resulting in increased monomer loss. The perturbation was minor and hard to detect; however consistently appeared in temperature ramps for Tm measurement. It is likely that the amino acid substation in this position impaired the conformational stability of part of the Fab domain as observed with mutants L116N and T118N.

5 Conclusion Protein aggregation is a complex challenge hindering the development and long-term stability of therapeutic mAbs. Using advanced computational tools, we identified aggregation prone regions and engineered adalimumab, a marketed therapeutic mAb, with additional N-glycans in the Fab domain, with the aim to sterically hinder aggregation and increase long-term stability. Our glycan- engineered variants displayed a range of different stabilities, with some mutations increasing susceptibility to aggregation, others enhancing the conformational stability of the Fab domain, and one showing minor improvements in resistance to aggregation. This approach has potential application for other antibodies where the Fab domain has poor stability or is a pre-cursor to aggregation. We demonstrated that the conformational stability of blockbuster therapeutic antibodies such as AdmAb can still be improved with glycan-engineering without compromising its antigen and Fc receptor binding affinity; these findings highlight the significant value of this engineering approach in the quest for developing more stable therapeutic mAbs with improved half- lives and therapeutic outcomes.

Acknowledgements The authors would like to acknowledge the Bosch Institute, Australia for technical help. E. Cruz would like to acknowledge the Ministry of Science, Technology and Telecommunications of the Republic of Costa Rica for the awarded postgraduate scholarship. M. Reslan is a recipient of the

155

Research Training Program stipend provided by the University of Sydney on behalf of the Department of Education and Training to support his research training.

156

Chapter 7 Discussion; Outcomes, Reflections and Future Prospects

157

1 Summary of Aims and Outcomes In this body of work, mutational projects were designed for enhanced affinity and reduced self- association and aggregation propensity, applied to Tmab and AdmAb as model mAbs for investigation. The rationale behind strategic mutagenesis for the successful modulation of Tmabs target affinity is the potential to reformat the antibody platform for a more efficacious therapeutic treatment. And similarly, the rationale behind strategic hyper-glycosylation for the successful reduction of self-association and aggregation propensity in both antibodies is the potential to improve manufacturability and formulation strategies, as well as prolonging mAb half-life for a more efficacious therapeutic treatment. The successful application of preliminary in silico analysis strategies to improve model mAbs by design for modulated functions can indeed be more broadly applied to other biotherapeutics, and is yet to be fully realised. The outcomes of this research has demonstrated the success of using in silico modelling and prediction for the preliminary elucidation and design of mAbs, having elected several lead mutants from the projects for further investigation.

Before commencing the design of mutational projects, appropriate expression systems for the lab- scale production of Tmab and AdmAb were required to be established and optimised. Several cloning strategies were implemented as to ascertain the most appropriate expression vector platforms for optimising antibody expression and purification, prior to biological assay and mutagenesis optimisation, and the design and expression of mutant libraries. Mutational projects were designed based on in silico analysis of crystal structures of the antibody Fabs complexed to a fragment of the target epitope. The Tmab affinity mutations were based on results from our collaborators from A/Prof Serdar Kuyucak’s Computational Biophysics Group, School of Physics, The University of Sydney, and the glycosylation mutations based on literature search of reported APRs, paying particular attention to spacial proximity of the APR and complexing interfaces. Mutant libraries were then produced and characterised before in vitro assessment to determine lead candidates for further investigation of modulated target binding for the affinity library specifically, or comparison of retained biological functions and stability profiling for the hyper-glycosylated libraries.

In silico prediction of Tmab affinity mutations was in agreement with in vitro analysis, highlighting the applicability and impact of producing lead candidates by design for more economical and efficient discovery and biobetter development. Furthermore, strategic introduction of glycosylation sites identified through in silico analysis for the steric hindrance of an APR showed considerable stability enhancement without compromising biological functions in both Tmab and AdmAb, which demonstrates the potential transferability of this engineering approach in the pursuit of biobetter

158 design and development. The results from these mutational projects show that the preliminary design of mAbs based on relevant computational tools can successfully elucidate key strategic modifications for in vitro validation, which exemplifies this approach in the pursuit and design of next generation biologic therapeutics.

2 Establishing Lab-Scale mAb Expression Platforms In order to design and produce mutational projects for in vitro assessment, the development of appropriate Tmab and AdmAb expression systems were required. The expression system for the mAb mutants was designed and optimised in a transient, serum-free, suspension 293F culture and captured through Protein A affinity chromatography.

A library of mammalian expression vectors was developed for Tmab and AdmAb expression, as reported in Chapter 2. Several expression vectors requiring different cloning strategies were considered to design and apply the assembly of Tmab and AdmAb expression vectors for transient and stable expression. A pilot transient expression system was performed to compare antibody expression between the vector platforms. The gWiz platform was elected as the most suitable for optimisation of transient expression due to its superior yield and cloning efficiency, and the pVITRO1 platform elected as the most suitable for stable expression of Tmab alone.

A high yielding stable Tmab expression system was established using the pVITRO1 platform, for preliminary characterisation and assay optimisations as reported in Chapter 3. The expression system was improved from traditional techniques through selection in adherent conditions prior to adaption to scaled-up serum free suspension culture, and the addition of surfactant and tryptone media supplements for enhanced protein expression and support of cell viability. The stable expression of Tmab was utilised to scout buffers for optimising the efficient capture of Tmab through Protein A affinity chromatography, and preliminary characterisation was performed to confirm the yield and purity of bulk Tmab product.

However, the composition of host cell proteins and DNA, leached Protein A ligand, mAb fragments, aggregates and ionic variant contaminants were not specifically profiled from the bulk Tmab product to elucidate the necessity of any further polishing steps. The significance of these potential contaminants in the bulk mAb product can cause product instability and loss in biological activity, which affects in vitro assay results for interpretation. As extensively discussed in Chapter 1, the profile of these contaminants can vary, being influenced by expression and capture conditions, and can be separated from the bulk mAb product with relevant ionic, hydrophobic and size-exclusion chromatographic technologies. Although not specifically addressed through the

159

Tmab mutant projects in Chapter 4 and Chapter 5, the AdmAb mutant project in Chapter 6 optimised a further size-exclusion polishing step for the isolation of pure AdmAb monomer product prior to stability profiling and biological assays, for confident interpretation of the in vitro analysis between AdmAb mutant variants.

As mAb mutant samples were equivalently produced using optimised expression conditions, variation in contaminant profiles between bulk mAb samples was likely minimal, hence mutation projects were comparable to their corresponding lab-produced wild-type mAb. In hindsight, profiling the contaminants in all mAb samples would have added value to provide more confidence behind the in vitro results from the Tmab mutation projects, to identify whether a polishing step was warranted and further ascertain the optimisation of mAb expression and capture conditions. However, this body of work has shown considerably interesting results from the Tmab mutation projects designed and expressed, with several lead candidates identified. Reproducing the Tmab projects with monomeric mAb samples would likely reflect similar in vitro properties as have been identified in the bulk mAb samples produced through this body of work; however, lead candidates warrant a further polishing step to reproduce and validate the current findings.

3 In Pursuit of Modulating Target Affinity As extensively discussed in Chapter 1, the affinity of a mAb to its biological target is a crucial parameter to be considered for the design of biobetter antibodies. In considering disease presentation and treatment strategies, modulation of a mAbs target affinity can improve its efficacy with an appropriate mAb format. ADCs and fragment mAbs favour enhanced target binding affinity; disease treatments requiring ADCC, tissue penetration and retention favour reduced target binding. Further to modulating target binding, mAb half-life extension and modulation of immune effector function has been demonstrated through strategic manipulation of mAb binding to specific immune receptors.

In this body of work, Tmab was used as a model antibody to design strategic mutations for modulated affinity to target HER2, as reported in Chapter 4. Our collaborators from A/Prof Serdar Kuyucak’s Computational Biophysics Group, School of Physics, The University of Sydney elucidated several substitutions in Tmabs Fab region that were predicted to modify its affinity to HER2 through in silico analysis. Specifically, substitution S50>E was disruptive to the Fab complex whilst reversal Y49/S50>S49/Y50 enhanced complex binding, substitutions D102>F and D28>K presented negligible alteration whilst D28>R showed considerable complex binding enhancement. Several mutagenesis strategies were implemented to optimise mutation yield and efficiency; however, of the five mutations designed only three were successfully produced for in

160 vitro validation, that being S50>E, D28>K and D102>F. Of the other predicted substitutions that were unsuccessfully isolated after several rounds of mutagenesis, the in silico results strongly suggest that they are lead enhanced affinity candidates to pursue for in vitro validation.

The affinity mutant library was assessed for binding to extracellularly expressed HER2 in a cell- binding titration assay, and surface plasmon resonance was performed to determine the binding kinetics of the affinity library to immobilized HER2 and FcγR1A. The findings from these results were in agreement to in silico prediction, in that substitution S50>E diminished binding to HER2, and that D28>K and D102>F showed negligible differences in binding to HER2 as compared to wild-type Tmab. Furthermore, binding to FcγR1A was assessed to validate that immunological activity was not compromised in the affinity mutant library. Interestingly, D102>F showed considerably enhanced binding to FcγR1A which would likely result in a more pronounced ADCC effect and is of interest for further investigation.

From the findings of this project, the diminished affinity exhibited from mutant S50>E to HER2 is of considerable interest for continued investigation as a potential platform for improved cancer treatment by coupling this mutation with enhanced ADCC technologies. S50>E mutant warrants further effector function and stability profiling, in vivo assays to elucidate retention of specificity and enhancement of tumour penetration and retention, and x-ray crystallography of the Fab complex to HER2 to corroborate the structure of the binding interface with in silico predictions. Producing substitution D28>K and reversal Y49/S50>S49/Y50 mutations was unfortunately not achieved in this body of work. The in silico predictions of these reported mutations strongly indicate these as enhanced affinity candidates and are of continued interest for in vitro validation.

4 In Pursuit of Reducing Self-Association and Aggregation Propensity As extensively discussed in Chapter 1, the propensity of a mAb to self-associate is a critical factor in its developability, as self-association induces mAb instability and irreversible aggregation leading to manufacturing loss, reduced efficacy and adverse immunogenicity in the drug product. mAb aggregation mainly occurs through the manufacture process, formulation development and long term storage, and remains a key challenge to be addressed for the design of biobetter formulations. Considerable efforts in determining self-association and APRs in mAbs have led to the design of in silico predictive tools that identify surface exposed hydrophobicity. The direct substitution of residues in APRs to remove hydrophobicity has shown considerably improved mAb stability and reduced aggregation propensity in vitro, validating this strategy of antibody engineering for the design of biobetters.

161

In this body of work, an extension of this mAb engineering technique was considered, in that the introduction of spatially proximate N-linked glycans to identified APRs would potentially sterically hinder colloidal APR associations and hence reduce self-association and aggregation propensity. Further to this, hyper-glycosylation technology has been utilised to increase solubility and enhance half-life of other biotherapeutics. Strategic hyper-glycosylation introduces orientated functional groups that can further be useful for conjugation technologies. Beyond self-association, the benefits for strategic hyper-glycosylation technology for biobetter mAb development are highly apparent and the potential is yet to be realised.

Tmab was used as a model antibody to design mutations introducing strategic N-linked glycans in the Tmab Fab to sterically hinder APR associations, as reported in Chapter 5. Design of the mutational projects was based on literature search of identified APR that were experimentally validated through direct substitution. Spatially proximate surface exposed residue substitutions that introduce an N-linked glycan were considered for glycosylation projects. One consolidated glycosylation APR project was elected out of the several projects designed. To minimise the possibility of the introduced N-glycan diminishing important biological functions, the mutations were restricted to the Fab region excluding CDRs. To demonstrate proof-of-concept of the introduced glycans reducing self-association and aggregation propensity, the glycosylation mutant library was transferred into AdmAb as reported in Chapter 6.

In both chapters, the mutation libraries were characterised through mass spectrometry to confirm the success of mutation and glycan addition, and interestingly one mutant (Q160>N) did not achieve glycosylation despite successful mutation in both mAbs and previous report of successful glycosylation of a bevacizumab mutant equivalent. Of the stability profiling, equivalent enhancement of intrinsic stability and aggregation resistance was observed between the Tmab and AdmAb libraries, although monomer retention was incomparable due to the markedly different parameters elected for the monomer loss studies. Tmab L177>N, L182>N and T198>N and the AdmAb equivalent L178>N, L183>N and T199>N mutants possessed enhanced intrinsic stability. Resistance to aggregation was observed in Tmab L177>N, A121>N, Q178>N, L182>N and T198>N the AdmAb equivalent L178>N, A122>N, Q179>N, L183>N and T199>N mutants. Tmab L115>N and T117>N and AdmAb equivalent L116>N and T118>N mutants equivalently performed poorly for intrinsic stability and resistance to aggregation. Highly accelerated monomer loss was performed on the Tmab library, of which candidates L177>N and L182>N exhibited superior monomer retention. Moderately accelerated monomer loss was performed on the AdmAb library, of which candidates A122>N and T199>N exhibited superior monomer retention, which

162 likely indicates a difference in the mechanism of aggregation. Tmab L177>N, L182>N and T198>N mutants and AdmAb L178>N, A122>N, L183>N and T199>N mutants were elected as lead candidates for more rigorous investigation.

Further to stability profiling, retention of biological activity was screened amongst the libraries to ascertain retention of biological functions. In general, mutants binding to tested targets were within 3-fold of the wild-type mAb, which may pose a small degree of significance; however, no real trend was observed between the two libraries. More specifically, Tmab mutant library binding to target

HER2 and FcγR3A were modestly compromised, and binding to FcγR1A and was modestly enhanced. Unlike the Tmab mutant library, the AdmAb mutant library showed more variability, less significant difference and no trends in comparison to the wild-type mAb binding to target receptors.

Interestingly, Tmab mutant L182>N displayed a drastic disruption in binding to FcγR2B and markedly enhanced affinity to FcγR3A, along with T198>N, which suggests a potentially enhanced ADCC effect. However, these modulations were not observed in the AdmAb equivalents, suggesting that the observed differences in binding may be negligible, although thorough in vivo assessment of modulated ADCC is warranted to clarify the case.

From the findings of these projects, several mutations were elected as lead candidates for perusal of more thorough stability screening in the lead up to formulation development, demonstrating the success and applicability of this hyper-glycosylation approach in designing stability enhanced biobetters. Specifically, Tmab L177>N, L182>N and T198>N, AdmAb equivalents L178>N, L183>N and T199>N, and further AdmAb A122>N mutants were elected as lead candidates for further investigation. Proof-of-concept for the transference of this technology for reducing self- association and aggregation propensity has been successfully demonstrated, in that three of four lead stability mutant candidates were elected from both antibody libraries. Characterisation of the addition of N-linked glycans was consistent between the two antibody libraries, further demonstrating the applicability of transference of this technology. The further size-exclusion polishing step in the AdmAb project provided more confidence behind the interpretation of the in vitro results to validate the current findings. The findings of the biological assays was marginally inconsistent between the two libraries, this is likely due to the additional size-exclusion purification step applied to the AdmAb library to produce monomeric species for robust in vitro results. In order to provide a more robust comparison, the Tmab library warrants reproduction following an appropriate polishing step to produce monomeric species. Following this, the lead stability mutant candidates warrant reproduction in several further antibodies for stability transferability comparison, in vivo assessment to observe retention of specificity and the potential for enhanced

163 half-life. X-ray crystallography of the glycosylated Fab is also warranted, to elucidate glycan orientation and structure for the potential to derive more significant findings through further in silico analysis. Furthermore, expression of the hyper-glycosylated mutants in different expression platforms for different glycan profiles is of further interest, to ascertain if particular glycosylation profiles may attribute to mAb stability, self-association and biological functions, in the aim of refining this technology for design-specific purposes.

In considering the findings and outcomes from the completed hyper-glycosylation project, designed based on the self-association of one specific APR, the design and in vitro validation of further hyper-glycosylation projects is further warranted. It is entirely possible that a few key glycan additions in combination would achieve highly favourable stability and solubility profiles that can apply the mAb to alternative, less invasive formulations for a more efficacious treatment strategy.

5 Concluding Remarks To recapitulate the findings from this body of work, in silico analysis was successfully applied to design model mAbs for a modulated affinity to biological target and disrupted colloidal self- association interactions, through residue substitution and hyper-glycosylation strategies. In vitro assessment of affinity mutations corroborated with in silico predictions, highlighting the success of this approach in elucidating lead candidates by design for more efficient biobetter development. In vitro assessment of hyper-glycosylated libraries produced several lead candidates with enhanced stability profiles, demonstrating the applicability of this approach in engineering stability enhanced mAbs by design. Furthermore, transference was demonstrated of this hyper-glycosylation technology, which implies the potential adaptability of this approach to the whole IgG class in the pursuit of designing next generation biologic therapeutics. These strategies address two key challenges that arise in mAb development and can indeed be more broadly applied to the design of other biologic therapeutics, showing immense promise for the application of the presented engineering strategies for the strategic development of biobetter therapies.

164

References

[1] Urquhart, L., Top drugs and companies by sales in 2017. Nat. Rev. Drug Discov. 2018, 17, 232. [2] Ecker, D., Dana Jones, S., L Levine, H., The Therapeutic Monoclonal Antibody Market. mAbs 2015, 7, 9-14. [3] An, Z., “Magic Bullets” at the center stage of immune therapy: a special issue on therapeutic antibodies. Protein Cell 2018, 9. [4] Santos, M. L. d., Quintilio, W., Manieri, T. M., Tsuruta, L. R., Moro, A. M., Advances and challenges in therapeutic monoclonal antibodies drug development. Bras. J. Pharm. Sci. 2018, 54. [5] Elgundi, Z., Reslan, M., Cruz, E., Sifniotis, V., Kayser, V., The state-of-play and future of antibody therapeutics. Adv. Drug Del. Rev. 2017, 122, 2-19. [6] Almagro, J. C., Daniels-Wells, T. R., Perez-Tapia, S. M., Penichet, M. L., Progress and challenges in the design and clinical development of antibodies for cancer therapy. Front. Immunol. 2018, 8. [7] Liu, B. N., Luo, J. H., Research and development of innovative antibody-based drugs. Yaoxue Xuebao 2017, 52, 1811-1819. [8] Awwad, S., Angkawinitwong, U., Overview of antibody drug delivery. Pharmaceutics 2018, 10. [9] Klein, C., Special Issue: Monoclonal Antibodies. Antibodies 2018, 7, 17. [10] Pawlowski, J. W., Bajardi-Taccioli, A., Houde, D., Feschenko, M., et al., Influence of glycan modification on IgG1 biochemical and biophysical properties. J. Pharm. Biomed. Anal. 2018, 151, 133-144. [11] Hoffmann, R. M., Coumbe, B. G. T., Josephs, D. H., Mele, S., et al., Antibody structure and engineering considerations for the design and function of Antibody Drug Conjugates (ADCs). OncoImmunology 2018, 7. [12] Wang, X., Mathieu, M., Brezski, R. J., IgG Fc engineering to modulate antibody effector functions. Protein Cell 2018, 9, 63-73. [13] Saldanha, J. W., Humanization Strategies, Handbook of Therapeutic Antibodies: Second Edition, Wiley Blackwell 2014, pp. 89-114. [14] Wang, C., Xu, P., Zhang, L., Huang, J., et al., Current strategies and applications for precision drug design. Front. Pharmacol. 2018, 9. [15] Strohl, W. R., Current progress in innovative engineered antibodies. Protein Cell 2018, 9, 86-120. [16] Fonseca, M. H. G., Furtado, G. P., Bezerra, M. R. L., Pontes, L. Q., Fernandes, C. F. C., Boosting half-life and effector functions of therapeutic antibodies by Fc- engineering: An interaction-function review. Int. J. Biol. Macromol. 2018, 119, 306- 311.

165

[17] Bray, F., Ferlay, J., Soerjomataram, I., Siegel, R. L., et al., Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J. Clin. 2018, 68, 394-424. [18] Waks, A. G., Winer, E. P., 69 - Chemotherapy and HER2-Directed Therapy for Metastatic Breast Cancer, in: Bland, K. I., Copeland, E. M., Klimberg, V. S., Gradishar, W. J. (Eds.), The Breast (Fifth Edition), Elsevier 2018, pp. 885-906.e888. [19] Carter, P., Presta, L., Gorman, C. M., Ridgway, J. B., et al., Humanization of an anti-p185HER2 antibody for human cancer therapy. Proceedings of the National Academy of Sciences 1992, 89, 4285-4289. [20] Lambertini, M., Pondé, N. F., Solinas, C., de Azambuja, E., Adjuvant trastuzumab: a 10-year overview of its benefit. Expert Rev. Anticancer Ther. 2017, 17, 61-74. [21] Dinh, P., Piccart, M. J., HER2-Targeted Therapy; Management of Breast Diseases, in: Jatoi, I., Rody, A. (Eds.), Management of Breast Diseases, Springer International Publishing, Cham 2016, pp. 391-410. [22] Baldo, B. A., Monoclonal Antibodies Approved for Cancer Therapy, Safety of Biologics Therapy: Monoclonal Antibodies, Cytokines, Fusion Proteins, Hormones, Enzymes, Coagulation Proteins, Vaccines, Botulinum Toxins, Springer International Publishing, Cham 2016, pp. 57-140. [23] Rudnick, S. I., Adams, G. P., Affinity and avidity in antibody-based tumor targeting. Cancer Biother. Radiopharm. 2009, 24, 155-161. [24] Guerin, M., Sabatier, R., Gonçalves, A., Autorisation de mise sur le marché du trastuzumab emtansine (Kadcyla®) dans les cancers du sein métastatiques HER2- positifs. Bull. Cancer 2015, 102, 390-397. [25] Hsu, J. L., Hung, M.-C., Mechanisms of Action and Resistance of Trastuzumab in Breast Cancer, in: Yarden, Y., Elkabets, M. (Eds.), Resistance to Anti-Cancer Therapeutics Targeting Receptor Tyrosine Kinases and Downstream Pathways, Springer International Publishing, Cham 2018, pp. 51-66. [26] Kourie, H. R., El Rassy, E., Clatot, F., de Azambuja, E., Lambertini, M., Emerging treatments for HER2-positive early-stage breast cancer: focus on neratinib. Onco. Targets Ther. 2017, 10, 3363-3372. [27] Chowdhury, R., Ellis, P., Trastuzumab (Herceptin®) and Ado-Trastuzumab Emtansine (Kadcyla®): Treatments for HER2-Positive Breast Cancer, Handbook of Therapeutic Antibodies: Second Edition, Wiley Blackwell 2014, pp. 2041-2068. [28] Genetech, in: FDA (Ed.), Centre for Drug Evaluation and Research, Package Insert Herceptin (Trastuzumab) 451798-0369-S03 2003, p. 26. [29] Phillips, G. D. L., Fields, C. T., Li, G., Dowbenko, D., et al., Dual Targeting of HER2-Positive Cancer with Trastuzumab Emtansine and Pertuzumab: Critical Role for Neuregulin Blockade in Antitumor Response to Combination Therapy. Clin. Cancer. Res. 2014, 20, 456-468. [30] Lin, N. U., Winer, E. P., Wheatley, D., Carey, L. A., et al., A phase II study of afatinib (BIBW 2992), an irreversible ErbB family blocker, in patients with HER2-

166 positive metastatic breast cancer progressing after trastuzumab. Breast Cancer Res. Treat. 2012, 133, 1057-1065. [31] Melichar, B., Studentova, H., Kalabova, H., Vitaskova, D., Role of subcutaneous formulation of trastuzumab in the treatment of patients with HER2-positive breast cancer. Immunotherapy 2014, 6, 811-819. [32] Jackisch, C., Kim, S. B., Semiglazov, V., Melichar, B., et al., Subcutaneous versus intravenous formulation of trastuzumab for HER2-positive early breast cancer: updated results from the phase III HannaH study. Ann. Oncol. 2015, 26, 320-325. [33] Pimentel, F. F., Morgan, G., Tiezzi, D. G., de Andrade, J. M., Development of New Formulations of Biologics: Expectations, Immunogenicity, and Safety for Subcutaneous Trastuzumab. Pharmaceut. Med. 2018, 32, 319-325. [34] Lefranc, M.-P., Giudicelli, V., Duroux, P., Jabado-Michaloud, J., et al., IMGT(®), the international ImMunoGeneTics information system(®) 25 years on. Nucleic Acids Res. 2015, 43, D413-D422. [35] Wishart, D. S., Feunang, Y. D., Guo, A. C., Lo, E. J., et al., DrugBank 5.0: a major update to the DrugBank database for 2018. Nucleic Acids Res. 2018, 46, D1074-D1082. [36] Burley, S. K., Berman, H. M., Christie, C., Duarte, J. M., et al., RCSB Protein Data Bank: Sustaining a living digital data resource that enables breakthroughs in scientific research and biomedical education. Protein Sci. 2018, 27, 316-330. [37] Vos, T., Allen, C., Arora, M., Barber, R. M., et al., Global, regional, and national incidence, prevalence, and years lived with disability for 310 diseases and injuries, 1990–2015: a systematic analysis for the Global Burden of Disease Study 2015. The Lancet 2016, 388, 1545-1602. [38] Van den Bosch, F., Coates, L., Clinical management of psoriatic arthritis. The Lancet 2018, 391, 2285-2294. [39] Ackerman, I. N., Pratt, C., Gorelik, A., Liew, D., Projected Burden of Osteoarthritis and Rheumatoid Arthritis in Australia: A Population-Level Analysis. Arthritis Care Res. (Hoboken) 2018, 70, 877-883. [40] Padjen, I., Chapter 3 - Drugs Used in Rheumatic Disease, in: Atzeni, F., Masala, I. F., Aletaha, D., Lee, M., Baraliakos, X. (Eds.), Handbook of Systemic Autoimmune Diseases, Elsevier 2018, pp. 39-76. [41] Smolen, J. S., Aletaha, D., McInnes, I. B., Rheumatoid arthritis. The Lancet 2016, 388, 2023-2038. [42] Ellinghaus, D., Jostins, L., Spain, S. L., Cortes, A., et al., Analysis of five chronic inflammatory diseases identifies 27 new associations and highlights disease- specific patterns at shared loci. Nat. Genet. 2016, 48, 510. [43] Reichert, J. M., Adalimumab (Humira®), Handbook of Therapeutic Antibodies: Second Edition, Wiley Blackwell 2014, pp. 1309-1322. [44] Kiely, P. D. W., Nikiphorou, E., Management of rheumatoid arthritis. Medicine 2018, 46, 216-221.

167

[45] Real-Fernández, F., Pregnolato, F., Cimaz, R., Papini, A. M., et al., Detection of anti-adalimumab antibodies in a RA responsive cohort of patients using three different techniques. Anal. Biochem. 2019, 566, 133-138. [46] Chimenti, M. S., Graceffa, D., Perricone, R., Anti-TNFα discontinuation in rheumatoid and psoriatic arthritis: Is it possible after disease remission? Autoimmun. Rev. 2011, 10, 636-640. [47] Abbot, in: FDA (Ed.), Centre for Drug Evaluation and Research, HumiraTM (adalimumab) FDA Label DN0735V7 CR22-05126 2002, p. 16. [48] Mizukami, A., Caron, A. L., Picanço-Castro, V., Swiech, K., Platforms for recombinant therapeutic glycoprotein production, Methods Mol. Biol., Humana Press Inc. 2018, pp. 1-14. [49] Thomson, C. A., IgG Structure and Function, Encyclopedia of Immunobiology, Elsevier Inc. 2016, pp. 15-22. [50] Pieracci, J. P., Armando, J. W., Westoby, M., Thommes, J., Chapter 9 - Industry Review of Cell Separation and Product Harvesting Methods, in: Jagschies, G., Lindskog, E., Łącki, K., Galliher, P. (Eds.), Biopharmaceutical Processing, Elsevier 2018, pp. 165-206. [51] Huang, C.-J., Lin, H., Yang, X., Industrial production of recombinant therapeutics in Escherichia coli and its recent advancements. J. Ind. Microbiol. Biotechnol. 2012, 39, 383-399. [52] Ulmer, N., Vogg, S., Müller-Späth, T., Morbidelli, M., Chapter 7 - Purification of Human Monoclonal Antibodies and Their Fragments, in: Steinitz, M. (Ed.), Human Monoclonal Antibodies: Methods and Protocols, Springer New York, New York, NY 2019, pp. 163-188. [53] Arosio, P., Barolo, G., Müller-Späth, T., Wu, H., Morbidelli, M., Aggregation Stability of a Monoclonal Antibody During Downstream Processing. Pharm. Res. 2011, 28, 1884-1894. [54] Bittner, B., Richter, W., Schmidt, J., Subcutaneous Administration of Biotherapeutics: An Overview of Current Challenges and Opportunities. Biodrugs 2018, 32, 425-440. [55] Viola, M., Sequeira, J., Seiça, R., Veiga, F., et al., Subcutaneous delivery of monoclonal antibodies: How do we get there? J. Controlled Release 2018, 286, 301- 314. [56] Otvos, L., 6.05 - Basic Principles of Formulation for Biotherapeutics: Approaches to Alternative Drug Delivery, in: Chackalamannil, S., Rotella, D., Ward, S. E. (Eds.), Comprehensive Medicinal Chemistry III, Elsevier, Oxford 2017, pp. 131-156. [57] Morales, J. O., Fathe, K. R., Brunaugh, A., Ferrati, S., et al., Challenges and Future Prospects for the Delivery of Biologics: Oral Mucosal, Pulmonary, and Transdermal Routes. The AAPS Journal 2017, 19, 652-668. [58] Muheem, A., Shakeel, F., Jahangir, M. A., Anwar, M., et al., A review on the strategies for oral delivery of proteins and peptides and their clinical perspectives. Saudi Pharm. J. 2016, 24, 413-428.

168

[59] Henry, K. A., Next-Generation DNA Sequencing of VH/VL Repertoires: A Primer and Guide to Applications in Single-Domain Antibody Discovery, in: Hust, M., Lim, T. S. (Eds.), Phage Display: Methods and Protocols, Springer New York, New York, NY 2018, pp. 425-446. [60] Fliedl, L., Grillari, J., Grillari-Voglauer, R., Human cell lines for the production of recombinant proteins: on the horizon. New Biotechnology 2015, 32, 673-679. [61] Hunter, M., Yuan, P., Vavilala, D., Fox, M., Optimization of Protein Expression in Mammalian Cells. Curr. Protoc. Protein Sci. 2019, 95, e77. [62] Lindskog, E. K., Fischer, S., Wenger, T., Schulz, P., Chapter 6 - Host Cells, in: Jagschies, G., Lindskog, E., Łącki, K., Galliher, P. (Eds.), Biopharmaceutical Processing, Elsevier 2018, pp. 111-130. [63] Hu, J., Han, J., Li, H., Zhang, X., et al., Human Embryonic Kidney 293 Cells: A Vehicle for Biopharmaceutical Manufacturing, Structural Biology, and Electrophysiology. Cells Tissues Organs 2018, 205, 1-8. [64] Jacobi, A., Enenkel, B., Garidel, P., Eckermann, C., et al., Process Development and Manufacturing of Therapeutic Antibodies, Handbook of Therapeutic Antibodies: Second Edition, Wiley Blackwell 2014, pp. 601-664. [65] Tada, M., Tatematsu, K.-I., Ishii-Watabe, A., Harazono, A., et al., Characterization of anti-CD20 monoclonal antibody produced by transgenic silkworms ( Bombyx mori ). mAbs 2015, 7, 1138-1150. [66] Maccani, A., Landes, N., Stadlmayr, G., Maresch, D., et al., Pichia pastoris secretes recombinant proteins less efficiently than Chinese hamster ovary cells but allows higher space-time yields for less complex proteins. Biotechnol. J. 2014, 9, 526-537. [67] Frenzel, A., Hust, M., Schirrmann, T., Expression of Recombinant Antibodies. Front. Immunol. 2013, 4. [68] Thoring, L., Kubick, S., Versatile Cell-Free Protein Synthesis Systems Based on Chinese Hamster Ovary Cells, in: Hacker, D. L. (Ed.), Recombinant Protein Expression in Mammalian Cells: Methods and Protocols, Springer New York, New York, NY 2018, pp. 289-308. [69] Matsuda, T., Ito, T., Takemoto, C., Katsura, K., et al., Cell-free synthesis of functional antibody fragments to provide a structural basis for antibody–antigen interaction. PLoS One 2018, 13, e0193158. [70] Tran, K., Gurramkonda, C., Cooper, M. A., Pilli, M., et al., Cell-free production of a therapeutic protein: Expression, purification, and characterization of recombinant streptokinase using a CHO lysate. Biotechnol. Bioeng. 2018, 115, 92-102. [71] Stech, M., Nikolaeva, O., Thoring, L., Stöcklein, W. F. M., et al., Cell-free synthesis of functional antibodies using a coupled in vitro transcription-translation system based on CHO cell lysates. Sci. Rep. 2017, 7, 12030. [72] Li, J., Wang, H., Kwon, Y.-C., Jewett, M. C., Establishing a high yielding streptomyces-based cell-free protein synthesis system. Biotechnol. Bioeng. 2017, 114, 1343-1353.

169

[73] Stech, M., Kubick, S., Cell-Free Synthesis Meets Antibody Production: A Review. Antibodies 2015, 4, 12. [74] Nakamura, T., Omasa, T., Optimization of cell line development in the GS-CHO expression system using a high-throughput, single cell-based clone selection system. J. Biosci. Bioeng. 2015, 120, 323-329. [75] Zhou, Y., Raju, R., Alves, C., Gilbert, A., Debottlenecking protein secretion and reducing protein aggregation in the cellular host. Curr. Opin. Biotechnol. 2018, 53, 151-157. [76] Inwood, S., Betenbaugh, M. J., Lal, M., Shiloach, J., Genome-Wide High- Throughput RNAi Screening for Identification of Genes Involved in Protein Production, in: Hacker, D. L. (Ed.), Recombinant Protein Expression in Mammalian Cells: Methods and Protocols, Springer New York, New York, NY 2018, pp. 209- 219. [77] Dangi, A. K., Sinha, R., Dwivedi, S., Gupta, S. K., Shukla, P., Cell Line Techniques and Gene Editing Tools for Antibody Production: A Review. Front. Pharmacol. 2018, 9. [78] Delic, M., Göngrich, R., Mattanovich, D., Gasser, B., Engineering of protein folding and secretion - Strategies to overcome bottlenecks for efficient production of recombinant proteins. Antioxid. Redox Signal. 2014, 21, 414-437. [79] Zhong, X., Wright, J. F., Biological Insights into Therapeutic Protein Modifications throughout Trafficking and Their Biopharmaceutical Applications. Int. J. Cell Biol. 2013, 2013, 19. [80] Altamirano, C., Berrios, J., Vergara, M., Becerra, S., Advances in improving mammalian cells metabolism for recombinant protein production. Electron. J. Biotechnol. 2013, 16. [81] Parola, C., Mason, D. M., Zingg, A., Neumeier, D., Reddy, S. T., Genome Engineering of Hybridomas to Generate Stable Cell Lines for Antibody Expression, in: Hacker, D. L. (Ed.), Recombinant Protein Expression in Mammalian Cells: Methods and Protocols, Springer New York, New York, NY 2018, pp. 79-111. [82] You, M., Yang, Y., Zhong, C., Chen, F., et al., Efficient mAb production in CHO cells with optimized signal peptide, codon, and UTR. Appl. Microbiol. Biotechnol. 2018, 102, 5953-5964. [83] Mauro, V. P., Chappell, S. A., Considerations in the Use of Codon Optimization for Recombinant Protein Expression, in: Hacker, D. L. (Ed.), Recombinant Protein Expression in Mammalian Cells: Methods and Protocols, Springer New York, New York, NY 2018, pp. 275-288. [84] Michael, I. P., Nagy, A., Inducible Protein Production in 293 Cells Using the piggyBac Transposon System, in: Hacker, D. L. (Ed.), Recombinant Protein Expression in Mammalian Cells: Methods and Protocols, Springer New York, New York, NY 2018, pp. 57-68. [85] Balasubramanian, S., Recombinant CHO Cell Pool Generation Using piggyBac Transposon System, in: Hacker, D. L. (Ed.), Recombinant Protein Expression in

170

Mammalian Cells: Methods and Protocols, Springer New York, New York, NY 2018, pp. 69-78. [86] Jäger, V., Büssow, K., Schirrmann, T., Transient Recombinant Protein Expression in Mammalian Cells, in: Al-Rubeai, M. (Ed.), Animal Cell Culture, Springer International Publishing, Cham 2015, pp. 27-64. [87] Wijesuriya, S. D., Pongo, E., Tomic, M., Zhang, F., et al., Antibody engineering to improve manufacturability. Protein Expression Purif. 2018, 149, 75-83. [88] L’Abbé, D., Bisson, L., Gervais, C., Grazzini, E., Durocher, Y., Transient Gene Expression in Suspension HEK293-EBNA1 Cells, in: Hacker, D. L. (Ed.), Recombinant Protein Expression in Mammalian Cells: Methods and Protocols, Springer New York, New York, NY 2018, pp. 1-16. [89] Arena, T. A., Harms, P. D., Wong, A. W., High Throughput Transfection of HEK293 Cells for Transient Protein Production, in: Hacker, D. L. (Ed.), Recombinant Protein Expression in Mammalian Cells: Methods and Protocols, Springer New York, New York, NY 2018, pp. 179-187. [90] Agrawal, V., Yu, B., Pagila, R., Yang, B., et al., A high-yielding, CHO-K1- based transient transfection system. BioProcess International 2013, 11, 28-35. [91] Hacker, D. L., Kiseljak, D., Rajendra, Y., Thurnheer, S., et al., Polyethyleneimine-based transient gene expression processes for suspension-adapted HEK-293E and CHO-DG44 cells. Protein Expression Purif. 2013, 92, 67-76. [92] Ritacco, F. V., Wu, Y., Khetan, A., Cell culture media for recombinant protein expression in Chinese hamster ovary (CHO) cells: History, key components, and optimization strategies. Biotechnol. Prog. 2018, 34, 1407-1426. [93] Whitford, W. G., Lundgren, M., Fairbank, A., Chapter 8 - Cell Culture Media in Bioprocessing, in: Jagschies, G., Lindskog, E., Łącki, K., Galliher, P. (Eds.), Biopharmaceutical Processing, Elsevier 2018, pp. 147-162. [94] Davami, F., Eghbalpour, F., Barkhordari, F., Mahboudi, F., Effect of Peptone Feeding on Transient Gene Expression Process in CHO DG44. Avicenna J. Med. Biotechnol. 2014, 6, 147-155. [95] Mahboudi, F., Abolhassan, M. R., Azarpanah, A., Aghajani-Lazarjani, H., et al., The Role of Different Supplements in Expression Level of Monoclonal Antibody against Human CD20. Avicenna J. Med. Biotechnol. 2013, 5, 140-147. [96] You, M., Liu, Y., Chen, Y., Guo, J., et al., Maximizing Antibody Production in Suspension-Cultured Mammalian Cells by the Customized Transient Gene Expression Method. Biosci., Biotechnol., Biochem. 2013, 77, 1207-1213. [97] Backliwal, G., Hildinger, M., Kuettel, I., Delegrange, F., et al., Valproic acid: a viable alternative to sodium butyrate for enhancing protein expression in mammalian cell cultures. Biotechnol. Bioeng. 2008, 101, 182-189. [98] Kim, S. J., Ha, G. S., Lee, G., Lim, S. I., et al., Enhanced expression of soluble antibody fragments by low-temperature and overdosing with a nitrogen source. Enzyme Microb. Technol. 2018, 115, 9-15. [99] Lalonde, M.-E., Durocher, Y., Therapeutic glycoprotein production in mammalian cells. J. Biotechnol. 2017, 251, 128-140.

171

[100] Rajendra, Y., Kiseljak, D., Baldi, L., Hacker, D. L., Wurm, F. M., A simple high-yielding process for transient gene expression in CHO cells. J. Biotechnol. 2011, 153, 22-26. [101] Codamo, J., Munro, T. P., Hughes, B. S., Song, M., Gray, P. P., Enhanced CHO Cell-Based Transient Gene Expression with the Epi-CHO Expression System. Mol. Biotechnol. 2011, 48, 109-115. [102] Codamo, J., Hou, J. J. C., Hughes, B. S., Gray, P. P., Munro, T. P., Efficient mAb production in CHO cells incorporating PEI-mediated transfection, mild hypothermia and the co-expression of XBP-1. J. Chem. Technol. Biotechnol. 2011, 86, 923-934. [103] Castan, A., Schulz, P., Wenger, T., Fischer, S., Chapter 7 - Cell Line Development, in: Jagschies, G., Lindskog, E., Łącki, K., Galliher, P. (Eds.), Biopharmaceutical Processing, Elsevier 2018, pp. 131-146. [104] Lindskog, E. K., Chapter 31 - The Upstream Process: Principal Modes of Operation, in: Jagschies, G., Lindskog, E., Łącki, K., Galliher, P. (Eds.), Biopharmaceutical Processing, Elsevier 2018, pp. 625-635. [105] Fisher, A. C., Kamga, M.-H., Agarabi, C., Brorson, K., et al., The Current Scientific and Regulatory Landscape in Advancing Integrated Continuous Biopharmaceutical Manufacturing. Trends Biotechnol. 2018. [106] Rathore, A. S., Agarwal, H., Sharma, A. K., Pathak, M., Muthukumar, S., Continuous Processing for Production of Biopharmaceuticals. Prep. Biochem. Biotechnol. 2015, 45, 836-849. [107] Shukla, A. A., Suda, E., Chapter 3 - Harvest and recovery of monoclonal antibodies: cell removal and clarification, in: Gottschalk, U. (Ed.), Process Scale Purification of Antibodies, John Wiley & Sons, Inc. 2017, pp. 55-79. [108] Thömmes, J., Twyman, R. M., Gottschalk, U., Chapter 10 - Alternatives to packed-bed chromatography for antibody extraction and purification, in: Gottschalk, U. (Ed.), Process Scale Purification of Antibodies, John Wiley & Sons 2017, pp. 215- 231. [109] Glynn, J., Chapter 11 - Process-scale precipitation of impurities in mammalian cell culture broth, in: Gottschalk, U. (Ed.), Process Scale Purification of Antibodies, John Wiley & Sons, Inc. 2017, pp. 233-246. [110] Singh, N., Arunkumar, A., Chollangi, S., Tan, Z. G., et al., Clarification technologies for monoclonal antibody manufacturing processes: Current state and future perspectives. Biotechnol. Bioeng. 2016, 113, 698-716. [111] Singh, N., Chollangi, S., Chapter 4 - Next-generation clarification technologies for the downstream processing of antibodies, in: Gottschalk, U. (Ed.), Process Scale Purification of Antibodies, John Wiley & Sons, Inc. 2017, pp. 81-112. [112] Kelley, B., Chapter 1 - Downstream processing of monoclonal antibodies: current practices and future opportunities, in: Gottschalk, U. (Ed.), Process Scale Purification of Antibodies, John Wiley & Sons, Inc. 2017, pp. 1-21.

172

[113] Danielsson, Å., Chapter 17 - Affinity Chromatography, in: Jagschies, G., Lindskog, E., Łącki, K., Galliher, P. (Eds.), Biopharmaceutical Processing, Elsevier 2018, pp. 367-378. [114] Kinna, A., Tolner, B., Rota, E. M., Titchener-Hooker, N., et al., IMAC capture of recombinant protein from unclarified mammalian cell feed streams. Biotechnol. Bioeng. 2016, 113, 130-140. [115] Ghose, S., Jin, M., Liu, J., Hickey, J., Lee, S., Chapter 14 - Integrated polishing steps for monoclonal antibody purification, in: Gottschalk, U. (Ed.), Process Scale Purification of Antibodies, John Wiley & Sons, Inc. 2017, pp. 303-323. [116] Joshi, V., Shivach, T., Kumar, V., Yadav, N., Rathore, A., Avoiding antibody aggregation during processing: Establishing hold times. Biotechnol. J. 2014, 9, 1195- 1205. [117] Liderfelt, J., Royce, J., Chapter 23 - Filtration Methods for Use in Purification Processes (Concentration and Buffer Exchange), in: Jagschies, G., Lindskog, E., Łącki, K., Galliher, P. (Eds.), Biopharmaceutical Processing, Elsevier 2018, pp. 441- 453. [118] Yu, M., Wu, J., Shi, J., Farokhzad, O. C., Nanotechnology for protein delivery: Overview and perspectives. J. Controlled Release 2016, 240, 24-37. [119] Garidel, P., Kuhn, A. B., Schäfer, L. V., Karow-Zwick, A. R., Blech, M., High- concentration protein formulations: How high is high? Eur. J. Pharm. Biopharm. 2017, 119, 353-360. [120] Kemter, K., Altrichter, J., Derwand, R., Kriehuber, T., et al., Amino Acid- Based Advanced Liquid Formulation Development for Highly Concentrated Therapeutic Antibodies Balances Physical and Chemical Stability and Low Viscosity. Biotechnol. J. 2018, 13. [121] Mandal, A., Pal, D., Agrahari, V., Trinh, H. M., et al., Ocular delivery of proteins and peptides: Challenges and novel formulation approaches. Adv. Drug Del. Rev. 2018, 126, 67-95. [122] Emami, F., Vatanara, A., Park, E. J., Na, D. H., Drying technologies for the stability and bioavailability of biopharmaceuticals. Pharmaceutics 2018, 10. [123] Izutsu, K. I., Adv. Exp. Med. Biol., Springer New York LLC 2018, pp. 371-383. [124] Schermeyer, M. T., Wöll, A. K., Kokke, B., Eppink, M., Hubbuch, J., Characterization of highly concentrated antibody solution - A toolbox for the description of protein long-term solution stability. mAbs 2017, 9, 1169-1185. [125] Dalziel, M., Beers, S. A., Cragg, M. S., Crispin, M., Through the barricades: Overcoming the barriers to effective antibody-based cancer therapeutics. Glycobiology 2018, 28, 697-712. [126] Liu, L., Pharmacokinetics of monoclonal antibodies and Fc-fusion proteins. Protein & Cell 2018, 9, 15-32. [127] Shi, J., Kantoff, P. W., Wooster, R., Farokhzad, O. C., Cancer nanomedicine: progress, challenges and opportunities. Nat. Rev. Cancer 2016, 17, 20. [128] Lambert, J. M., Berkenblit, A., Annu. Rev. Med., Annual Reviews Inc. 2018, pp. 191-207.

173

[129] Lambert, J. M., Morris, C. Q., Antibody–Drug Conjugates (ADCs) for Personalized Treatment of Solid Tumors: A Review. Adv. Ther. 2017, 34, 1015-1035. [130] Schweizer, D., Serno, T., Goepferich, A., Controlled release of therapeutic antibody formats. Eur. J. Pharm. Biopharm. 2014, 88, 291-309. [131] Ickenstein, L. M., Garidel, P., Hydrogel formulations for biologicals: Current spotlight from a commercial perspective. Ther. Deliv. 2018, 9, 221-230. [132] Levin, D., Golding, B., Strome, S. E., Sauna, Z. E., Fc fusion as a platform technology: potential for modulating immunogenicity. Trends Biotechnol. 2015, 33, 27-34. [133] Saxena, A., Wu, D., Advances in Therapeutic Fc Engineering – Modulation of IgG-Associated Effector Functions and Serum Half-life. Front. Immunol. 2016, 7. [134] Jefferis, R., Glycosylation of Antibody Molecules, Handbook of Therapeutic Antibodies: Second Edition, Wiley Blackwell 2014, pp. 171-200. [135] Zheng, K., Bantog, C., Bayer, R., The impact of glycosylation on monoclonal antibody conformation and stability. mAbs 2011, 3, 568-576. [136] Booth, B. J., Ramakrishnan, B., Narayan, K., Wollacott, A. M., et al., Extending human IgG half-life using structure-guided design. mAbs 2018, 10, 1098- 1110. [137] Kellner, C., Otte, A., Cappuzzello, E., Klausz, K., Peipp, M., Modulating Cytotoxic Effector Functions by Fc Engineering to Improve Cancer Therapy. Transfus. Med. Hemoth. 2017, 44, 327-336. [138] Wang, Q., Chen, Y., Pelletier, M., Cvitkovic, R., et al., Enhancement of antibody functions through Fc multiplications. mAbs 2017, 9, 393-403. [139] Rabia, L. A., Desai, A. A., Jhajj, H. S., Tessier, P. M., Understanding and overcoming trade-offs between antibody affinity, specificity, stability and solubility. Biochem. Eng. J. 2018, 137, 365-374. [140] van de Bovenkamp, F. S., Derksen, N. I. L., van Breemen, M. J., de Taeye, S. W., et al., Variable Domain N-Linked Glycans Acquired During Antigen-Specific Immune Responses Can Contribute to Immunoglobulin G Antibody Stability. Front. Immunol. 2018, 9. [141] Kuhn, A. B., Kube, S., Karow-Zwick, A. R., Seeliger, D., et al., Improved Solution-State Properties of Monoclonal Antibodies by Targeted Mutations. J. Phys. Chem. B 2017, 121, 10818-10827. [142] Singh, S. N., Yadav, S., Shire, S. J., Kalonia, D. S., Dipole-dipole interaction in antibody solutions: Correlation with viscosity behavior at high concentration. Pharm. Res. 2014, 31, 2549-2558. [143] Ionescu, R. M., Vlasak, J., Price, C., Kirchmeier, M., Contribution of Variable Domains to the Stability of Humanized IgG1 Monoclonal Antibodies. J. Pharm. Sci. 2008, 97, 1414-1426. [144] Lefranc, M. P., Ehrenmann, F., Kossida, S., Giudicelli, V., Duroux, P., Methods Mol. Biol., Humana Press Inc. 2018, pp. 35-69.

174

[145] Martin, A. C. R., Allen, J., Bioinformatics Tools for Analysis of Antibodies, Handbook of Therapeutic Antibodies: Second Edition, Wiley Blackwell 2014, pp. 201-228. [146] Wishart, D. S., Knox, C., Guo, A. C., Shrivastava, S., et al., DrugBank: a comprehensive resource for in silico drug discovery and exploration. Nucleic Acids Res. 2006, 34, D668-D672. [147] Śledź, P., Caflisch, A., Protein structure-based drug design: from docking to molecular dynamics. Curr. Opin. Struct. Biol. 2018, 48, 93-102. [148] Pujols, J., Peña-Díaz, S., Ventura, S., AGGRESCAN3D: Toward the prediction of the aggregation propensities of protein structures, Methods Mol. Biol., Humana Press Inc. 2018, pp. 427-443. [149] Pandya, A., Howard, M. J., Zloh, M., Dalby, P. A., An evaluation of the potential of nmr spectroscopy and computational modelling methods to inform biopharmaceutical formulations. Pharmaceutics 2018, 10, 1.-24. [150] Rathore, D., Faustino, A., Schiel, J., Pang, E., et al., The role of mass spectrometry in the characterization of biologic protein products. Expert Rev. Proteomic. 2018, 15, 431-449. [151] Wang, X., An, Z., Luo, W., Xia, N., Zhao, Q., Molecular and functional analysis of monoclonal antibodies in support of biologics development. Protein Cell 2018, 9, 74-85. [152] Baker Edward, N., Crystallography and the development of therapeutic medicines. IUCrJ 2018, 5, 118-119. [153] Vénien-Bryan, C., Li, Z., Vuillard, L., Boutin, J. A., Cryo-electron microscopy and X-ray crystallography: Complementary approaches to structural biology and drug discovery. Acta Crystallogr. Sect. F Struct. Biol. Commun. 2017, 73, 174-183. [154] Brader, M. L., Baker, E. N., Dunn, M. F., Laue, T. M., Carpenter, J. F., Using X-Ray Crystallography to Simplify and Accelerate Biologics Drug Development. J. Pharm. Sci. 2017, 106, 477-494. [155] Nero, T. L., Parker, M. W., Morton, C. J., Protein structure and computational drug discovery. Biochem. Soc. Trans. 2018, 46, 1367-1379. [156] Kumar, S., Plotnikov, N. V., Rouse, J. C., Singh, S. K., Biopharmaceutical Informatics: supporting biologic drug development via molecular modelling and informatics. J. Pharm. Pharmacol. 2018, 70, 595-608. [157] Westbrook, J. D., Burley, S. K., How Structural Biologists and the Protein Data Bank Contributed to Recent FDA New Drug Approvals. Structure 2018. [158] Kuroda, D., Tsumoto, K., Methods Mol. Biol., Humana Press Inc. 2018, pp. 15- 34. [159] Fischman, S., Ofran, Y., Computational design of antibodies. Curr. Opin. Struct. Biol. 2018, 51, 156-162. [160] Adolf-Bryfogle, J., Kalyuzhniy, O., Kubitz, M., Weitzner, B. D., et al., RosettaAntibodyDesign (RAbD): A general framework for computational antibody design. PLoS Comp. Biol. 2018, 14.

175

[161] Branco, R. J., Dias, A. M., Roque, A. C., Understanding the molecular recognition between antibody fragments and protein A biomimetic ligand. J. Chromatogr. A 2012, 1244, 106-115. [162] Huang, B., Liu, F. F., Dong, X. Y., Sun, Y., Molecular mechanism of the affinity interactions between protein A and human immunoglobulin G1 revealed by molecular simulations. J. Phys. Chem. B 2011, 115, 4168-4176. [163] van der Kant, R., van Durme, J., Rousseau, F., Schymkowitz, J., Methods Mol. Biol., Humana Press Inc. 2019, pp. 317-333. [164] Gil-Garcia, M., Banó-Polo, M., Varejao, N., Jamroz, M., et al., Combining Structural Aggregation Propensity and Stability Predictions to Redesign Protein Solubility. Mol. Pharm. 2018, 15, 3846-3859. [165] van der Kant, R., Karow-Zwick, A. R., Van Durme, J., Blech, M., et al., Prediction and Reduction of the Aggregation of Monoclonal Antibodies. J. Mol. Biol. 2017, 429, 1244-1261. [166] Lerch, T. F., Sharpe, P., Mayclin, S. J., Edwards, T. E., et al., Infliximab crystal structures reveal insights into self-association. mAbs 2017, 9, 874-883. [167] Dobson, C. L., Devine, P. W. A., Phillips, J. J., Higazi, D. R., et al., Engineering the surface properties of a human monoclonal antibody prevents self- association and rapid clearance in vivo. Sci. Rep. 2016, 6. [168] Courtois, F., Agrawal, N. J., Lauer, T. M., Trout, B. L., Rational design of therapeutic mAbs against aggregation through protein engineering and incorporation of glycosylation motifs applied to bevacizumab. mAbs 2016, 8, 99-112. [169] Courtois, F., Schneider, C. P., Agrawal, N. J., Trout, B. L., Rational Design of Biobetters with Enhanced Stability. J. Pharm. Sci. 2015, 104, 2433-2440. [170] Seeliger, D., Schulz, P., Litzenburger, T., Spitz, J., et al., Boosting antibody developability through rational sequence optimization. mAbs 2015, 7, 505-515. [171] Majumder, S., Jones, M. T., Kimmel, M., Alphonse Ignatius, A., Probing Conformational Diversity of Fc Domains in Aggregation-Prone Monoclonal Antibodies. Pharm. Res. 2018, 35. [172] Nakamura, H., Oda-Ueda, N., Ueda, T., Ohkuri, T., Introduction of a glycosylation site in the constant region decreases the aggregation of adalimumab Fab. Biochem. Biophys. Res. Commun. 2018, 503, 752-756. [173] Pepinsky, R. B., Silvian, L., Berkowitz, S. A., Farrington, G., et al., Improving the solubility of anti‐LINGO‐1 monoclonal antibody Li33 by isotype switching and targeted mutagenesis. Protein Sci. 2010, 19, 954-966. [174] Wu, S.-J., Luo, J., O'Neil, K. T., Kang, J., et al., Structure-based engineering of a monoclonal antibody for improved solubility. Protein Eng. Des. Sel. 2010, 23, 643- 651. [175] InvivoGen, Technical Data Sheet 05G06-MT pVITRO1-hygro-mcs v05G06- MT, pp. 1-5. [176] InvivoGen, Technical Data Sheet 05C23-MT pVITRO1-blasti-mcs v05C23- MT, pp. 1-5.

176

[177] InvivoGen, Technical Data Sheet 05C23-MT pVITRO1-neo-mcs v05C23-MT, pp. 1-5. [178] Dodev, T. S., Karagiannis, P., Gilbert, A. E., Josephs, D. H., et al., A tool kit for rapid cloning and expression of recombinant antibodies. Sci. Rep. 2014, 4, 5885. [179] Jones, M. L., Seldon, T., Smede, M., Linville, A., et al., A method for rapid, ligation-independent reformatting of recombinant monoclonal antibodies. J. Immunol. Methods 2010, 354, 85-90. [180] Trent Phillip Munro, M. L. J., Matthew George Smede, Acyte Biotech Pty Ltd, USA 2012, p. 26. [181] Acyte, mAbXpress Vectors Users Guide, National Biologics Facility, AIBN, pp. 1-9. [182] Genlantis, Technical Data Sheet MV032206 gWiz Blank Mammalian Expression Vector, Gene Therapy Systems, p. 1. [183] Nelson, A. L., Reichert, J. M., Development trends for therapeutic antibody fragments. Nat. Biotechnol. 2009, 27, 331-337. [184] Drake, P. M., Rabuka, D., An emerging playbook for antibody-drug conjugates: lessons from the laboratory and clinic suggest a strategy for improving efficacy and safety. Curr. Opin. Chem. Biol. 2015, 28, 174-180. [185] Kontermann, R. E., Brinkmann, U., Bispecific antibodies. Drug Discov. Today 2015, 20, 838-847. [186] Beck, A., Wurch, T., Bailly, C., Corvaia, N., Strategies and challenges for the next generation of therapeutic antibodies. Nat. Rev. Immunol. 2010, 10, 345-352. [187] Food and Drug Administration (FDA), in: U.S. Department of Health and Human Services, Center for Drug Evaluation and Research (CDER), Center for Biologics Evaluation and Research (CBER) (Eds.), Quality Considerations in Demonstrating Biosimilarity of a Therapeutic Protein Product to a Reference Product; Guidance for Industry, Silver Spring (MD) 2015, p. 22. [188] European Medicines Agency (EMA), in: Committee for Medicinal Products for Human Use (CHMP) (Ed.), Guideline on similar biological medicinal products containing biotechnology-derived proteins as active substance: non-clinical and clinical issues, London (UK) 2014, p. 13. [189] Lopez-Morales, C. A., Miranda-Hernandez, M. P., Juarez-Bayardo, L. C., Ramirez-Ibanez, N. D., et al., Physicochemical and biological characterization of a biosimilar trastuzumab. Biomed Res Int 2015, 2015, 427235. [190] Jung, S. K., Lee, K. H., Jeon, J. W., Lee, J. W., et al., Physicochemical characterization of Remsima®. mAbs 2014, 6, 1163-1177. [191] Visser, J., Feuerstein, I., Stangler, T., Schmiederer, T., et al., Physicochemical and functional comparability between the proposed biosimilar rituximab GP2013 and originator rituximab. Biodrugs 2013, 27, 495-507. [192] Li, J., Menzel, C., Meier, D., Zhang, C., et al., A comparative study of different vector designs for the mammalian expression of recombinant IgG antibodies. J. Immunol. Methods 2007, 318, 113-124.

177

[193] Fitzgerald, K. D., Semler, B. L., Bridging IRES elements in mRNAs to the eukaryotic translation apparatus. Biochim. Biophys. Acta 2009, 1789, 518-528. [194] Ho, S. C., Bardor, M., Feng, H., Mariati, et al., IRES-mediated Tricistronic vectors for enhancing generation of high monoclonal antibody expressing CHO cell lines. J. Biotechnol. 2012, 157, 130-139. [195] Walsh, G., Post-translational modifications of protein biopharmaceuticals. Drug Discov. Today 2010, 15, 773-780. [196] Backliwal, G., Hildinger, M., Chenuet, S., Wulhfard, S., et al., Rational vector design and multi-pathway modulation of HEK 293E cells yield recombinant antibody titers exceeding 1 g/l by transient transfection under serum-free conditions. Nucleic Acids Res. 2008, 36, e96. [197] Bebbington, C. R., Renner, G., Thomson, S., King, D., et al., High-level expression of a recombinant antibody from myeloma cells using a glutamine synthetase gene as an amplifiable selectable marker. Biotechnology. (N. Y.) 1992, 10, 169-175. [198] Kaufman, R. J., Wasley, L. C., Spiliotes, A. J., Gossels, S. D., et al., Coamplification and coexpression of human tissue-type plasminogen activator and murine dihydrofolate reductase sequences in Chinese hamster ovary cells. Mol. Cell. Biol. 1985, 5, 1750-1759. [199] Dalton, A. C., Barton, W. A., Over-expression of secreted proteins from mammalian cell lines. Protein Science : A Publication of the Protein Society 2014, 23, 517-525. [200] Pham, P. L., Perret, S., Cass, B., Carpentier, E., et al., Transient gene expression in HEK293 cells: peptone addition posttransfection improves recombinant protein synthesis. Biotechnol. Bioeng. 2005, 90, 332-344. [201] Pham, P. L., Perret, S., Doan, H. C., Cass, B., et al., Large-scale transient transfection of serum-free suspension-growing HEK293 EBNA1 cells: peptone additives improve cell growth and transfection efficiency. Biotechnol. Bioeng. 2003, 84, 332-342. [202] Yang, W. C., Lu, J., Nguyen, N. B., Zhang, A., et al., Addition of valproic acid to CHO cell fed-batch cultures improves monoclonal antibody titers. Mol. Biotechnol. 2014, 56, 421-428. [203] Jiang, Z., Sharfstein, S. T., Sodium butyrate stimulates monoclonal antibody over-expression in CHO cells by improving gene accessibility. Biotechnol. Bioeng. 2008, 100, 189-194. [204] Jordan, M., Wurm, F., Transfection of adherent and suspended cells by calcium phosphate. Methods 2004, 33, 136-143. [205] Dakshinamurthy, P., Mukunda, P., Prasad Kodaganti, B., Shenoy, B. R., et al., Charge variant analysis of proposed biosimilar to Trastuzumab. Biologicals 2017, 46, 46-56. [206] Lopez-Morales, C., P Miranda-Hernández, M., Bayardo, C., D Ramírez-Ibáñez, N., et al., Physicochemical and Biological Characterization of a Biosimilar Trastuzumab. BioMed Res. Int. 2015, 2015, 10.

178

[207] Lee, J., Kang, H. A., Bae, J. S., Kim, K. D., et al., Evaluation of analytical similarity between trastuzumab biosimilar CT-P6 and reference product using statistical analyses. mAbs 2018, 10, 547-571. [208] Kim, S., Song, J., Park, S., Ham, S., et al., Drifts in ADCC-related quality attributes of Herceptin®: Impact on development of a trastuzumab biosimilar. mAbs 2017, 9, 704-714. [209] Park, B. W., Zhang, H. T., Wu, C., Berezov, A., et al., Rationally designed anti-HER2/neu peptide mimetic disables P185HER2/neu tyrosine kinases in vitro and in vivo. Nat. Biotechnol. 2000, 18, 194-198. [210] Wittrup, K. D., Thurber, G. M., Schmidt, M. M., Rhoden, J. J., Chapter ten - Practical Theoretic Guidance for the Design of Tumor-Targeting Agents; Methods in Enzymology, in: Wittrup, K. D., Verdine, G. L. (Eds.), Methods Enzymol., Academic Press 2012, pp. 255-268. [211] Rudnick, S. I., Lou, J., Shaller, C. C., Tang, Y., et al., Influence of affinity and antigen internalization on the uptake and penetration of Anti-HER2 antibodies in solid tumors. Cancer Res. 2011, 71, 2250-2259. [212] Zahnd, C., Kawe, M., Stumpp, M. T., de Pasquale, C., et al., Efficient tumor targeting with high-affinity designed ankyrin repeat proteins: effects of affinity and molecular size. Cancer Res. 2010, 70, 1595-1605. [213] Adams, G. P., Schier, R., McCall, A. M., Simmons, H. H., et al., High affinity restricts the localization and tumor penetration of single-chain fv antibody molecules. Cancer Res. 2001, 61, 4750-4755. [214] Thurber, G. M., Schmidt, M. M., Wittrup, K. D., Antibody tumor penetration: Transport opposed by systemic and antigen-mediated clearance. Adv. Drug Del. Rev. 2008, 60, 1421-1434. [215] Antikainen, N. M., Martin, S. F., Altering protein specificity: techniques and applications. Bioorg. Med. Chem. 2005, 13. [216] Edelheit, O., Hanukoglu, A., Hanukoglu, I., Simple and efficient site-directed mutagenesis using two single-primer reactions in parallel to generate mutants for protein structure-function studies. BMC Biotechnol. 2009, 9, 61. [217] Li, B., Zhao, L., Wang, C., Guo, H., et al., The Protein-Protein Interface Evolution Acts in a Similar Way to Antibody Affinity Maturation. J. Biol. Chem. 2010, 285, 3865-3871. [218] Lorenz, T. C., Polymerase chain reaction: basic protocol plus troubleshooting and optimization strategies. JoVE 2012, e3998-e3998. [219] Silva, D., Santos, G., Barroca, M., Collins, T., Inverse PCR for Point Mutation Introduction, in: Domingues, L. (Ed.), PCR: Methods and Protocols, Springer New York, New York, NY 2017, pp. 87-100. [220] van Pelt-Verkuil, E., van Belkum, A., Hays, J. P., PCR Primers, Principles and Technical Aspects of PCR Amplification, Springer Netherlands, Dordrecht 2008, pp. 63-90. [221] New England BioLabs, Phusion® High-Fidelity PCR Kit Instruction Manual v1.4, pp. 1-16.

179

[222] Hunter, S. A., Cochran, J. R., Cell-Binding Assays for Determining the Affinity of Protein-Protein Interactions: Technologies and Considerations. Methods Enzymol. 2016, 580, 21-44. [223] Liao-Chan, S., Daine-Matsuoka, B., Heald, N., Wong, T., et al., Quantitative assessment of antibody internalization with novel monoclonal antibodies against Alexa fluorophores. PLoS One 2015, 10, e0124708-e0124708. [224] Yang, Y., Zhao, J., Geng, S., Hou, C., et al., Improving Trastuzumab's Stability Profile by Removing the Two Degradation Hotspots. J. Pharm. Sci. 2015, 104, 1960- 1970. [225] Gervais, D., Protein deamidation in biopharmaceutical manufacture: understanding, control and impact. J. Chem. Technol. Biotechnol. 2016, 91, 569-575. [226] Schasfoort, R. B. M., Handbook of Surface Plasmon Resonance, Royal Society of Chemistry 2017. [227] Hayes, J. M., Frostell, A., Cosgrave, E. F. J., Struwe, W. B., et al., Fc Gamma Receptor Glycosylation Modulates the Binding of IgG Glycoforms: A Requirement for Stable Antibody Interactions. J. Proteome Res. 2014, 13, 5471-5485. [228] Karlsson, R., Pol, E., Frostell, Å., Comparison of surface plasmon resonance binding curves for characterization of protein interactions and analysis of screening data. Anal. Biochem. 2016, 502, 53-63. [229] Kamat, V., Rafique, A., Extending the throughput of Biacore 4000 biosensor to accelerate kinetic analysis of antibody-antigen interaction. Anal. Biochem. 2017, 530, 75-86. [230] Geuijen, K. P. M., Oppers‐Tiemissen, C., Egging, D. F., Simons, P. J., et al., Rapid screening of IgG quality attributes – effects on Fc receptor binding. FEBS Open Bio 2017, 7, 1557-1574. [231] Hayes, J. M., Frostell, A., Karlsson, R., Muller, S., et al., Identification of Fc Gamma receptor glycoforms that produce differential binding kinetics for rituximab. Mol. Cell. Proteomics 2017, 16, 1770-1788. [232] Wang, W., Roberts, C. J., Protein aggregation – Mechanisms, detection, and control. Int. J. Pharm. 2018, 550, 251-268. [233] Shah, M., Commentary: New perspectives on protein aggregation during Biopharmaceutical development. Int. J. Pharm. 2018, 552, 1-6. [234] Uchiyama, S., Liquid formulation for antibody drugs. Biochim. Biophys. Acta Proteins Proteomics 2014, 1844, 2041-2052. [235] Sauerborn, M., The Immunogenicity of Therapeutic Antibodies, Handbook of Therapeutic Antibodies: Second Edition, Wiley Blackwell 2014, pp. 665-680. [236] Beck, A., Wagner-Rousset, E., Ayoub, D., Van Dorsselaer, A., Sanglier- Cianférani, S., Characterization of Therapeutic Antibodies and Related Products. Anal. Chem. 2013, 85, 715-736. [237] Hofmann, M., Gieseler, H., Predictive Screening Tools Used in High- Concentration Protein Formulation Development. J. Pharm. Sci. 2018, 107, 772-777.

180

[238] Pouya, M. A., Daneshmand, B., Aghababaie, S., Faghihi, H., Vatanara, A., Spray-Freeze Drying: a Suitable Method for Aerosol Delivery of Antibodies in the Presence of Trehalose and Cyclodextrins. AAPS PharmSciTech 2018, 19, 2247-2254. [239] Galm, L., Amrhein, S., Hubbuch, J., Predictive approach for protein aggregation: Correlation of protein surface characteristics and conformational flexibility to protein aggregation propensity. Biotechnol. Bioeng. 2017, 114, 1170- 1183. [240] Baumann, P., Schermeyer, M. T., Burghardt, H., Dürr, C., et al., Prediction and characterization of the stability enhancing effect of the Cherry-Tag™ in highly concentrated protein solutions by complex rheological measurements and MD simulations. Int. J. Pharm. 2017, 531, 360-371. [241] Bauer, K. C., Hämmerling, F., Kittelmann, J., Dürr, C., et al., Influence of structure properties on protein–protein interactions—QSAR modeling of changes in diffusion coefficients. Biotechnol. Bioeng. 2017, 114, 821-831. [242] Chow, C. K., Allan, B. W., Chai, Q., Atwell, S., Lu, J., Therapeutic Antibody Engineering to Improve Viscosity and Phase Separation Guided by Crystal Structure. Mol. Pharm. 2016, 13, 915-923. [243] Arzenšek, D., Kuzman, D., Podgornik, R., Hofmeister Effects in Monoclonal Antibody Solution Interactions. J. Phys. Chem. B 2015, 119, 10375-10389. [244] Roberts, D., Keeling, R., Tracka, M., Van Der Walle, C. F., et al., The role of electrostatics in protein-protein interactions of a monoclonal antibody. Mol. Pharm. 2014, 11, 2475-2489. [245] Zhang, C., Samad, M., Yu, H., Chakroun, N., et al., Computational Design to Reduce Conformational Flexibility and Aggregation Rates of an Antibody Fab Fragment. Mol. Pharm. 2018, 15, 3079-3092. [246] Chennamsetty, N., Voynov, V., Kayser, V., Helk, B., Trout, B. L., Prediction of Aggregation Prone Regions of Therapeutic Proteins. The Journal of Physical Chemistry B 2010, 114, 6614-6624. [247] Kayser, V., Chennamsetty, N., Voynov, V., Forrer, K., et al., Glycosylation influences on the aggregation propensity of therapeutic monoclonal antibodies. Biotechnol. J. 2011, 6, 38-44. [248] Pepinsky, R. B., Laura, S., A., B. S., Graham, F., et al., Improving the solubility of anti-LINGO-1 monoclonal antibody Li33 by isotype switching and targeted mutagenesis. Protein Sci. 2010, 19, 954-966. [249] Pincetic, A., Bournazos, S., DiLillo, D. J., Maamary, J., et al., Type I and type II Fc receptors regulate innate and adaptive immunity. Nat. Immunol. 2014, 15, 707. [250] Kayser, V., Chennamsetty, N., Voynov, V., Helk, B., et al., Evaluation of a Non-Arrhenius Model for Therapeutic Monoclonal Antibody Aggregation. J. Pharm. Sci. 2011, 100, 2526-2542. [251] Temel, D. B., Landsman, P., Brader, M. L., Methods Enzymol., Academic Press Inc. 2016, pp. 359-389.

181

[252] Nemergut, M., Žoldák, G., Schaefer, J. V., Kast, F., et al., Analysis of IgG kinetic stability by differential scanning calorimetry, probe fluorescence and light scattering. Protein Sci. 2017, 26, 2229-2239. [253] Kangwa, M., Yelemane, V., Ponnurangam, A., Fernández-Lahore, M., An engineered Staphylococcal Protein A based ligand: Production, characterization and potential application for the capture of Immunoglobulin and Fc-fusion proteins. Protein Expression Purif. 2019, 155, 27-34. [254] Levy, N. E., Valente, K. N., Choe, L. H., Lee, K. H., Lenhoff, A. M., Identification and characterization of host cell protein product-associated impurities in monoclonal antibody bioprocessing. Biotechnol. Bioeng. 2014, 111, 904-912. [255] Ecker, D. M., Jones, S. D., Levine, H. L., The therapeutic monoclonal antibody market. mAbs 2015, 7, 9-14. [256] Elgundi, Z., Reslan, M., Cruz, E., Sifniotis, V., Kayser, V., The state-of-play and future of antibody therapeutics. Adv. Drug Delivery Rev. 2017, 122, 2-19. [257] Fleischman, M. L., Chung, J., Paul, E. P., Lewus, R. A., Shipping-Induced Aggregation in Therapeutic Antibodies: Utilization of a Scale-Down Model to Assess Degradation in Monoclonal Antibodies. J. Pharm. Sci. 2017, 106, 994-1000. [258] Beck, A., Goetsch, L., Dumontet, C., Corvaïa, N., Strategies and challenges for the next generation of antibody–drug conjugates. Nat. Rev. Drug Discov. 2017, 16, 315. [259] Bessa, J., Boeckle, S., Beck, H., Buckel, T., et al., The Immunogenicity of Antibody Aggregates in a Novel Transgenic Mouse Model. Pharm. Res. 2015, 32, 2344-2359. [260] Ahmadi, M., Bryson, C. J., Cloake, E. A., Welch, K., et al., Small Amounts of Sub-Visible Aggregates Enhance the Immunogenic Potential of Monoclonal Antibody Therapeutics. Pharm. Res. 2015, 32, 1383-1394. [261] Reslan, M., Demir, Y. K., Trout, B. L., Chan, H.-K., Kayser, V., Lack of a synergistic effect of arginine–glutamic acid on the physical stability of spray-dried bovine serum albumin. Pharm. Dev. Technol. 2016, 1-7. [262] Zhang, J., Frey, V., Corcoran, M., Zhang-van Enk, J., Subramony, J. A., Influence of Arginine Salts on the Thermal Stability and Aggregation Kinetics of Monoclonal Antibody: Dominant Role of Anions. Mol. Pharm. 2016, 13, 3362-3369. [263] Respaud, R., Marchand, D., Parent, C., Pelat, T., et al., Effect of formulation on the stability and aerosol performance of a nebulized antibody. mAbs 2014, 6, 1347- 1355. [264] Reslan, M., Kayser, V., The effect of deuterium oxide on the conformational stability and aggregation of bovine serum albumin. Pharm. Dev. Technol. 2016, 1-7. [265] Sridhar, B. V., Janczy, J. R., Hatlevik, Ø., Wolfson, G., et al., Thermal Stabilization of Biologics with Photoresponsive Hydrogels. Biomacromolecules 2018, 19, 740-747. [266] Lee, J., Ko, J. H., Mansfield, K. M., Nauka, P. C., et al., Glucose-Responsive Trehalose Hydrogel for Insulin Stabilization and Delivery. Macromol. Biosci. 2018, 18, 1700372.

182

[267] Bansal, R., Dhawan, S., Chattopadhyay, S., Maurya, G. P., et al., Peptide Dendrons as Thermal-Stability Amplifiers for Immunoglobulin G1 Monoclonal Antibody Biotherapeutics. Bioconj. Chem. 2017, 28, 2549-2559. [268] Reslan, M., Kayser, V., Ionic liquids as biocompatible stabilizers of proteins. Biophys. Rev. 2018, 10, 781-793. [269] Chennamsetty, N., Voynov, V., Kayser, V., Helk, B., Trout, B. L., Design of therapeutic proteins with enhanced stability. Proc. Natl. Acad. Sci. USA 2009, 106, 11937-11942. [270] Kulshreshtha, S., Chaudhary, V., Goswami, G. K., Mathur, N., Computational approaches for predicting mutant protein stability. J. Comput.-Aided Mol. Des. 2016, 30, 401-412. [271] Zambrano, R., Jamroz, M., Szczasiuk, A., Pujols, J., et al., AGGRESCAN3D (A3D): server for prediction of aggregation properties of protein structures. Nucleic Acids Res. 2015, 43, W306-W313. [272] Kuyucak, S., Kayser, V., Biobetters From an Integrated Computational/Experimental Approach. Comput. Struc. Biotechnol. J. 2017, 15, 138- 145. [273] Nakamura, H., Oda-Ueda, N., Ueda, T., Ohkuri, T., Introduction of a glycosylation site in the constant region decreases the aggregation of adalimumab Fab. Biochem. Biophys. Res. Commun. 2018. [274] Kaymakcalan, Z., Sakorafas, P., Bose, S., Scesney, S., et al., Comparisons of affinities, avidities, and complement activation of adalimumab, infliximab, and etanercept in binding to soluble and membrane tumor necrosis factor. Clin. Immunol. 2009, 131, 308-316. [275] Magnenat, L., Palmese, A., Fremaux, C., D'Amici, F., et al., Demonstration of physicochemical and functional similarity between the proposed biosimilar adalimumab MSB11022 and Humira®. mAbs 2017, 9, 127-139.

183

Appendix

1 E. coli Protocols 1.1 Media and Culture Conditions In LB broth: 37 °C, 250 rpm O/N, on LB agar: 37 °C O/N. O/N (broadly means) 16 – 22 hrs Inoculate media using a sterile 10 µL inoculation loop. From a plate: Pick a single colony isolate. From liquid culture: Pick a loopful of liquid culture. To a plate: Over the surface of the agar plate, rub the inoculated loop in a small region towards the circumference. To one corner of the rubbed area, rub out in a separate region using the flipped side of the loop (first dilution). Making sure to only contact the edge of the prior rubbed region, rub 2 further dilutions over the plate with a fresh loop. The final dilution to be a broad squiggle that takes the majority of the space of the plate, where single colonies can be isolated. To liquid media: Stir the loopful of liquid culture into the fresh media. For larger cultures, inoculate with larger volumes of liquid culture with a pipette (100 µL – 1 mL). Small scale (miniprep extraction): 15 mL culture tube containing 3 – 5 mL LB broth Medium scale (midiprep extraction): 50 mL centrifuge tube containing 15 – 35 mL LB broth Large Scale (maxiprep extraction): 1L baffled culture flask containing 150 – 350 mL LB broth Luria-Bertani (LB) Broth

Deionised H2O containing:  10 g/L tryptone (casein peptone) (LP0042B Thermo Fisher Scientific)  5 g/L yeast extract (70161 Sigma-Aldrich)  10 g/L NaCl (S9888 Sigma-Aldrich) Combine ingredients to the desired volume, adjust to pH 7.5 and autoclave. Keep media sealed and refrigerated until use. For antibiotic media, wait for the autoclaved media to cool to < 50 °C before adding desired antibiotic. LB Agar LB broth containing 15 g/L agar (20767.232 VWR International) Prepare media as per LB Broth section.

184

Once autoclaved, allow the media to cool to a lukewarm temperature for plate pouring. Approximately 15 – 20 mL media can be poured on a conventional sized plate (9 - 10 cm diameter), enough to cover the entire surface, having 5 – 7 mm depth. Allow a minimum of 20 min for the poured plates to set at room temperature. Once set, cover and label the plates. Refrigerate until use. For antibiotic media, wait for the autoclaved media to cool to < 50 °C before adding desired antibiotic. Mix gently to limit bubble formation before pouring plates. SOC Media

Freshly prepare 100 x filter sterilised solutions in deionised H2O:  2 M glucose (G5767 Sigma-Alrich)  1 M MgCl2 (M4880 Sigma-Aldrich)  1 M MgSO4 (M2643 Sigma-Aldrich) SOC media preparation in deionised H2O containing:  20 g/L tryptone (casein peptone) (LP0042B Thermo Fisher Scientific)  5 g/L yeast extract (70161 Sigma-Aldrich)  0.5 g/L NaCl (S9888 Sigma-Aldrich)  20 mM glucose  10 mM MgCl2  10 mM MgSO4 Combine tryptone, yeast extract and NaCl to the desired volume and autoclave. Once autoclaved, allow the media to cool to < 50 °C before adding freshly filter sterilised 2 M glucose, 1 M MgCl2 and 1 M MgSO4 solutions. Prepare as 1.2 mL aliquots in microfuge tubes, keep media aliquots at -20 °C until use. 1.2 Glycerol Stock Preparation and Resuscitation Prepare 80% Glycerol solution (v/v) in deionised H2O and autoclave. Keep solution sealed and refrigerated until use. Glycerol stock preparation: To a labelled cryovial, add 200 µL of autoclaved 80% glycerol solution and 800 µL liquid culture. Mix gently but thoroughly via pipetting. Immediately transfer to -80 °C until use. Glycerol stock resuscitation: Take desired cryovial from -80 °C. Whist still frozen, use a sterile 10 µL inoculation loop to scrape some frozen stock from the vial and inoculate an appropriate agar plate as per 1.1. Immediately transfer the cryovial to -80 °C until further use; transfer the plate for O/N incubation as per 1.1. 1.3 Calcium Chloride Preparation of Competent E. coli This method was adapted from Xiaowei, Li (2013) “An improved calcium chloride method preparation and transformation of competent cells” African Journal of Biotechnology 9(50) 8549- 8554 This method was used on E. coli strain JM109, however this method can be applied to other E. coli strains. JM109 was a kind donation from Dr Fanfan Zhou (Pharmacy School, Faculty of Medicine and Health, The University of Sydney).

185

1.3.1 Reagents Prepare LB broth and LB agar (1.1).

Prepare 15% glycerol solution (v/v) in deionised H2O for immediate use. Freshly prepare buffers and autoclave:

Buffer 1: 75 mM CaCl2 (C4901 Sigma Aldrich) in deionised H2O.

Buffer 2: 85 mM CaCl2 (C4901 Sigma Aldrich) in 15% glycerol solution. Keep media and buffers sealed and refrigerated until use. 1.3.2 Calcium Chloride Method of Preparing In-house Competent JM109 Inoculate a LB agar plate with JM109 for O/N culture. Isolate a single colony from the plate to inoculate 3 mL LB broth for O/N culture. Inoculate 200 mL LB broth with 2 mL of liquid culture and place in incubation for monitoring. After 1.5 hrs incubation, measure optical density at 600 nm (OD600) from the culture every 15 min until the culture OD600 resides between 0.35 – 0.45. When the OD is reached, remove the culture from incubation and transfer into an ice bath for 30 min. Transfer the culture to chilled 50 mL centrifuge tubes and chill in an ice bath for 5 min. Pellet the cells (1,000 g, 5 min, 4 °C), resuspend the cells in 20 mL chilled Buffer 1 and chill on ice for 5 min. Pellet the cells (1,000 g, 5 min, 4 °C), resuspend the cells in 20 mL chilled Buffer 2 and chill on ice for 5 min. Pellet the cells (1,000 g, 5 min, 4 °C), resuspend the cells in fresh 4 mL chilled Buffer 2 and prepare as 100 µL aliquots in labelled 1.5 mL microcentrifuge tubes (keeping cells chilled on ice at all times). Immediately transfer aliquots to -80 °C for storage, aliquots to be thawed on ice for single use in heat-shock transformations. 1.4 Transformation of Competent Cells by Heat-Shock method In-house JM109 competent cells and commercial Stellar™ competent E. coli were transformed using the following heat-shock method. Briefly per competent cell reaction, incubate 50 µL competent E. coli with product on ice for 30 min. Heat-pulse the competent cell reaction in a 42 ºC waterbath for 45 sec, then return cells to ice for 2 min. Add 950 µL of pre-warmed (37 ºC) SOC media to the cells and incubate for 1 hr culture at 37 ºC, 250 rpm. After incubation, pellet the cells (12,000 g, 1 min) then resuspend the cells in 200 µl fresh SOC media. Transfer the entire cell suspension and spread onto an LB agar plate containing appropriate selection antibiotic, for O/N incubation. 2 DNA Protocols 2.1 DNA Extraction Kits All plasmid preparations, purified PCR products and digestion products were eluted in H2O. All DNA elutions were quantified using a NanoDrop ND-1000 spectrophotometer and stored at 2 – 8 °C for immediate use, or -20 °C for long term storage. For plasmid extraction, in the event that bacterial cell pellets could not be immediately prepared for extraction, the pellets were stored at -80 °C.

186

*Several different microcentrifuges were used depending on availability; top speed ranges from 21,000 g to 30,000 g. The Sigma Refrigerated Centrifuge 4K15 was used for the larger volume high speed refrigerated centrifugation, located in W312, Anderson Stuart Building. 2.1.1 GenElute™ Plasmid Miniprep Kit Plasmid minipreps were performed using the GenElute™ Plasmid Miniprep Kit (PLN70 Sigma- Aldrich) as per manufacturer’s instructions. Resuspension Solution and Wash Solution were prepared before first use of the kit as per manufacturer’s instructions. All elutions were prepared in 100 µL deionised, filter sterilized H2O. Briefly, 3 – 5 mL of O/N culture was pelleted (12,000 g, 1 min). The cell pellet was resuspended in 200 µL Resuspension Solution, lysed with 200 µL Lysis Solution then neutralised within 5 min with the addition 350 µL of Neutralisation Solution. The sample was centrifuged at top speed* for 10 mins to precipitate cell debris. A capture column was prepared in a sample tube by passing 500 µL Column Preparation Solution through (12,000 g, 1 min). Once the sample had finished centrifugation, the entire lysate was carefully transferred to the column (omitting precipitate) and passed through (15,000 g, 1 min). The column was washed by passing 750 µL Wash Solution through the column (15,000 g, 1 min) followed by complete wash solution removal (15,000 g, 2 min). The column was transferred to a clean microfuge tube, 100 µL H2O was added to the column and incubated at RT for 5 min before elution (15,000 g, 1 min). 2.1.2 PureLink® HiPure Plasmid Midiprep Kit Plasmid midipreps were performed using the PureLink® HiPure Plasmid Midiprep Kit (K210004 Life Technologies) as per manufacturer’s instructions. Resuspension Buffer and Wash Buffer were prepared before first use of the kit as per manufacturer’s instructions. The alcohol precipitation process was conducted in aliquots of microcentrifuge tubes due to accessibility of refrigerated microcentrifuges. All elutions were prepared as 3 x 100 µL aliquots in endotoxin free, commercially filter sterilized H2O that were pooled. Briefly, 10 mL of Equilibration Buffer was passed through a midi column via gravity flow, and 25 - 40 mL of O/N culture was pelleted (15,000 g, 1 min). The cell pellet was resuspended in 4 mL Resuspension Buffer, lysed with 4 mL Lysis Buffer then neutralised within 5 min with the addition 4 mL of Precipitation Buffer. The sample was aliquoted evenly to microfuge tubes and centrifuged at top speed* for 10 mins to precipitate cell debris. Once the sample had finished centrifugation, the entire lysate was carefully transferred to the column (omitting precipitate) and passed through via gravity flow. The column was washed twice with 10 mL Wash Buffer until completely dry. The column was transferred to a clean 15 mL tube, 5 mL Elution Buffer was added to the column and the elution captured for alcohol precipitation. 3.5 mL isopropanol was added to the elution, the elution was evenly distributed amongst 6 microfuge tubes (i.e. 1.4 mL each tube) and centrifuged at top speed* at 4 °C for 30 min to pellet the plasmid DNA. The supernatant was carefully removed from the elution tubes, the DNA pellets were washed and pooled into 3 microfuge tubes in 70% (v/v) ethanol (total volume 3 mL i.e. 1 mL each tube). The elution tubes were again centrifuged at top speed* at 4 °C for 5 min and the supernatants carefully removed. The DNA pellets were left to air dry for 10 – 20 mins, then each tube resuspended in 100 µL H2O. Plasmid elutions of the same sample were pooled prior to quantification. 2.1.3 JetStar® LFU Plasmid Maxiprep Kit Plasmid maxipreps from earlier work (Jan 2015 – Dec 2016) were performed using the JetStar® LFU Plasmid Maxiprep Kit (G221020 Astral Scientific) as per manufacturer’s instructions. Cell

187

Resuspending Buffer and Wash Buffer were prepared before first use of the kit as per manufacturer’s instructions. All elutions were prepared in 800 µL deionised autoclaved H2O. Briefly, 30 mL of Equilibration Buffer was passed through a maxi column / filter cartridge unit via gravity flow, and 100 - 250 mL of O/N culture was pelleted (3,000 g, 15 min). The cell pellet was resuspended in 10 mL Cell Resuspending Buffer, lysed with 10 mL Lysis Buffer then neutralised within 5 min with the addition 10 mL of Precipitation Buffer. The entire precipitated lysate was transferred to the column / filter unit and passed through via gravity flow. The column / filter unit was washed with 50 mL Wash Buffer until completely dry. The filter unit was disposed, the column was transferred to a clean 50 mL tube, 15 mL Elution Buffer was added to the column and the elution captured for alcohol precipitation. 10.5 mL isopropanol was added to the elution and centrifuged at 25,000 g, 4 °C for 30 min to pellet the plasmid DNA. The supernatant was carefully removed from the elution tube and the DNA pellet was washed in 5 mL 70% (v/v) ethanol. The elution tube was again centrifuged at 25,000 g, 4 °C for 5 min and the supernatant carefully removed. The DNA pellet was left to air dry for 10 – 20 mins, then resuspended in 800 µL H2O in a rotating mixer O/N at 2 - 8 °C. 2.1.4 Genelute™ HP Plasmid Maxiprep Kit Plasmid maxipreps from later work (Jan 2017 onward) were performed using the Genelute™ HP Plasmid Maxiprep Kit (NA0310-1KT Sigma-Aldrich) as per manufacturer’s instructions. Resuspension Solution and Wash Solution 2 were prepared before first use of the kit as per manufacturer’s instructions. All elutions were prepared in 3 mL endotoxin free, commercially filter sterilized H2O. Briefly, 100 - 350 mL of O/N culture was pelleted (3,000 g, 15 min). The cell pellet was resuspended in 12 mL Resuspension Solution, lysed with 12 mL Lysis Solution then neutralised within 5 min with the addition 12 mL of chilled Neutralisation Solution. 9 mL of Binding Solution was added to the lysate and the entire precipitated lysate was transferred to a syringe / filter unit and left to stand at RT for 5 min, to separate cell debris aggregate from the solution. 12 mL of Column Preparation Solution was passed through a column (3,000 g, 2 min) then the cell lysate was passed through the syringe / filter unit into the column (3,000 g, 2 min). The column was washed with 12 mL Wash Solution 1 (3,000 g, 2 mins) and 12 mL Wash Solution 2 (3,000 g, 5 min) until completely dry. The column was transferred to a clean 50 mL tube, 3 mL H2O was added to the column and incubated at RT for 5 min before elution (3,000 g, 5 min). 2.1.5 Nucleospin® DNA Purification kit DNA purifications from PCR reactions, RE digestions or agarose gels were performed using the Nucleospin® DNA Purification kit (740609.1 Scientifix) as per manufacturer’s instructions. Buffer NT3 was prepared before first use of the kit as per manufacturer’s instructions. All elutions were prepared in 30 µL deionised, filter sterilized H2O. Briefly, DNA from PCR/digestion reaction, or excised from agarose gel bands were combined with equal volumes of Buffer NT1 and deionised, filter sterilized H2O. The sample was well mixed, then transferred and passed through a capture column (11,000 g, 30 sec). The column was washed twice by passing 700 µL Buffer NT3 through the column (11,000 g, 30 sec) followed by complete Buffer NT3 removal (11,000 g, 1 min). The column was transferred to a clean microfuge tube, 30 µL H2O was added to the column and incubated at RT for 1 min before elution (11,000 g, 1 min). 2.2 Sample Preparation for Sequencing (AGRF) Plasmid elutions were prepared for Sanger Sequencing at AGRF as purified DNA (PD) samples as follows:  1 µL of [10 µM] sequence primer

188

 1 µg purified plasmid DNA (volume dependant on DNA concentration)  H2O up to total volume 12 µL AGRF additional recommendations for sample preparation:  0.6 – 1.5 µg total plasmid double stranded DNA.  Primer Tm to be between 55 – 60 °C.  Repetitive freeze-thaw of primers / template DNA causes degradation.  DNA quantitation via agarose is preferred over spectrophotometry as values from absorbance ratios can overestimate concentration in contaminated samples. Forward and reverse sequencing primers were prepared as 2 separate sequencing samples for each plasmid isolate. Few isolates had a very low DNA concentration and were prepared neat (i.e. 11 µL, no additional H2O) for sequencing. 2.3 Agarose Gel Electrophoresis DNA fragments were analysed for size and quality using agarose gel electrophoresis. Equipment to cast and run the gels was sourced from the Bosch Institute, The University of Sydney. Bio-Rad Mini Gel casters were used to prepare gels. Gels were prepared as 0.8% - 1.2% (w/v) agarose in 1 x TAE buffer, stained with 0.004% (v/v) SYBR® Safe nucleotide stain (S33102 Life Technologies). A Bio-Rad Mini-Sub Cell GT system was used to perform electrophoresis, with the gels submerged in Tris-acetate-EDTA (1 x TAE) buffer, running at 70 - 80V for 15 - 30 mins. Gel electrophoresis was monitored and stopped when the samples gel loading dye migrated between ⅓ to ½ through the gel. Gels were imaged (1.5 sec exposure) on a Bio-Rad Gel Doc™ XR system and analysed using Bio- Rad Image Lab 5.2.1 software. For resolution of larger fragments (1 – 10 kb; 10 kb ladder), 0.8% agarose was used. For resolution of smaller fragments (0.1 – 1 kb; 100 bp ladder), 1.2% agarose was used. Intermediate fragment sizes were run on 1 % agarose. Stock TAE buffer (25x)

Deionised H2O containing:  1 M (121 g/L) Tris base (BIOTB0196 Astral Scientific)  25 mM (7.3 g/L) EDTA (E6758 Sigma-Aldrich)  0.5 M (28.6 mL/L) Glacial acetic acid (27225 Sigma-Aldrich) Combine ingredients to the desired volume. Keep stock buffer sealed until use.

Prepare 1 x TAE buffer freshly for use to the desired volume in deionised H2O. 2.3.1 Agarose Gel Casting Using clean, dry casting equipment, seal the sides of a mini-gel caster with masking and tape and place desired well combs into the caster holders. To a glass flask, add 25 mL 1 x TAE and agarose to the desired concentration i.e. 0.2 g (0.8 %), 0.25 g (1 %) or 0.3 g (1.2 %). Microwave the solution in 30 sec increments until agarose fully dissolves and solution is gently bubbling (gently swirl the solution between increments). Take precautions for handling super- heated fluid and glassware. 189

Set solution to cool on the bench until it can be handled by hand. Once cooled, add 1 µL SYBR® Safe nucleotide stain and gently swirl through the solution. Pour the solution into the casting tray, remove bubbles with a disposable pipette and allow the gel to set for 30 mins, RT. Once set, remove well combs and masking tape from gel. In preparation for electrophoresis, place gel in Mini-Sub Cell tank and fill tank with 1 x TAE until the gel is completely submerged. If the gel is not for immediate use, store the gel in a zip-lock bag with 20 mL 1 x TAE and refrigerate until use. 2.3.2 DNA Preparation for Loading into Gels For a size and resolution reference in each prepared gel, 2 µL of prepared DNA ladder (i.e. 200 ng DNA) was directly loaded into gel wells. 100 bp (N3231S Genesearch) and 1 kb (N3232S Genesearch) DNA ladders were prepared to 100 ng/µL with 6 x gel loading dye (made to 1 x) and deionised H2O. Colony PCR products from OneTaq Quick-Load mastermix (M0486L Genesearch) may be loaded directly as the PCR mastermix contains loading dye that does not interfere with the PCR reaction. 5 µL of colony PCR product was directly loaded into gel wells. Purified DNA samples and digestion/ligation products require to be mixed with 6 x gel loading dye (B7025 Genesearch) to be visualised in the gel through the electrophoresis run. 20 – 100 ng of purified sample DNA is clearly visible and resolvable; digestion/ligation products may be added without prior measurement. To prepare DNA samples for loading: On a sheet of parafilm load 1 µL of 6 x gel loading dye, an appropriate volume of DNA sample (e.g. 1.5 µL) and a volume of deionised H2O to make up to 6 µL. Mix well via pipette, then transfer the entire sample to be loaded into a gel well. 3 Tissue Culture Protocols 293F and CHOS cells were routinely cultured in suspension, serum free media (SFM) for protein expression. SK-BR3 and CHO-K1 cells were routinely cultured in adherent conditions, with 10% serum addition to media. For certain experiments, 293F and CHOS were also cultured in adherent conditions for stable transfection studies, and were adapted to suspension SFM culture. 3.1 Maintenance Media, Reagents and Culture Conditions All tissue cultures were maintained at 37 °C, 5% CO2, humidified conditions. Suspension cultures were further maintained at 120 rpm on an orbital shaker. Suspension cultures were routinely maintained in 125 mL single use sterile baffled Erlenmeyer culture flasks (CLS431405 Sigma-Aldrich) or scaled up in pre-autoclaved, sealed 250 mL / 500 mL baffled flasks for protein expression experiments. Small scale (maintenance): 125 mL flask containing 10 – 35 mL culture (routinely 30 mL) Medium scale (scale up): 250 mL flask containing 40 – 80 mL culture (routinely 60 mL) Large Scale (protein expression): 500 mL flask containing 90 - 150 mL culture (generally 120 mL)

190

Adherent cultures were routinely maintained in T75 flasks (CORN430641U Bio-Strategy), small volume work and resuscitation performed in T25 flasks and scale up for experiments and cryopreservation in T182 flasks. Small scale (resuscitation): T25 flask containing 3 - 8 mL culture (routinely 5 mL) Medium scale (maintenance): T75 flask containing 10 - 20 mL culture (routinely 15 mL) Large Scale (cryopreservation): T182 mL flask containing 25 - 50 mL culture (routinely 35 mL) 100 x Surfactant Supplement: Prepare 10% (w/v) Kolliphor P 188 Lutrol F68 (Pluronic F-68) stock solution (K4894 Sigma- Aldrich) in deionised H2O. Sterilize by passing through a 0.22 µm filter, keep sealed and refrigerate until use. Pen-Strep: Penicillin-Streptomycin (10,000 U/mL) solution (15140-122 Life Technologies). Prepare 5 mL aliquots and store at – 20 °C until use. L-Glutamine: GlutaMAX™ Supplement (35050-061 Life Technologies); i.e. 200 mM L-Glutamine. Store at – 20 °C until use. Trypsin-EDTA stock: Trypsin-EDTA (0.5%) solution (15400-054 Life Technologies). Store at – 20 °C until use. 293F EM: FreeStyle™ 293 Expression Medium (12338018 Life Technologies) supplemented with:  Additional 1% (v/v) 100 x surfactant supplement  Additional 1% (v/v) Pen-Strep

CHOS EM: FreeStyle™ CHO Expression Medium (12651014 Life Technologies) supplemented with:  Additional 1% (v/v) 100 x surfactant supplement  Additional 0.5% (v/v) Pen-Strep  Additional 8 mM L-Glutamine

CD CHO EM: CD CHO Medium (10743029 Life Technologies) supplemented with:  Additional 1% (v/v) 100 x surfactant supplement  Additional 1% (v/v) Pen-Strep  Additional 6 mM L-Glutamine

PBS:

191

Deionised H2O containing 1 phosphate buffered saline (PBS) tablet per 100 mL (09-2051-100 Astral Scientific). Sterilize by autoclave or by passing through a 0.22 µm filter, keep sealed and refrigerate until use. 1 x Trypsin-EDTA: Prepare PBS containing 10% v/v Trypsin-EDTA stock. Sterilize by passing through a 0.22 µm filter, keep sealed and refrigerate until use.

RPMI: RPMI 1640 Medium (21870076 Life Technologies) supplemented with:  Additional 10% (v/v) FBS (10082-147 Life Technologies)  Additional 1% (v/v) Pen-Strep  Additional 6 mM L-Glutamine

DMEM/F-12:

Prepare 50% DMEM (11995-065 Life Technologies) and 50% Ham's F-12K Media (21127-022 Life Technologies) supplemented with:  Additional 10% (v/v) FBS (10082-147 Life Technologies)  Additional 1% (v/v) Pen-Strep Keep all in use stocks and prepared media sealed and refrigerated until use. 3.2 Cell Counting All cell counting was performed using trypan blue exclusion method, using 0.4% Trypan Blue solution (AMK940 Astral Scientific). Live cells are identified as clear with a halo against a light blue background, dead cells are stained dark blue. Live cells, dead cells, total cells and viability are calculated from cell counts. Cell counts from earlier work (Jan – Dec 2015) were performed on a Countess automated cell counter (Invitrogen) using Countess Chamber slides (C10228 Life Technologies). Briefly, an aliquot of cell suspension was mixed with an equal volume of trypan blue solution in an Eppendorf tube. 10 µL of stained cells was dispensed into a countess chamber and the chamber was loaded into the countess for measurement. The countess was turned on and focused prior to taking measurement; viability and total, live and dead cell concentration was reported directly from the machine. Measurements were conducted in duplicate to get an average set of values. Cell counts from later work (Jan 2016 onwards) were performed on a haemocytometer, read under a microscope at 10 x magnification. Briefly, 10 µL of cell suspension was mixed with an appropriate volume of trypan blue solution in an Eppendorf tube (the dilution factor to be noted for calculation). That is, routinely 40 µL for adherent cells, 90 µL for suspension maintenance cultures and 190 µL for transfection harvest cultures. The haemocytometer is set up with a coverslip, 10 µL of stained cells was added to a chamber of the haemocytometer and the chamber was examined under a microscope at 10 x magnification.

192

Of the 9 squares that comprise the grid pattern in the chamber, 4 full squares were examined and counted for live and dead cells to take average values and derive total cells and viability. To convert to concentration (cells/mL), the cell values were multiplied by 104, then by the dilution factor. Viability was calculated as the percentage of live cells from total cells. 3.3 Suspension 293F / CHOS Culture Maintenance FreeStyle™ 293F cells (293F) were a kind donation from Dr Mario Torrado del Rey, Prof Joel Mackay’s research group. Freestyle™ CHO-S cells (CHOS) were purchased from Life Technologies (R80007). Suspension 293F and CHOS cells are passaged when cell density reaches 2 – 3 x 106 cells/mL, seeding the new culture at 2 – 3 x 105 cells/mL. 293F culture has a 24 hr cell doubling rate, and is recommended to passage every 3 – 4 days at a seed of 3 x 105 cells/mL. 293F culture is maintained in 293F EM. CHOS culture has an 18 hr cell doubling rate and is recommended to passage every 2 – 3 days at a seed of 2 x 105 cells/mL. CHOS culture is maintained in CHOS EM and/or CD CHO EM. Maintenance culture (30 mL) passage protocol When a culture is due for passage:  Take a cell count to derive number of live cells per mL  Calculate the volume of culture required to seed a new passage (30 mL)  Transfer the volume of culture to a centrifuge tube  Pellet the cells via centrifugation (300 g, 5 min)  Remove the supernatant, agitate the cell pellet (flicking)  Resuspend the cell pellet in 5 mL fresh maintenance EM  Transfer the suspended cells to a new 125 mL flask with a further 25 mL fresh maintenance EM  Label the flask with cells, passage number, passage date, initial  Incubate the culture 3.4 Suspension 293F / CHOS Cryopreservation and Resuscitation Suspension 293F and CHOS cells require to be preserved at 1 x 107 cells/mL in cryopreservation media. Cryopreservation media Maintenance EM containing 10% (v/v) DMSO; filter sterilised Suspension cells cryopreservation protocol For a culture that is ready to be cryopreserved:  Take a cell count to quantify the total amount of live cells in the culture  Pellet the cells of the entire culture via centrifugation (300 g, 5 min)  Remove the supernatant; agitate the cell pellet (flicking)  Calculate amount of cryopreservation media needed and prepare  Resuspend cells evenly in cryopreservation media  Aliquot 1 mL cell suspension into pre-labelled cryovials  Transfer cryovials to a Mr Frosty rack (prepared with fresh isopropanol)  Transfer the Mr Frosty rack to -80 °C freezer for O/N storage

193

 Transfer the vials from the -80 °C Mr Frosty rack to liquid nitrogen for long term storage For resuscitation of cells from a cryovial, the vial needs to be kept frozen (minimum in dry ice) until such time that it is ready for immediate thaw and transfer into media, as DMSO is toxic to cells and can reduce cell viability over prolonged exposure time. Resuscitation of cells for suspension culture protocol:  Prepare a tube with 4 mL pre-warmed maintenance EM  Retrieve vial from liquid nitrogen and thaw in 37 °C water-bath  Gently mix the thawed cell suspension with a pipette then transfer to the tube containing media  Pellet the cells via centrifugation (300 g, 5 min)  Remove the supernatant; agitate the cell pellet (flicking)  Resuspend the cell pellet in 5 mL fresh maintenance EM  Transfer the suspended cells to a new 125 mL flask with a further 10 mL (293F) or 20 mL (CHOS) fresh maintenance EM  Label the flask with cells, passage number, passage date, initial  Incubate the culture; monitor cell counts daily to determine next passage 3.5 Adherent SK-BR3 / CHO-K1 WT Culture Maintenance SK-BR3 cells were a kind donation from A/Prof Thomas Grewal’s research group. CHO-K1 cells were a kind donation from Dr Alryel Yang, Dr Ingrid Gelissen’s research group. Adherent cultures have their media replenished when the media colour changes from red to orange/yellow and are passaged when confluency reaches 70 – 90%. SK-BR3 culture has a growth doubling rate of 7 days, and is recommended to replenish media after 4 days and passage at 7 days at a 50% seed in RMPI. CHO-K1 culture has a growth doubling rate of 24 hrs, as is recommended to passage at 3 days at 25% seed in DMEM/F-12. Maintenance culture (15 mL) passage protocol When a culture is due for passage:  Remove the culture media and wash the cell layer twice with 5 mL PBS  Add 3 mL 1 x Trypsin-EDTA to cell layer  Incubate flask at 37 °C for 5 min  Add 9 mL maintenance media to the flask  Gently dissociate the cell layer via pipette  Transfer seed amount of cell suspension to new T75 flask (i.e. 3 mL for 25% seed, 6 mL for 50%)  Add further maintenance media to flask to total 15 mL volume  Label the flask with cells, passage number, passage date, initial  Incubate the culture 3.6 Adherent SK-BR3/CHO-K1 WT Cryopreservation and Resuscitation Adherent SK-BR3 and CHO-K1 cells require to be preserved at 3 x 106 cells/mL in cryopreservation media. Cryopreservation media: Maintenance media containing 5% (v/v) DMSO; filter sterilised

194

Adherent culture cryopreservation protocol For a culture that is ready to be cryopreserved:  Prepare cell suspension as outlined above for cell passage  Take a cell count to quantify the total amount of live cells in the culture  Pellet the cells of the entire culture via centrifugation (300 g, 5 min)  Remove the supernatant; agitate the cell pellet (flicking)  Calculate amount of cryopreservation media needed and prepare  Resuspend cells evenly in cryopreservation media  Aliquot 1 mL cell suspension into pre-labelled cryovials  Transfer cryovials to a Mr Frosty rack (prepared with fresh isopropanol)  Transfer the Mr Frosty rack to -80 °C freezer for O/N storage  Transfer the vials from the -80 °C Mr Frosty rack to liquid nitrogen for long term storage For resuscitation of cells from a cryovial, the vial needs to be kept frozen (minimum in dry ice) until such time that it is ready for immediate thaw and transfer into media, as DMSO is toxic to cells and can reduce cell viability over prolonged exposure time. Resuscitation of cells for suspension culture protocol:  Prepare a tube with 4 mL pre-warmed maintenance media  Retrieve vial from liquid nitrogen and thaw in 37 °C water-bath  Gently mix the thawed cell suspension with a pipette then transfer to the tube containing media  Pellet the cells via centrifugation (300 g, 5 min)  Remove the supernatant; agitate the cell pellet (flicking)  Resuspend the cell pellet in 1 mL fresh maintenance media  Transfer the suspended cells to a new T25 flask with a further 4 mL fresh maintenance media  Label the flask with cells, passage number, passage date, initial  Incubate the culture; monitor cell counts daily to determine next passage 3.7 Reagents for Protein Expression Commercial reagents for transfections include: OptiPRO™ Serum free media (SFM) (12309-050), Freestyle™ MAX Transfection reagent (16447- 100) and Lipfectamine™ 2000 Transfection reagent (11668-027) from Life Technologies.

Along with these commercial reagents, 1 mg / mL PEI solution is prepared for transfection as per 0. Reagents prepared as supplements to support cell viability and protein expression include Tryptone (0) and VPA (0), prepared as 40 x stock solutions. 1 mg/mL Polyethylenimine, 25 kDa, linear (PEI) PEI is not readily dissolved in H2O at RT, it forms a white suspension. To prepare a PEI solution, it is recommended to heat the H2O up to 80 °C prior to addition of PEI and to adjust the suspension to pH 7.0 with HCl.

Prepare a PEI (23966-2 BioScientific) 1 g/L stock solution in deionised H2O. Adjust to pH 7.0 with HCl and sterilize by passing through a 0.22 µm filter. Prepare as 1.2 mL aliquots in microfuge tubes, keep aliquots at -80 °C until use.

195

40 x Tryptone stock Prepare 20% (w/v) tryptone (casein peptone) (LP0042B Thermo Fisher Scientific) stock solution in PBS and autoclave. Keep stock solution sealed and refrigerated until use. 40 x Valproic Acid (VPA) stock Prepare valproic acid sodium salt (P4543 Sigma-Aldrich) 140 mM stock solution in PBS and sterilize by passing through a 0.22 µm filter. Keep stock solution sealed and refrigerated until use. 4 Antibody Purification and Characterisation Protocols Transfection cultures were harvested for antibody purification by centrifugation (3,000 g, 15 mins). The culture supernatant was decanted and further clarified by vacuum filtration, using Corning® 0.22 μm bottle-top vacuum filter units (CLS430513 Sigma-Aldrich). Antibodies were purified from clarified supernatant through Protein A affinity chromatography, performed on an ÄKTApurifier 100 FLPC system using Unicorn 5.11 system and analysis software (GE Healthcare). Antibody purifications from earlier work (Jan 2015 – Mar 2018) were performed on a 5 mL HiTrap® Protein A High Performance column (GE17-0403-01 Sigma-Aldrich). Antibody purifications from later work (Apr 2018 – onwards) were performed on a 1 mL HiTrap® MabSelect Sure column (29-0491-04 GE Healthcare). Purified antibodies were buffer exchanged to PBS and concentrated in centrifugal filters prior to further characterisation, stability and biological assays. Antibody samples were quantified through UV absorption at 280 nm prior to normalisation of sample concentrations for assays. Antibody samples were characterised through reducing and non-reducing SDS-PAGE for purity, LC-MS/MS for confirmation of peptide sequence and glycosylation of antibody samples and SE-HPLC for size distribution analysis. 4.1 FPLC Buffers and Protocol: The FPLC system was set up prior to purification with the method program, column connected to the system and washed, supernatant samples, buffers and elution collection tubes as detailed in Chapter 3. Antibody purification was performed on the FPLC system under the following conditions:  All system operations at 1 mL/min  Pass 5 column volumes (CV) of binding buffer through the system to equilibrate the column  Allowing 5 mL dead volume, pass the entire sample volume through the system; the antibody is captured in the column  Pass 5 CV of binding buffer through the system to wash the column  Pass 5 CV elution buffer through the system to disassociate the captured antibody from the column  Isocratic fractionation of elutions;  100 μL neutralisation buffer added to each elution collection tube in preparation of capturing the elution fractions  Pass 5 column volumes (CV) of binding buffer through the system to equilibrate the column FPLC was conducted with continuous monitoring of change in system pH, conductivity and UV absorption at 280 nm, producing chromatograms that demonstrate the performance of antibody capture and elution through the run.

196

Prepared buffers were sterilized and degassed by passing through a 0.22 µm vacuum filter unit prior to use on FPLC. Binding Buffer Deionised H2O containing 1 phosphate buffered saline (PBS) tablet per 100 mL (09-2051-100 Astral Scientific), adjusted to pH 7.4. Elution Buffer ® Elution buffer for the HiTrap Protein A column was deionised H2O containing 0.1 M glycine HCl (G2879 Sigma-Aldrich), adjusted to pH 2.7.

® Elution buffer for the HiTrap MabSelect Sure column was deionised H2O containing 0.1 M glycine HCl (G2879 Sigma-Aldrich), adjusted to pH 3. Neutralisation Buffer Antibody purifications from earlier work (Jan 2015 – Apr 2017) used neutralisation buffer 1 M Tris-HCl pH 9 (BIOSD814 Astral Scientific). Antibody purifications from later work (May 2017 – onwards) used neutralisation buffer 1 M Tris- HCl pH 8 (15568-025 Life Technologies). 4.2 Buffer Exchange, concentration and quantification Elution fractions from FPLC runs were measured on a Shimadzu UV-Visible Spectrophotometer (UV-Vis); an absorbance peak at 280 nm indicated the presence of protein. Fractions that contained protein were pooled together (and the remainder discarded). Pooled fractions were buffer exchanged and concentrated in PBS using Amicon® 50 kDa centrifugal filters (Z740194, Z740177 Sigma-Aldrich) following manufacturing instructions for centrifugation conditions. Concentrated antibody elutions were measured on a UV-Vis for absorbance at 280 nm and their concentrations were derived from the Beer Lambert Law.

A = ε.c.ℓ → c = A/ε.ℓ where: c = concentration (M) A = absorbance (abs) ε = molar attenuation coefficient (M-1cm-1) ℓ = pathlength (cm) For unit conversions from molarity to weight, the mass attenuation (aka extinction) coefficient for each species is derived by dividing ε by the species molecular mass (i.e. g-1L-1). The above formula was used with extinction coefficients 1.48 g-1L-1 for trastuzumab and 1.39 g-1L-1 for adalimumab as reported from original patents and publications. Antibody samples were sterilized by passing through a 0.22 µm filter, prepared as small labelled aliquots in microfuge tubes and stored at -80 °C until use. The concentration of samples derived from the UV absorbance was used as reference for all further experiments. 4.3 Purity Determined by SDS PAGE Analysis Purity of antibody samples (protein content) was determined by 10% (acrylamide) reducing and non-reducing SDS PAGE analysis and visualised with Coomassie Blue R-250 stain (786-498 Astral Scientific).

197

All SDS-PAGE chemicals were purchased from Astral Scientific. Gels were cast manually and separation performed on a Bio-Rad Mini-PROTEAN® Tetra Vertical Electrophoresis System. Imaging was performed on a Bio-Rad Gel Doc™ XR system and analysed using Bio-Rad Image Lab 5.2.1 software.

Antibody expected reduced bands at 50 – 60 kDa (TH/AH) and 25 - 30 kDa (TL/AL) and non- reduced band at 150 - 165 kDa (whole antibody). Further bands indicate possible contaminants such as leached protein A, host cell protein and antibody degradation products. 4.4 Sequence and Glycosylation Confirmed by LC-MS/MS Analysis Purified antibodies and mutants were confirmed of the predicted amino acid sequence and introduction of glycosylation motifs through site-specific liquid chromatography tandem mass spectrometry (LC-MS/MS). Briefly, purified antibody samples were separated on reducing SDS-PAGE and digested with trypsin and chymotrypsin. For glycosylated antibody mutants, a fraction of the digestions was further treated with PNGase F for glycan release. The digestion products were analysed by LC- MS/MS using higher-energy collisional dissociation (HCD), collision induced dissociation (CID) and electron transfer dissociation (ETD) methods. 4.4.1 Reducing SDS PAGE Separation and Peptide Digestion Trypsin

Trypsin from porcine pancreas Proteomics Grade (T6567 Sigma-Aldrich) Incubate 1 µg trypsin to every 50 µg protein at 37 °C, O/N. Chymotrypsin

Chymotrypsin Sequencing Grade from bovine pancreas (11418467001 Sigma-Aldrich) Incubate at RT, O/N. PNGase F

PNGase F Recombinant (P0708 Genesearch) Incubate 400 units PNGase F to every 1 µg protein at 37 °C, O/N. 10 μg samples were run on a 10% reducing SDS-PAGE and in-gel trypsin digestion was performed on the heavy and light chain bands. The bands were excised, diced and washed with 50% (v/v) acetonitrile (MeCN) in 100 mM NH4HCO3. The gel pieces were reduced with 10 mM DTT (56 °C, 45 min) then alkylated with 55 mM iodoacetamide (IAA) in 100 mM NH4HCO3 (25 °C, 30 min).

The alkylated polypeptide samples were washed with 50% (v/v) MeCN in 100 mM NH4HCO3, then incubated with trypsin in 50 mM NH4HCO3 pH 6.8. The tryptic peptides were further digested with chymotrypsin in 100 mM Tris-HCl, 10 mM CaCl2 pH 7.8. For glycosylation mutant samples, a fraction of the chymotryptic peptides was incubated with PNGase F in 50 mM sodium phosphate buffer pH 7.5.

All chymotryptic peptides and de-glycosylated fractions were desalted with C18 Millipore Ziptips (Z720070 Sigma-Aldrich), eluted in 80% (v/v) MeCN, dried by vacuum centrifugation then redissolved in 0.1% formic acid (FA) for LC-MS/MS analysis.

198

4.4.2 LC-MS/MS Operation and Analysis Glycopeptides and de-glycosylated peptides were separated on an in-house packed 20 cm × 75 μm Reprosil-Pur C18-AQ column, 3 μm, 120 A (r13.aq Dr. Maisch GmbH) using a Thermo Fisher Dionex™ 3500RS HPLC over a 90-min 0-40% solvent B gradient at a flow rate of 300 nL/min, at 60 °C (solvent A was 0.1% (v/v) formic acid; solvent B was 80% (v/v) acetonitrile and 0.1% formic acid). MS analysis was performed on an Orbitrap Fusion™ Tribid™ mass spectrometer (Thermo Scientific) in positive mode. The initial MS scan was acquired in the Orbitrap mass analyzer (350–2000 m/z; MS AGC = 6 × 105) with a resolution of 60,000 at 400 m/z. MS1 was followed by data-dependent HCD using top speed, where as many dependant scans as possible are acquired in a specified time. HCD fragmentation was followed by re-isolation of the top two most intense precursors, and fragmentation with CID and ETD MS/MS. Instrument parameters:

 Source voltage = 2.3 kV  S-lens RF level = 68%  Capillary temperature = 275 °C

HCD parameters:

 Activation time = 0.1 ms  Resolution = 30,000  Maximum injection time = 200 ms  Dynamic exclusion = enabled with repeat count 1  Normalized energy = 40  Exclusion duration = 20 s  Default charge state = 2  MSn AGC 2.0e5

CID parameters:

 30,000 resolution in orbitrap  Activation time = 10 ms  Max injection time = 300 ms  Dynamic exclusion = enabled with repeat count 1  Normalized energy = 35  Exclusion duration = 20 s  Fefault charge state = 2  MSn AGC 5.0e4

ETD parameters:

 30,000 resolution in orbitrap  5.0e4 AGC

199

 ETD reaction time = 50 ms  Max injection time = 300 ms

Peptide spectrum matches (PSM) were generated through a search against the corresponding fasta file (native antibody or predicted mutation amino acid sequence) utilising the search engine Byonic (Protein Metrics Inc.). Annotation of the HCD and ETD spectra were automated through Byonic and the PSM assignments were validated by manually annotating the aforementioned spectra and the complementary CID spectrum. Further confirmation of the identity assignment was achieved through analysis of the de-glycosylated chymotryptic peptides utilising the same LC-MS/MS approach as for the glycosylated fraction. 4.5 Size Distribution of Antibody Samples Determined by SE-HPLC Antibody samples prepared for size distribution analysis were characterised by size-exclusion high- performance liquid chromatography (SE-HPLC) performed on an Agilent™ 1200 Liquid Chromatography system, using ChemStation system and analysis software (Agilent Technologies). For earlier work (2015 – 2017) SE-HPLC was performed on a 7.8 x 300 mm, 5 µm particle size ® ® TSKgel G3000SWXL HPLC column (0008541) coupled to a 6 x 40 mm TSKgel SWXL guard column (0008543 Tosoh Bioscience GmbH), at a flow rate of 1 mL/min. For later work (2018) SE-HPLC was performed on a 4.6 x 250 mm, 4 µm particle size Zorbax GF- 250 HPLC column (884973-701) coupled to a 4.6 x 12.5 mm, 6 µm particle size Zorbax Diol guard column (82095-911 Agilent Technologies) at a flow rate of 0.5 mL/min. SE-HPLC was performed on the Agilent™ 1200 LC system running at 22 °C, with a mobile phase buffer of 150 mM potassium phosphate, pH 6.5. Samples were loaded into the system at varied injection volume (5 – 100 µL) depending on the assay requirements. Total protein in the sample was detected through UV absorption at 280 nm, producing chromatograms where different sized protein species eluted separately through the system. Higher molecular weight species eluted ahead of smaller species; antibody soluble aggregates, dimers, monomers and fragments were separately distinguished on chromatograms for analysis. Protein content of individual chromatogram peaks was determined by the area under the curve (AUC). Relative comparisons of monomer content were made between the AUC of each species in the sample and between samples, described as relative percentage monomer content or monomer loss (%). 4.5.1 Reagents for HPLC Stock Solution 1: Prepare 1.5 M potassium phosphate monobasic (AJA391Thermo Fisher Scientific) in deionised H2O Stock Solution 2:

Prepare 1.5 M potassium phosphate dibasic (AJA391Thermo Fisher Scientific) in deionised H2O Mobile Phase Buffer (150 mM potassium phosphate, pH 6.5): Combine 66.3 mL Stock Solution 1 with 33.7 mL Stock Solution 2, dilute to 1L with deionised H2O and adjust to pH 6.5. Buffers were freshly prepared, sterilized and degassed by passing through a 0.22 µm vacuum filter unit prior to use on HPLC. Solutions were kept sealed and refrigerated until use.

200

5 Affinity and Binding Kinetic Assays 5.1 ELISA Antibody samples such as transfection culture supernatant and bulk mAb product were quantified through an ELISA protocol. This protocol was derived from a commercial ELISA kit: Affymetrix Human IgG total ELISA Ready-SET-Go!® (88-50550-22 Jomar Life Research). Samples required appropriate dilution (within the range of 1 – 100 ng/mL) for robust fitting to standard curve. All samples were measured in duplicate, including the standard curve which was generated from Herceptin. ELISA was performed in batches, each batch including a 1:2 serial dilution of Herceptin standard (from 100 > 1.5625 ng/mL concentration). ELISAs were performed in Corning high protein binding 8-well microplate strips (734-1514 Bio-Strategy), shaking incubations conducted on a Biosan PSU-10i orbital shaker (Fisher Biotech) at 400 rpm, and photometric measurements conducted on a Victor X plate reader (Perkin Elmer) at 450 nm. All reagents and sample dilutions were prepared to slight excess, as to minimise loss or load inaccuracy through serial pipetting. ELISA antibodies were pre-aliquoted and stored at -20 °C until use, all working reagents were refrigerated until use. Coating Antibody

Prepare anti-human IgG (Fab specific) antibody produced in goat (I9010 Sigma Aldrich) as a 1:10,000 dilution in PBS to the total volume required for the batch of samples (i.e. 100 µL per sample). Wash Buffer

Prepare PBS with 0.05% (v/v) Tween 20 (P9416 Sigma-Aldrich). Keep solution in squidgy bottle for efficient aspiration into wells. Stock Buffer

Deionised H2O containing:  20 x PBS (09-2051-100 Astral Scientific)  1% (v/v) Tween 20 (P9416 Sigma-Aldrich)  10% w/v BSA (BSAS 0.1 Bovogen) This stock will be diluted for the further buffers used in this assay. Blocking Buffer

Dilute 1:10 Stock Buffer in deionised H2O Assay Buffer

Dilute 1:20 Stock Buffer in deionised H2O Detection Antibody

Prepare anti-human IgG (Fc specific) − Peroxidase antibody produced in goat (A0170 Sigma Aldrich) as a 1:20,000 dilution in Assay Buffer to the total volume required for the batch of samples (i.e. 100 µL per sample). Substrate Solution

TMB ELISA substrate solution (T4444 Sigma Aldrich); bring to RT before use.

201

Stop Solution

1M H3PO4 in deionised H2O (i.e. 1.5 mL 85 % (w/v) phosphoric acid (695017 Sigma Aldrich) in 21 mL deionised H2O. Herceptin® Preparation for Standard Curves; sample dilutions

Herceptin® 21 mg/mL was a kind donation from Genentech (USA). Prepare 500 µL 100 ng/mL Herceptin in Assay Buffer, then perform 6 subsequent 1:2 serial dilutions in 250 µL Assay Buffer. Assay Buffer is included for a blank [0 ng/mL] sample, totalling 8 standard samples for the standard curve. Samples are diluted in Assay Buffer to a minimum volume of 220 µL for duplicate measurement. Appropriate dilutions for supernatants of good yield span 1:200 – 1:500, and bulk mAb product diluted to 50 ng/mL as calculated from UV measurement at 280 nm. However, if fairly uncertain of the sample concentration then perform several different dilutions for testing (e.g. 1:2, 1:20, and 1:200). 5.1.1 ELISA Protocol  Set up 8-well strips in a plate according the amount needed for the batch  Prepare appropriate volume of Coating Antibody, load 100 µL to each well  Seal the wells and refrigerate the plate 2 - 8 °C O/N  Aspirate wells, wash wells twice with Wash Buffer, aspirate and tap dry  Prepare appropriate volume of Blocking Buffer, load 250 µL to each well  Seal the wells and incubate at RT for 2 hrs (or 2 - 8 °C O/N)  Aspirate wells, wash wells twice with Wash Buffer, aspirate and tap dry  Prepare sample and standard dilutions in Assay Buffer  Load 100 µL standard / sample to designated wells  Seal the wells and incubate at RT for 2 hrs, 400 rpm  Aspirate wells, wash wells four times with Wash Buffer, aspirate and tap dry  Prepare appropriate volume of Detection Antibody, load 100 µL to each well  Seal the wells and incubate at RT for 1 hr, 400 rpm  Aspirate wells, wash wells four times with Wash Buffer, aspirate and tap dry  Add 100 µL Substrate Solution to each well, incubate at RT for 10 – 20 mins  Add 100 µL Stop Solution to each well, measure on plate reader at 450 nm 5.1.2 ELISA Analysis Measurements on the Victor X plate reader will give absorbance values for each well as depicted in A, with columns 1 and 2 elected for the standard curve samples, descending in concentration from the top row. Absorbance directly correlates to concentration; samples are comparable to the standard curve in the same batch if their absorbance values reside within the absorbance range of the standard curve. However, due to variability with incubation and loading times, the standard curve of one batch cannot be compared to another, therefore a curve must be generated with each batch.  Calculate the average absorbance value of the blank samples, then subtract that blank value across all the sample values.  Produce a 4-PL sigmoidal curve of the blank subtracted standard values (absorbance versus concentration) as depicted in B

202

 Interpolate all blank subtracted sample values to the standard curve  Multiply the interpolated concentrations by the appropriate sample dilution factor, to derive the original sample concentration

A B

5.2 HER2 Cell-binding Assay Reagents and Protocols Binding of trastuzumab antibodies to extracellularly expressed HER2 was detected through a cell- binding endpoint affinity assay. Overexpressing HER2 cell line SK-BR3 was used for the assay. Briefly, antibody samples were incubated with SK-BR3 cells, the cells were washed and tagged with a FITC-anti human IgG secondary antibody. The cells were again washed and ran though flow cytometry to detect the level of green fluorescence per cell. Cell populations were defined as positive or negative binding fractions and plotted against antibody concentration on a sigmoidal binding curve. Versene Solution

Prepare EDTA (E6758 Sigma-Aldrich) 0.5 mM stock solution in PBS and sterilize by passing through a 0.22 µm filter. Keep stock solution sealed and refrigerated until use. Dilution Buffer

Prepare 2 mM EDTA (E6758 Sigma-Aldrich), 0.5% (w/v) BSA (BSAS 0.1 Bovogen) stock solution in PBS. Sterilize and degas by passing through a 0.22 µm vacuum filter. Keep stock solution sealed and refrigerated until use. Secondary Antibody

Prepare anti-Human IgG (Fc specific) − FITC antibody produced in goat (F9512 Sigma Aldrich) as a 1:200 dilution in Dilution Buffer to the total volume required for the batch of samples (i.e. 100 µL per sample). Isotype (negative) Control

203

Prepare 100 µL of IgG1, Kappa from human myeloma plasma purified immunoglobulin (I5154 Sigma-Aldrich) to 20 µg/mL in Dilution Buffer. Herceptin (positive) Control

Herceptin® was a kind donation from Genentech (USA). Herceptin was buffer exchanged from its formulation to PBS in preparation of experiments (PBS- Herc). Prepare 100 µL of PBS-Herc to 20 µg/mL in Dilution Buffer. 5.2.1 SK-BR3 Cell Harvesting and Preparation SK-BR3 cells were harvested for the HER2 binding assay when the culture was due for passage (as per 3.5). As the culture is a 50% split for passage, the other 50% was used for the assay. When a culture is due for passage and harvest:  Remove the culture media and wash the cell layer twice with 5 mL PBS  Add 3 mL Versene Solution to cell layer  Incubate flask at 37 °C for 5 min  Observe for detachment and single cell release under the microscope  Add 9 mL maintenance media to the flask  Gently dissociate the cell layer via pipette  From the cell suspension, prepare 50% split for passage: o Transfer 6 mL of cell suspension to new T75 flask o Add 9 mL maintenance media to flask (total 15 mL volume) o Label the flask with cells, passage number, passage date, initial o Incubate the culture  With the remaining 50% (6 mL) cell suspension: o Take a cell count, calculate the total number of cells o Transfer to centrifuge tube and pellet the cells at 300 g, 5 min o Remove the supernatant, agitate the cell pellet (flicking) o Resuspend the cell pellet in 5 mL Dilution Buffer o Repeat pellet and wash with dilution buffer as per the last 3 steps o Repeat pellet, remove supernatant and resuspend the cell pellet in Dilution Buffer to a total concentration of 2 x 106 cells/mL o Prepare 50 µL aliquots of cells for samples in assay o Keep cells on ice / in the fridge for the duration of the assay

5.2.2 Antibody Titrations and Cell Binding Assay For each antibody tested, an 8 sample titration series was performed in 1:2 serial dilutions from 20 µg/mL. All sample dilutions were prepared to 100 µL in Dilution Buffer. In each batch, PBS-Herc was prepared as a positive control and a polyclonal IgG1/κ isotype was prepared as a negative control. 50 µL of each sample dilution was added to an aliquot (50 µL) of cells. An additional wash control was prepared by adding 50 µL Dilution Buffer to an aliquot of cells. All cell aliquots were incubated at 4 °C, 20 mins.

204

The cell aliquots were washed twice as follows: the cells were pelleted (300g, 5 mins), the supernatant removed and the cell pellet resuspended in 1 mL Dilution Buffer. After performing 2 wash steps, the cells were again pelleted and supernatant removed. The secondary antibody was prepared as a 1:200 dilution in Dilution Buffer according to the volume required for the batch (i.e. 100 µL per sample). All cell pellets were resuspended in 100 µL diluted secondary antibody and were further incubated at 4 °C, 20 mins. The cell aliquots were washed twice as described previously. After performing 2 wash steps, the cells were resuspended in 250 µL for detection on flow cytometry. Cell suspension samples remained on ice for the duration of the measurements. 5.2.3 Detection and Analysis with Flow Cytometry Flow cytometry measurements for the cell-binding assay were performed on a Guava® easyCyte 6- 2L benchtop flow cytometer using InCyte 3.1 analysis software (Merck). Prior to sample measurements on flow cytometry, excess SK-BR3 cells were used to calibrate the detection of the cells with the following parameters: Threshold: 1000 - 2000 Forward scatter: 4 – 4.5 Side scatter: 8.85 – 9.25 Flow rate: Medium Collection: 5000 events The forwards and side scatter plot was used to gate the population of single cells (example as depicted in A, single cells gated in the red ellipse). The gated region was used for all further data collection. A histogram of cell counts versus log green fluorescence from the gated region was used to adjust the lasers for optimal detection of green fluorescence – i.e. the cell population was detected at 1 – 2 x 101 green fluorescence units emitted per cell. The control samples were assayed through flow cytometry. Successful batches were confirmed if the Herceptin treated cell population was detected at 1 – 3 x 102 green fluorescence units emitted per cell, and that the isotype and wash controls resided in the same region as the cells alone as detected through the calibration. The control samples were used to define regions in the histogram for positive and negative binding as depicted in B, Herceptin treated cells (green peak), IgG1/κ A

Count 0 10 20 30 40 50 60 70 80 90 100 90 80 70 60 50 40 30 20 10 0

Green-B Fluorescence (Log)

205 isotype treated cells (aqua peak), dilution buffer treated cells (brown peak) and untreated cells (yellow peak). All other titration samples were assayed in the same parameters and data collection method to generate the percentage of cell population residing in the negative (unbound) or positive (bound) regions. A sigmoidal binding curve of antibody concentration was plotted against the percentage of fraction bound (positive counts) in a 5PL non-linear fit, to extrapolate KD at 50% positive. 5.3 Binding Kinetic Assay Materials and Protocols Binding kinetic measurements were obtained though surface plasmon resonance (SPR) performed on a Biacore™ T200, using Biacore T200 Control and Evaluation software (GE HealthCare). All measurements were obtained through operation of single cycle kinetic assays at 25 °C with running buffer HBS-T. All receptors were prepared at 4 nM, and all antibody dilutions in Running Buffer HBS-T. Receptors as ligand were injected at 5 µL/min for 5 min for HIS-tagged capture. Antibody samples as analyte were injected at increasing concentrations at 20 µL/min for 2 min. The chip surface was regenerated with 10 mM glycine HC1 pH 1.5 at 30 µL/min for 1 min. Antibody sample as ligand was prepared as 10 nM and injected at 20 µL/min for 10 sec for ProA capture. TNF-α as analyte was injected at increasing concentrations at 20 µL/min for 2 min. The chip surface was regenerated with 50 mM NaOH at 30 µL/min for 30 sec. Analyte concentrations and disassociation times differed between the assays, each assay is detailed more thoroughly below. Stock Buffer

Prepare PBS containing 0.1% (w/v) BSA (BSAS 0.1 Bovogen). Combine ingredients to the desired volume. Sterilise and degas by passing through a 0.22 µm vacuum filter. Keep buffer sealed and store at – 20 °C until use. Running Buffer (HBS-T)

Deionised H2O containing:  10 mM HEPES (54457 Sigma-Aldrich)  150 mM NaCl (S9888 Sigma-Aldrich)  0.05% (v/v) Tween 20 (P9416 Sigma-Aldrich) Combine ingredients to the desired volume and adjust to pH 7.4. Sterilise and degas by passing through a 0.22 µm vacuum filter. Keep buffer sealed and refrigerated until use. ProA chip

Series S Sensor Chip Protein A (29-1275-55 GE Healthcare) The ProA chip was used for the capture of antibody samples (adalimumab specifically) to screen against soluble homotrimeric TNF-α. CM5 chip

206

Series S Sensor Chip CM5 (29-1049-88 GE Healthcare) Amine Coupling Kit

Amine Coupling Kit (BR-1000-50 GE Healthcare) Anti HIS capture Kit

HIS Capture Kit (28-9950-56 GE Healthcare) CM5 chips were prepared for amine coupling with the anti-HIS capture kit (antiHIS chip) as per manufacturer’s instructions for the capture of HIS-tagged receptors: HER2, FcγR1a, FcγR2B, FcγR3A and FcγR3B. HIS-tagged HER2 receptor

Prepare erbB2/HER2/CD340 human recombinant (SRP6405 Sigma-Aldrich) as 4 µM stock in Stock Buffer. HIS-tagged TNF-α receptor

Prepare Tumor Necrosis Factor-α human TNF-α, recombinant (H8916 Sigma-Aldrich) as 4 µM stock in Stock Buffer.

HIS-tagged FcγR1A receptor

Prepare FCGR1A (CD64) Recombinant Human (10256-H08H-5 Life Technologies) as 4 µM stock in Stock Buffer.

HIS-tagged FcγR2B receptor

Prepare Fc-gamma RIIB/CD32b human recombinant (SRP6396 Sigma-Aldrich) as 4 µM stock in Stock Buffer.

HIS-tagged FcγR3A receptor

Prepare FCGR3A/CD16a human recombinant (SRP6436 Sigma-Aldrich) as 4 µM stock in Stock Buffer.

HIS-tagged FcγR3B receptor

Prepare Fc gamma RIIIB/CD16b human recombinant (SRP6451 Sigma-Aldrich) as 4 µM stock in Stock Buffer. Prepare all receptor stocks as 5 µL aliquots and store at – 80 °C until use. Keep in-use aliquots refrigerated, for use within 1 week. HER2 Captured Binding Assay

HIS-tagged HER2 receptor was prepared as 4 nM and captured over an antiHIS chip at 5 µL/min for 5 min. Trastuzumab samples (specifically) were prepared as 2-fold serial dilutions from 0.5 – 8 nM and injected over the HER2 captured surface in increasing concentrations at 20 µL/min for 2 min, with dissociation time of 60 min. Soluble Homotrimeric TNF-a Binding Assay

207

Adalimumab samples (specifically) were prepared as 10 nM and captured over a ProA chip at 20 µL/min for 10 sec. TNF-α was prepared as 2-fold serial dilutions from 2.5 – 40 nM and injected over the antibody captured surface at increasing concentrations at 20 µL/min for 2 min, with dissociation time of 60 min.

FcγR1A Captured Binding Assay

HIS-tagged FcγR1A receptor was prepared as 4 nM and captured over an antiHIS chip at 5 µL/min for 5 min. Antibody samples were prepared as 2-fold serial dilutions from 2.5 – 40 nM and injected over the FcγR1A captured surface in increasing concentrations at 20 µL/min for 2 min, with dissociation time of 30 min.

FcγR2B, FcγR3A and FcγR3B Captured Binding Assays

HIS-tagged FcγR2B, FcγR3A and FcγR3B receptors were prepared as 4 nM and captured over an antiHIS chip at 5 µL/min for 5 min. Antibody samples were prepared as 2-fold serial dilutions from 25 – 400 nM and injected over the FcγR captured surfaces in increasing concentrations at 20 µL/min for 2 min, with dissociation time of 30 min. Analysis

All sensograms were analysed globally using a Langmuir 1:1 binding model to determine Ka, Kd and KD. All samples were conducted in duplicate to obtain average values and standard deviation. Fluctuations produced from phase transitions through the cycle were excised from the sensograms to produce more robust curve fittings. Kinetic rate and binding constants determined from sensograms were accepted if the goodness of fit value (χ2) was within 5% of the maximum response level (Rmax) of the sensograms and KD was calculated as Kd/Ka.

208

6 Additional data for Chapter 4 and Chapter 5 6.1 LC-MS/MS Representative Data of Tmab Glycosylation Library

Figure 42 Representative data; glycan release profiles of Tmab glycosylation library analysed through LC-MS/MS. Tmab WT (A), L115>N (B), T117>N (C), A121>N (D), Q160>N (E), L177>K (F), L177>N (G), Q178>N (H), L182>N (I), T198>N (J). Glycosylation was confirmed in all mutation sites aside for L177>K and Q160>N.

209

6.2 SPR Sensorgrams of Tmab Glycosylation Library

Figure 43 Analysed SPR sensorgrams of single cycle kinetic assays of the Tmab glycosylation mutation library to surface captured HER2. PBS-Herc (A), Tmab WT (B), L177>K (C), L177>N (D), Q160>N (E), L115>N (F), representative T117>N sensorgram of unreliable analysis (G), A121>N (H), Q178>N (I), single sensorgram of L182>N (J) and T198>N (K).

Figure 44 Analysed SPR sensorgrams of single cycle kinetic assays of the Tmab glycosylation

mutation library to surface captured FcγR1A. PBS-Herc (A), Tmab WT (B), L177>K (C), L177>N (D), Q160>N (E), L115>N (F), representative T117>N sensorgram of unreliable analysis (G), A121>N (H), Q178>N (I), L182>N (J) and T198>N (K). 210

Figure 45 Analysed SPR sensorgrams of single cycle kinetic assays of the Tmab glycosylation mutation library to surface captured FcγR2B. PBS-Herc (A), Tmab WT (B), L177>K (C), L177>N ( D), Q160>N (E), L115>N (F), representative sensorgram of unreliable analysis for T117>N (G), A121>N (H), Q178>N (I), L182>N (J) and T198>N (K).

Figure 46 Analysed SPR sensorgrams of single cycle kinetic assays of the Tmab glycosylation mutation library to surface captured FcγR3A. PBS-Herc (A), Tmab WT (B), L177>K (C), L177>N (D), Q160>N (E), L115>N (F), representative T117>N sensorgram of unreliable analysis (G), A121>N (H), Q178>N (I), L182>N (J) and T198>N (K).

211

6.3 SPR Residual data of Tmab Affinity Library

Figure 47 Residuals of analysed SPR sensorgrams from single cycle kinetic assays of the Tmab affinity mutation library to surface captured HER2 (A) and FcγR1A (B). Tmab samples PBS-Herc (1), Tmab pVITRO1 (2), D28>K (3), S50>E (4) and D102>F (5). Sensorgram data (red/green), curve fitting (black).

212

6.4 SPR Residual data of Tmab Glycosylation Library

Figure 48 Residuals of analysed SPR sensorgrams of single cycle kinetic assays of the Tmab glycosylation mutation library to surface captured HER2. PBS-Herc (A), Tmab WT (B), L177>K (C), L177>N (D), Q160>N (E), L115>N (F), representative T117>N sensorgram of unreliable analysis (G), A121>N (H), Q178>N (I), single sensorgram of L182>N (J) and T198>N (K).

Figure 49 Residuals of analysed SPR sensorgrams of single cycle kinetic assays of the Tmab glycosylation mutation library to surface captured FcγR1A. PBS-Herc (A), Tmab WT (B), L177>K (C), L177>N (D), Q160>N (E), L115>N (F), representative T117>N sensorgram of unreliable analysis (G), A121>N (H), Q178>N (I), L182>N (J) and T198>N (K).

213

Figure 50 Residuals of analysed SPR sensorgrams of single cycle kinetic assays of the Tmab glycosylation mutation library to surface captured FcγR2B. PBS-Herc (A), Tmab WT (B), L177>K (C), L177>N (D), Q160>N (E), L115>N (F), representative sensorgram of unreliable analysis for T117>N (G), A121>N (H), Q178>N (I), L182>N (J) and T198>N (K).

Figure 51 Residuals of analysed SPR sensorgrams of single cycle kinetic assays of the Tmab glycosylation mutation library to surface captured FcγR3A. PBS-Herc (A), Tmab WT (B), L177>K (C), L177>N (D), Q160>N (E), L115>N (F), representative T117>N sensorgram of unreliable analysis (G), A121>N (H), Q178>N (I), L182>N (J) and T198>N (K).

214

7 Additional data for Chapter 6 7.1 Primers; Mutation Efficiency of AdmAb Glycosylation Library

Table 46 List of additional primers designed for AH /AL glycosylation mutations. Nucleotides that mismatch the AdmAb gene sequences, introducing codon mutation (bold).

Sequence Primer 5’ 3’

SDM2 AHV L116N Fwd GGCCAAGGTACCAATGTCACCGTCTCGAGTGC SDM2 AHV L116N Rev GCACTCGAGACGGTGACATTGGTACCTTGGCC

SDM2 AHV T118N Fwd GGTACCCTGGTCAACGTCTCGAGTGC SDM2 AHV T118N Rev GCACTCGAGACGTTGACCAGGGTACC

SDM0 AH1 A122N Fwd CCGTCTCGAGTAATAGCACCAAGGGC SDM0 AH1 A122N Rev GCCCTTGGTGCTATTACTCGAGACGG

Table 47 Mutation efficiencies of the AdmAb glycosylation mutant library.

Isolates Mutation Primer Ligation Mutation Yielded Efficiency Artefacts

L178 > K 41 1/10 (20 %) 1/10 (10 %)

L178 > N 281 1/6 (16.7 %) 0/6 (0 %)

Q160 > N 2 1/2 (50 %) 0/2 (0 %)

L116 > N 17 2/10 (20 %) 1/10 (10 %)

T118 > N 91 3/6 (50 %) 0/6 (0 %)

A122 > N > 300 2/4 (50 %) 0/4 (0 %)

Q179 > N 45 1/10 (10 %) 1/10 (10 %)

L183 > N > 300 1/9 (11 %) 0/9 (0 %)

T199 > N 40 1/8 (12.5 %) 5/8 (62.5 %)

7.2 LC-MS/MS Representative Data of AdmAb Glycosylation Library

215

Intensity C

Figure 52 CID spectrum (A) and ETD spectrum (B) of the chymotryptic peptide NSGALTSGVHTFPAVK for L178>K AdmAb mutant confirming successful aa substitution. HCD spectrum (C) and ETD spectrum (D) of the chymotryptic peptide PAVNQSSGLY for L178>N AdmAb mutant confirming successful aa substitution. 216

Intensity B

Figure 53 HCD spectrum (C) and ETD spectrum (D) of the chymotryptic peptide KVDNALQSGNSNESSVTEQDSKDSTY for Q160>N AdmAb mutant confirming successful aa substitution; glycosylation of this mutant could not be confirmed.

217

Intensity

Figure 54 HCD spectrum (A) and ETD spectrum (B) of the chymotryptic peptide LSTASSLDYWGQGTNVTVSSASTK for L116>N AdmAb mutant confirming successful aa substitution. HCD spectrum of the de-glycosylated chymotryptic peptide GQGTNVTVSSASTKGPAVFPLAPSSK for the L116>N AdmAb mutant (C) confirming successful glycosylation. 218

C Intensity

Figure 55 HCD spectrum (A) and ETD spectrum (B) of the chymotryptic peptide LSTASSLDYWGQGTLVNVSSASTK for T118>N AdmAb mutant confirming successful aa substitution. CID spectrum (C) and ETD spectrum (D) of the de-glycosylated chymotryptic peptide GQGTLVNVSSASTK for the T118>N AdmAb mutant confirming successful glycosylation. 219

Intensity C

Figure 56 HCD spectrum (A) and ETD spectrum (B) of the chymotryptic peptide GQGTLVTVSSNSTK for A122>N AdmAb mutant confirming successful aa substitution. CID spectrum (C) and ETD spectrum (D) of the de-glycosylated chymotryptic peptide LSTASSLDYWGQGTLVTVSSNSTK for the A122>N AdmAb mutant confirming successful glycosylation. 220

Intensity

Figure 57 HCD spectrum (A) and ETD spectrum (B) of the chymotryptic peptide NSGALTSGVHTFPAVLQSSGNY for L183>N AdmAb mutant confirming successful aa substitution. ETD spectrum of the chymotryptic peptide (C) and de-glycosylated chymotryptic peptide (D) SLSSVVTVPSSSLGNQTY for T198>N AdmAb mutant confirming successful aa substitution and glycosylation. 221

7.3 SPR Sensorgrams of Monomer AdmAb Glycosylation Library

Figure 58 Single cycle kinetic sensorgrams of monomeric AdmAb variants binding to soluble homotrimeric TNF-α. Humira® (A), AdmAb WT (B), L178K (C), L178N (D), Q160N (E), L116N (F), T118N (G), A122N (H), Q179N (I), L183N (J) and T199N (K).

Figure 59 Single cycle kinetic sensorgrams of monomeric AdmAb variants binding to FcγR1A. Humira® (A), AdmAb WT (B), L178K (C), L178N (D), Q160N (E), L116N (F), T118N (G), A122N (H), Q179N (I), L183N (J) and T199N (K). 222

Figure 60 Single cycle kinetic sensorgrams of monomeric AdmAb variants binding to FcγR2B. Humira® (A), AdmAb WT (B), L178K (C), L178N (D), L116N (E), T118N (F), A122N (G), Q179N (H), L183N (I) and T199N (J).

Figure 61 Single cycle kinetic sensorgrams of monomeric AdmAb variants binding to FcγR3A. Humira® (A), AdmAb WT (B), L178K (C), L178N (D), Q160N (E), L116N (F), T118N (G), A122N (H), Q179N (I), L183N (J) and T199N (K).

223

7.4 SPR Results and Sensorgrams of AdmAb Glycosylation Library, pre-SEC Purification Table 48 Binding kinetics of the AdmAb glycosylation mutant library to surface captured HER2 * and FcγRs 1A, 2B and 3A as measured through SPR (±SD), one value only, no data (ND).

TNF-α FcγR1A

Ka Kd KD Ka Kd KD (x 105) M-1.s-1 (x 10-5) s-1 pM (x 105) M-1.s-1 (x 10-4) s-1 nM

Humira 11.66 ± 0.01 3.59 ± 0.80 30.77 ± 6.83 5.83 ± 0.38 10.72 ± 0.33 1.84 ± 0.18

AdmAb WT 11.63 ± 0.70 3.95 ± 0.30 33.98 ± 0.59 3.24 ± 0.04 8.63 ± 0.13 2.66 ± 0.08

L178>K 8.91 ± 0.07 4.19 ± 2.20 47.02 ± 25.22 3.02 ± 0.07 9.34 ± 0.03 3.09 ± 0.08

L178>N 9.38 ± 0.71 3.99 ± 0.59 42.56 ± 9.51 3.45 ± 0.09 9.26 ± 0.32 2.68 ± 0.16

Q160>N 9.75 ± 0.58 3.32 ± 0.58 34.05 ± 3.96 3.04 ± 0.31 7.91 ± 0.04 2.61 ± 0.27

L116>N 9.61 ± 0.57 3.69 ± 0.50 38.39 ± 2.88 2.71 ± 0.13 8.83 ± 0.40 3.26 ± 0.29

T118>N 8.70 ± 0.90 6.29 ± 5.41 72.33 ± 75.75 2.59 ± 0.14 2.61 ± 0.16 1.01 ± 0.01

A122>N 9.60 ± 0.02 5.61 ± 3.52 58.45 ± 36.09 3.25 ± 0.18 8.78 ± 0.10 2.70 ± 0.11

Q179>N 12.04 ± 0.24 3.25 ± 2.17 27.01 ± 18.99 2.59 ± 0.16 10.06 ± 0.10 3.88 ± 0.27

L183>N 8.87 ± 0.52 5.21 ± 3.84 58.74 ± 37.75 2.77 ± 0.13 9.23 ± 0.01 3.33 ± 0.16

T199>N 11.25 ± 1.36 6.10 ± 0.38 54.19 ± 9.77 3.10 ± 0.12 8.24 ± 0.29 2.66 ± 0.01

FcγR2B FcγR3A

Ka Kd KD Ka Kd KD (x 104) M-1.s-1 (x 10-4) s-1 nM (x 104) M-1.s-1 (x 10-4) s-1 nM

Humira 56.28 ± 9.59 14.36 ± 8.04 2.55 ± 1.67 38.93 ± 5.41 7.16 ± 2.06 1.84 ± 0.68

AdmAb WT 1.72 ± 0.85 1.64 ± 0.74 9.55 ± 15.96 3.98 ± 0.42 5.69 ± 0.21 14.30 ± 1.99

L178>K 6.87 ± 0.14 8.75 ± 3.16 12.74 ± 4.34 4.00 ± 0.11 5.66 ± 0.19 14.13 ± 0.86

L178>N 1.95* 1.17* 5.96* 4.06 ± 0.46 5.20 ± 0.30 12.81 ± 2.02

Q160>N 1.95 ± 0.21 ND ND 4.30 ± 0.46 6.98 ± 0.59 16.24 ± 3.05

T118>N 8.48 ± 0.36 1.96 ± 1.43 2.31 ± 1.81 9.77 ± 3.88 2.96 ± 0.14 3.03 ± 1.44

A122>N 1.17 ± 0.12 0.88 ± 0.42 7.53 ± 2.74 2.32 ± 0.68 4.17 ± 0.55 18.01 ± 7.42

L183>N 1.71 ± 0.57 0.92 ± 1.21 5.41 ± 9.67 ND ND ND

T199>N 1.38 ± 0.12 1.13* 8.55* 2.51 ± 0.09 4.71 ± 0.17 18.75 ± 1.14 224

Figure 62 Single cycle kinetic sensorgrams of pre-SEC purification AdmAb variants binding to TNF-α. Humira® (A), AdmAb WT (B), L178K (C), L178N (D), Q160N (E), L116N (F), T118N (G), A122N (H), Q179N (I), L183N (J) and T199N (K). A deviation in binding and dissociation was observed for T118N, A122N, Q179N and L183N in the repeated cycle which affected the calculated values and SD. Values are as reported although they warrant repeat testing.

Figure 63 Single cycle kinetic sensorgrams of pre-SEC purification AdmAb variants binding to ® FcγR1A. Humira (A), AdmAb WT (B), L178K (C), L178N (D), Q160N (E), L116N (F), T118N (G), A122N (H), Q179N (I), L183N (J) and T199N (K). 225

Figure 64 Single cycle kinetic sensorgrams of pre-SEC purification AdmAb variants binding to ® FcγR2B. Humira (A), AdmAb WT (B), L178K (C), L178N (D), Q160N (E), T118N (F), A122N (G), L183N (H) and T199N (I). T118N displayed markedly enhanced binding, FcγR2B assay requires optimisation.

Figure 65 Single cycle kinetic sensorgrams of pre-SEC purification AdmAb variants binding to ® FcγR3A. Humira (A), AdmAb WT (B), L178K (C), L178N (D), Q160N (E), T118N (F), A122N (G), L183N (H) and T199N (I). 226

7.5 SPR Residual data of AdmAb Glycosylation Library

Figure 66 Residuals of analysed SPR sensorgrams of single cycle kinetic assays of monomeric AdmAb variants binding to soluble homotrimeric TNF-α. Humira® (A), AdmAb WT (B), L178K (C), L178N (D), Q160N (E), L116N (F), T118N (G), A122N (H), Q179N (I), L183N (J) and T199N (K).

Figure 67 Residuals of analysed SPR sensorgrams of single cycle kinetic assays of monomeric ® AdmAb variants binding to FcγR1A. Humira (A), AdmAb WT (B), L178K (C), L178N (D), Q160N (E), L116N (F), T118N (G), A122N (H), Q179N (I), L183N (J) and T199N (K). 227

Figure 68 Residuals of analysed SPR sensorgrams of single cycle kinetic assays of monomeric ® AdmAb variants binding to FcγR2B. Humira (A), AdmAb WT (B), L178K (C), L178N (D), L116N (E), T118N (F), A122N (G), Q179N (H), L183N (I) and T199N (J).

Figure 69 Residuals of analysed SPR sensorgrams of single cycle kinetic assays of monomeric ® AdmAb variants binding to FcγR3A. Humira (A), AdmAb WT (B), L178K (C), L178N (D), Q160N (E), L116N (F), T118N (G), A122N (H), Q179N (I), L183N (J) and T199N (K). 228

Figure 70 Residuals of analysed SPR sensorgrams of single cycle kinetic assays of pre-SEC purification AdmAb variants binding to TNF-α. Humira® (A), AdmAb WT (B), L178K (C), L178N (D), Q160N (E), L116N (F), T118N (G), A122N (H), Q179N (I), L183N (J) and T199N (K).

Figure 71 Residuals of analysed SPR sensorgrams of single cycle kinetic assays of pre-SEC ® purification AdmAb variants binding to FcγR1A. Humira (A), AdmAb WT (B), L178K (C), L178N (D), Q160N (E), L116N (F), T118N (G), A122N (H), Q179N (I), L183N (J) and T199N (K). 229

Figure 72 Residuals of analysed SPR sensorgrams of single cycle kinetic assays of pre-SEC ® purification AdmAb variants binding to FcγR2B. Humira (A), AdmAb WT (B), L178K (C), L178N (D), Q160N (E), T118N (F), A122N (G), L183N (H) and T199N (I).

Figure 73 Residuals of analysed SPR sensorgrams of single cycle kinetic assays of pre-SEC ® purification AdmAb variants binding to FcγR3A. Humira (A), AdmAb WT (B), L178K (C), L178N (D), Q160N (E), T118N (F), A122N (G), L183N (H) and T199N (I).

230

7.6 Representative Thermal Melting and Aggregation curves of Monomer AdmAb Glycosylation Library

Figure 74 Melting curves of the AdmAb glycosylation mutant library generated from the UNcle™ system. (A) Full temperature ramp of all samples highlighting the unfolding events in reference ® Humira that correspond to the calculated Tm1 and Tm2, (B) curve magnified over the melting regions for comparison of mutant Tms against WT.

231

Figure 75 Thermal aggregation curves of the AdmAb glycosylation mutant library generated from the UNcle™ system, measured at 266 nm. (A) Full temperature ramp of all samples highlighting the ® initiated aggregation event in the reference Humira that correspond to the calculated Tagg, (B) curve magnified over the region where aggregation initiates for comparison of Tagg of mutants against WT.

232

Authorship Attribution Statement

Chapter 3 of this thesis is published as: Elgundi, Z., Sifniotis, V., Reslan, M., Cruz, E., Kayser, V. (2017). Laboratory Scale Production and Purification of a Therapeutic Antibody. Journal of Visualized Experiments, 119: e55153. V. Sifniotis co-designed the project and conducted the majority experimental work including establishing and optimising the expression system, preparing vector DNA, stable transfection, culture adaption to suspension, culture scale up for expression, harvest and antibody titre quantification (BLItz). She also assisted with the writing of the manuscript sections relating to the experimental work that she conducted and was the lead author for the script and videography.

Chapter 6 of this thesis is prepared for publication as: Reslan, M., Sifniotis, V., Cruz, E., Sumer-Bayraktar, Z., Cordwell, S., Kayser, V. (2018). Enhancing the stability of adalimumab by engineering additional glycosylation motifs. European Journal of Pharmaceutics and Biopharmaceutics. V. Sifniotis is the co-first author for this publication. She designed this project, optimised the experiments in the lead up to this project and conducted the majority of the experimental work including cloning the expression vectors, optimising the expression system, mutagenesis and sequencing analysis, production and biological characterisation (SPR) of the mutants. She also wrote a significant portion of the manuscript including the mutation library literature review and rationale, expression of the mutant library methods and biological assay methods, results and discussion. She performed analysis for those sections to produce the relevant figures and tables in the publication. In addition to the statements above, in cases where I am not the corresponding author of a published item, permission to include the published material has been granted by the corresponding author.

Vicki Sifniotis A/Prof Veysel Kayser 13/02/2019 13/02/2019

233

Poster and Publications

234

Review Current Advancements in Addressing Key Challenges of Therapeutic Antibody Design, Manufacture, and Formulation

Vicki Sifniotis, Esteban Cruz, Barbaros Eroglu and Veysel Kayser *

School of Pharmacy, Faculty of Medicine and Health, The University of Sydney, Sydney 2006, Australia; [email protected] (V.S.); [email protected] (E.C.); [email protected] (B.E.) * Correspondence: [email protected]; Tel.: +61-2-9351-3391

Received: 16 April 2019; Accepted: 31 May 2019; Published: 3 June 2019

Abstract: Therapeutic antibody technology heavily dominates the biologics market and continues to present as a significant industrial interest in developing novel and improved antibody treatment strategies. Many noteworthy advancements in the last decades have propelled the success of antibody development; however, there are still opportunities for improvement. In considering such interest to develop antibody therapies, this review summarizes the array of challenges and considerations faced in the design, manufacture, and formulation of therapeutic antibodies, such as stability, bioavailability and immunological engagement. We discuss the advancement of technologies that address these challenges, highlighting key antibody engineered formats that have been adapted. Furthermore, we examine the implication of novel formulation technologies such as nanocarrier delivery systems for the potential to formulate for pulmonary delivery. Finally, we comprehensively discuss developments in computational approaches for the strategic design of antibodies with modulated functions.

Keywords: therapeutic antibody; stability; aggregation; manufacture challenges; formulation

1. Introduction

Since the first therapeutic monoclonal antibody (mAb) Orthoclone OKT3® (Janssen Biotech, Horsham, PA, USA) was approved by the USA Food and Drug Administration in 1986, whole antibody therapeutics have become and persistently remain the most dominant and significant biologic therapeutic platform in the pharmaceutical industry [1,2]. To date, therapeutic antibodies treat a plethora of indications including cancers, infections, autoimmune disorders, and cardiovascular and neurological diseases [3]. The whole antibody therapeutics platform is regarded as the most promising class of pharmaceutical technology to date; it is continually being applied to newly identified biological targets and implemented in many formats to produce strategically engineered next generation antibody therapeutics, otherwise termed “biobetters” [4–9]. The international ImMunoGeneTics information system® (IMGT®, Montpellier, France) database reveals that as of December 2018, 65 whole antibodies and 18 next generation fragment or recombinant fusion antibody-based therapies are approved for clinical use, with hundreds more in clinical trials expected to reach market. Whole therapeutic mAbs are presently the dominant antibody platform approved for clinical use (Figure 1), although antibody engineering technologies have advanced in recent years to produce highly optimized, strategically engineered biobetter therapies, along with biosimilar mAbs reaching market to compete against their originator. Further whole mAb formats include antibody–drug conjugates (ADCs), bispecifics, isotype-switched, and glycoengineered. These additional formats

Antibodies 2019, 8, 36; doi:10.3390/antib8020036 www.mdpi.com/journal/antibodies Antibodies 2019, 8, 36 2 of 22 have been strategically designed to introduce exceptional potency, to engage dual biological targets, and to modulate Fc effector functions. Fragments of mAbs such as the crystallizable fragment (Fc), antigen binding fragment (Fab), and single-chain variable fragment (scFv) possess key functions such as specificity to a biological target or immunological activation. The isolation of these fragments for fusion with other mAb fragments, biologically functional proteins, cytotoxic drugs, or drug carriers has been the crux of ingenuity in developing the next generation of biobetter therapies [4–15]. Figure 2 depicts several examples of prominent biobetter formats, providing a general representation of current fragment mAbs, whole mAb bispecifics, fragment mAb multispecifics, and fragment mAb fusion therapeutics.

Whole mAb formats 65 Naked Whole Naked whole mAbs 60 mAbs ADCs 4 ADCs Whole mAb bispecifics 1

Whole mAb Bispecifics Fragment mAb formats 18 Biosimilars Fabs 4 Fc fusions 10 Fabs scFv fusions 3 scFv bispecifics (BiTE) 1 Originator mAbs with 6 approved biosimilars

(a) (b)

Figure 1. The proportions of therapeutic antibody formats approved for therapeutic use as of December 2018, IMGT® depicted through (a) a pie chart and (b) a table format.

Figure 2. Schematic representation of a whole monoclonal antibody (mAb), a fragment mAb, and prominent fusion mAb formats that have been developed for strategic therapeutic uses. Proteins fused to mAb fragments are depicted as blue ovals for a general representation; however, fusion proteins may vary in size and structure. Fragment formats include the crystallizable (Fc), antigen binding (Fab and F(ab)2), and single-chain variable (scFv) fragments. Further whole mAb formats

Antibodies 2019, 8, 36 3 of 22

include the antibody–drug conjugate (ADC), triomab, dual variable domain immunoglobulin (DVD- Ig), and immunoglobulin–scFv fusion (IgG-scFv). Multispecific fragment formats include the F(ab)2 bispecific, bispecific T-cell engager (BiTE), dual affinity re-targeting molecule (DART), and tandem diabody (tandAb).

2. Overview of mAb Production Challenges and Considerations Whole therapeutic mAbs require a mammalian expression system to produce the biologically functional product; however, a mAb fragment and recombinant fusion mAb products with simplified (or lacking) glycosylation are suitable for lower organism expression platforms [16]. Unlike oligopeptides, which can be chemically synthesized, whole therapeutic mAbs are considerably larger (with monomer ranging from 140–160 kDa) and comprise of four peptide chains (two heavy and two light chains) bound together by disulfide bonds and interchain non-covalent interactions. Further to this, antibodies contain glycosylation in a conserved region of the Fc (N297) that contributes to its stability and immune effector functions [17,18]. Aside from peptide synthesis, cell machinery is required to glycosylate, fold, orient, and covalently bind the antibody peptide chains in order to produce the complete, biologically functional antibody product. Manufacturing biologically functional whole mAb product is therefore commercially unfeasible through chemical synthesis and insufficient in lower organism expression platforms such as bacteria, yeast, insect, and plant cells that may not have the machinery to produce the equivalent tertiary structure and glycosylation profiles. In particular, many industrially relevant bacterial strains such as E. coli are completely deficient in the machinery to add post-translational glycosylations; yeasts hyper-mannosylate glycans, which cause immunogenicity; and insect cells which are deficient in sialylation machinery and produce immunogenic glycan structures [16]. A secretion of the mAb product for purification is suboptimal for several lower organism expression platforms such as E. coli due to poor productivity and harsh culture conditions that promote product degradation. Protein is therefore produced intracellularly, as inclusion bodies and harvest involves further processing steps such as cell lysis, inclusion body recovery, protein solubilization, and renaturation prior to further downstream purification steps [19,20]. Despite these pitfalls, the development of lower organism expression systems is of high commercial interest due to the simplified culture conditions, cheaper media requirements, rapid organism growth, and higher product yield as compared to mammalian expression systems [16,21]. In considering the requirements through the entire process of mAb discovery, manufacture, formulation, and disease treatment, several key challenges arise which have sparked overwhelming interest in pursuit of achieving better mAb manufacturing outcomes and treatment strategies. As with all biotherapeutics, mAbs and mAb-based therapeutics are limited to production in cell-based expression systems, which is considerably costly and inefficient, can have varied yields depending on the product and expression system, and requires downstream processing to remove biological contaminants introduced from the expression system. Despite affinity chromatography being a robust technology for the initial capture of a mAb for purification, the capture process and further downstream processes such as viral inactivation applies the mAb product to harsh pH and salt conditions, which can chemically degrade the mAb, leading to product instability and loss [5,22,23]. Many factors through the manufacture process influence glycosylation and charge heterogeneity of mAbs, which affects their biophysical and pharmacological properties. Though not specifically discussed in this review, the improvement and control mAb production technologies address these variations to reduce formulation heterogeneity and off-target cytotoxicities. A common challenge, as seen with all biotherapeutics, is that mAbs and mAb-based therapeutics are currently restricted to lyophilised and liquid-based formulations for intravenous (IV) or subcutaneous (SC) delivery to achieve maximum bioavailability. Protein self-association and intrinsic stability drive this limitation, in that viscosity and propensity to aggregate are dependent on mAb concentration. Formulations are optimized to achieve the highest dosing concentration at the minimum achievable volume for injection, without compromising the quality of the mAb in formulation [24]. Viscosity remains a key limiting factor for formulating as a SC administration— certain mAb therapies are suitable and others not based on their solubility, self-association, and

Antibodies 2019, 8, 36 4 of 22 aggregation profiles. Alternative non-invasive administration strategies such as pulmonary delivery causes additional mechanical stress that further contribute to mAb instability and loss. Furthermore, oral delivery is unsuitable due to chemical and enzymatic degradation, as well as poor absorption in the gastric and intestinal environments [5,25–29]. A brief overview of considerations through the different concept stages of therapeutic mAb development is depicted in Figure 3. The main challenges and considerations in the manufacture and formulation of mAb therapeutics are briefly summarized in Table 1.

Figure 3. Schematic representation of the concept stages of mAb drug development in which considerations follow on from one process to the next in the design and manufacture of mAb-based therapeutics.

Antibodies 2019, 8, 36 5 of 22

Table 1. Summary of key technological advancements that address challenges and considerations in mAb design, manufacture, and formulation strategies.

Challenges Advancements Manufacture Considerations

Hybridoma technology produces immunogenic mAbs Humanization technologies [4–6]

Yield from hybridoma technology is variable Commercial cell line development and recombinant technology [4–6]; lower organism systems for Significance of post-translational modifications and higher-order fragment mAbs, mammalian systems for whole mAbs, and Fc fusions [16,30,31] structure in mAb product CHO expressed mAbs contain an immunogenic glycosylation Human-based expression systems [16,30–32] profile HEK 293 expressed mAbs are prone to aggregation HKB-11 and PER.C6 cell lines [32] Undesirable byproducts produced in the manufacture process in vitro cell-free synthesis technology [33,34] Stability of mAb affects manufacture yield due to product loss Enhancing mAb stability through framework mutations and hyperglycosylation [5,35–40] through aggregation in downstream processing steps

Treatment Considerations for Drug Design and Formulation

Susceptibility of mAb to degradation limits delivery to intravenous and subcutaneous only Enhancing mAb stability through framework mutations, hyperglycosylation [5,35–40], and nanocarrier technologies [8,41] Stability of mAb limits concentration of formulation

Concentration of mAb affects viscosity and injection pressure for Excipients, fragment mAbs, PEGylation, and hyperglycosylation [25–27,41] subcutaneous delivery Poor tissue penetration and biodistribution Fragment mAbs and nanocarrier technologies [8,41,42]; high affinity ADCs [43–45]

PEGylation, hyper-glycosylation and Fc fusion proteins [8,41,46]; modulation of FcRn recycling Reduced half-life in low MW species through Fc mutation [11,13]

Modulation of immunological engagement Isotype switching, glycoengineering, and Fc mutations [11,13,15,43,46]

Antibodies 2019, 8, 36 6 of 22

3. mAb Discovery and Manufacture Technologies Traditional technology for whole therapeutic antibody discovery required the immunization of animals—primarily mice with a target antigen for the generation of a mixed population of B- lymphocytes-producing antibody against the target. The B-lymphocytes would be isolated for immortalization to produce monoclonal hybridoma cell lines that secrete antibody candidates of interest for further panning through display technologies to isolate potential leads [5]. This method has led to the generation of highly specific mAb libraries of non-human origin that were potentially immunogenic and less efficacious at eliciting an immune response as compared to wholly human mAbs. Technologies arose to address immunogenicity by producing chimeric and humanized antibodies through the grafting of the variable domains (chimeric) or complementarity-determining region (CDR) residues (humanized) isolated from lead non-human antibodies to a human antibody framework. To further improve the humanization strategy, transgenic mice were developed with their murine antibody heavy and light chain genes replaced with equivalent human genes; they consequently expressed wholly human antibodies for discovery [4–6]. Phage display technology remains the most prominent selection technology for panning antibody gene pools for specificity to a target antigen, as it effectively couples the expression of mAb proteins to the genes that encode them for the panning of high affinity leads. Antibody variable genes from B-lymphocyte pools are isolated through polymerase chain reaction (PCR), and cloned into a phage expression vector that presents the expressed mAb on the surface of the phage, the library of vectors is rescued in phage, and the phage library is screened against a specific target antigen. Phage pools undergo several rounds of selection to isolate high affinity candidates, the variable genes of the lead mAb candidates can then be isolated and sequenced for the design of a mAb drug format and the development of a suitable expression platform [5,47]. Error-prone PCR-based mutagenesis has propelled display technology for the generation of enormous libraries for target affinity screening. Yeast surface display further improves these screening technologies, as glycosylation sites introduced in CDRs are expressed through this platform [5,48]. The expression, yield, and quality of a mAb can vary quite substantially in hybridoma cell lines. High yielding mammalian expression systems have been developed to meet the commercial need for high expression efficiency, scalability, quality, and reproducibility. Antibody genes of interest are introduced into suitable expression vectors and transfected into highly efficient mammalian cell lines for antibody expression and secretion; a mAb can then be directly captured from the culture supernatant through affinity chromatography. Currently, the majority of mammalian expression systems for commercial whole therapeutic antibody expression are based on stable chinese hamster ovary (CHO), mouse myeloma (NS0), and mouse hybridoma (Sp2/0) cell lines [4,5]. However, in the context of research and development, the transient expression in human cell lines such as embryonic kidney (HEK 293), amniotic (CAP), a hybrid of HEK 293 and lymphoma (HKB-11), and embryonic retina (PER.C6) are favored over stable CHO expression due to the ease and speed of production of workable quantities of antibody for preliminary studies [16,30–32,49,50]. The generated mAb library undergoes relevant in vitro testing along with formulation stability screening to exclude candidates that have poor manufacturability attributes [51]. Lead mAb candidates that show potential for further investigation would then be developed for high yielding stable expression and more rigorous characterization leading up to therapeutic development. However, a drawback from changing from human to hamster expression systems is the difference in glycosylation profile. CHO expressed mAbs produce immunogenic non-human glycoform Neu5Gc and have a higher composition of sialylation, which may result in reduced antibody-dependent cellular cytotoxicity (ADCC). Therefore in the early stages of development, leads need to be identified and thoroughly characterized in the expression system to be developed for commercial manufacture before committing to industrial scale up [16,30,32]. Amongst other biotherapeutics, approved mAb Fc fusion therapeutics dulglutide (Trulicity®, Eli Lilly, Indianapolis, IN USA), efmoroctocog alpha and eftrenonacog alpha (Eloctate® and Alprolix® , Biogen, Cambridge, MA, USA), are manufactured in a HEK 293 expression system, which is giving rise to the acceptance of human-based expression systems for the production of mAb- based therapeutics [4]. However, HEK 293 expression systems are prone to inducing mAb

Antibodies 2019, 8, 36 7 of 22 aggregation in cultures, which is detrimental to cell viability and creates a loss of product during manufacture. Hence, HKB-11 and PER.C6 are the preferred commercial human cell lines for mAb expression in human cell lines, specifically [16,31,32]. Established lower organism expression systems for fragment or recombinant fusion mAb therapeutics include bacteria such as E. coli, yeasts such as S. cerevisiae and P. pastoris, plants such as tobacco, algae, and insects such as silkworm [5,16,52–54]. Expression platforms that use E. coli in particular are considered high risk due to the potential endotoxin contamination in the mAb product, of which complete endotoxin removal requires further purification steps [19]. However, several approved mAb fragment- and recombinant-based therapies are produced in an E. coli-based expression system, including pegol conjugated Fab’ certolizumab pegol (Cimzia®, Celltech UCB, Brussels, Belgium), Fab ranibizumab (Lucentis®, Genentech, San Francisco, CA, USA), and recombinant Fc fusion romiplostim (Nplate®, Amgen, Thousand Oaks, CA, USA). Lower organism expression platforms such as E. coli and P. pastoris continue to be the preferred option for the manufacture of fragment mAb formats due to the relative ease, high yield, and reduced cost for manufacture as compared to mammalian expression platforms. An emerging in vitro cell-free synthesis technology is being developed with bacterial and CHO cell lysates which have the potential to alleviate formation of undesirable biological byproducts, as the machinery from the cell lysates purely express protein from the mAb genes that are introduced to the system. Though this technology is not currently applicable for industrial scale manufacture, it holds much promise as an alternative to live culture. For instance, culture maintenance and highly defined media are no longer necessary, reactions can run continuously, lysates can be recycled with reproducible results, unmodified linear DNA is suitable for the system (thus alleviating a need for multiple cloning steps), and additional enzymes can be introduced to the system for the engineering of specific post-translational modifications [33,34,55–58]. Despite the apparent advantages of this technology for mAb manufacture, cell lysates encounter the same challenges as their host cell expression system. That is, post-translational modifications of mAb product from lysates from lower level organisms are limited by the endogenous machinery of that host organism. Furthermore, preparation of mammalian cell lysates is challenging, and yield produced from these lysates is considerably low [59,60]. Current bioprocess engineering and cell line development strategies have been crucial in enhancing the manufacturability of mAbs. Many current mAb therapeutic expression platforms make use of CHO cell lines that have been commercially developed with dihydrofolate reductase or glutamine synthetase deficiency for enhanced stability selection through resistance to methotrexate and methionine sulfoximine inhibition, respectively [50,61]. Mammalian cell lines have been adapted to suspension culture, as they support higher cell densities and mAb titers in the absence of serum from the culture media, for large scale fed-batches, perfusion systems, or continuous culture systems. Further to this, cell lines are continually being engineered for enhanced metabolic functions, introducing glycosylation pathways, superior secretion, and resistance to apoptosis for prolonged survival, in efforts to generate super-producer cell lines that can sustain a continuous culture [62–68]. Gene and expression vector design and development are specifically tailored to the expression system to maximize mAb expression and cell line stabilization through host cell codon optimization and the addition of highly efficient transcription, secretion, selection, and integration elements [69– 73]. Vectors can contain a single site for gene insertion for the subsequent co-transfection of vectors that carry the antibody heavy and light chains, or both chains can be cloned into a dual expression vector. In more elaborately designed recombinant mAb drug formats that require the expression of several genes, vector technologies have implemented multicistronic expression to enhance the efficiency of the vector system. The stable or transient transfection of vectors is performed on highly viable cell cultures with high cell density. Antibody heavy and light chain ratios may require adjustment in transfection for optimizing expression and secretion. A reporter vector that expresses green fluorescent protein is typically included to observe transfection efficiency during optimization [74,75]. Liposome-mediated transfection is preferential over all other mammalian transfection strategies and is induced chemically through the complexing of DNA to a cationic lipid prior to or

Antibodies 2019, 8, 36 8 of 22 during addition to a culture. Polyethylenimine is the most prevalently used transfection reagent despite the development of superior reagents such as Lipofectamine™ 2000 and Freestyle™ MAX (Thermo Fisher Scientific, Waltham, MA, USA), LyoVec™ (InvivoGen, San Diego, CA, USA), FuGENE 6™ (Promega, Madison, WI, USA), and TransIT-PRO® (Mirus, Madison, WI, USA), owing to its lower cost and relative efficiency [73,75–78]. The commercial manufacture of mAbs has transitioned to serum-free, chemically defined, and animal-free media which has removed a source of expression variability and has been driven particularly from the advent of mad cow disease [50,79]. Several media supplements such as plant and yeast digests (peptones/hydrolysates), surfactants (e.g., Pluronic F-68), DNA methyltransferase (azacytidine), and histone deacetylase (sodium butyrate, valproic acid) inhibitors have been found to support cell viability and enhance mAb expression [79–85]. Mild hypothermia of the culture has also demonstrated to reduce cell expansion, support prolonged cell viability, and enhance mAb expression [73,86–90]. Further to this, continuous supplementation in expression cultures of uridine, manganese chloride, and galactose demonstrated successful the glycan engineering of expressed mAb, where mAb hyper-galactosylation was promoted to enhance complement-dependent cytotoxicity (CDC) activity [91]. In the manufacturing process of fed-batch, perfusion, and continuous culture systems, media is continuously fed in the culture system, whilst parameters such as cell viability, cell density, and metabolite levels are monitored in real time until the parameters indicate that it is the optimal time to harvest the expressed mAb from the culture supernatant. In perfusion and continuous culture systems, a feed is removed from the culture simultaneously to the media addition, the difference being that the cell dilution rate in continuous culture is optimized to remain equal or higher than the cell growth rate, which allows the perpetuation of mAb expression and requires the continuous harvest of the supernatant [92–94]. The harvest of the supernatant requires clarification technologies such as the use of precipitants/flocculants (e.g., polyethylene glycol, diethylaminoethyl dextran, caprylic acid, and polyethylenimine), high throughput centrifugation, and filtration methods to remove cells and biological debris, for the lowering the loading burden in the lead up to mAb capture through affinity chromatography [19,95–100]. The mechanical stresses from centrifugation and filtration are unavoidable, although they can induce mild mAb product loss through fragmentation and aggregation. Protein A-ligand-based affinity chromatography is the most robust, efficient, and prevalently used mAb capture technology, owing to its selectivity and high affinity to various human Fc, as well as its efficient dissociation of captured mAbs at a low pH for reuse. Fragment mAbs and recombinant formats based on Fabs and single chain variable fragments (scFv) are unable to take advantage of protein A affinity capture, and alternatives have been developed, such as capturing proteins G, M, and L, which bind at different mAb epitopes; ion exchange chromatography; and polyhistidine tagged capture through immobilized metal chromatography [100–102]. A further advantage of affinity chromatography in manufacture is the integration of a viral inactivation step by holding the low pH-eluted mAb product prior to further purification steps; however, this applies the mAb product to low pH at high concentrations, which can induce mild mAb instability, aggregation, and the formation of acidic variants [5,22,100,103,104]. Post mAb capture, further polishing steps are required to remove further contaminants such as host cell proteins and DNA, as well as leached affinity chromatography ligand- and mAb- degradation products such as fragments, aggregates, and ionic variants. The polishing steps are designed based on the specific properties of the mAb product, such as the isoelectric point and molecular weight (MW), to implement appropriate chromatographic technologies such as anion and cation exchanges, hydrophobic interaction, and multimodal and size exclusion that will separate the mAb product from the manufacture-introduced impurities [22,100,103]. The additional chromatographic steps unavoidably apply the mAb product to buffers of varying ionic strength and high concentrations, which can again induce mild mAb instability and aggregation. Post polishing steps, the purified mAb product undergoes a further viral removal step, such as filtration, and then

Antibodies 2019, 8, 36 9 of 22 it undergoes buffer exchange and concentration through ultrafiltration or diafiltration methods to prepare the bulk mAb product for formulation [100,105].

4. Formulation Strategies and Considerations Strategies to improve the formulation of mAbs and mAb-based therapies continues to be an ongoing challenge, as is faced with all biotherapeutics. Firstly, whole therapeutic mAbs are unsuitable for non-invasive oral, nasal, or pulmonary routes of administration as they are susceptible to chemical and enzymatic degradation in the gastrointestinal tract. The bioavailability of mAbs through these routes is poor, owing to mAbs’ polar surface charge and relatively large MW, limiting transport through mucosal membranes. Fragment mAb platforms, together with excipient and PEGylation technologies, have been developed to help circumvent transportation limitations for pulmonary delivery, specifically. However, physical stresses applied to the mAb (e.g., shear stresses from the aerosolization of liquid formulations for a pressurized meter dose and a nebulization delivery, or from the production of dry powder for inhalation) pose further challenges of mAb instability, leading to degradation and reduced efficacy. [8,26,28]. Few biologics have been successfully approved for pulmonary delivery, most notably insulin formulation Afrezza® (MannKind, Westlake Village, CA, USA), and, currently, an erythropoietin-Fc fusion and a nanobody-targeting respiratory syncytial virus are in clinical trials, showing promise for the further development of this administration strategy for both the systemic and localized delivery of mAb-based therapeutics [5,28]. Several drug carrier technologies such as microencapsulation, liposome, and nanoparticle formulations inherently enhance the stability and control the release of mAbs, which can prolong their half-life. These nanocarrier formulation strategies are of intense interest, as they hold promise for developing less invasive inhaled formulations of mAb-based therapeutics with superior attributes to currently established formulations [41,106–109]. The majority of currently approved mAb therapeutics are formulated for IV, although commercial interest has directed technologies to develop injectable mAb formulations which has seen success in many approved mAb therapies thus far, primarily SC for systemic delivery, along with a few intravitreal and intramuscular formulations for tissue-specific indications. Amongst others, anti- HER2 antibody trastuzumab was originally developed as an IV formulation and was successfully repurposed as a SC formulation [110–112], whereas next generation therapies have directly moved towards SC formulation, such as with anti-TNF-α antibody adalimumab [113]. Injectable administrations offer several advantages over IV administration, especially in the treatment of chronic diseases in regards to a reduced burden to allied health services, patient tolerance, and adherence to treatment; however, the intrinsic physicochemical properties of mAbs may be undesirable for injectable formulation. In considering the low volume, injection pressure, and the typically high (>100 mg) effective dose of mAb for injectable delivery, the viscosity and aggregation propensity of mAbs become key formulation challenges to address, as they are dependent on mAb concentration. Certain mAb therapies are suitable, and others are not based on their solubility, viscosity, self-association, intrinsic stability, aggregation, and precipitation profiles. Injectable mAb formulations can be further improved with the use of excipients to increase solubility, reduce viscosity, and enhance the stability of mAbs. Excipients are considered for a formulation based on their physicochemical properties, pharmacokinetics, and safety. For example, polysorbates are commonly used in biologics as a stabilizing agent; however, their addition in high concentrations can denature proteins and cause adverse side effects such as injection site reactions [114–116]. Injectable mAb formulations are co-formulated with recombinant human hyaluronidase, specifically as a permeation enhancer for more efficient absorption into tissue, although the inclusion of this additional biologic adds further burden to the formulation’s viscosity and propensity to aggregate [5,8,25–28,117–123]. Antibody therapies for IV administration are prepared as lyophilised powder for reconstitution and further dilution, and injectable administrations are prepared as liquid- based formulations in pre-filled syringes. Liquid formulations of mAbs are more susceptible to physiochemical degradation, are less stable, and have a reduced shelf-life as compared to lyophilised

Antibodies 2019, 8, 36 10 of 22 formulations. However, drying technologies to produce lyophilised formulations apply the mAb to physical stresses that induce instability and degradation, leading to reduced efficacy [124–126]. Long-term stability predictions are elucidated from formulation screening in accelerated aggregation studies, as part of the preliminary screening process of generated mAb libraries [127]. The stability profiles of mAbs can be greatly affected by the amino acid composition, structure, and potential glycosylation in their variable region CDRs that are isolated through the screening and maturation process. Elucidating the causes for reduced solubility or increased aggregation propensity has to be considered together with the molecular interactions between the mAb and the biological target as to not compromise affinity. Computational tools have been developed to elucidate amino acids within the mAb CDR structure that have a high propensity to aggregate, and strategies for improving solubility and aggregation profiles have been developed through direct amino acid substitutions and the strategic addition of glycosylation sites [128–132]. The characterization of mAb stability and target interaction for optimization is considered a fundamental step as part of preliminary drug discovery, as the applicability for development, manufacture, and formulation of the mAb therapeutic is governed by the mAbs stability profile.

5. Improving mAb Tissue Penetration for Cancer Treatment Antibody therapies are currently restricted to invasive parenteral routes of administration for maximum bioavailability and systemic distribution; however, delivery to the specific target tissue, such as tumors for cancer treatments, continues to be a challenge. Penetration to tissue from blood vessels is again poor, owing to mAbs’ polar surface charge and relatively large MW, limiting transport through physiological barriers. The efficiency of tissue penetration is further influenced by systemic and local mAb clearance rates. Enhancement to mAbs’ affinity to their biological target through maturation and strategic mutation technologies have led to faster diffusion rates of mAb to target for increased efficacy. However, for tumourous tissue specifically, penetration through tumors is restricted by the tumor’s binding site barrier, in which antigens expressed on the tumors periphery capture the majority of the mAb released to surrounding tissue; this effect is exaggerated with overexpressed antigen. The enhancement of whole mAbs’ affinity beyond 1 nM has specifically shown that there is no further improvement of tumor diffusion rates, tissue penetration, and accumulation [42,43,133]. An increased affinity of mAbs to cellular targets also leads to the increased uptake, internalization, and catabolism of mAbs, which reduces ADCC and increases the mAb clearance rate. This mechanism is exploited through the development of high affinity ADCs to target the delivery and release of potent drugs, inducing targeted cell death [43–45]. Fragment mAb platforms have shown better tissue penetration and biodistribution than whole mAb therapeutics; however, a pitfall of smaller peptides lacking an Fc region is a highly reduced in vivo half-life and poor retention times. Technologies to improve the half-life of mAb fragments have been primarily through PEGylation, along with strategic Fc mutation and glycosylation engineering to enhance neonatal Fc receptor (FcRn) recycling. Furthermore, hyper-glycosylation technology has been successfully applied in other biotherapeutics, such as with glycosylated erythropoietin Aranesp® (Amgen, Thousand Oaks, CA, USA). Conversely, nanocarrier platforms that are relatively larger as compared to whole mAbs have demonstrated superior pharmacokinetics and tumor retention, as well as previously mentioned enhanced stability and the sustained, controlled release of mAbs [41]. These enhanced properties have generated considerable interest in developing the systemic delivery of mAb-nanoparticle platforms for superior tumor penetration, as well as targeted and sustained therapeutic drug delivery. Further to nanocarriers, additional formulation strategies developed for the controlled release of mAbs include hydrogels and crystalline antibodies, which have shown success as stable, injectable formulations for development [5,8,15,41–43,134–136]. In addition, to further enhance the stability and half-life of other biotherapeutics, recombinant technologies allowed their fusion to mAb Fc as to introduce FcRn recycling as a protection mechanism, which has innovatively expanded the applicability of mAb-based therapeutic platforms. Notable examples include the TNF Receptor–Fc fusion protein Enbrel® (Amgen, Thousand Oaks, CA, USA), which acts as an inhibitor to overexpressed TNF-α in autoimmune diseases, and the Factor IX–

Antibodies 2019, 8, 36 11 of 22

Fc fusion protein Alprolix® (Biogen, Cambridge, MA, USA), which is a blood factor supplement for hemophiliacs [11,46,133].

6. Strategic Modulation of mAb Immune Effector Functions The modulation of mAbs effector functions through isotype switching, glycoengineering, and strategic mutations have proved advantageous for the development of more effective mAb treatment strategies. Antibody binding to Cq1 promotes the complement cascade, FcγR1A/B, FcγR2A, and FcγR3A/B receptors to activate immune effector functions; FcγR2B counter-balances the effector response, and FcRn prolongs mAb half-life. Binding to these receptors is primarily done through sites in the Chinge, CH2, and the conserved glycosylation region in the Fc (N297). IgG3 does not efficiently bind to FcRn, which reduces its half-life to approximately seven days, as opposed to a 21 day half-life for the other IgG isotypes. In most instances of mAb design, an extended half-life is preferred as to prolong the effective dose of a mAb in serum [15]. IgG2 and IgG4 mAbs have a reduced effector function as compared to IgG1 and are thus used in instances where minimal engagement to the immune system is warranted to increase safety of the mAb therapy— with ADCs reducing off-target cytotoxicity, for instance. Eculizumab (Soliris®, Alexion Pharmaceuticals, New Haven, CT, USA) is the first (and thus far only) approved recombinant IgGκ 2(CH1/Chinge)–4(CH2/CH3) hybrid whole mAb. Further cross-sub-class variant mAb therapies are in development, with key amino acid substitutions from the IgG2 and IgG4 sub-classes introduced for intentionally suppressed effector functions [4,13]. The removal of the conserved glycosylation site through amino acid substitution of N297 or T299 (aglycosylation) has also demonstrated reduced effector function; however, this happens at the expense of mAb stability and is therefore not a suitable strategy for whole mAb therapeutics [11,14,15,137–140]. On the other hand, enhancing ADCC and CDC is useful for engaging the immune system to tumor tissues in the absence of a conjugated cytotoxic drug. Certain glycoengineered modifications to the conserved glycosylation region in the Fc, such as deficiency in core fucose (afucosylation) and hyper-galactosylation, have demonstrated enhanced mAb binding to FcγR3A and Cq1, specifically for an enhanced ADCC and CDC effect [13,43]. Afucosylated mAbs benralizumab (Fasenra™, AstraZeneca, London, UK) and mogamulizumab (Poteligeo®, Kyowa Hakko Kirin, Tokyo, Japan), as well as low fucose content mAb obinutuzumab (Gazyva®, Genentech, San Francisco, CA, USA), are currently approved therapies, with several further in development (along with clinical trials), which demonstrates the commercial interest and applicability of this technology in improving mAb therapeutics through enhancing ADCC. However, the significance of manipulating sialylation in mAb therapeutics is a controversial topic, with several reports demonstrating a reduction in ADCC and CDC and others reporting no observable difference. This highlights a clear need to characterize and define the in vivo efficacy of such manipulation [11,13,14,43]. No currently approved mAb therapies contain amino acid substitutions for an improved half- life through modulated binding properties to FcRn, improved CDC through binding to complement factor Cq1, or improved ADCC through enhanced binding affinity to FcγR1A and 3A. However, several mutations have been reported, patented, and are in clinical development, demonstrating the commercial interest in this technology for improving mAb therapeutics [6,11,13,15,133,137,140–143]. For a more comprehensive understanding of mAb engineering strategies for modulated immune effector functions, we direct the reader to the following reviews [13–15,140].

7. Computational Approaches for Aggregation Prediction and Rational Design of mAbs The advancement of in silico analysis of mAb peptide sequences, structures, conformation, and their associated biological interaction has been integral to the development of various computational tools for characterizing, designing, and optimizing mAb-based therapeutics. The generation and continued pursuit of mAb structural data for in silico analysis has led to the development of several integral databases, notably IMGT®, serving as a key resource for data mining [144–146]. Molecular dynamic (MD) simulation analysis has driven the computational analysis of the discovery and characterization of relevant molecular interactions within the mAb molecule, mAb binding to

Antibodies 2019, 8, 36 12 of 22 biological targets and mAb surface association with the surrounding environment. These interactions can infer intrinsic stability, target binding associations, solubility and aggregation propensity of the mAb. MD simulations and free energy calculations from crystal structures remain key for the highly specific elucidation of mAb-target intermolecular interactions that correlate to binding affinity, as significant interactions such as hydrogen-bond formation can be predicted and determine the strength of the molecular associations [147]. However, elucidation of mAb self-association, solubility, and aggregation propensity have been driven by the development of computational modelling and simulation tools that simultaneously analyses mAb topography and surface polarity. Notably, AGGRESCAN3D, TANGO, and PASTA are the most prominent tools for predicting the site specific aggregation propensity of mAbs [35,148]. The preliminary elucidation of mAb structural data—that being amino acid sequences and higher order structure (HOS)—is experimentally derived to produce crystal structures that model the solid-state 3D structure of the mAb. Mass spectrometry technologies have come of age to produce high throughput and orthogonal analysis to elucidate mAb peptide sequences, oxidation, deamidation, and glycosylation heterogeneity, as well as, more recently, for native, destabilized, and aggregated HOS elucidation [149,150]. X-ray crystallography technologies have been the underlying workhorse for elucidating the crystal structure of mAbs. Producing crystalline mAbs is challenging, owing to the complexity of mAb HOS and the degree of mAb conformational heterogeneity. However, several complementing technologies have evolved in recent years to ascertain structure and interactions, including circular dichroism (CD), infrared (IR) and raman spectroscopy, cryogenic- electron microscopy, and nuclear magnetic resonance (NMR) spectroscopy [148,150–157]. Of the technologies available for HOS elucidation, 2D-NMR and X-Ray crystallography (followed by MS) provide the highest sensitivity and local specificity. In comparison, CD, IR, and raman are high- throughput methods, although they have much lower sensitivity [150,157]. Thus far, only four whole IgG antibodies have a successfully determined crystal structure (PDB ID: 1HZH, 1IGT, 1IGY, and 5DK3), notably 1HZH as a wholly human IgG1 and 5DK3 as a humanized IgG4/κ, both of which have been used as model structures for MD simulations [158]. However, fragment mAbs yield much higher success with producing crystal structures, which has led to much pursuit and contribution in generating mAb fragments and targetting complexing crystal structure libraries for data mining and in silico analysis [159,160]. Upon obtaining relevant crystal structures for a candidate mAb, detailed crystal structure and MD simulation analysis has been extensively used to elucidate the mAbs intramolecular interactions that confer intrinsic stability and intermolecular interactions to confer interactions to biological targets and self-association. The particular binding interface of interest is analyzed, and amino acids in the mAb structure are identified for substitution that can disrupt molecular interactions in the interface in the instance of diminishing target binding and identify potentially unutilized bonding sites and polarity mismatches in the interface that can be improved in the instance of enhancing target binding. MD simulations are then carried out for the native and substituted mAb structures to elucidate any relevant bond formations, interactions, and more favorable binding free energy produced by the substituted structure, of which promising candidates can be synthesized and validated through in vitro analysis. This predictive approach has been successfully applied to identify mutations in mAbs for modulating effector functions, target binding, optimizing affinity capture for manufacturing, and, more recently, for designing antibodies de novo from targets of interest [11,161– 165]. Mutations have been specifically identified for enhanced binding to FcγR2A, FcγR3A, and Cq1, as well as a reduced binding to FcγR2B for a more pronounced ADCC and CDC effect, an enhanced binding to FcRn for an extended half-life, and a reduced binding to FcγRs and Cq1 for a diminished ADCC and CDC effect, all of which are transferrable to mAbs of the same isotype sub-class [137,141,142]. Further to this, the molecular interactions of glycoengineered mAb variants to FcRs have been characterized through MD simulations to corroborate predictions with the observed modulated effector functions [13,15,137,141,142]. Self-association, solubility, and aggregation propensity are further evaluated by computational modelling tools that specifically characterize the topography and surface polarity of mAbs,

Antibodies 2019, 8, 36 13 of 22 concurrently analyzing the local spatial arrangement of amino acids in the mAb structure, solvent accessibility, and local surface charge [5]. In particular, surface exposed hydrophobicity is sought as the lead mechanism for protein self-association driving aggregation propensity, in which aggregation-prone regions (APRs) are identified in the mAb structure. The substitution of identified APRs to gatekeeper (i.e., polar) amino acids has experimentally demonstrated resistance to aggregation and improved solubility, which validates this strategy for improving mAb stability [36– 40,166–170]. The majority of APRs identified are located in biologically relevant mAb regions—those being the CDRs and the Fc. Specifically targeting these APRs through mutation may disrupt important biological functions and mAb structure. Analyses of the mAbs intermolecular interactions are necessary to validate the intrinsic stability of substituted mAb variants; if conformational fluctuations are elucidated, then the biologically active interface is likely disrupted. Interestingly, mutations for enhancing biological functions have demonstrated a negative impact on mAb solubility and stability, further suggesting that the self-association profile of mAbs is linked to its biological activity [36,128,171]. Aside from direct residue substitution in mAb structures, other strategies reported to profoundly interfere with the effects of APRs include isotype switching and the strategic addition of N-linked glycans [39,128,172,173]. Additionally, N-linked glycosylation in the CDR regions of mAbs, although uncommon, has demonstrated improved solubility and stability profiles of mAbs without impacting their target affinity, as compared to the same mAbs with removed CDR glycans [128,174]. The rationale behind strategic glycan addition is based on the election of N-linked glycosylation sites that have apparent spacial proximity to APRs, with the introduced glycan therefore sterically hindering the APR from self-association interactions. The benefits of extending mAb half-life with strategic glycan addition, as seen with hyper-glycosylated biotherapeutics (as well as improving mAb solubility and stability for improved formulation strategies) has yet to be realized and suggests a very intriguing and highly relevant technology for perusal.

8. Concluding Remarks Therapeutic antibodies have come of age as continuing to be a key, dominant technology in the biopharmaceutical industry. The repurpose of antibodies to many formats have made them versatile to design and tailor highly specialized treatments, including ADCs as a targeted drug delivery system, bispecific and fragment mAb platforms for tailored engagement and increased bioavailability, and recombinant Fc-fusion proteins for an increased half-life and introduced immunological engagement. Further advancements include the modulation of Fc effector functions through manipulations of the Fc, either through isotype switching, glycoengineering, or strategic mutations in the Fc region, along with the PEGylation of fragment mAbs for enhanced half-life. Advancements in discovery, manufacture, and formulation technologies have further propelled the success of therapeutic antibodies, notably through expression system development and the transition from IV to SC formulations. Human-based expression systems have been extensively used in mAb development and are becoming an accepted manufacturing platform for mAb therapeutics. Furthermore, cell-free synthesis technology is giving rise to the potential for higher efficiency in the manufacture process. Though developments for further generation antibody therapies have led to great strides in producing improved therapeutic outcomes, many facets of the manufacture process and formulation development strategy pose as challenges to be considered. Antibody-based therapies are susceptible to chemical and enzymatic degradation through oral, nasal, or pulmonary routes of administration and are therefore currently restricted IV or SC delivery. Despite achieving maximum bioavailability through IV/SC administration, tissue penetration of mAb-based therapies is poor, which limits their local bioavailability, requiring high concentrations to achieve an effective dose. The stability of mAbs- based therapeutics is a highly pronounced and recurring challenge to be considered, as it affects manufacture yield and formulation considerations. Several strategies are in development to improve the stability of mAbs in order to potentially produce formulations for pulmonary or oral administration. Notably, computational tools have come of age, complementing experimental

Antibodies 2019, 8, 36 14 of 22 techniques to derive antibody structure and aggregation prediction. Through these methods, the stability and aggregation propensity of mAb-based therapies have demonstrated improvement through rational mutation and glycosylation within the framework region, which is potentially translatable to all mAbs within the same isotype. Furthermore, nanocarrier technologies have been shown to enhance the stability and potentially control the release of mAbs. The refinement of rational mAb design coupled with nanocarrier technologies has the potential to overcome these challenges, to develop superior treatment strategies, and ultimately to formulate for non-invasive administration routes such as pulmonary delivery.

Author Contributions: V.S. revised the figure and prepared the scope, literature review, and original manuscript draft. E.C prepared the original figure and reviewed the manuscript. B.E. and V.K. equally reviewed and edited the manuscript.

Funding: This research received no external funding.

Conflicts of Interest: The authors declare no conflict of interest.

References

1. Urquhart, L. Top drugs and companies by sales in 2017. Nat. Rev. Drug Discov. 2018, 17, 232. 2. Ecker, D.; Dana Jones, S.; L Levine, H. The therapeutic monoclonal antibody market. mAbs 2015, 7, 9–14. 3. An, Z. “Magic bullets” at the center stage of immune therapy: A special issue on therapeutic antibodies. Protein Cell 2018, 9, 1–2. 4. Santos, M.L.d.; Quintilio, W.; Manieri, T.M.; Tsuruta, L.R.; Moro, A.M. Advances and challenges in therapeutic monoclonal antibodies drug development. Bras. J. Pharm. Sci. 2018, 54, doi:10.1590/s2175- 97902018000001007. 5. Elgundi, Z.; Reslan, M.; Cruz, E.; Sifniotis, V.; Kayser, V. The state-of-play and future of antibody therapeutics. Adv. Drug Del. Rev. 2017, 122, 2–19. 6. Almagro, J.C.; Daniels-Wells, T.R.; Perez-Tapia, S.M.; Penichet, M.L. Progress and challenges in the design and clinical development of antibodies for cancer therapy. Front. Immunol. 2018, 8, 1751. 7. Liu, B.N.; Luo, J.H. Research and development of innovative antibody-based drugs. Yaoxue Xuebao 2017, 52, 1811–1819. 8. Awwad, S.; Angkawinitwong, U. Overview of antibody drug delivery. Pharmaceutics 2018, 10, 83. 9. Klein, C. Special issue: Monoclonal antibodies. Antibodies 2018, 7, 17. 10. Wang, C.; Xu, P.; Zhang, L.; Huang, J.; Zhu, K.; Luo, C. Current strategies and applications for precision drug design. Front. Pharmacol. 2018, 9, 787. 11. Strohl, W.R. Current progress in innovative engineered antibodies. Protein Cell 2018, 9, 86–120. 12. Hoffmann, R.M.; Coumbe, B.G.T.; Josephs, D.H.; Mele, S.; Ilieva, K.M.; Cheung, A.; Tutt, A.N.; Spicer, J.F.; Thurston, D.E.; Crescioli, S.; et al. Antibody structure and engineering considerations for the design and function of antibody drug conjugates (adcs). OncoImmunology 2018, 7, e1395127. 13. Wang, X.; Mathieu, M.; Brezski, R.J. Igg fc engineering to modulate antibody effector functions. Protein Cell 2018, 9, 63–73. 14. Pawlowski, J.W.; Bajardi-Taccioli, A.; Houde, D.; Feschenko, M.; Carlage, T.; Kaltashov, I.A. Influence of glycan modification on igg1 biochemical and biophysical properties. J. Pharm. Biomed. Anal. 2018, 151, 133– 144. 15. Fonseca, M.H.G.; Furtado, G.P.; Bezerra, M.R.L.; Pontes, L.Q.; Fernandes, C.F.C. Boosting half-life and effector functions of therapeutic antibodies by fc-engineering: An interaction-function review. Int. J. Biol. Macromol. 2018, 119, 306–311. 16. Mizukami, A.; Caron, A.L.; Picanço-Castro, V.; Swiech, K. Platforms for recombinant therapeutic glycoprotein production. In Methods mol. Biol.; Humana Press Inc.: New York, NY, USA, 2018; Volume 1674, pp. 1–14. 17. Thomson, C.A. Igg structure and function. In Encyclopedia of Immunobiology; Elsevier Inc.: 2016; Volume 2, pp. 15–22. 18. Kayser, V.; Chennamsetty, N.; Voynov, V.; Forrer, K.; Helk, B.; Trout, B.L. Glycosylation influences on the aggregation propensity of therapeutic monoclonal antibodies. Biotechnol. J. 2011, 6, 38–44.

Antibodies 2019, 8, 36 15 of 22

19. Pieracci, J.P.; Armando, J.W.; Westoby, M.; Thommes, J. Chapter 9—Industry review of cell separation and product harvesting methods. In Biopharmaceutical Processing; Jagschies, G., Lindskog, E., Łącki, K., Galliher, P., Eds.; Elsevier: Amsterdam, The Netherlands, 2018; pp. 165–206. 20. Huang, C.-J.; Lin, H.; Yang, X. Industrial production of recombinant therapeutics in escherichia coli and its recent advancements. J. Ind. Microbiol. Biotechnol. 2012, 39, 383–399. 21. Behme, S. Manufacturing of Pharmaceutical Proteins: From Technology to Economy: Second, Revised and Expanded Edition; Wiley: Hoboken, NJ, USA, 2015; pp. 1–427. 22. Ulmer, N.; Vogg, S.; Müller-Späth, T.; Morbidelli, M. Chapter 7—Purification of human monoclonal antibodies and their fragments. In Human monoclonal Antibodies: Methods and Protocols; Steinitz, M., Ed.; Springer: New York, NY, USA, 2019; pp. 163–188. 23. Arosio, P.; Barolo, G.; Müller-Späth, T.; Wu, H.; Morbidelli, M. Aggregation stability of a monoclonal antibody during downstream processing. Pharm. Res. 2011, 28, 1884–1894. 24. Kayser, V.; Chennamsetty, N.; Voynov, V.; Helk, B.; Trout, B.L. Conformational stability and aggregation of therapeutic monoclonal antibodies studied with ans and thioflavin t binding. mAbs 2011, 3, 408–411. 25. Bittner, B.; Richter, W.; Schmidt, J. Subcutaneous administration of biotherapeutics: An overview of current challenges and opportunities. Biodrugs 2018, 32, 425–440. 26. Viola, M.; Sequeira, J.; Seiça, R.; Veiga, F.; Serra, J.; Santos, A.C.; Ribeiro, A.J. Subcutaneous delivery of monoclonal antibodies: How do we get there? J. Control. Release 2018, 286, 301–314. 27. Otvos, L. 6.05—Basic principles of formulation for biotherapeutics: Approaches to alternative drug delivery. In Comprehensive Medicinal Chemistry III; Chackalamannil, S., Rotella, D., Ward, S.E., Eds.; Elsevier: Oxford, UK, 2017; pp. 131–156. 28. Morales, J.O.; Fathe, K.R.; Brunaugh, A.; Ferrati, S.; Li, S.; Montenegro-Nicolini, M.; Mousavikhamene, Z.; McConville, J.T.; Prausnitz, M.R.; Smyth, H.D.C. Challenges and future prospects for the delivery of biologics: Oral mucosal, pulmonary, and transdermal routes. AAPS J. 2017, 19, 652–668. 29. Muheem, A.; Shakeel, F.; Jahangir, M.A.; Anwar, M.; Mallick, N.; Jain, G.K.; Warsi, M.H.; Ahmad, F.J. A review on the strategies for oral delivery of proteins and peptides and their clinical perspectives. Saudi Pharm. J. 2016, 24, 413–428. 30. Hunter, M.; Yuan, P.; Vavilala, D.; Fox, M. Optimization of protein expression in mammalian cells. Curr. Protoc. Protein Sci. 2019, 95, e77. 31. Lindskog, E.K.; Fischer, S.; Wenger, T.; Schulz, P. Chapter 6—Host cells. In Biopharmaceutical Processing; Jagschies, G., Lindskog, E., Łącki, K., Galliher, P., Eds.; Elsevier: Amsterdam, The Netherlands, 2018; pp. 111–130. 32. Fliedl, L.; Grillari, J.; Grillari-Voglauer, R. Human cell lines for the production of recombinant proteins: On the horizon. New Biotechnol. 2015, 32, 673–679. 33. Matsuda, T.; Ito, T.; Takemoto, C.; Katsura, K.; Ikeda, M.; Wakiyama, M.; Kukimoto-Niino, M.; Yokoyama, S.; Kurosawa, Y.; Shirouzu, M. Cell-free synthesis of functional antibody fragments to provide a structural basis for antibody–antigen interaction. PLoS ONE 2018, 13, e0193158. 34. Stech, M.; Kubick, S. Cell-free synthesis meets antibody production: A review. Antibodies 2015, 4, 12. 35. Pujols, J.; Peña-Díaz, S.; Ventura, S. Aggrescan3d: Toward the prediction of the aggregation propensities of protein structures. In Methods Mol. Biol.; Humana Press Inc.: New York, NY, USA, 2018; Volume 1762, pp. 427–443. 36. Van der Kant, R.; Karow-Zwick, A.R.; Van Durme, J.; Blech, M.; Gallardo, R.; Seeliger, D.; Aßfalg, K.; Baatsen, P.; Compernolle, G.; Gils, A.; et al. Prediction and reduction of the aggregation of monoclonal antibodies. J. Mol. Biol. 2017, 429, 1244–1261. 37. Lerch, T.F.; Sharpe, P.; Mayclin, S.J.; Edwards, T.E.; Lee, E.; Conlon, H.D.; Polleck, S.; Rouse, J.C.; Luo, Y.; Zou, Q. Infliximab crystal structures reveal insights into self-association. mAbs 2017, 9, 874–883. 38. Dobson, C.L.; Devine, P.W.A.; Phillips, J.J.; Higazi, D.R.; Lloyd, C.; Popovic, B.; Arnold, J.; Buchanan, A.; Lewis, A.; Goodman, J.; et al. Engineering the surface properties of a human monoclonal antibody prevents self-association and rapid clearance in vivo. Sci. Rep. 2016, 6, 38644. 39. Courtois, F.; Agrawal, N.J.; Lauer, T.M.; Trout, B.L. Rational design of therapeutic mabs against aggregation through protein engineering and incorporation of glycosylation motifs applied to bevacizumab. mAbs 2016, 8, 99–112. 40. Courtois, F.; Schneider, C.P.; Agrawal, N.J.; Trout, B.L. Rational design of biobetters with enhanced stability. J. Pharm. Sci. 2015, 104, 2433–2440.

Antibodies 2019, 8, 36 16 of 22

41. Yu, M.; Wu, J.; Shi, J.; Farokhzad, O.C. Nanotechnology for protein delivery: Overview and perspectives. J. Control. Release 2016, 240, 24–37. 42. Shi, J.; Kantoff, P.W.; Wooster, R.; Farokhzad, O.C. Cancer nanomedicine: Progress, challenges and opportunities. Nat. Rev. Cancer 2016, 17, 20. 43. Dalziel, M.; Beers, S.A.; Cragg, M.S.; Crispin, M. Through the barricades: Overcoming the barriers to effective antibody-based cancer therapeutics. Glycobiology 2018, 28, 697–712. 44. Lambert, J.M.; Berkenblit, A. Antibody-drug conjugates for cancer treatment. Annu. Rev. Med. 2018, 69, 191– 207. 45. Lambert, J.M.; Morris, C.Q. Antibody–drug conjugates (adcs) for personalized treatment of solid tumors: A review. Adv. Ther. 2017, 34, 1015–1035. 46. Levin, D.; Golding, B.; Strome, S.E.; Sauna, Z.E. Fc fusion as a platform technology: Potential for modulating immunogenicity. Trends Biotechnol. 2015, 33, 27–34. 47. Henry, K.A. Next-generation DNA sequencing of vh/vl repertoires: A primer and guide to applications in single-domain antibody discovery. In Phage Display: Methods and Protocols; Hust, M., Lim, T.S., Eds.; Springer: New York, NY, USA, 2018; pp. 425–446. 48. Frenzel, A.; Roskos, L.; Klakamp, S.; Liang, M.; Arends, R.; Green, L. Antibody affinity. In Handbook of Therapeutic Antibodies, 2nd ed.; Wiley Blackwell: Hoboken, NJ, USA, 2014; Volume 1–4, pp. 115–140. 49. Hu, J.; Han, J.; Li, H.; Zhang, X.; Liu, L.; Chen, F.; Zeng, B. Human embryonic kidney 293 cells: A vehicle for biopharmaceutical manufacturing, structural biology, and electrophysiology. Cells Tissues Organs 2018, 205, 1–8. 50. Jacobi, A.; Enenkel, B.; Garidel, P.; Eckermann, C.; Knappenberger, M.; Presser, I.; Kaufmann, H. Process development and manufacturing of therapeutic antibodies. In Handbook of Therapeutic Antibodies, 2nd ed.; Wiley Blackwell: Hoboken, NJ, USA, 2014; Volume 2–4, pp. 601–664. 51. Kayser, V.; Chennamsetty, N.; Voynov, V.; Helk, B.; Forrer, K.; Trout, B.L. A screening tool for therapeutic monoclonal antibodies: Identifying the most stable protein and its best formulation based on thioflavin t binding. Biotechnol. J. 2012, 7, 127–132. 52. Tada, M.; Tatematsu, K.-I.; Ishii-Watabe, A.; Harazono, A.; Takakura, D.; Hashii, N.; Sezutsu, H.; Kawasaki, N. Characterization of anti-cd20 monoclonal antibody produced by transgenic silkworms (bombyx mori). mAbs 2015, 7, 1138–1150. 53. Maccani, A.; Landes, N.; Stadlmayr, G.; Maresch, D.; Leitner, C.; Maurer, M.; Gasser, B.; Ernst, W.; Kunert, R.; Mattanovich, D. Pichia pastoris secretes recombinant proteins less efficiently than chinese hamster ovary cells but allows higher space-time yields for less complex proteins. Biotechnol. J. 2014, 9, 526–537. 54. Frenzel, A.; Hust, M.; Schirrmann, T. Expression of recombinant antibodies. Front. Immunol. 2013, 4, 217. 55. Thoring, L.; Kubick, S. Versatile cell-free protein synthesis systems based on chinese hamster ovary cells. In Recombinant Protein Expression in Mammalian Cells: Methods and Protocols; Hacker, D.L., Ed.; Springer: New York, NY, USA, 2018; pp. 289–308. 56. Tran, K.; Gurramkonda, C.; Cooper, M.A.; Pilli, M.; Taris, J.E.; Selock, N.; Han, T.-C.; Tolosa, M.; Zuber, A.; Peñalber-Johnstone, C.; et al. Cell-free production of a therapeutic protein: Expression, purification, and characterization of recombinant streptokinase using a cho lysate. Biotechnol. Bioeng. 2018, 115, 92–102. 57. Stech, M.; Nikolaeva, O.; Thoring, L.; Stöcklein, W.F.M.; Wüstenhagen, D.A.; Hust, M.; Dübel, S.; Kubick, S. Cell-free synthesis of functional antibodies using a coupled in vitro transcription-translation system based on cho cell lysates. Sci. Rep. 2017, 7, 12030. 58. Li, J.; Wang, H.; Kwon, Y.-C.; Jewett, M.C. Establishing a high yielding streptomyces-based cell-free protein synthesis system. Biotechnol. Bioeng. 2017, 114, 1343–1353. 59. Jaroentomeechai, T.; Stark, J.C.; Natarajan, A.; Glasscock, C.J.; Yates, L.E.; Hsu, K.J.; Mrksich, M.; Jewett, M.C.; DeLisa, M.P. Single-pot glycoprotein biosynthesis using a cell-free transcription-translation system enriched with glycosylation machinery. Nat. Commun. 2018, 9, 2686. 60. Gurramkonda, C.; Rao, A.; Borhani, S.; Pilli, M.; Deldari, S.; Ge, X.; Pezeshk, N.; Han, T.-C.; Tolosa, M.; Kostov, Y.; et al. Improving the recombinant human erythropoietin glycosylation using microsome supplementation in cho cell-free system. Biotechnol. Bioeng. 2018, 115, 1253–1264. 61. Nakamura, T.; Omasa, T. Optimization of cell line development in the gs-cho expression system using a high-throughput, single cell-based clone selection system. J. Biosci. Bioeng. 2015, 120, 323–329. 62. Zhou, Y.; Raju, R.; Alves, C.; Gilbert, A. Debottlenecking protein secretion and reducing protein aggregation in the cellular host. Curr. Opin. Biotechnol. 2018, 53, 151–157.

Antibodies 2019, 8, 36 17 of 22

63. Inwood, S.; Betenbaugh, M.J.; Lal, M.; Shiloach, J. Genome-wide high-throughput rnai screening for identification of genes involved in protein production. In Recombinant Protein Expression in Mammalian Cells: Methods and Protocols; Hacker, D.L., Ed.; Springer: New York, NY, USA, 2018; pp. 209–219. 64. Dangi, A.K.; Sinha, R.; Dwivedi, S.; Gupta, S.K.; Shukla, P. Cell line techniques and gene editing tools for antibody production: A review. Front. Pharmacol. 2018, 9, 630. 65. Delic, M.; Göngrich, R.; Mattanovich, D.; Gasser, B. Engineering of protein folding and secretion— Strategies to overcome bottlenecks for efficient production of recombinant proteins. Antioxid. Redox Signal. 2014, 21, 414–437. 66. Zhong, X.; Wright, J.F. Biological insights into therapeutic protein modifications throughout trafficking and their biopharmaceutical applications. Int. J. Cell Biol. 2013, 2013, 19. 67. Altamirano, C.; Berrios, J.; Vergara, M.; Becerra, S. Advances in improving mammalian cells metabolism for recombinant protein production. Electron. J. Biotechnol. 2013, 16, doi:10.2225/vol16-issue3-fulltext-2. 68. Parola, C.; Mason, D.M.; Zingg, A.; Neumeier, D.; Reddy, S.T. Genome engineering of hybridomas to generate stable cell lines for antibody expression. In Recombinant Protein Expression in Mammalian Cells: Methods and Protocols; Hacker, D.L., Ed.; Springer: New York, NY, USA, 2018; pp. 79–111. 69. You, M.; Yang, Y.; Zhong, C.; Chen, F.; Wang, X.; Jia, T.; Chen, Y.; Zhou, B.; Mi, Q.; Zhao, Q.; et al. Efficient mab production in cho cells with optimized signal peptide, codon, and utr. Appl. Microbiol. Biotechnol. 2018, 102, 5953–5964. 70. Mauro, V.P.; Chappell, S.A. Considerations in the use of codon optimization for recombinant protein expression. In Recombinant Protein Expression in Mammalian Cells: Methods and Protocols; Hacker, D.L., Ed.; Springer: New York, NY, USA, 2018; pp. 275–288. 71. Michael, I.P.; Nagy, A. Inducible protein production in 293 cells using the piggybac transposon system. In Recombinant Protein Expression in Mammalian Cells: Methods and Protocols; Hacker, D.L., Ed.; Springer: New York, NY, USA, 2018; pp. 57–68. 72. Balasubramanian, S. Recombinant cho cell pool generation using piggybac transposon system. In Recombinant Protein Expression in Mammalian Cells: Methods and Protocols; Hacker, D.L., Ed.; Springer: New York, NY, USA, 2018; pp. 69–78. 73. Jäger, V.; Büssow, K.; Schirrmann, T. Transient recombinant protein expression in mammalian cells. In Animal Cell Culture; Al-Rubeai, M., Ed.; Springer International Publishing: Cham, Switzerland, 2015; pp. 27–64. 74. Wijesuriya, S.D.; Pongo, E.; Tomic, M.; Zhang, F.; Garcia-Rodriquez, C.; Conrad, F.; Farr-Jones, S.; Marks, J.D.; Horwitz, A.H. Antibody engineering to improve manufacturability. Protein Expression Purif. 2018, 149, 75–83. 75. L’Abbé, D.; Bisson, L.; Gervais, C.; Grazzini, E.; Durocher, Y. Transient gene expression in suspension hek293-ebna1 cells. In Recombinant Protein Expression in Mammalian Cells: Methods and Protocols; Hacker, D.L., Ed.; Springer: New York, NY, USA, 2018; pp. 1–16. 76. Arena, T.A.; Harms, P.D.; Wong, A.W. High throughput transfection of hek293 cells for transient protein production. In Recombinant Protein Expression in Mammalian Cells: Methods and Protocols, Hacker, D.L., Ed.; Springer: New York, NY, USA, 2018; pp. 179–187. 77. Agrawal, V.; Yu, B.; Pagila, R.; Yang, B.; Simonsen, C.; Beske, O. A high-yielding, cho-k1-based transient transfection system. BioProcess Int. 2013, 11, 28–35. 78. Hacker, D.L.; Kiseljak, D.; Rajendra, Y.; Thurnheer, S.; Baldi, L.; Wurm, F.M. Polyethyleneimine-based transient gene expression processes for suspension-adapted hek-293e and cho-dg44 cells. Protein Expression Purif. 2013, 92, 67–76. 79. Ritacco, F.V.; Wu, Y.; Khetan, A. Cell culture media for recombinant protein expression in chinese hamster ovary (cho) cells: History, key components, and optimization strategies. Biotechnol. Prog. 2018, 34, 1407– 1426. 80. Whitford, W.G.; Lundgren, M.; Fairbank, A. Chapter 8—cell culture media in bioprocessing. In Biopharmaceutical Processing; Jagschies, G., Lindskog, E., Łącki, K., Galliher, P., Eds.; Elsevier: Amsterdam, The Netherlands, 2018; pp. 147–162. 81. Davami, F.; Eghbalpour, F.; Barkhordari, F.; Mahboudi, F. Effect of peptone feeding on transient gene expression process in cho dg44. Avicenna J. Med. Biotechnol. 2014, 6, 147–155.

Antibodies 2019, 8, 36 18 of 22

82. Mahboudi, F.; Abolhassan, M.R.; Azarpanah, A.; Aghajani-Lazarjani, H.; Sadeghi-Haskoo, M.A.; Maleknia, S.; Vaziri, B. The role of different supplements in expression level of monoclonal antibody against human cd20. Avicenna J. Med. Biotechnol. 2013, 5, 140–147. 83. You, M.; Liu, Y.; Chen, Y.; Guo, J.; Wu, J.; Fu, Y.; Shen, R.; Qi, R.; Luo, W.; Xia, N. Maximizing antibody production in suspension-cultured mammalian cells by the customized transient gene expression method. Biosci. Biotechnol. Biochem. 2013, 77, 1207–1213. 84. Backliwal, G.; Hildinger, M.; Kuettel, I.; Delegrange, F.; Hacker, D.L.; Wurm, F.M. Valproic acid: A viable alternative to sodium butyrate for enhancing protein expression in mammalian cell cultures. Biotechnol. Bioeng. 2008, 101, 182–189. 85. Elgundi, Z.; Sifniotis, V.; Reslan, M.; Cruz, E.; Kayser, V. Laboratory scale production and purification of a therapeutic antibody. JoVE 2017, doi:10.3791/55153. 86. Kim, S.J.; Ha, G.S.; Lee, G.; Lim, S.I.; Lee, C.M.; Yang, Y.H.; Lee, J.; Kim, J.E.; Lee, J.H.; Shin, Y.; et al. Enhanced expression of soluble antibody fragments by low-temperature and overdosing with a nitrogen source. Enzyme Microb. Technol. 2018, 115, 9–15. 87. Lalonde, M.-E.; Durocher, Y. Therapeutic glycoprotein production in mammalian cells. J. Biotechnol. 2017, 251, 128–140. 88. Rajendra, Y.; Kiseljak, D.; Baldi, L.; Hacker, D.L.; Wurm, F.M. A simple high-yielding process for transient gene expression in cho cells. J. Biotechnol. 2011, 153, 22–26. 89. Codamo, J.; Munro, T.P.; Hughes, B.S.; Song, M.; Gray, P.P. Enhanced cho cell-based transient gene expression with the epi-cho expression system. Mol. Biotechnol. 2011, 48, 109–115. 90. Codamo, J.; Hou, J.J.C.; Hughes, B.S.; Gray, P.P.; Munro, T.P. Efficient mab production in cho cells incorporating pei-mediated transfection, mild hypothermia and the co-expression of xbp-1. J. Chem. Technol. Biotechnol. 2011, 86, 923–934. 91. Castan, A.; Schulz, P.; Wenger, T.; Fischer, S. Chapter 7—Cell line development. In Biopharmaceutical Processing; Jagschies, G., Lindskog, E., Łącki, K., Galliher, P., Eds.; Elsevier: Amsterdam, The Netherlands, 2018; pp. 131–146. 92. Lindskog, E.K. Chapter 31—the upstream process: Principal modes of operation. In Biopharmaceutical Processing; Jagschies, G., Lindskog, E., Łącki, K., Galliher, P., Eds.; Elsevier: Amsterdam, The Netherlands, 2018; pp. 625–635. 93. Fisher, A.C.; Kamga, M.-H.; Agarabi, C.; Brorson, K.; Lee, S.L.; Yoon, S. The current scientific and regulatory landscape in advancing integrated continuous biopharmaceutical manufacturing. Trends Biotechnol. 2018, 37, 253–267. 94. Rathore, A.S.; Agarwal, H.; Sharma, A.K.; Pathak, M.; Muthukumar, S. Continuous processing for production of biopharmaceuticals. Prep. Biochem. Biotechnol. 2015, 45, 836–849. 95. Shukla, A.A.; Suda, E. Chapter 3—Harvest and recovery of monoclonal antibodies: Cell removal and clarification. In Process Scale Purification of Antibodies; Gottschalk, U., Ed.; John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2017; pp. 55–79. 96. Thömmes, J.; Twyman, R.M.; Gottschalk, U. Chapter 10—Alternatives to packed-bed chromatography for antibody extraction and purification. In Process Scale Purification of Antibodies; Gottschalk, U., Ed.; John Wiley & Sons: Hoboken, NJ, USA, 2017; pp. 215–231. 97. Glynn, J. Chapter 11—Process-scale precipitation of impurities in mammalian cell culture broth. In Process Scale Purification of Antibodies; Gottschalk, U., Ed.; John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2017; pp. 233–246. 98. Singh, N.; Arunkumar, A.; Chollangi, S.; Tan, Z.G.; Borys, M.; Li, Z.J. Clarification technologies for monoclonal antibody manufacturing processes: Current state and future perspectives. Biotechnol. Bioeng. 2016, 113, 698–716. 99. Singh, N.; Chollangi, S. Chapter 4—Next-generation clarification technologies for the downstream processing of antibodies. In Process scale Purification of Antibodies; Gottschalk, U., Ed.; John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2017; pp. 81–112. 100. Kelley, B. Chapter 1—Downstream processing of monoclonal antibodies: Current practices and future opportunities. In Process Scale Purification of Antibodies; Gottschalk, U., Ed.; John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2017; pp. 1–21. 101. Danielsson, Å. Chapter 17—Affinity chromatography. In Biopharmaceutical Processing; Jagschies, G., Lindskog, E., Łącki, K., Galliher, P., Eds.; Elsevier: Amsterdam, The Netherlands, 2018; pp. 367–378.

Antibodies 2019, 8, 36 19 of 22

102. Kinna, A.; Tolner, B.; Rota, E.M.; Titchener-Hooker, N.; Nesbeth, D.; Chester, K. Imac capture of recombinant protein from unclarified mammalian cell feed streams. Biotechnol. Bioeng. 2016, 113, 130–140. 103. Ghose, S.; Jin, M.; Liu, J.; Hickey, J.; Lee, S. Chapter 14—Integrated polishing steps for monoclonal antibody purification. In Process Scale Purification of Antibodies; Gottschalk, U., Ed.; John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2017; pp. 303–323. 104. Joshi, V.; Shivach, T.; Kumar, V.; Yadav, N.; Rathore, A. Avoiding antibody aggregation during processing: Establishing hold times. Biotechnol. J. 2014, 9, 1195–1205. 105. Liderfelt, J.; Royce, J. Chapter 23—Filtration methods for use in purification processes (concentration and buffer exchange). In Biopharmaceutical Processing; Jagschies, G., Lindskog, E., Łącki, K., Galliher, P., Eds.; Elsevier: Amsterdam, The Netherlands, 2018; pp. 441–453. 106. Anselmo, A.C.; Gokarn, Y.; Mitragotri, S. Non-invasive delivery strategies for biologics. Nat. Rev. Drug Discov. 2018, 18, 19. 107. Sousa, D.; Ferreira, D.; Rodrigues, J.L.; Rodrigues, L.R. Chapter 14—Nanotechnology in targeted drug delivery and therapeutics. In Applications of Targeted Nano Drugs and Delivery Systems; Mohapatra, S.S., Ranjan, S., Dasgupta, N., Mishra, R.K., Thomas, S., Eds.; Elsevier: Amsterdam, The Netherlands, 2019; pp. 357–409. 108. Jani, R.; Krupa, G.; Rupal, J. Active targeting of nanoparticles: An innovative technology for drug delivery in cancer therapeutics. J. Drug Deliv. Ther. 2019, doi:10.22270/jddt.v9i1-s.2356. 109. Abdelaziz, H.M.; Gaber, M.; Abd-Elwakil, M.M.; Mabrouk, M.T.; Elgohary, M.M.; Kamel, N.M.; Kabary, D.M.; Freag, M.S.; Samaha, M.W.; Mortada, S.M.; et al. Inhalable particulate drug delivery systems for lung cancer therapy: Nanoparticles, microparticles, nanocomposites and nanoaggregates. J. Control. Release 2018, 269, 374–392. 110. Jackisch, C.; Kim, S.B.; Semiglazov, V.; Melichar, B.; Pivot, X.; Hillenbach, C.; Stroyakovskiy, D.; Lum, B.L.; Elliott, R.; Weber, H.A.; et al. Subcutaneous versus intravenous formulation of trastuzumab for her2- positive early breast cancer: Updated results from the phase iii hannah study. Ann. Oncol. 2015, 26, 320– 325. 111. Lambertini, M.; Pondé, N.F.; Solinas, C.; de Azambuja, E. Adjuvant trastuzumab: A 10-year overview of its benefit. Expert Rev. Anticancer Ther. 2017, 17, 61–74. 112. Sanford, M. Subcutaneous trastuzumab: A review of its use in her2-positive breast cancer. Target. Oncol. 2014, 9, 85–94. 113. Reichert, J.M. Adalimumab (humira®). In Handbook of Therapeutic Antibodies, 2nd ed.; Wiley Blackwell: Hoboken, NJ, USA, 2014; Volume 3–4, pp. 1309–1322. 114. Crommelin, D.J.A.; Hawe, A.; Jiskoot, W. Formulation of biologics including biopharmaceutical considerations. In Pharmaceutical Biotechnology: Fundamentals and Applications; Crommelin, D.J.A., Sindelar, R.D., Meibohm, B., Eds.; Springer International Publishing: Cham, Switzerland, 2019; pp. 83–103. 115. Singh, S.K.; Mahler, H.-C.; Hartman, C.; Stark, C.A. Are injection site reactions in monoclonal antibody therapies caused by polysorbate excipient degradants? J. Pharm. Sci. 2018, 107, 2735–2741. 116. Rayaprolu, B.M.; Strawser, J.J.; Anyarambhatla, G. Excipients in parenteral formulations: Selection considerations and effective utilization with small molecules and biologics. Drug Dev. Ind. Pharm. 2018, 44, 1565–1571. 117. Pimentel, F.F.; Morgan, G.; Tiezzi, D.G.; de Andrade, J.M. Development of new formulations of biologics: Expectations, immunogenicity, and safety for subcutaneous trastuzumab. Pharmaceut. Med. 2018, 32, 319– 325. 118. Garidel, P.; Kuhn, A.B.; Schäfer, L.V.; Karow-Zwick, A.R.; Blech, M. High-concentration protein formulations: How high is high? Eur. J. Pharm. Biopharm. 2017, 119, 353–360. 119. Kemter, K.; Altrichter, J.; Derwand, R.; Kriehuber, T.; Reinauer, E.; Scholz, M. Amino acid-based advanced liquid formulation development for highly concentrated therapeutic antibodies balances physical and chemical stability and low viscosity. Biotechnol. J. 2018, 13, e1700523. 120. Mandal, A.; Pal, D.; Agrahari, V.; Trinh, H.M.; Joseph, M.; Mitra, A.K. Ocular delivery of proteins and peptides: Challenges and novel formulation approaches. Adv. Drug Del. Rev. 2018, 126, 67–95. 121. Reslan, M.; Kayser, V. The effect of deuterium oxide on the conformational stability and aggregation of bovine serum albumin. Pharm. Dev. Technol. 2016, 1–7, doi:10.1080/10837450.2016.1268157. 122. Reslan, M.; Ranganathan, V.; Macfarlane, D.R.; Kayser, V. Choline ionic liquid enhances the stability of herceptin(r) (trastuzumab). Chem. Commun. (Camb.) 2018, 54, 10622–10625.

Antibodies 2019, 8, 36 20 of 22

123. Reslan, M.; Kayser, V. Ionic liquids as biocompatible stabilizers of proteins. Biophys. Rev. 2018, 10, 781–793. 124. Emami, F.; Vatanara, A.; Park, E.J.; Na, D.H. Drying technologies for the stability and bioavailability of biopharmaceuticals. Pharmaceutics 2018, 10, 131. 125. Izutsu, K.I. Applications of freezing and freeze-drying in pharmaceutical formulations. In Adv. Exp. Med. Biol.; Springer: New York, NY, USA, 2018; Volume 1081, pp. 371–383. 126. Schermeyer, M.T.; Wöll, A.K.; Kokke, B.; Eppink, M.; Hubbuch, J. Characterization of highly concentrated antibody solution—A toolbox for the description of protein long-term solution stability. mAbs 2017, 9, 1169– 1185. 127. Kayser, V.; Chennamsetty, N.; Voynov, V.; Helk, B.; Forrer, K.; Trout, B.L. Evaluation of a non-arrhenius model for therapeutic monoclonal antibody aggregation. J. Pharm. Sci. 2011, 100, 2526–2542. 128. Rabia, L.A.; Desai, A.A.; Jhajj, H.S.; Tessier, P.M. Understanding and overcoming trade-offs between antibody affinity, specificity, stability and solubility. Biochem. Eng. J. 2018, 137, 365–374. 129. van de Bovenkamp, F.S.; Derksen, N.I.L.; van Breemen, M.J.; de Taeye, S.W.; Ooijevaar-de Heer, P.; Sanders, R.W.; Rispens, T. Variable domain n-linked glycans acquired during antigen-specific immune responses can contribute to immunoglobulin g antibody stability. Front. Immunol. 2018, 9, 740. 130. Kuhn, A.B.; Kube, S.; Karow-Zwick, A.R.; Seeliger, D.; Garidel, P.; Blech, M.; Schäfer, L.V. Improved solution-state properties of monoclonal antibodies by targeted mutations. J. Phys. Chem. B 2017, 121, 10818– 10827. 131. Singh, S.N.; Yadav, S.; Shire, S.J.; Kalonia, D.S. Dipole-dipole interaction in antibody solutions: Correlation with viscosity behavior at high concentration. Pharm. Res. 2014, 31, 2549–2558. 132. Ionescu, R.M.; Vlasak, J.; Price, C.; Kirchmeier, M. Contribution of variable domains to the stability of humanized igg1 monoclonal antibodies. J. Pharm. Sci. 2008, 97, 1414–1426. 133. Liu, L. Pharmacokinetics of monoclonal antibodies and fc-fusion proteins. Protein Cell 2018, 9, 15–32. 134. Schweizer, D.; Serno, T.; Goepferich, A. Controlled release of therapeutic antibody formats. Eur. J. Pharm. Biopharm. 2014, 88, 291–309. 135. Rudnick, S.I.; Adams, G.P. Affinity and avidity in antibody-based tumor targeting. Cancer Biother. Radiopharm. 2009, 24, 155–161. 136. Ickenstein, L.M.; Garidel, P. Hydrogel formulations for biologicals: Current spotlight from a commercial perspective. Ther. Deliv. 2018, 9, 221–230. 137. Saxena, A.; Wu, D. Advances in therapeutic fc engineering—Modulation of igg-associated effector functions and serum half-life. Front. Immunol. 2016, 7, 580. 138. Jefferis, R. Glycosylation of antibody molecules. In Handbook of Therapeutic Antibodies, 2nd ed.; Wiley Blackwell: Hoboken, NJ, USA, 2014; Volume 1–4, pp. 171–200. 139. Zheng, K.; Bantog, C.; Bayer, R. The impact of glycosylation on monoclonal antibody conformation and stability. mAbs 2011, 3, 568–576. 140. Lei, C.; Gong, R.; Ying, T. Editorial: Antibody fc engineering: Towards better therapeutics. Front. Immunol. 2018, 9, 2450. 141. Booth, B.J.; Ramakrishnan, B.; Narayan, K.; Wollacott, A.M.; Babcock, G.J.; Shriver, Z.; Viswanathan, K. Extending human igg half-life using structure-guided design. mAbs 2018, 10, 1098–1110. 142. Kellner, C.; Otte, A.; Cappuzzello, E.; Klausz, K.; Peipp, M. Modulating cytotoxic effector functions by fc engineering to improve cancer therapy. Transfus. Med. Hemoth. 2017, 44, 327–336. 143. Wang, Q.; Chen, Y.; Pelletier, M.; Cvitkovic, R.; Bonnell, J.; Chang, C.Y.; Koksal, A.C.; O’Connor, E.; Gao, X.; Yu, X.Q.; et al. Enhancement of antibody functions through fc multiplications. mAbs 2017, 9, 393–403. 144. Lefranc, M.P.; Ehrenmann, F.; Kossida, S.; Giudicelli, V.; Duroux, P. Use of imgt® databases and tools for antibody engineering and humanization. In Methods Mol. Biol.; Humana Press Inc.: New York, NY, USA, 2018; Volume 1827, pp. 35–69. 145. Lefranc, M.-P.; Giudicelli, V.; Duroux, P.; Jabado-Michaloud, J.; Folch, G.; Aouinti, S.; Carillon, E.; Duvergey, H.; Houles, A.; Paysan-Lafosse, T.; et al. Imgt(®), the international immunogenetics information system(®) 25 years on. Nucleic Acids Res. 2015, 43, D413–D422. 146. Martin, A.C.R.; Allen, J. Bioinformatics tools for analysis of antibodies. In Handbook of Therapeutic Antibodies, 2nd ed.; Wiley Blackwell: Hoboken, NJ, USA, 2014; Volume 1–4, pp. 201–228. 147. Śledź, P.; Caflisch, A. Protein structure-based drug design: From docking to molecular dynamics. Curr. Opin. Struct. Biol. 2018, 48, 93–102.

Antibodies 2019, 8, 36 21 of 22

148. Pandya, A.; Howard, M.J.; Zloh, M.; Dalby, P.A. An evaluation of the potential of nmr spectroscopy and computational modelling methods to inform biopharmaceutical formulations. Pharmaceutics 2018, 10, 1–24. 149. Rathore, D.; Faustino, A.; Schiel, J.; Pang, E.; Boyne, M.; Rogstad, S. The role of mass spectrometry in the characterization of biologic protein products. Expert Rev. Proteomic. 2018, 15, 431–449. 150. Wang, X.; An, Z.; Luo, W.; Xia, N.; Zhao, Q. Molecular and functional analysis of monoclonal antibodies in support of biologics development. Protein Cell 2018, 9, 74–85. 151. Baker Edward, N. Crystallography and the development of therapeutic medicines. IUCrJ 2018, 5, 118–119. 152. Vénien-Bryan, C.; Li, Z.; Vuillard, L.; Boutin, J.A. Cryo-electron microscopy and X-ray crystallography: Complementary approaches to structural biology and drug discovery. Acta Crystallogr. Sect. F Struct. Biol. Commun. 2017, 73, 174–183. 153. Brader, M.L.; Baker, E.N.; Dunn, M.F.; Laue, T.M.; Carpenter, J.F. Using x-ray crystallography to simplify and accelerate biologics drug development. J. Pharm. Sci. 2017, 106, 477–494. 154. Nero, T.L.; Parker, M.W.; Morton, C.J. Protein structure and computational drug discovery. Biochem. Soc. Trans. 2018, 46, 1367–1379. 155. Blaffert, J.; Haeri, H.H.; Blech, M.; Hinderberger, D.; Garidel, P. Spectroscopic methods for assessing the molecular origins of macroscopic solution properties of highly concentrated liquid protein solutions. Anal. Biochem. 2018, 561–562, 70–88. 156. Brinson, R.G.; Marino, J.P.; Delaglio, F.; Arbogast, L.W.; Evans, R.M.; Kearsley, A.; Gingras, G.; Ghasriani, H.; Aubin, Y.; Pierens, G.K.; et al. Enabling adoption of 2d-nmr for the higher order structure assessment of monoclonal antibody therapeutics. mAbs 2019, 11, 94–105. 157. Young, J.A.; Gabrielson, J.P. Higher order structure methods for similarity assessment. In Biosimilars: Regulatory, Clinical, and Biopharmaceutical Development; Gutka, H.J., Yang, H., Kakar, S., Eds.; Springer International Publishing: Cham, Switzerland, 2018; pp. 321–337. 158. Kumar, S.; Plotnikov, N.V.; Rouse, J.C.; Singh, S.K. Biopharmaceutical informatics: Supporting biologic drug development via molecular modelling and informatics. J. Pharm. Pharmacol. 2018, 70, 595–608. 159. Westbrook, J.D.; Burley, S.K. How structural biologists and the protein data bank contributed to recent fda new drug approvals. Structure 2018, 27, 211–217. 160. Burley, S.K.; Berman, H.M.; Christie, C.; Duarte, J.M.; Feng, Z.; Westbrook, J.; Young, J.; Zardecki, C. Rcsb protein data bank: Sustaining a living digital data resource that enables breakthroughs in scientific research and biomedical education. Protein Sci. 2018, 27, 316–330. 161. Kuroda, D.; Tsumoto, K. Antibody affinity maturation by computational design. In Methods Mol. Biol.; Humana Press Inc.: New York, NY, USA, 2018; Volume 1827, pp. 15–34. 162. Fischman, S.; Ofran, Y. Computational design of antibodies. Curr. Opin. Struct. Biol. 2018, 51, 156–162. 163. Adolf-Bryfogle, J.; Kalyuzhniy, O.; Kubitz, M.; Weitzner, B.D.; Hu, X.; Adachi, Y.; Schief, W.R.; Dunbrack, R.L., Jr. Rosettaantibodydesign (rabd): A general framework for computational antibody design. PLoS Comp. Biol. 2018, 14, e1006112. 164. Branco, R.J.; Dias, A.M.; Roque, A.C. Understanding the molecular recognition between antibody fragments and protein a biomimetic ligand. J. Chromatogr. A 2012, 1244, 106–115. 165. Huang, B.; Liu, F.F.; Dong, X.Y.; Sun, Y. Molecular mechanism of the affinity interactions between protein a and human immunoglobulin g1 revealed by molecular simulations. J. Phys. Chem. B 2011, 115, 4168–4176. 166. van der Kant, R.; van Durme, J.; Rousseau, F.; Schymkowitz, J. Solubis: Optimizing protein solubility by minimal point mutations. In Methods Mol. Biol.; Humana Press Inc.: New York, NY, USA, 2019; Volume 1873, pp. 317–333. 167. Gil-Garcia, M.; Banó-Polo, M.; Varejao, N.; Jamroz, M.; Kuriata, A.; Díaz-Caballero, M.; Lascorz, J.; Morel, B.; Navarro, S.; Reverter, D.; et al. Combining structural aggregation propensity and stability predictions to redesign protein solubility. Mol. Pharm. 2018, 15, 3846–3859. 168. Seeliger, D.; Schulz, P.; Litzenburger, T.; Spitz, J.; Hoerer, S.; Blech, M.; Enenkel, B.; Studts, J.M.; Garidel, P.; Karow, A.R. Boosting antibody developability through rational sequence optimization. mAbs 2015, 7, 505– 515. 169. Chennamsetty, N.; Voynov, V.; Kayser, V.; Helk, B.; Trout, B.L. Prediction of aggregation prone regions of therapeutic proteins. J. Phys. Chem. B 2010, 114, 6614–6624. 170. Chennamsetty, N.; Voynov, V.; Kayser, V.; Helk, B.; Trout, B.L. Design of therapeutic proteins with enhanced stability. Proc. Natl. Acad. Sci. USA 2009, 106, 11937–11942.

Antibodies 2019, 8, 36 22 of 22

171. Majumder, S.; Jones, M.T.; Kimmel, M.; Alphonse Ignatius, A. Probing conformational diversity of fc domains in aggregation-prone monoclonal antibodies. Pharm. Res. 2018, 35, 220. 172. Nakamura, H.; Oda-Ueda, N.; Ueda, T.; Ohkuri, T. Introduction of a glycosylation site in the constant region decreases the aggregation of adalimumab fab. Biochem. Biophys. Res. Commun. 2018, 503, 752–756. 173. Pepinsky, R.B.; Silvian, L.; Berkowitz, S.A.; Farrington, G.; Lugovskoy, A.; Walus, L.; Eldredge, J.; Capili, A.; Mi, S.; Graff, C.; et al. Improving the solubility of anti-lingo-1 monoclonal antibody li33 by isotype switching and targeted mutagenesis. Protein Sci. 2010, 19, 954–966. 174. Wu, S.-J.; Luo, J.; O’Neil, K.T.; Kang, J.; Lacy, E.R.; Canziani, G.; Baker, A.; Huang, M.; Tang, Q.M.; Raju, T.S.; et al. Structure-based engineering of a monoclonal antibody for improved solubility. Protein Eng. Des. Sel. 2010, 23, 643–651.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Rational addition of glycosylation sites on trastuzumab Glycoengineering a model antibody for enhanced stability

Vicki Sifniotis1, Mouhamad Reslan1, Esteban Cruz1, Veysel Kayser1|| ||Corresponding author 1School of Pharmacy, Faculty of Medicine and Health, The University of Sydney

Table 2: Melting temperature (Tm) and onset of aggregation (Tagg) INTRODUCTION Antibodies like many biologics, are lost in production of Tmabs and mutants. Intrinsic tryptophan fluorescence and static and storage due to degradation and aggregation. Current formulation light scattering measurements were taken simultaneously over a strategies to enhance stability include addition of stabilising excipients temperature ramp. Mutant E became unstable compared to Tmab, and lyophilisation. whereas mutants B, F and G showed enhanced intrinsic stability and Trastuzumab (Herceptin®) is used to treat HER2 positive breast resistance to aggregation. cancers and possesses homology to many antibody therapeutics. Modifications in the peptide sequence of trastuzumab that prove Tm1 (°C) Tm2 (°C) Tagg (°C) successful in enhancing stability may be applied to the whole class of Tmab 68.5 81.3 81.54 IgG1 antibodies, which can lead to improved formulation, administration and treatment strategies. A 69.0 79.3 78.51 B 70.5 78.8 82.13 AIM This study reports on the strategic addition of N-linked glycans to D 71.0 81.9 74.15 the Fab region of trastuzumab for the purpose of enhancing its E 54.9 68.3 ND intrinsic stability. F 69.6 80.5 82.41 G 69.4 80.5 82.22 Table 1: Surface exposed amino acid residues identified in H 68.6 81.3 64.83 trastuzumab Fab region for substitution. I 69.8 82.5 80.61

Mutant Region Proposed Function D A Remove APR by introducing polar amino acid [1] E C B H1 Remove APR and introduce glycosylation site [1] C C kappa Sterically hinder APR by introducing glycosylation D A/B C site [1] F E join I F Improve solubility by introducing glycosylation site [2] G Sterically hinder APR by introducing glycosylation H H C site [1, 2] G H1 C Improve solubility and sterically hinder self- I association by introducing glycosylation site [2, 3] Figure 1: Crystal structure 1n8z of trastuzumab Fab fragment complexed to extracellular domain of

HER2 receptor. Trastuzumab VH/CH1 (blue), VL/CL (green), HER2 fragment (orange), aggregation prone region (APR) elected for this study (red) and surface exposed amino acids identified for substitution to an N- kDa Tmab A B G I E H F linked glycosylation site (purple). 250 150 Figure 3: Titration of Tmab and mutants to SK-BR 3. Affinity of Table 3: Kinetics and binding affinity of Tmab mutants to HER2, FcyR1A and FcγR3A (± SD) measured Tmab and mutants to surface receptor HER2 were tested in situ 100 through surface plasmon resonance. Single cycle kinetics was performed to assay Tmab and mutants against SK-BR 3 by an endpoint titration assay detected with flow 75 over a receptor captured surface. With the exception of mutant E (not shown) Tmab and mutants showed cytometry. All Tmab mutants retained affinity to HER2 exposed on the similarity in binding kinetics; the biological functions of the mutants had not been compromised. cell surface. HER2 K (pM) Fc R1A K (nM) Fc R3A K (nM) 50 D γ D γ D DISCUSSION It was found that glycosylated mutants B, F and G Tmab 12.11 ± 2.72 5.05 ± 0.33 1.13 ±0.32 exhibited enhanced intrinsic stability and that all mutants (except E) retained their biological functions as compared to Tmab. Also, despite 37 A 72.97 ± 3.64 2.90 ± 0.01 1.00 ±0.25 successful mutation, glycan addition in mutant C was unsuccessful, B 34.28 ± 2.01 3.00 ± 0.15 10.67 ±1.03 which is contrary to previous reports on this mutation [1]. C 33.53 ± 15.86 2.59 ± 0.02 3.78 ±0.46 25 D 47.56 ± 25.89 2.34 ± 0.26 8.17 ±1.60 CONCLUSION The successful mutants warrant further investigation and demonstrate the effectiveness of our approach for the F 54.24 ± 10.24 2.29 ± 0.08 7.01 ±3.84 Figure 2: Reducing SDS-PAGE gel of several trastuzumab (Tmab) development of highly stable whole therapeutic antibodies. This can mutants. Mutants B, D (not shown), E, F G, H and I showed G 31.88 ± 2.11 1.25 ± 0.02 5.04 ±1.47 lead to improved formulation, administration and treatment strategies, glycosylation in the heavy chain. Mutant C was not glycosylated in the H ND 3.37 ± 0.22 6.24 ±1.30 and may increase antibody half-life and yield from production light chain (not shown). processes. I 83.44 ± 77.77 1.72 ± 0.03 3.81 ±0.41 Successful mutation and glycan addition was confirmed through mass spectrometry (not shown). Glycan addition was unsuccessful for REFERENCES [1] Courtois, F., et al. mAbs, 2016. 8(1): p. 99-112. [2] Pepinsky, R.B., et al. Protein Science, 2010. 19(5): p. 954-966. [3] Lerch, T.F., et al. mAbs, 2017. 9(5): p. 874-883. mutant C, so was excluded from further stability assessment. Advanced Drug Delivery Reviews 122 (2017) 2–19

Contents lists available at ScienceDirect

Advanced Drug Delivery Reviews

journal homepage: www.elsevier.com/locate/addr

The state-of-play and future of antibody therapeutics☆

Zehra Elgundi, Mouhamad Reslan, Esteban Cruz, Vicki Sifniotis, Veysel Kayser ⁎

Faculty of Pharmacy, The University of Sydney, Pharmacy and Bank Building, Science Road, NSW 2006, Australia article info abstract

Article history: It has been over four decades since the development of monoclonal antibodies (mAbs) using a hybridoma cell Received 5 September 2016 line was first reported. Since then more than thirty therapeutic antibodies have been marketed, mostly as oncol- Received in revised form 26 November 2016 ogy, autoimmune and inflammatory therapeutics. While antibodies are very efficient, their cost-effectiveness has Accepted 28 November 2016 always been discussed owing to their high costs, accumulating to more than one billion dollars from preclinical Available online 2 December 2016 development through to market approval. Because of this, therapeutic antibodies are inaccessible to some pa-

Keywords: tients in both developed and developing countries. The growing interest in biosimilar antibodies as affordable Monoclonal antibodies (mAbs) versions of therapeutic antibodies may provide alternative treatment options as well potentially decreasing Therapeutic antibodies costs. As certain markets begin to capitalize on this opportunity, regulatory authorities continue to refine the re- Biologicals quirements for demonstrating quality, efficacy and safety of biosimilar compared to originator products. In addi- Biosimilars tion to biosimilars, innovations in antibody engineering are providing the opportunity to design biobetter Biobetters antibodies with improved properties to maximize efficacy. Enhancing effector function, antibody drug conjugates Antibody engineering (ADC) or targeting multiple disease pathways via multi-specific antibodies are being explored. The manufactur- Antibody drug conjugates (ADC) ing process of antibodies is also moving forward with advancements relating to host cell production and purifi- Bispecific antibodies cation processes. Studies into the physical and chemical degradation pathways of antibodies are contributing to Manufacture Degradation the design of more stable proteins guided by computational tools. Moreover, the delivery and pharmacokinetics Stability of antibody-based therapeutics are improving as optimized formulations are pursued through the implementa- Formulation tion of recent innovations in the field. © 2016 Elsevier B.V. All rights reserved.

Contents

1. Introduction...... 3 2. Antibodydiscoverystrategies...... 3 3. Novelantibodiesinapprovalandpreclinicaldevelopmentstages...... 4 4. Biosimilars,guidelinesandchallenges...... 5 5. Biobetterantibodies...... 6 5.1. Fcengineeredantibodiesforenhancedeffectorfunctions...... 6 5.2. Antibodydrugconjugates(ADC)...... 8 5.3. Bispeciﬁcs...... 9 6. Manufactureofantibodies...... 11 7. Physicalandchemicaldegradationofantibodies...... 12 7.1. Aggregation...... 12 7.2. Denaturation...... 12 7.3. Fragmentation...... 12 7.4. Deamidation...... 12 7.5. Oxidation...... 13 7.6. Computationaldesigntools...... 13 8. Optimizationofantibodybioavailabilityanddelivery...... 13 9. Conclusions...... 14 Disclosures...... 14

☆ This review is part of the Advanced Drug Delivery Reviews theme issue on “Editor Collection 2017”. ⁎ Corresponding author. E-mail address: [email protected] (V. Kayser).

Acknowledgements...... 14 References...... 14

1. Introduction The concept of display technology provides a direct physical link between a gene (genotype) and the encoding antibody fragment (pheno- Over the past twenty years, therapeutic antibodies have rapidly be- type) to allow selection of genes that encode a protein with the desired come the leading product within the biopharmaceutical market. In binding function. Phage display technology remains the most widely 2013, therapeutic antibodies represented 50% of the $140 billion taken used in vitro method for the display of large repertoires and for the se- by the biopharmaceutical market with sales growing from $39 billion lection of high affinity antibodies to biologically relevant targets [6]. in 2008 to $75 billion in 2013 [1]. There are currently more than thirty Phage display involves the expression of proteins on the surface of fila- therapeutic antibodies approved for established markets such as the mentous phage via fusion with phage coat protein with the genetic se- United States and Europe with over three hundred antibody-based quence packaged within, linking phenotype to genotype selection. products in clinical development [1–3]. Therapeutic antibodies are no When combined with antibody libraries, phage display allows for longer full-length, naked mouse antibodies; advancements in antibody rapid in vitro selection of antigen-specific antibodies and recovery of engineering technologies, novel antigen discovery strategies and prog- their corresponding coding sequence [12–14]. This system is highly ef- ress in deciphering disease pathways have all generated robust interest, fective, robust and amenable to high throughput processes for screening resources and investment in antibody development. In this review, we of N1010 specificities [15]. The diversity of phage display libraries is dis- will discuss developments in the field of therapeutic antibodies, the tinguishable by source and design: naïve [16], immune [12], synthetic growth of biosimilars and pay particular attention to targeting degrada- [17] and semi-synthetic [18]. The technology was first demonstrated tion pathways of antibodies to produce more stable biobetter antibodies for a single chain variable fragment (scFv) [13] with screening of other and formulations. formats also introduced including human antigen binding fragments (Fabs) [19], domain antibodies [20], camelid domain antibodies [21], 2. Antibody discovery strategies single domain shark antibodies [22], diabodies [23] and even whole IgG [24].Thefirst approved human antibody isolated by phage display The generation of early antibodies relied on the immunization of technology was adalimumab (Humira®) which binds the cytokine mice or other mammals with the desired antigen target. This resulted TNF. This antibody was first selected as a scFv expressed on the surface in multiple antibodies directed at different epitopes of the antigen se- of phage and was further engineered in human IgG1 format, providing creted by a mixed population of B cells with each cell secreting only major validation for phage display technologies [25]. Adalimumab re- one specific antibody (i.e. polyclonal). Unfortunately, secreting B cells mains the most lucrative antibody product generating global sales of can only replicate a limited number of times, therefore rendering mass $11 billion in 2013 [1]. The number of phage display-derived candidates production all but impossible. The ground-breaking hybridoma technol- currently in clinical development further demonstrates the value of ogy developed by Kohler and Milstein allowed antibody secreting cells phage display as an established and reliable drug discovery platform from the spleen of immunized animals to be fused with immortalized [26]. non-antibody secreting cells, thus resulting in cells that would divide Other cell surface methodologies such as bacterial, baculovirus and continuously when cultivated in permissive conditions [4,5]. Although yeast display present thousands of copies of the displayed protein on the first recombinant antibodies were produced using this technology, the cell surface thereby allowing for quantitative screening such as including the first approved therapeutic antibody muromonab-CD3 flow cytometry [27–29]. Yeast display has the additional feature of eu- (Orthoclone OKT®3) in 1986 for preventing kidney transplant rejection, karyotic protein folding pathways and as a result, is perhaps the most hybridoma production presented some drawbacks. Hybridomas can be ideally suited system for the surface display of mammalian secreted labor intensive, low yielding or genetically unstable [6]. More impor- proteins such as antibodies. Alternatively, ribosome display is used to tantly though, the antibody sequences originated from an immunized screen large scFv and single variable domain libraries containing up to animal and consequently had the potential of triggering an immune re- N1013 mutant clones [30]. The main distinguishing feature of this meth- sponse in humans. Therefore, further improvements were needed to od is that large libraries can be generated because the entire procedure yield antibodies more human-like and safe. These technologies have is performed in vitro without the need for cell transformations. The ge- evolved from chimeric antibodies that is, grafting essential mouse notype and phenotype are linked through ribosomal complexes, amino acids needed for antigen binding onto a human antibody frame- consisting of messenger RNA (mRNA), ribosome and encoded protein. work [7,8],tobothin vitro and in vivo techniques for generating human- The in vitro selection of antibodies from antibody libraries based on tar- ized antibodies. get affinity is termed “panning”. The method is an iterative process The XenoMouse™ (Abgenix) and HuMab-Mouse® (Medarex) are whereby the population of target specificantibodiesisenrichedrelative transgenic mice developed in parallel and in both, the endogenous mu- to the number of panning rounds. rine heavy and kappa light chain genes are inactivated and replaced With the advent of next-generation sequencing (NGS), analysis of with the equivalent human germline sequences [9,10]. Injection of anti- the natural and synthetic repertoires from which libraries have been gens into these mice leads to development of ‘fully human’ antibodies constructed has become possible [31–33]. In one application, NGS was that have undergone mouse somatic hypermutation and selection to used to identify antigen-specific IgG antibodies from polyclonal serum relatively high affinity. Validation of this technology came with the reg- of immunized rabbits and mice when used in combination with affinity ulatory approval of panitumumab (Vectibix®) in 2006; a fully human chromatography coupled to LC-MS/MS [34]. The technology has been antibody directed against epidermal growth factor receptor (EGFR) as used to demonstrate its utility in selecting antibodies with favorable treatment for advanced colorectal cancer [11]. Since then, RANK properties. For example, Reddy et al. found that 21/27 (or 78%) of scFv ligand-specific denosumab (Prolia®) has been approved for bone loss they constructed could bind the target antigen with nanomolar affinity and TNF-specific golimumab (Simponi®) for rheumatoid arthritis. after using NGS data to pair together the most abundant variable heavy An in vitro method for generating fully human antibodies can be ac- (VH) and variable light (VL) genes from immunized mice [35]. Other complished by cloning and screening large libraries of sufficiently di- studies have seen similar findings with natural VH:VL pairing limita- verse human antibody genes in combination with display technology. tions [36] and without the limitation [37].Interestingly,themethodof 4 Z. Elgundi et al. / Advanced Drug Delivery Reviews 122 (2017) 2–19 repertoire mining of VH and VL abundances using NGS of splenocytes (BLA) submission and awaiting FDA approval or have completed isolated from immunized mice was compared with a phage panning ap- Phase III trials with endpoints met (Table 1). proach of the same cDNA [38]. While both methods provided complete- There is expected to be an influx of antibodies in the coming years ly different sets of antibodies, specificity and affinities of the antibody for the treatment of inflammatory skin conditions including psoriasis, clones were comparable. encompassing various cytokine targets or their respective receptors; IL-17a, IL-17R, IL-1α, IL-4R and IL-23 (Table 1). Other promising antibodies based on recent clinical trial results include ocrelizumab and 3. Novel antibodies in approval and preclinical development stages atezolizumab. Ocrelizumab is a humanized antibody which selectively targets CD20-positive B cells which based on emerging evidence play In 2015, eight therapeutic antibodies were granted market approval a major role in activating T cells, the key cell type responsible for inflam- by the United States Food and Drug Administration (FDA) (Table 1). matory damage within central nervous system lesions in multiple scle- While full-length and IgG1 antibodies still dominate, a humanized Fab rosis (MS) [39]. Data from three Phase III studies (OPERA I, II and also gained approval. Five of the eight approvals were to non- ORATORIO) show positive results in patients with relapsing MS and pri- oncology indications with two antibodies entering ‘first-in-class’ for mary progressive MS (PPMS) with superiority to well-established ther- the treatment of hyperlipidemia. The outlook for 2016 has been positive apy (Rebif®), by reducing the three major markers of disease activity. with six antibodies approved (as of November 2016), including a first- Atezolizumab is a monoclonal antibody designed to target PD-L1 in-class antibody to bacterial target Bacillus anthraxis.Anumberof expressed on T cells and tumor-infiltrating immune cells, preventing full-length antibodies are currently under biologics license application binding to PD-1 and B7.1 to mediate T cell immune responses. Positive

Table 1 Novel full-length antibodies and fragments developed in recent years and their clinical status.

Name Commercial Company Target Indication (FDA approved) Format name

Approved by FDA in 2015 Secukinumab Costentyx® Novartis IL-17 Psoriasis Whole IgG1 (fully human) Dinutuximab Unituxin® United Therapeutics GD2 Neuroblastoma (pediatric Whole IgG1 (chimeric) Corporation patients) Alirocumab Praulent® Sanoﬁ PCSK9 Hyperlipidemia Whole IgG1 (fully human) Evolocumab Repatha® Amgen PCSK9 Hyperlipidemia Whole IgG2 (fully human) Idarucizumab Praxbind® Boehringer Ingelheim Dabigatran Anti-coagulation reversal Fab fragment (humanized) Mepolizumab Nucala® GlaxoSmithKline IL-5 Asthma Whole IgG1 (humanized) Necitumumab Portrazza® Eli Lilly EGFR Non-small cell lung cancer Whole IgG1 (fully human) (NSCLC) Daratumumab Darzalex® Janssen Biotech CD38 Multiple myeloma Whole IgG1 (fully human)

Approved by FDA in 2016 (as of November) Reslizumab Cinquil® Teva IL-5 Asthma Whole IgG4 (humanized) Ixekizumab Taltz® Eli Lilly IL-17a Psoriasis Whole IgG4 (humanized) Obiltoxaximab Anthim® Elusys Therapeutics Bacillus anthrax Anthrax Whole IgG3 (humanized) Atezolizumab Tecentriq® Genentech PD-L1 NSCLC, Bladder cancer Whole IgG1 (humanized) Olaratumab (+ Lartruvo® Eli Lilly PDGFRalpha Soft tissue carcinoma Whole IgG1 Doxorubicin) Bezlotoxumab Zinplava® Merck Clostridium difﬁcile Clostridium difﬁcile infection Whole IgG1 (fully human) toxin B

Not yet approved, trials completed Brodalumab Valeant Pharmaceuticals IL-17R Psoriasis Whole IgG1 (fully human) Bimagrumab Novartis ACVR2B Muscle loss and weakness Whole IgG1 (fully human) MABp1 Xilonix XBiotech IL-1alpha Colorectal cancer, Cachexia Whole IgG1 (‘true’ human) Catumaxomab Removab Trion Pharma EPCAM/CD3 Malignant ascites Whole hybrid of rat IgG2 variable and mouse IgG2a Fc regions Guselkumab Janssen Biotech IL-23/p19 Psoriasis Whole IgG1 (fully human) Dupilumab Sanoﬁ/Regeneron IL-4R Atopic Dermatitis Whole IgG4 (fully human)

Ongoing Phase III trials Ocrelizumab Ocrevus Roche CD20 Multiple Sclerosis Whole IgG1 (humanized) Sirukumab GlaxoSmithKline IL-6 Rheumatoid arthritis Whole IgG1 Sarilumab Sanoﬁ/Regeneron IL-6R Rheumatoid arthritis Whole IgG1 (fully human) Tildrakizumab Sun Pharmaceutical IL-23/p19 Psoriasis Whole IgG1 (humanized) Romosozumab Amgen/UCB Pharma Sclerostin Osteoporosis Whole IgG2 (humanized) Racotumomab Vaxira Recombio NGcGM3 NSCLC Whole IgG1 (mouse) Clivatuzumab hPAM4-Cide Immunomedics Mucin1 Pancreatic cancer Whole IgG1 (humanized) tetraxetan Ublituximab TG Therapeutics CD20 Chronic Lymphocytic Leukemia Whole IgG1 (chimeric) Benralizumab AstraZeneca IL-5R Asthma Whole IgG1 (humanized) Caplacizumab Ablynx vWF Thrombotic thrombocytopenic VH-VH (humanized) purpura Lampalizumab Roche CFD Macular degeneration Fab fragment (humanized) Avelumab Merck/Pfzier PD-L1 Merkel Cell Carcinoma Whole IgG1 (fully human) Aducanumab Biogen Beta amyloid Alzheimer’s Disease Whole IgG1 (fully human)

Abbreviations: GD = disialoganglioside; PCSK = proprotein convertase subtilisin/kexin; EGFR = epidermal growth factor receptor; PD-L = programmed death-ligand; PDGFR = platelet- derived growth factor receptor; ACVR = activin receptor; EPCAM = epithelial cell adhesion molecule; NGcGM = N-glycolyl (NGc) gangliosides; vWF = Von Willebrand factor; CF = complement factor. Z. Elgundi et al. / Advanced Drug Delivery Reviews 122 (2017) 2–19 5 results announced from two Phase II studies (POPLAR and BIRCH) in pa- degree of similarity between the biosimilar and originator reference tients with advanced non-small cell lung cancer (NSCLC) show strong must be established through analytical and preclinical evaluation in correlation between PD-L1 expression and response rates, suggesting order to warrant abbreviated clinical studies [43,44]. that measuring PD-L1 may help identify people most likely to respond The European Medicines Agency (EMA) Committee for Medicinal to antibody treatment. Atezolizumab has since been granted FDA ap- Products for Human Use (CHMP) and the FDA are the leading interna- proval (after priority review) for treatment of a specifictypeofbladder tional authorities for regulating sales and administration of therapeu- cancer and NSCLC. tics. Other regulating bodies in the world adopt their guidelines, along A number of novel formats and treatment options are also likely to with clinical context outlined from the World Health Organization enter the market in 2017. Of note based on novelty, are racotumomab, (WHO) or International Conference of Harmonization (ICH). In ublituximab, MABp1 and caplacizumab. Racotumomab is an anti- Australia, the Therapeutic Goods Administration (TGA) of the idiotypic mouse monoclonal antibody that mimics N-glycolyl (NGc) Australian Department of Health adopts the scientific guidelines from gangliosides, thus triggering responses against tumor antigen the European Union (EU) and ICH in approving biosimilars. A key NGcGM3 [40]. The idiotypic antibody acts as a therapeutic vaccine by in- issue for the biosimilar approval pathway has been the uncertainty ducing the immune system to elicit a specific response. A Phase III clin- and evolving nature of the regulatory process. The inconsistencies be- ical trial is ongoing in advanced NSCLC patients. tween the agencies has led to a marked difference in the numbers of Ublituximab is a chimeric monoclonal antibody which targets CD20 biosimilar antibodies approved across the different markets (Table 2). and has been fragment of crystallization (Fc) engineered with low fucose content to enhance affinity to all allelic variants of FcγIIIa receptors, Clinical context from WHO guidelines [45,46] demonstrating greater antibody dependent cell cytotoxicity (ADCC) than rituximab (Rituxan®) and ofatumumab (Arzerra®). This will see The WHO guidelines express as a minimum the requirement of clin- the second glyco-engineered antibody to enter the market (discussed ically relevant endpoints including PK, PD, efficacy, safety, and immuno- later in Fc engineered section). Ublituximab is currently being assessed genicity data to assess biosimilarity. In some circumstances, efficacy in Phase III trials in combination with small molecule ibrutinib in pa- studies are waived if there is sufficient evidence of biosimilarity from tients with previously treated high-risk Chronic Lymphocytic Leukemia all other non-clinical data. More extensive toxicology and/or clinical fi (CLL). Indeed, 2015 witnessed the rst FDA approval of a combination studies are required when differences are identified to address residual regime of anti-PD-L1 antibodies nivolumab (Opdivo®) and ipilimumab uncertainties and undesirable immunogenicity. (Yerzoy®) and is quickly becoming a focus of ongoing research with both emerging and established therapeutics. EMA approach to assessing biosimilarity [47–52] MABp1 is a ‘natural’ antibody cloned from an affinity-matured, in vivo human immune response, with no sequence modifications. The fi “ antibody was derived from Epstein-Barr-virus-immortalized B lympho- The EMA guidelines de ne a biosimilar as a biological medicinal cytes derived from an individual with circulating anti-IL-1α, a method product that contains a version of the active substance of an already au- previously described by Garrone and colleagues [41]. This potential thorized original biological medicinal product (reference medicinal ” “ breakthrough first-in-class ‘true human™’ monoclonal antibody product) and that similarity to the reference medicinal product in fi targeting IL-1α is being assessed in Phase III clinical trials for late stage terms of quality characteristics, biological activity, safety and ef cacy ” colorectal cancer. The safety results to date suggest a unique safety pro- based on comprehensive comparability exercise needs to be file for this antibody, among the best tolerated therapies used in oncol- established. ogy, making it ideally suited for treating advanced cancer patients with The EMA discusses that the stepwise approach for assessing a reduced tolerance for toxic therapy [42]. The antibody is also being in- biosimilar include a comprehensive physicochemical and biological vestigated in a Phase II trial to correct the metabolic dysregulation un- characterization and in vitro pharmaco-toxicological studies as derlying the wasting phenotype associated with malignancy by paramount to establishing the quality of the biosimilar. In vivo animal specifically targeting IL-1α signaling in the hypothalamus. toxicology, immunogenicity, PK and PD studies may be required de- A Phase III clinical trial (HERCULES) has been initiated for bivalent pending on the collated non-clinical in vitro evidence. Clinical studies fi nanobody® (or VHH) caplacizumab targeting anti-von Willebrand fac- are to further evaluate safety, PK, PD and ef cacy endpoint data. The tor (vWF) to treat acquired Thrombotic Thrombocytopenic Purpura EMA review the presented data in a totality of evidence approach. (TTP). This could see the first nanobody® to be approved and the – smallest antibody fragment on the market at approximately fifteen FDA approach to assessing biosimilarity [53 55] kilodaltons (kDa). The FDA scientific and quality guidelines define biosimilarity as “that 4. Biosimilars, guidelines and challenges the biological product is highly similar to the reference product not- withstanding minor differences in clinically inactive components” and Over the last decade there has been a growing interest from the bio- that “there are no clinically meaningful differences between the biolog- pharmaceutical industry to heavily invest in the development of ical product and the reference product in terms of safety, purity, and po- biosimilars to compete with blockbuster antibodies that are about to tency of a product.” The FDA considers the following differences come off patent. Biosimilar antibodies are ‘highly similar’ replicate prod- between protein products: “(1) primary amino acid sequence; (2) mod- ucts of a therapeutic antibody that has already received marketing ap- ification to amino acids, such as sugar moieties (glycosylation) or other proval. The proposed biosimilar must be comparable with the side-chains; and (3) higher order structure (protein folding and originator reference with respect to its structure, function, animal toxic- protein-protein interactions).” ity, human pharmacokinetics (PK), pharmacodynamics (PD), clinical The FDA further discusses the requirement of interchangeability: “To safety, immunogenicity and effectiveness. Biosimilars are therefore ex- meet the additional standard of “interchangeability,” an applicant must pected to be produced, formulated and administered in the same regi- provide sufficient information to demonstrate biosimilarity, and also men as the approved originator reference, without access to the demonstrate that the biological product can be expected to produce proprietary manufacturing information; adding to the technical com- the same clinical results as the reference product in any given patient plexity and challenge of producing a highly similar product. The approv- and, if the biological product is administered more than once in an indi- al process of a biosimilar is less extensive than for a new therapeutic vidual, the risk in terms of safety or diminished efficacy of alternating or antibody; and while there is less emphasis on clinical trials, a high switching between the use of the biological product and the reference 6 Z. Elgundi et al. / Advanced Drug Delivery Reviews 122 (2017) 2–19 product is not greater than the risk of using the reference product with- interchangeability status. Nevertheless, reports from recent clinical out such alteration or switch.” studies are laying the groundwork for the feasibility of ‘switching’ The FDA discusses that the stepwise approach for assessing from an originator to a biosimilar antibody [57,58]. Moving forward, biosimilarity include a comprehensive structural and functional charac- the regulatory approval of biosimilar antibodies will be a global evolu- terization, animal toxicity, human PK and PD, clinical immunogenicity, tion requiring both biosimilar developers and regulatory agencies to clinical safety and effectiveness. The FDA adopts a totality-of-the-evi- continuously monitor and adapt. At this stage, biosimilar antibodies dence approach to reviewing the presented evidence, and they require offer cost reductions of 10–30% which has been calculated to lead to at least one clinical PK study and, if appropriate, one PD study in the rep- considerable cost savings to healthcare authorities in countries that sub- ertoire of evidence that supports biosimilarity. Clinically meaningful dif- sidize costs [59]. The competition offered by lower-cost biosimilars is ferences to be addressed include the expected range of safety, purity expected to increase volume that may further drive a reduction in over- and potency of the biosimilar. all cost of both originator and biosimilar antibodies [60]. Certainly, A couple of notable aspects between the authorities is the clinical ev- biosimilar antibodies are starting to surge in developing countries idence of interchangeability proposed by the FDA, which almost all EU such as India and South America (Table 2). countries have ruled out with the exception of France [56]. In the context of one or more indications, both authorities advise that if 5. Biobetter antibodies biosimilarity has been demonstrated for one indication, extrapolation of clinical data to other indications of the originator reference may be Despite its success, antibody-based therapy still presents a long list acceptable. Similarly, due to the structural complexity and heterogene- of important shortcomings that need to be overcome to fully exploit ity that comes from the production and purification of antibodies, pro- their full therapeutic potential. Typical drawbacks in antibody-based ducing a biosimilar that is structurally identical to its reference therapy include limited efficacy due to poor tissue and tumor penetra- product is extremely unlikely. Both authorities therefore are aware tion, low in vivo efficacy, cumbersome administration, antibody aggre- that manufacturing conditions can influence differences in protein gation, solubility as well as high production costs [61]. The use of modifications and higher order structure, such as the glycosylation pro- antibody engineering to improve the properties of therapeutic antibod- file, which can affect the safety and effectiveness of the protein in its ies has advanced greatly in the last decades giving rise to a varied set of clinical context. Biosimilarity can be accepted as long as structural and novel formats that offer enhanced attributes for therapeutic and re- functional differences are recognized and thoroughly evaluated to pro- search purposes. These novel molecules are often referred to as vide sufficient evidence that they are of no clinical relevance. Further- biobetters or next-generation antibodies and include platforms such more, both authorities have considered differences in formulation as a as: engineered antibodies for enhanced effector functions, antibody possible area of consideration during evaluation of a biosimilar. It is pos- drug-conjugates (ADC), multi-specific antibodies and single-domain sible that improvements in formulation and/or excipients have oc- antibody fragments (sdAb or nanobodies®) [62,63]. curred since the originator was approved. A biobetter can have modifications to its chemical structure such as To date, four biosimilar antibodies have been approved by the EMA humanization, fusion/conjugation or be glyco-engineered to be less im- with the first approval in 2013 for the infliximab biosimilar CT-P13 munogenic or more efficacious. For instance, anti-CD20 ocrelizumab (Table 2). In comparison, the FDA only recently approved three which is currently under review for FDA approval for MS is a humanized biosimilar antibodies almost three-years later; and without version of rituximab (Rituxan®). A biobetter may be modified to have an improved formulation for improved treatment regimen so treatment is less evasive or have a simplified manufacturing process. This was demonstrated with novel subcutaneous (SC) formulations of Table 2 trastuzumab (Herceptin®) and rituximab (Mabthera®) developed by Summary of approved biosimilar antibodies worldwide. co-formulating the antibody with hyaluronidase, an enzyme that in- Originator Biosimilars in Approving country Year of creases the absorption and distribution of the injected product [64].A market approval biobetter may be engineered to have higher target affinity, bind at a dif- Abciximab Clotinab® Europe 2013 ferent epitope or stronger effector function to enhance efficacy and po- (ReoPro®) tentially reduce any off-target side effects. Any such improvement Rituximab AcellBia® (BCD-20) Russia 2014 means that the therapeutic has been modified and incomparable to (Rituxan®, MabTas India fi MabThera®) Reditux® South America the original therapeutic; it is therefore classi ed as a new biological Etanar® therapeutic and requires more laborious testing to obtain regulatory ap- Kikuzubam® proval before market entry. In the next section, Fc engineering, ADC and Etanercept BRENZYS® (SB4) Korea 2015 multi-specific antibody-based therapies will be discussed. (Enbrel®) Benepali® Europe 2016 Etacept India Erelzi® United States 2016 5.1. Fc engineered antibodies for enhanced effector functions Infliximab Remsima® (CT-P13) Korea, South 2012 (Remicade®) America Inflectra® (CT-P13) Europe, United 2013, Antibody engineering has sought to improve the effector function States 2016 of antibodies via the Fc region; namely ADCC, complement depen- INFIMAB® India 2014 dent cytotoxicity (CDC) and PK profile. Specifically, this has been (BOW015) mostly explored through modifications in the amino acid sequence Infliximab Biosimilar Japan fi Flixabi® Europe 2016 or the glycosylation pattern in the Fc region to enhance the af nity Trastuzumab HERTRAZ® India 2013 towards Fcγ receptors (FcγR) on effector cells [62,65].Themost (Herceptin®) CANMAB® prominent and successful technology so far has been the glycosyla- HERZUMA® Korea 2014 tion approach, which has seen nearly twenty glyco-engineered (CT-P06) CREDIMA antibodies enter clinical trials with two already approved for clinical HERtiCAD® Korea, Russia 2014, use [66–68].Thefirst success came with the approval of 2016 mogamulizumab (Poteligeo®) in Japan in 2012, developed using Adalimumab Exemptia® India 2014 the POTELLIGENT (Kyowa Hakko Kirin) platform which is indicated (Humira®) Amjevita® United States 2016 for CC-chemokine receptor 4 (CCR4)-expressing T cell leukaemia- Z. Elgundi et al. / Advanced Drug Delivery Reviews 122 (2017) 2–19 7 lymphoma and peripheral T cell lymphoma (PTCL) in adult patients GlycoMAb) and is approved for CCL [67]. Both POTELLIGENT® and [69]. GlycArt platforms are based on the manufacture of products using An ADCC enhanced version of rituximab, obinituzumab engineered cell lines that yield defucosylated antibodies; a structural (Gazyva®) gained approval by the FDA in 2013 under breakthrough modification that removes the core 1,6 fucose from the N-glycans at- therapy designation [70]. Obinituzumab is an anti-CD20 monoclonal tached to asparagine at amino acid position 297 of human IgG1 antibody that was originated by GlycArt Biotechnology (now thereby greatly enhancing the affinity towards FcγRIII, and

(a)

(b)

(c)

Fig. 1. Schematic representation of biobetter antibody formats being pursued for clinical development. (a) Schematic structure of an IgG1 antibody with glycosylation sites at asparagine 297 (Asn-297) in CH2 domains indicated by hexagons. The general structure of N-linked glycosylation is shown inset; core structures indicated by solid lines and variable structures by dotted lines comprising of Fucose (Fuc), N-Acetylglucosamine (GlcNAc), Mannose (Man), Galactose (Gal) or N-Acetylneuraminic acid (Neu5Ac). Glyco-engineering of antibodies can involve defucosylation which refers to the removal of core fucose to enhance Fc-mediated effector functions. (b) Schematic structure of an antibody drug conjugate (ado-trastuzumab emtansine; Kadcyla®) including N-maleimidomethyl) cyclohexane-1-carboxylate (MCC) linker and maytansinoid 1 (DM1) payload. (c) Prominent bispecific formats in development as discussed in the text including: Triomab or Trifunctional antibody, Dual variable domain immunoglobulin (DVD-Ig), Dock-and-Lock (DNL) antigen binding fragments (Fabs), Bispecific T cell engager (BITE), Dual affinity re-targeting (DART) molecule, Tandem diabody (tandAb) and Immune-mobilizing monoclonal TCRs against cancer (ImmTAC). 8 Z. Elgundi et al. / Advanced Drug Delivery Reviews 122 (2017) 2–19 increasing ADCC induction by NK cells (Fig. 1a) [71,72]. NK cells are IgG1 otelixizumab directed against CD3 which is being assessed in not abundant in the tumor site, and human endogenous IgG can in- Phase II trials for type I diabetes mellitus. hibit the elicitation of ADCC by therapeutic antibodies. [73] There- fore, increasing the affinity towards FcγRIII to preferentially 5.2. Antibody drug conjugates (ADC) interact with the therapeutic antibody is a valuable approach to improve the efficacy profile. This approach was demonstrated by Zhang ADC comprise one of the foremost antibody-based platforms cur- et al. by producing an anti-HER2 antibody with an identical sequence rently being pursued for clinical implementation. Indeed, the FDA ap- to trastuzumab but with superior binding affinity to FcyRIIIa and proval of brentuximab vedotin (Adcetris®) and a cytotoxin- greater ADCC activity [74]. The antibody was produced using glyco- conjugated biobetter of trastuzumab, ado-trastuzumab emtansine engineered Pichia pastoris and the differences were presumed to be (Kadcyla®) in 2011 and 2013 respectively, have eclipsed the clinical due to absence of fucose in the N-glycans attached to the IgG1 failures of first generation ADC. As of early 2016, in addition to the product. two ADC available in the market, over forty antibody drug conjugates Whereas the glyco-engineering approach generates defucosylated are undergoing clinical trials and a plethora of such formats are in pre- antibodies with FcγRIIIa-specificaffinity improvement, variants gener- clinical development [3] (Table 3). The achievement of such milestones ated by mutagenesis of Fc amino acid sequence can be enhanced for has been facilitated by the advent of novel ADC technologies and multiple FcγR interactions. A number of publications have identified designs. specific Fc mutations to improve binding to the activating receptor While the targeted nature of ADC therapy is expected to reduce FcγIIIa and reduce binding to the inhibitory receptor FcγIIb with corre- treatment side effects, toxicity still remains an important concern for sponding improvement to ADCC activity [75–77].TheXmAb®concept ADC. In fact, gemtuzumab ozogamicin (Mylotarg®), a humanized anti- is currently being investigated as an anti-CD30 in Phase I trials for Hodg- CD33 antibody conjugated to calicheamicin was the first ADC to receive kin lymphoma. FDA approval in 2000 for treatment of acute myeloid leukemia (AML). In an alternative approach, aglycosylated antibodies (without However, it was voluntarily withdrawn from the market after results glycan structures) can be engineered to display effector functions of a Phase III study raised toxicity concerns and showed no benefitcom- that are distinct from those of glycosylated counterparts [78–81]. pared to standard therapies [84]. Brentuximab vedotin and ado- The use of aglycosylated therapeutic antibodies offers manufactur- trastuzumab emtansine also present significant toxicity; both are re- ing advantages by bypassing glycosylation and hence production quired by the FDA to carry a black box warning due to the risk (although can be performed in prokaryotic hosts. On the other hand, the impor- only seen in rare cases) of developing progressive multifocal tance of glycosylation on the structural stability of antibodies has leukoencephalopathy (brentuximab vedotin) and risk of hepatotoxicity, been demonstrated [82,83].Thefirst aglycosylated antibody to cardiac toxicity, and embryo-fetal toxicity (ado-trastuzumab enter clinical trials, produced in yeast, is humanized rat-derived emtansine). Consequently, there remains a challenge in the field to

Table 3 Overview of ADC developed in recent years and their clinical status.

ADC Format Target Payload Linker Lead indications

FDA approved Brentuximab vedotin (Adcetris®) IgG1 (chimeric) CD30 MMAE Valine-Citrulline Hodgkin lymphoma, Anaplastic large cell lymphoma Ado-trastuzumab emtansine (Kadcyla®) IgG1 HER2 DM1 Valine-Cirtulline HER2+ breast cancer (humanized) Gemtuzumab ozogamicin (Mylotarg® - IgG4 CD33 Calicheamicin SS/hydrazone Acute myeloid leukemia withdrawn) (humanized)

Phase III Inotuzumab ozogamicin IgG4 CD22 Calicheamicin SS/hydrazone Acute lymphoblastic leukemia (humanized)

Phase II Enfortumab vedotin IgG1 (human) Nectin-4 MMAE Valine-Citrulline Bladder cancer (ASG-22ME) PSMA ADC IgG1 (human) PSMA MMAE Valine-Citrulline Prostate cancer Glembatumumab vedotin (CDX-011) IgG2 (human) GPNMB MMAE Valine-Citrulline Breast cancer Lifastuzumab vedotin IgG1 NaPi2b MMAE Valine-Citrulline Ovarian cancer (DNIB0600A) (humanized) Pinatuzumab vedotin (DCDT2980S) IgG1 CD22 MMAE Valine-Citrulline Non-Hodgkin’s lymphoma (NHL) (humanized) Polatuzumab vedotin (DCDS4501A) IgG1 CD79b MMAE Valine-Citrulline NHL, diffuse large B cell lymphoma (humanized) Lorvotuzumab mertansine (IMGN-901) IgG1 CD56 DM1 SPP Hematological malignancies (humanized) IMMU-130 (Labetuzumab-SN-38) IgG1 CEACAM5 SN-38 Carbonate Colorectal cancer (humanized) IMMU-132 (hRS7-SN38ADC) IgG1 (human) TROP-2 SN-38 Carbonate Breast cancer Milatuzumab doxorubicin (IMMU-110) IgG1 CD74 Doxorubicin Hydrazone Multiple myeloma (humanized) Indatuximab ravtansine (BT-062) IgG1 CD138 DM4 SPDB Multiple myeloma (humanized) Rovalpituzumab tesirine Undisclosed Fyn3 D6.5 Undisclosed Small cell lung cancer (SC16LD6.5) Depatuxizumab mafodotin IgG1 (human) EGFRvIII MMAF Maleimidocaproyl Glioblastoma (ABT-414)

Abbreviations: MMAE = monomethylauristatin; DM = maytansinoid; PSMA = prostate-specific membrane antigen; GPNMB = transmembrane glycoprotein NMB; NaPi = sodium-dependent inorganic phosphate; CEACAM = carcinoembryonic antigen-related cell adhesion molecule; EGFR = epidermal growth factor receptor. Z. Elgundi et al. / Advanced Drug Delivery Reviews 122 (2017) 2–19 9 further validate the superior therapeutic index that ADC can deliver functional group in the linker and a reactive moiety in native amino over traditional chemotherapy and other targeted therapies. In this re- acid residues, predominantly the thiol side-chains of partially reduced gard, great efforts are being directed to implement refinements in cysteines and the epsilon-amino end of lysine residues [107,108].In each of the components (targeted antigen, targeting moiety, linker, pay- these cases, the conjugation sites and the number of drug molecules load and conjugation chemistry) to boost the therapeutic potential of affixed to the antibody are stochastic, resulting in heavily heteroge- ADC. neous conjugates. This heterogeneity undermines the clinical potential While most conjugates in clinical and preclinical development target of ADC, since it impacts severely on their toxicity and safety profiles. cell surface receptors that are overexpressed in solid tumors and B-cell In terms of potency, the efficacy is compromised by the reduced subset malignancies (Table 3), novel approaches such as targeting antigens of ADC with an optimal amount of attached drug to exert a significant present in neovasculature or extracellular stromal tissue are being ex- cytotoxic effect. Besides the fact that conjugates with low drug-to- plored. One main advantage of such approaches is that perfusion into antibody ratios (DAR) have limited efficacy, high DAR also possess the tumor is not required since blood serum is in direct contact with higher toxicity and reduced plasma stability due to increased clearance vascular endothelial cells, with exposure of the conjugate to its target rates [109]. This clearly illustrates the critical importance of achieving tissue considerably greater than for traditional ADC targeting neoplastic an optimal ratio. Despite the shortcomings of stochastic attachments cells [85–87]. Another attractive feature is that vascular endothelial cells methods, both cysteine and lysine conjugation techniques have been are less likely to develop resistance mechanisms. Others have also ex- utilized in the development of brentuximab vedotin and ado- plored the possibility of targeting cancer stromal cells, aiming to sup- trastuzumab emtansine, respectively. A set of alternate novel tech- press tumor proliferation, neovascularization, invasion, and metastasis niques have been developed to allow for site-specific conjugation in- which are typically promoted by the tumor microenvironment [88, cluding the THIOMAB platform [110], introduction of unnatural amino 89]. Overall, the factors affecting the internalization rate of ADC suggest acids or enzymatic modification of native amino acid side-chains that antigen selection is of high importance in determining the efficacy [111–113] and even glycan attachment through glyco-engineering of the conjugate in a physiological environment. [114]. Conversely, the efficacy of a conjugate is predominantly determined by its affixed payload. Current ADC feature antineoplastic agents that 5.3. Bispecifics have an extremely high potency, usually in the sub-nanomolar range. Most of these molecules are derivatives of highly potent cytotoxic The next novel wave of therapeutic antibodies is predicted to be in drugs that have exquisite potency but at the same time, make them un- the format of bispecific antibodies. The concept of bispecificantibodies amenable for development as chemotherapeutic agents alone. The cyto- was first demonstrated more than twenty years ago, initially by chemi- toxic drugs being employed can be classified into three major groups cal conjugation of two antibodies [115] then by fusing two hybridoma namely, calicheamicins (DNA disruption), maytansinoids and cultures [116]. There are now more than fifty different bispecific for- auristatins (cell cycle interference) [90–93]. Monomethylauristatin E mats in development, enabling researchers to adjust and control param- (MMAE) and monomethylauristatin F (MMAF) are the most widely eters such as size, half-life, stability, flexibility and orientation to achieve used payloads in the current pipeline with another twenty auristatin the desired therapeutic outcome. The assembly of antibody chains to ac- conjugates under clinical development [94,95]. Among the maytansine complish the various antibody formats has been made possible with an- analogs, DM1 and DM4 are being pursued with more than ten ADC in tibody engineering techniques including variations of the general clinical trials (Table 3; Fig. 1b). Alternative drug classes such as approach [117–119] to novel approaches such as ‘knobs-into-holes’ duocarmycins and pyrrolobenzodiazepines may also gain popularity [120], CrossMab [121], ‘Dock-and-Lock’ (DNL) [122,123] and even hy- due to their remarkable potency [96,97]. brid domains derived from other immunoglobulin antibodies, via a The linker component is a major determinant of plasma stability of a technology referred to as strand-exchange engineered domain (SEED) conjugate, thus impacting on the efficacy and safety profile. One of the [124]. main reasons for off-target effects is the premature release of cytotoxic The majority of bispecific antibodies are currently being assessed in activity outside of target cells. A commonly used approach for coupling clinical trials and are detailed in Table 4 with a few examples discussed the conjugate to an antibody is via a linker that contains a lysosomal- below based on their mode of action. Referred to as a trifunctional or specific protease cleavage site. In this format, the linker is intended to triomab is catumaxomab (Removab®), bispecific to tumor antigen be cleaved in the lysosome by proteases such as cathepsin B, plasmin EPCAM and T cell marker CD3 (Fig. 1c). The antibody acts by recruiting and β-glucoronidase [98,99]. Indeed, the valine-citrulline (Val-Cit) di- and activating immune cells; EPCAM targets the tumor, CD3 recruits T peptide linker, recognized and cleaved by cathepsin B, is currently the effector cells and the Fc region recruits and activates monocytes, macro- most widely implemented linker technology [100,101].Anovelap- phages, dendritic cells (DC) and NK cells by FcγRbinding[125]. proach using non-cleavable linkers has gained popularity since they Catumaxomab is a hybrid antibody comprised of rat IgG2b binding do- could substantially increase blood serum stability. In this modality, the mains and mouse IgG2a Fc region. Catumaxomab has not yet received antibody is digested inside the lysosome upon internalization, releasing FDA approval (BLA submitted) but has EMA approval for the treatment the payload attached to an amino acid residue, yet still retaining its cy- of malignant ascites since 2009. The antibody is also in clinical trials for totoxic activity [102,103]. For example, ado-trastuzumab emtansine, application in ovarian cancer (II), gastric cancer (II) and epithelial can- employs the hetero-bifunctional crosslinker N-succinimidyl-4-(N- cer (I). Given that catumaxomab is a rat-mouse hybrid, some anti-rat maleimidomethyl) cyclohexane-1-carboxylate (SMCC), creating a or anti-mouse IgG responses are observed in patients, though treatment non-reducible thioether linker (MCC) as a spacer between the anti- does not appear to be significantly affected. In fact, development of HER2 antibody and the maytansine DM1 [104]. Presumably due to in- human anti-mouse antibodies (HAMA) have been shown to contribute creased stability, reduced toxicity has been observed with non- to greater clinical benefit [126]. cleavable linker design relative to conjugates with cleavable linker Another example that achieves effector cell recruitment via a con- [105]. Remarkably, ado-trastuzumab emtansine has displayed superior ceptually different format without a Fc region is bispecific T cell engager in vivo activity in mouse breast cancer models compared to a similar (BiTE®). BiTEs are variable regions in the form of two scFv connected by trastuzumab-DM1 conjugate that differs only in that it contains a disul- flexible linker peptides (Fig. 1c) [127,128]. One scFv is directed at a cell fide reducible linker [106]. surface tumor antigen (with higher affinity) and the other scFv binds The chemical reactions used to couple ADC components can also CD3 (with lower affinity). The first marketed BiTE was blinatumomab play a role in the therapeutic index. The first conjugation techniques de- (Blincyto®) which targets CD19 on acute lymphoblastic leukemia veloped to create ADC involved a chemical reaction between a (ALL), approved by the FDA in late 2014 under the accelerated approval 10 Z. Elgundi et al. / Advanced Drug Delivery Reviews 122 (2017) 2–19

Table 4 Bispeciﬁc antibodies developed in recent years and their clinical status.

Name Company Targets Antibody type Mode of action Indications

FDA approved Blinatumomab Blincyto® Amgen CD19/CD3 BiTE T cell Philadelphia chromosome-negative recruitment precursor B-cell acute lymphoblastic leukemia

Clinical trials completed Catumaxomab Removab® Trion Pharma EPCAM/CD3 Trifunctional/Triomab T cell Malignant ascites recruitment/Fc effector

Phase III Emicizumab/RG6013/ACE910 Roche Factors XI/X CLC-IgG Two factor Haemophilia A dimerisation MT111/MEDI565 MedImmune CEA/CD3 BiTE T cell Gastrointestinal cancer recruitment

Phase II Vanucizumab Roche Angiopoietin2/VEGF Crossmab Two ligand Colorectal cancer inactivation RG7716 Roche Angiopoietin2/VEGF Crossmab Two ligand Macular degeneration inactivation TF2 IBC CEA/Hapten DNL Fab3 Payload Small cell lung cancer, Colorectal cancer, Pharmaceutical/Immunomedics delivery Thyroid cancer Duligotuzumab Genentech/Roche HER1/HER3 DAF-IgG Two ligand Head and neck cancer, Colorectal cancer inactivation ABT122 Abbott Laboratories TNF/IL-17 DVD-IgG Two ligand Rheumatoid arthritis, Psoriatic arthritis inactivation ABT981 Abbott Laboratories IL-1α/IL-1β DVD-IgG Two ligand Osteoarthritis inactivation SAR156597 Sanoﬁ-Aventis IL-4/IL-13 DVD-IgG Two ligand Idiopathic pulmonary ﬁbrosis inactivation Istiratumab/MM141 Merrimack Pharmaceuticals IGF1R/HER3 IgG-scFv Two ligand Pancreatic cancer inactivation IMCgp100 Immunocore MHC ImmTAC T cell Malignant melanoma

peptide280-88/CD3 recruitment AFM13 Afﬁmed Therapeutics CD30/CD16 TandAb NK cell Hodgkin Lymphoma recruitment

Phase I BI1034020 Boehringer Ingelheim Beta amyloid 2 Bi-nanobody Undisclosed Alzheimer's Disease Pharmaceuticals epitopes ALX0761 Merck Sereno IL-17A/IL-17F Bi-nanobody Two ligand Psoriasis inactivation LY3164530 Eli Lilly HER1/cMET orthoFab-IgG Two ligand Metastatic cancer inactivation Pasotuxizumab Bayer PSMA/CD3 BiTE T cell Prostate cancer recruitment MGD006 Servier CD123/CD3 DART T cell Acute Myeloid Leukemia recruitment AFM11 Afﬁmed Therapeutics CD19/CD3 TandAb T cell Acute Lymphoblastic Leukemia, recruitment Non-Hodgkin's Lymphoma MGD007 Servier GPA33/CD3 DART-Fc T cell Colorectal cancer recruitment/Fc effector

Abbreviations: BiTE = bispecific T cell engager; CLC = common light chain; DNL = dock-and-lock; Fab = antigen binding fragment; DAF = dual acting Fab; DVD-IgG = dual variable domain immunoglobulin, scFv = single chain variable fragment; ImmTAC = immune-mobilizing monoclonal TCRs against cancer; TandAb = tandem diabody; DART = dual affinity re-targeting molecule; Fc = fragment of crystallization; EPCAM = epithelial cell adhesion molecule; CEA = carcinoembryonic antigen; VEGF = vascular endothelial growth factor; TNF = tumor necrosis factor; IGFR = insulin growth factor receptor; MHC = major histocompatibility complex; PSMA = prostate-specific membrane antigen; GP = glycoprotein; NK = natural killer. program. Blinatumomab is currently undergoing multiple Phase II and the variety of potential tumor-specific antigens, by incorporating solu- Phase III clinical trials for other B-cell related malignancies [129].A ble T cell receptors (TCRs) that can target intracellular antigens through major setback for BiTEs is the requirement of continuous intravenous recognition of peptide-HLA antigen complexes on the cell surface infusion due to the short half-life inherent to the small size of the mol- (Fig. 1c) [135]. Under current investigation in Phase II clinical trial for ecule (55 kDa) and the lack of Fc region. Additionally, the treatment malignant melanoma is IMCgp100. The engineered TCR portion of the produces significant side effects, notably neurotoxicity and symptoms molecule targets the gp100 peptide280–288 antigen, which is of cytokine-release syndrome [130]. Despite the challenges, this strate- overexpressed and presented by HLA-A2 on the surface of melanoma gy is being investigated for other targets (CEA, EPCAM, PSMA) and indi- cells. The anti-CD3 scFv portion captures and redirects T cells to kill cations. Other examples of molecules with immune cell recruitment are the melanoma cells. dual affinity re-targeting (DART®) [131,132] and tandem diabodies With the knowledge gained on validated targets and receptor signal- (tandAb®) (Fig. 1c) [133,134]. ing of existing therapeutic antibodies, targeting two antigens to simul- Another novel class of bispecific molecules called immune- taneously interfere with two or more signaling pathways is being mobilizing monoclonal TCRs against cancer (ImmTAC) further expands pursued to improve therapeutic efficacy. The most advanced molecule Z. Elgundi et al. / Advanced Drug Delivery Reviews 122 (2017) 2–19 11 in clinical development is dugliotuzumab, a human IgG which targets advanced antibodies otelixizumab (refer to Fc engineered section) the combination of HER1 and HER3 as a dual action Fab (DAF)-IgG and clazakizumab in Phase II trials for psoriatic arthritis and rheuma- [136]. Duligotuzumab is being evaluated in Phase II trial in patients toid arthritis. with head and neck cancer. The dual interference can also be applied Plant systems are also generating interest as an alternative pro- to inflammatory conditions. Dual variable domain immunoglobulin duction platform due to the absence of mammalian pathogens but (DVD-Ig™) formats directed at TNF + IL-17 and IL-1α +IL-1β are un- too are subject to disadvantages given the different glycosylation dergoing assessment in Phase I studies in patients with rheumatoid ar- profile to humans, subcellular localisation issues and decreased thritis (Fig. 1c) [137]. Other formats being investigated to interfere with yield because of proteolytic degradation. Nevertheless, a number of signaling pathways include CrossMabs [121] and bi-nanobodies [138]. antibodies have reached clinical development. For instance, CaroRX Bispecifics can also serve as vehicles to deliver payloads to tumor is an antibody that targets Streptococcus mutans and used for treating cells. TF2 is a DNL(Fab3) bispecific antibody that binds to CEA present cavities, produced in Nicotiana tabacum andapprovedforusein on the surface of many solid tumors (Fig. 1c). The antibody comprises Europe [146]. MB66 (or MAPP66) contains a combination of three three Fab modules that are stably coupled to each other in a triangular monoclonal antibodies produced in Nictiana benthamiana designed fashion using the DNL technology. Without a Fc region, these molecules to block distinct mechanisms of HIV which has entered Phase I trials. have a rather short serum half-life which is an advantage for pre- The clear advantages of using plant systems is the low cost, scalabil- targeting approaches. TF2 is being evaluated in a Phase I trial in patients ity and relative ease to deploy in developing countries [147,148].In- with colorectal cancer using Lutetium-177 or Indium-111 payload for terestingly, a biosimilar of trastuzumab produced in N.benthamiana imaging. is entering Phase I clinical trials. An efficacy study in mice has shown that biosimilar trastuzumab was as effective as originator 6. Manufacture of antibodies trastuzumab (Herceptin®) in reducing the size and growth rate of breast cancer tumours [149]. The discovery and development of a therapeutic antibody can be a The choice of expression system therefore depends on many factors lengthy process, starting with early candidates (“hits”) then selection in order to guarantee cost-effective high yields and meet safety criteria; of advanced candidates (“leads”). Progress to commercial production from discovery stages with the precise sequence and mode of action of is also lengthy, costly, scientifically and technically complex, on average the specific antibody to development of the manufacturing cell line. requiring at least 18 months to complete at a cost of approximately Manufacturing cell lines to yield a therapeutic antibody are initially gen- seven million dollars [139,140]. The chemistry, manufacturing, and con- erated by vector construction and transfection. A variety of different ex- trols (CMC) activities are a critical element of the antibody development pression systems and transfection methods exist [150]. Following process, necessary to enable human clinical testing. These tasks include transfection, cells are subjected to a selection procedure where stable construction and testing of a production cell line, manufacturing process cell lines are cloned and expanded. Several strategies have since evolved for production, as well as development of suitable analytical methods to from the limiting dilution approach that utilise automated cell sorting characterize the antibody and ensure that it is safe and has the desired equipment [151,152]. The screening and selection of stable clones functional properties [139]. In the following section, certain aspects of with desired growth and production characteristics is a critical albeit CMC are discussed in the context of new headway and the challenges time-consuming step in the process. Once conditions are defined, the faced with the manufacture of novel antibody formats briefly process is often transferred to a pilot scale to test scalability and produce mentioned. material for preclinical toxicology studies. Larger scale manufacturing Therapeutic antibodies are mainly produced in mammalian host cell for production of clinical material is performed under current good lines because of their ability to introduce post-translational modifica- manufacturing practices (cGMP) regulations involving either fed- tions. Chinese hamster ovary (CHO) cells remain the most widely used batch or continuous perfusion culture [140,153]. For validation pur- cell line for the manufacture of antibodies (20 of 39 FDA approved), poses, three (United States) or five (Europe) consecutive full-scale cul- followed by SP2/0 (8/39) and NS0 (7/39) mouse cell lines and hybrid- ture runs are required for BLA approval [154]. omas (2/39). Two of the three Fabs are produced in the bacterial strain Therapeutic antibodies must have very high purity and be sterile, Escherichia coli. As the clinical success of therapeutic antibodies grows, with the concentration of host cell proteins reduced to parts per million alternative and more cost-effective production platforms are being pur- (ppm), DNA to parts per billion (ppb) and virus levels to less than one sued including microbial hosts and plant systems. Microbial cells such virus particle per million doses [139,155]. The stringent purification of as bacteria and yeast possess many advantages including fast growth, antibodies from mammalian cells is typically accomplished using a well-characterised genetics and low cultivation costs. three column chromatography process; protein A affinity chromatogra- Antibody production in bacterial systems has focused on frag- phy as an initial capture step followed by cation exchange (CEX) and ments due to the lack of post-translation glycosylation required on anion exchange (AEX) chromatography [156]. These are followed by a the Fc region of full length IgG antibodies, specifically required for ef- virus clearance step after which purified product is concentrated and ficient FcγR interactions. A major advancement was recently diafiltered into the final formulation buffer. While the antibody is pri- achieved with the expression of active, full-length IgG in the cyto- marily captured during the affinity step, the consecutive ion exchange plasm of E. coli to generate ‘cyclonals’.Robinsonetal.usean processes act as ‘polishing’ steps to remove impurities such as host engineered strain with an oxidative cytoplasmic environment to fa- cell protein, DNA, aggregates, endotoxin, adventitious and endogenous cilitate and promote the formation of disulfide bonds and efficient viruses [154]. With the growing demand and increasing market compe- IgG folding and assembly [141]. Moreover, molecular engineering tition, attention is now being focused on reducing manufacturing costs of cyclonal Fc domains with previously identified Fc mutations en- and improving process efficiency. For example, Protein A affinity chro- abled FcγRinteractions[78,79,142]. Conversely, yeast expression of- matography contributes a major cost to the purification process, there- fers the advantages of post-translational modifications, disulfide fore improved chromatography matrices, specifically non-affinity bond formation as well as secretion of correctly-folded proteins. An- methods are gaining popularity such as; hydrophobic charge induction tibody yields surpassing mammalian systems have been reported for chromatography (HCIC), high performance tangential flow filtration a glyco-engineered strain of P. pastoris capable of producing human (HPTFF) and MEP HyperCel [157–160]. N-linked glycoproteins [143,144]. Additionally, a series of experi- New manufacturing challenges are being faced as ADC and multi- ments have demonstrated production of IgG at industrial scale to specific antibodies move towards clinical use. In the case of full-length be robust and commercially viable [145]. The potential of yeast ex- bispecific IgG molecules or IgG-like antibodies (i.e. those containing Fc pression systems has not yet been fully realised with the most region), production then purification can typically follow that of other 12 Z. Elgundi et al. / Advanced Drug Delivery Reviews 122 (2017) 2–19

IgG antibodies. When affinity chromatography methods are not an op- reverse. Mechanisms of protein aggregation include: (1) aggregation tion, such as the purification of variable regions and other antibody frag- of native state monomers; (2) aggregation of monomers with a modi- ments without Fc region, then capture using ion exchange fied conformation (non-native); (3) aggregation of chemically- chromatography (IEX) or hydrophobic interaction chromatography modified monomers; (4) aggregation via a nucleation-dependant pro- (HIC) have been reported [161]. In another example, blinutonumab cess; and (5) surface-induced aggregation via adsorption of protein to has been engineered to express a histidine tag which allows the use of glass-liquid or air-liquid interfaces [181,182]. immobilized metal affinity chromatography (IMAC) for the initial chromatography step [162]. ADC are complex products comprised of an an- 7.2. Denaturation tibody component, chemical linker and a cytotoxic drug. Several reaction steps have to be carefully controlled to manufacture ADC prod- Protein denaturation refers to the partial or complete unfolding ucts consistently. Perhaps the greatest challenge in ADC manufacturing, of the native three-dimensional folded protein structure. A dena- is the design of a manufacturing facility capable of generating each part tured antibody often loses its tertiary and perhaps secondary struc- alone then together and to perform as required. Manufacturing the an- ture leading to loss of binding affinity and activity if an active site tibody component is the relatively easy part, because these technologies domain is affected, and may expose aggregation-prone regions lead- are now well-established. Linker chemical technology is rather special- ing to further degradation [178,183,184]. Several intermediate states ized with chemical synthesis involved. Manufacturing the cytotoxin may exist between the folded native structure of an antibody mole- component is the dangerous, complex, and costly phase. Toxin produc- cule and the denatured state, with some intermediates thought to tion, both up- and downstream, needs full isolation to prevent release of actasprecursorsor‘nuclei’, attracting other protein species to ex- toxic product; this would involve extensive use of isolator cabinets and posed hydrophobic sites and forming irreversible aggregates. Dena- rooms with careful environmental monitoring [163,164]. Moreover, the turation may be induced by a number of stress conditions that arise use of disposable technology and closed purification systems are recom- during antibody manufacture including changes in solution pH or mended due to the high level of toxicity of the chemicals used in these temperature, use of organic solvents or chaotropes, high salt concen- processes [165]. trations, or shear force [177,178,185]. In general, the CH3 domain of an antibody is often the most stable against denaturation at high 7. Physical and chemical degradation of antibodies temperatures (highest Tm) while the CH2 domain is least stable and denatures first (lowest Tm) [186]. During manufacturing and storage, therapeutic antibodies are at risk of degradation via a number of pathways. Though these reactions may 7.3. Fragmentation be kept under control by appropriate storage and formulation conditions of the final product, degradation that occurs during culture, down- Fragmentation of therapeutic antibodies can be a product of enzy- stream processing and in vivo cannot be controlled sufficiently. These matic or non-enzymatic hydrolysis that occurs at the peptide backbone degradation events may affect antigen recognition, hamper functionali- of a number of regions, such as the hinge region, the CH2-CH3 interface ty and in severe cases lead to immunogenic responses [166–170].Each or a region containing aspartic acid (Asp) or tryptophan (Trp) residues antibody molecule seems to have a unique personality related to its re- [177]. Asp-associated hydrolysis is affected by pH and the n + 1 residue; quirements for stability; a phenomenon derived from the fact that dif- for instance, a serine (Ser), valine (Val) or tyrosine (Tyr) adjacent to an ferences in the CDR between antibodies are primarily dictated by the Asp may increase the rate of Asp-hydrolysis. Hinge region hydrolysis surface exposed amino acids that define antigen specificity [169]. The can occur in the absence of Asp, and occurs most commonly in the identification of degradation prone or unstable regions early in the anti- IgG1 isoform [177,187,188]. The rate of hydrolysis is dependent on the body development process would ideally permit re-engineering of flexibility and the peptide sequence at the hinge, occuring within a nar- these problematic areas [171–174]. This approach is aided by recent de- row range of residues. Hinge hydrolysis rates are affected by solution velopments in computational modelling tools that predict regions of in- pH, with a minimum rate of hydrolysis observed near pH 6, and higher terest susceptible to physical, chemical degradation or influence other ratesatalowerorhigherpH[177,188]. Fragmentation of full-length an- biophysical properties of antibodies. In the next section, degradation tibodies is a common occurrence and generally, cleaved forms are pres- pathways commonly observed in therapeutic antibodies and an over- ent in such low amounts that an effect on efficacy would not likely be view of predictive tools are discussed. seen.

7.1. Aggregation 7.4. Deamidation

Protein aggregation is the most common and significant type of Deamidation is the most common chemical degradation pathway of physical degradation associated with therapeutic antibodies, often lead- therapeutic antibodies and results from the hydrolysis of the amide ing to reduced activity and in some cases, formation of immunogenic side-chain of amino acids glutamine (Gln) or asparagine (Asn) [177, products [175,176]. Initial preparations of therapeutic antibodies were 181,189,190]. Hydrolysis of the side-chain can occur at acidic pH administered by intravenous (IV) infusion formulated at lower protein (pH b 4) resulting in the conversion of Asn to Asp and Gln to glutamic concentrations (1–25 mg/ml). As antibody-based therapeutics became acid (Glu) [177]. However at higher pH, deamidation occurs predomi- more widely used, high concentration formulations that allow SC injec- nantly (and more slowly) via the formation of a cyclic imide intermedi- tion (N50 mg/ml) became desirable giving rise to aggregation issues. ate. For Asn, the cyclic intermediate (succinimide) leads to either the Proteins are folded in such a way as to internalize hydrophobic domains formation of Asp or an isomer of Asp. A similar process results in the and surface expose more hydrophilic domains. As protein-protein con- deamidation of Gln to Glu; although Gln deamidation is much less fre- tact frequency increases at high concentrations, the opportunity for ag- quent, due to the lower stability of the 6-membered cyclic intermediate gregation formation increases proportionally. Changes in extrinsic formed. In short, deamidation events lead to more acidic forms of the conditions including temperature, pH, salt, shaking, viscosity and con- antibody through the acquisition of additional carboxylic acid groups. centration can transiently expose hydrophobic domains which pro- Conversely, it is also possible for Asp residues to undergo modification motes protein-protein interactions that lead to aggregation events to a succinimide intermediate that produces a basic form of the anti- [177–180]. Non-covalent aggregates can be formed via hydrophobic body by removal of a carboxylic acid group [189]. and/or electrostatic interactions and may be reversible, while covalent A number of factors can affect the rate of deamidation. For instance, aggregates are usually formed by disulfide bonds and are difficult to Asn residues are more prone to deamidation if they are present in Z. Elgundi et al. / Advanced Drug Delivery Reviews 122 (2017) 2–19 13 solvent-accessible or structurally-flexible regions, especially if followed Table 5 by a small or flexible residue such as Gly, Ser, threonine (Thr) or Asn A representative list of computational tools for prediction of protein aggregation hot spots. [177,191]. Deamidation rate is also affected by extrinsic conditions in- Name Properties cluding pH, temperature, buffer composition and concentration [192]. Sequence-based methods Gln deamidation is thought to be less common than Asp deamidation TANGO Determines the secondary structure formation propensity due to the lower stability of the 6-membered cyclic ring intermediate, [204] which results in a much slower reaction rate [177].AlthoughGln Aggrescan Uses amino acid aggregation propensity value deamidation occurs less frequently, a study found that following incu- [205] Zyggregator Compares a new peptide sequence to the database bation at pH 9, 7.8% of Gln82 of a recombinant IgG1 mAb had undergone [207] deamidation, despite no deamidation of this residue occurring at neu- PASTA Predicts amyloid structure aggregation by looking into tral pH [190]. [208] sequences that are likely to stabilize the cross-beta core of fibrils Deamidation of therapeutic antibodies is well characterized both PAGE Uses physicochemical properties for prediction [209] in vitro and in vivo, and has been shown to decrease the potency, activity and stability of antibodies [189,191–195]. The deamidation events ap- Structure-based methods pear to be highly selective for individual antibodies. For example, Harris SAP Determines spatial effective surface accessible area [173] et al. performed accelerated stability studies at elevated temperatures LIP Measures ratio of polar surface area to apolar surface area of with rhuMAB HER2 antibody and found three labile Asn residues in [211] buried interfaces the CDR region (Asn55, Asn30 and Asn102) [189]. These residues either AGGRESCAN3D Based of original AGGRESCAN server with input from 3D formed aspartate, isoasparatate or a stable succinimide intermediate, [212] structure and spatial arrangement of residues resulting in a total of 7 species of the antibody being resolved. fi Deamidation of these Asn residues was shown to signi cantly affect Some of the early computational tools used for protein modelling fi the speci c in vitro activity and potency of rhuMAb HER2. include TANGO, PAGE, AGGRESCAN, PASTA and Zyggregator all of which rely on the sequence of the protein of interest (Table 5) 7.5. Oxidation [173,204–209]. These computational tools use force fields such as CHARMM or AMBER and exploit chemical properties of the amino Oxidation is another common degradation pathway which can acids such as hydrophobicity, β-sheet propensity, charge, and aro- occur during antibody production, formulation or storage. A number matic content to predict aggregation hot-spots and residues suscep- of amino acid residues may be affected, including methionine (Met), tible to chemical degradation. In some cases, multiple tools can be fi cysteine (Cys), histidine (His), tyrosine (Tyr) and Trp [177].Speci c used in combination to improve the predictive power. For example, Met residues within the Fc region (up to four residues) are prone to Wang et al. combined TANGO and PAGE to identify aggregation- oxidation resulting in production of methionine sulfoxide [181,196, prone motifs and residues susceptible to deamidation or oxidation 197]. Oxidation of these residues may affect the stability of an anti- of 22 commercial and 20 non-commercial therapeutic antibodies fi body, Fc-mediated effector function or Protein A binding af nity [210]. fi which is often used for puri cation from cell culture supernatant Some structure-based computational tools have also been devel- [198]. Wang et al. demonstrated that oxidation of Met252 can result oped. One such method, Spatial Aggregation Propensity (SAP), predicts N in 4-fold reduction in the half-life of an antibody in transgenic mice surface exposed aggregation-prone regions of a protein based on hydro- expressing human Fc neonatal receptor (FcRn). However, this was phobicity, dynamic fluctuations and solvent accessibility of residues and only observed when 80% of the antibody existed in the oxidized regions [172]. The tool has been used to simulate entire antibodies and form, and not at 40% [199]. develop IgG1 antibody variants with enhanced physical stability to the Oxidation can be dependent on intrinsic factors such as the de- wild type, by performing single or multiple mutations in either the gree of surface exposed residues as well as extrinsic factors including Fab or Fc regions [172,174]. buffer composition, light exposure and pH, although Met oxidation Another method developed by Angarica and Sancho predicts aggre- appears to be almost pH-independent [177,198,200].Oxidative gation propensity based on the packing density and polarity ratio (ratio stress has been observed during antibody production in mammalian of polar surface area to apolar surface area) of buried interfaces [211]. expression systems where the formation of reactive oxygen species The tool was designed to characterise “Light Interfaces of high Polarity” as a result of hypoxic conditions caused fragmentation of an IgG1 an- (LIPs) considered to be intrinsically unstable cores. The technology tibody [201]. shows promise as a tool for engineering antibody variants with in- Tryptophan oxidation of antibodies has been reported following creased aggregation-resistance, especially as a complimentary method light exposure. Sreedhara et al. found that light induced oxidation of to surface- or sequence-based tools mentioned above. surface exposed Trp residues (Trp53, Trp108 and Trp94) in the Fab re- One of the most recent developments in computational modelling is gion of an IgG1 antibody led to a loss of potency accompanied with a so- the evolution of AGGRESCAN to AGGRESCAN3D (A3D), an improved lution colour change [196]. In another example, oxidation of a Trp server which addresses many of the limitations of AGGRESCAN and residue in the H3 CDR loop (Trp135) of a humanized anti-respiratory other sequence-based methods. A3D takes into account the three- syncytial virus (RSV) therapeutic antibody resulted in loss of antigen dimensional structure of the protein and the spatial arrangement of binding and biological function [202]. the residues when the protein is in its native folded state [212]. With the incorporation of a mutation module that allows the easy modelling 7.6. Computational design tools of the detected aggregation-prone and surrounding residues, A3D looks to be a promising tool for predicting problematic regions and the same In recent years, computational methods used to simulate and devel- time, allowing for re-design of more stable proteins. op structural models of proteins have transformed into practical design tools for the development of biobetter and next-generation antibody therapeutics [172,203]. Current computational design tools have 8. Optimization of antibody bioavailability and delivery evolved to allow rapid identification of specific amino acid sequences or regions on a protein of interest, that contribute to its observed Antibody-based therapies are predominantly delivered intrave- in vivo properties such as binding affinity, efficacy, stability and half- nously (IV), though an increasing number are now being formulated life [203]. and administered subcutaneously (SC). While the IV route offers 14 Z. Elgundi et al. / Advanced Drug Delivery Reviews 122 (2017) 2–19

100% bioavailability, systemic distribution and physiological barriers microspheres have shown potential to increase bioavailability of respi- greatly reduce the actual concentration of antibody achieved in tar- ratory delivery by preventing proteolysis [225]. The PEGylation of anti- get tissues [213]. What is more, IV infusions are time-consuming body fragments was shown to increase lung lumen residence time in a and inconvenient. Ideally, an antibody formulation should be non- murine model through decreased clearance of alveolar macrophages invasive and increase local bioavailability. Limited alternatives and increased mucoadhesion [226]. Other strategies being investigated have been implemented in the clinic since formulation requirements to deliver antibodies via the respiratory tract include IgG-loaded lipid for such delivery often pose significant hurdles. For other parenteral microparticles and nano-micelles [227–229]. The most advanced mole- administration routes, such as SC or intramuscular (IM), the most cule in clinical development is ALX-0171, an anti-RSV nanobody admin- common limitation is poor antibody solubility at the high concentra- istered through inhalation; demonstrating a positive safety and tions required, given that maximum volume of injection is restricted tolerability profile in a first-in-infant Phase I/II study with an anti-viral to 2 ml and 5 ml, respectively. When compared to IV formulations effect observed [230]. which range from 1 to 25 mg/ml, SC and IM products often require Overall, the IV route will likely remain the most prominent adminis- concentrations N100 mg/ml to deliver an effective dose. Further- tration route in development due to ease of formulation. Still, the im- more, these delivery routes involve an absorption step to enable sys- provement of local bioavailability is an obvious requirement to fulfil temic circulation. Two main approaches are being pursued to the potential of antibody-based therapy, both in terms of efficacy and optimize bioavailability: 1) the design of aggregation-resistant anti- patient convenience. As such, strategies such as pursuing alternative ad- bodies with higher solubility to prevent precipitation at higher con- ministration routes and developing appropriate controlled release sys- centrations; 2) the use of polymer matrix systems to develop tems will gain relevance as their therapeutic potential continues to be controlled release formulations and improve PK profile. explored in preclinical studies. As discussed in the previous section, the design of aggregation- resistant antibodies albeit challenging, has become possible with 9. Conclusions the implementation of computational tools and increasing understanding of antibody structure and degradation pathways. Antibody-based therapeutics have evolved from murine antibodies Formulation stability has also been improved by using stabilizing to humanized and fully human antibodies, developed with innovative additives such as salts (e.g. citrates, sulfates), amino acids (e.g. gly- technologies such as transgenic mice and phage display. In the coming cine) and sugars (e.g. sorbitol, sucrose, trehalose) [177,214–217]. years, next-generation antibodies with improved properties and for- The development of SC and IM products has been successful in mats including ADC and bispecifics are expected to gain popularity as the last decade because of the formulation of large doses in signif- biobetter antibody therapeutics. Major hurdles are being overcome for icantly smaller volumes without aggregation issues [64].SCandIM biosimilar development following the marketing approval of the first formulations have significantly improved patient convenience en- biosimilar antibody by the FDA earlier this year. The advancement of abling self-administration through the use of pre-filled syringes, a next-generation biobetters and biosimilars is critical to; 1) reduce the feature that is highly advantageous for treatment of chronic dis- high cost of therapeutic antibodies encountered over the last thirty eases. A novel approach to improving SC formulation and PK has years, 2) address the shortcomings confronted with the use of these an- been proposed by Yang et al. whereby, crystalline antibody prepa- tibodies such as poor efficacy and stability and most importantly, rations of infliximab (Remicade®) and trastuzumab (Herceptin®) (3) provide greater patient benefit. were formulated at 200 mg/ml while maintaining low viscosities suitable for this delivery route [218]. Animal studies in rats showed Disclosures a 2-fold increase of antibody half-life compared to non-crystallized antibody, demonstrating the potential of crystalline preparations as The authors declare that they have no competing financial interests. controlled release systems. Alginate polymers have also been explored for controlled release. Schweizer et al. developed two Acknowledgements polyanionic alginate matrices loaded with antibody through electrostatic interactions [219]. Both matrices were delivered to rats The authors would like to acknowledge the Faculty of Pharmacy of subcutaneously as hydrogels. After comparison with its liquid anti- The University of Sydney for financial contribution. EC acknowledges fi body counterpart, no signi cant differences in bioavailability were the Ministry of Science, Technology and Telecommunications of the reported. Republic of Costa Rica for postgraduate scholarship. For many pathologies, local delivery could increase efficacy and reduce systemic exposure. In such cases, administration routes such as References oral, topical, respiratory and intraocular become highly relevant. Con- trolled release systems based on polymer matrices are also being tested [1] D.M. Ecker, S.D. Jones, H.L. Levine, The therapeutic monoclonal antibody market, in preclinical studies for these delivery routes [213,220]. For instance, li- MAbs 7 (2015) 9–14. [2] R.E. Kontermann, U. Brinkmann, Bispecific antibodies, Drug Discov. Today 20 posomes and chitosan-alginate microparticles have been employed for (2015) 838–847. oral delivery in order to protect antibodies from gastric inactivation [3] P.M. Drake, D. Rabuka, An emerging playbook for antibody-drug conjugates: les- and allow release in the small intestine [221,222].Thisisbeingexplored sons from the laboratory and clinic suggest a strategy for improving efficacy and safety, Curr. Opin. Chem. Biol. 28 (2015) 174–180. in combination with the conjugation of targeting ligands to improve de- [4] G. Kohler, C. Milstein, Continuous cultures of fused cells secreting antibody of livery and absorption in the gastro-intestinal tract. [223]. For topical ap- predefined specificity, Nature 256 (1975) 495–497. plication, a hydrofiber dressing/adhesive sheet has been used to apply [5] G. Kohler, C. Milstein, Derivation of specific antibody-producing tissue culture and – infliximab as a gel formulation for wound healing with improvements tumor lines by cell fusion, Eur. J. Immunol. 6 (1976) 511 519. [6] W. Li, N.B. Caberoy, New perspective for phage display as an efficient and versatile in 7/8 patients tested [224]. In another example, a Phase I trial has technology of functional proteomics, Appl. Microbiol. Biotechnol. 85 (2010) been completed with positive results for BIL-010t, a topically adminis- 909–919. tered, sheep antibody therapy to treat Basal Cell Carcinoma (BCC). BIL- [7] P.T. Jones, P.H. Dear, J. Foote, M.S. Neuberger, G. Winter, Replacing the complementarity-determining regions in a human antibody with those from a 010t ointment was self-applied for 28 days; it was noted that 13/20 pa- mouse, Nature 321 (1986) 522–525. tients had decreases in the sizes of their lesions with only mild localized [8] L. Riechmann, M. Clark, H. Waldmann, G. Winter, Reshaping human antibodies for skin reactions reported. The respiratory route has been extensively therapy, Nature 332 (1988) 323–327. [9] N. Lonberg, Human antibodies from transgenic animals, Nat. Biotechnol. 23 (2005) studied for treatment of chronic obstructive pulmonary disease, lung 1117–1125. cancer, asthma and other pulmonary pathologies. Liposomes and [10] C.T. Scott, Mice with a human touch, Nat. Biotechnol. 25 (2007) 1075–1077. Z. Elgundi et al. / Advanced Drug Delivery Reviews 122 (2017) 2–19 15

[11] A. Jakobovits, R. Amado, X. Yang, L. Roskos, G. Schwab, From XenoMouse technol- Smith, C.G. Willson, A.D. Ellington, G. Georgiou, High-throughput sequencing of ogy to panitumumab, the first fully human antibody product from transgenic mice, the paired human immunoglobulin heavy and light chain repertoire, Nat. Nat.Biotechnol.25(2007)1134–1143. Biotechnol. 31 (2013) 166–169. [12] W.D. Huse, L. Sastry, S.A. Iverson, A.S. Kang, M. Alting-Mees, D.R. Burton, S.J. [38] I. Saggy, Y. Wine, L. Shefet-Carasso, L. Nahary, G. Georgiou, I. Benhar, Antibody iso- Benkovic, R.A. Lerner, Generation of a large combinatorial library of the immuno- lation from immunized animals: comparison of phage display and antibody globulin repertoire in phage lambda, Science 246 (1989) 1275–1281. discovery via V gene repertoire mining, Protein Eng. Des. Sel. 25 (2012) [13] J. McCafferty, A.D. Griffiths, G. Winter, D.J. Chiswell, Phage antibodies: filamentous 539–549. phage displaying antibody variable domains, Nature 348 (1990) 552–554. [39] K. Lehmann-Horn, H.C. Kronsbein, M.S. Weber, Targeting B cells in the treatment of [14] C.F. Barbas III, A.S. Kang, R.A. Lerner, S.J. Benkovic, Assembly of combinatorial anti- multiple sclerosis: recent advances and remaining challenges, Ther. Adv. Neurol. body libraries on phage surfaces: the gene III site, Proc. Natl. Acad. Sci. 88 (1991) Disord. 6 (2013) 161–173. 7978–7982. [40] A.M.Vazquez,A.M.Hernandez,A.Macias,E.Montero,D.E.Gomez,D.F.Alonso, [15] D.J. Schofield, A.R. Pope, V. Clementel, J. Buckell, S.D.J. Chapple, K.F. Clarke, J.S. M.R. Gabri, R.E. Gomez, Racotumomab: an anti-idiotype vaccine related to N- Conquer, A.M. Crofts, S.R.E. Crowther, M.R. Dyson, G. Flack, G.J. Griffin, Y. Hooks, glycolyl-containing gangliosides - preclinical and clinical data, Front. Oncol. 2 W.J. Howat, A. Kolb-Kokocinski, S. Kunze, C.D. Martin, G.L. Maslen, J.N. Mitchell, (2012) 150. M. O'Sullivan, R.L. Perera, W. Roake, S.P. Shadbolt, K.J. Vincent, A. Warford, W.E. [41] P. Garrone, O. Djossou, F. Fossiez, J. Reyes, S. Ait-Yahia, C. Maat, S. Ho, T. Hauser, J.M. Wilson, J. Xie, J.L. Young, J. McCafferty, Application of phage display to high Dayer, J. Greffe, P. Miossec, S. Lebecque, F. Rousset, J. Banchereau, Generation and throughput antibody generation and characterization, Genome Biol. 8 characterization of a human monoclonal autoantibody that acts as a high (2007). affinity interleukin-1 alpha specific inhibitor, Mol. Immunol. 33 (1996) [16] J.D. Marks, H.R. Hoogenboom, T.P. Bonnert, J. McCafferty, A.D. Griffiths, G. Winter, 649–658. By-passing immunization. Human antibodies from V-gene libraries displayed on [42] D.S. Hong, D. Hui, E. Bruera, F. Janku, A. Naing, G.S. Falchook, S. Piha-Paul, J.J. phage, J. Mol. Biol. 222 (1991) 581–597. Wheler, S. Fu, A.M. Tsimberidou, M. Stecher, P. Mohanty, J. Simard, R. Kurzrock, [17] C.F.I. Barbas, J.D. Bain, D.M. Hoekstra, R.A. Lerner, Semisynthetic combinatorial an- MABp1, a first-in-class true human antibody targeting interleukin-1alpha in re- tibody libraries: a chemical solution to the diversity problem, Proc. Natl. Acad. Sci. fractory cancers: an open-label, phase 1 dose-escalation and expansion study, Lan- 89 (1992) 4457–4461. cet Oncol. 15 (2014) 656–666. [18] H.R. Hoogenboom, G. Winter, By-passing immunisation. Human antibodies from [43] Y.S. Kim, B.W. Choi, S.W. Yang, S.M. Shin, S.W. Nam, Y.S. Roh, J.Y. Lee, K.J. Lee, Y.J. synthetic repertoires of germline VH gene segments rearranged in vitro, J. Mol. Kim, J.-Y. Kwon, D.-I. Kim, Biosimilars: challenges and path forward, Biotechnol. Biol. 227 (1992) 381–388. Bioprocess Eng. 19 (2014) 755–765. [19] H.R. Hoogenboom, A.D. Griffiths, K.S. Johnson, D.J. Chiswell, P. Hudson, G. Winter, [44] A. Berghout, Clinical programs in the development of similar biotherapeutic prod- Multi-subunit proteins on the surface of filamentous phage: methodologies for ucts: rationale and general principles, Biologicals 39 (2011) 293–296. displaying antibody (Fab) heavy and light chains, Nucleic Acids Res. 19 (1991) [45] A. Kudrin, I. Knezevic, J. Joung, H.-N. Kang, Case studies on clinical evaluation of 4133–4137. biosimilar monoclonal antibody: scientific considerations for regulatory approval, [20] J. Davies, L. Riechmann, An antibody variable heavy domain with a lox-Cre site in- Biologicals 43 (2015) 1–10. tegrated into its coding region: bacterial recombination within a single polypep- [46] WHO, Guidelines on evaluation of similar biotherapeutic products (SBPs), World tide chain, FEBS Lett. 377 (1995) 92–96. Health Organization (WHO) Expert Committee on Biological Standardization, [21] M.M. Harmsen, H.J. De Haard, Properties, production, and applications of camel- 2009. id single-domain antibody fragments, Appl. Microbiol. Biotechnol. 77 (2007) [47] European Medicines Agency, Committee for Medicinal Products for Human Use 13–22. (CHMP), Guideline on Developement, Production, Characterisation and Specifica- [22] S.D. Nuttall, U.V. Krishnan, L. Doughty, K. Pearson, M.T. Ryan, N.J. Hoogenraad, M. tion for Monoclonal Antibodies and Related Products, European Medicines Agency, Hattarki, J.A. Carmichael, R.A. Irving, P.J. Hudson, Isolation and characterization of London (UK), 2008. an IgNAR variable domain specific for the human mitochondrial translocase recep- [48] Committee for Medicinal Products for Human Use, Guideline on Similar Biological tor Tom70, Eur. J. Biochem. 270 (2003) 3543–3554. Medicinal Products Containing Monoclonal Antibodies—Non-clinical and Clinical [23] K. Nakano, T. Kojima, K. Kasutani, C. Senoh, O. Natori, S. Ishii, H. Tsunoda, K. Hattori, Issues, European Medicines Agency, London (UK), 2012. Effective screening method of agonistic diabodies based on autocrine growth, J. [49] Committee for Medicinal Products for Human Use, Guideline on Immunogenicity Immunol. Methods 347 (2009) 31–35. Assessment of Monoclonal Antibodies Intended for In Vivo Clinical Use, [24] Y. Mazor, T. Van Blarcom, R. Mabry, B.L. Iverson, G. Georgiou, Isolation of European Medicines Agency, London (UK), 2012. engineered, full-length antibodies from libraries expressed in Escherichia coli, [50] Committee for Medicinal Products for Human Use, Guideline on Similar Biological Nat. Biotechnol. 25 (2007) 563–565. Medicinal Products Containing Biotechnology-derived Proteins as Active Sub- [25] L.S. Jespers, A. Roberts, S.M. Mahler, G. Winter, H.R. Hoogenboom, Guiding the se- stance: Non-clinical and Clinical Issues, European Medicines Agency, London lection of human antibodies from phage display repertoires to a single epitope of (UK), 2014. an antigen, Biotechnology (N Y) 12 (1994) 899–903. [51] Committee for Medicinal Products for Human Use, Guideline on Similar [26] A.E. Nixon, D.J. Sexton, R.C. Ladner, Drugs derived from phage display: from candi- Biological Medicinal Products, European Medicines Agency, London (UK), date identification to clinical practice, MAbs 6 (2014) 73–85. 2014. [27] P.S. Daugherty, G. Chen, M.J. Olsen, B.L. Iverson, G. Georgiou, Antibody affinity mat- [52] Committee for Medicinal Products for Human Use, Guideline on Similar Biological uration using bacterial surface display, Protein Eng. 11 (1998) 825–832. Medicinal Products Containing Biotechnology-derived Proteins as Active Sub- [28] M.J. Feldhaus, R.W. Siegel, L.K. Opresko, J.R. Coleman, J.M. Feldhaus, Y.A. Yeung, J.R. stance: Quality Issues (Revision 1), European Medicines Agency, London (UK), Cochran, P. Heinzelman, D. Colby, J. Swers, C. Graff, H.S. Wiley, K.D. Wittrup, Flow- 2014. cytometric isolation of human antibodies from a non immune Sacchromyces [53] Food and Drug Administration, Scientific Considerations in Demonstrating cerevisiae surface display library, Nat. Biotechnol. 21 (2003) 163–170. Biosimilarity to a Reference Product, Guidance for Industry, Silver Spring (MD), [29] D.G. Mottershead, K. Alfthan, K. Ojala, K. Takkinen, C. Oker-Blom, Baculoviral dis- 2015. play of functional scFv and synthetic IgG-binding domains, Biochem. Biophys. [54] Food and Drug Administration, Quality Considerations in Demonstrating Res. Commun. 275 (2000) 84–90. Biosimilarity of a Therapeutic Protein Product to a Reference Product, Guidance [30] J. Hanes, C. Schaffitzel, A. Knappik, A. Pluckthun, Picomolar affinity antibodies from for Industry, Silver Spring (MD), 2015. a fully synthetic naive library selected and evolved by ribosome display, Nat. [55] Food and Drug Administration, Guidance for Industry, Clinical Pharmacology Data Biotechnol.18(2000)1287–1292. to Support a Demonstration of Biosimilarity to a Reference Product (Draft Guid- [31] G. Georgiou, G.C. Ippolito, J. Beausang, C.E. Busse, H. Wardemann, S.R. Quake, The ance), 2014. promise and challenge of high-throughput sequencing of the antibody repertoire, [56] P.K. Thimmaraju, R. Rakshambikai, R. Farista, K. Juluru, Legislations on biosimilar Nat.Biotechnol.32(2014)158–168. interchangeability in the US and EU – developments far from visibility, Generics [32] U. Ravn, G. Didelot, S. Venet, K.T. Ng, F. Gueneau, F. Rousseau, S. Calloud, M. Kosco- Biosimilars Initiative (2015) http://www.gabionline.net/Sponsored-Articles/Legis- Vilbois, N. Fischer, Deep sequencing of phage display libraries to support antibody lations-on-biosimilar-interchangeability-in-the-US-and-EU-developments-far- discovery, Methods 60 (2013) 99–110. from-visibility. [33] P. Mathonet, C.G. Ullman, The application of next generation sequencing to the un- [57] L.J. Smits, L.A. Derikx, D.J. de Jong, R.S. Boshuizen, A.A. van Esch, J.P. Drenth, F. derstanding of antibody repertoires, Front. Immunol. 4 (2013). Hoentjen, Clinical outcomes following a switch from remicade to the biosimilar [34] W.C. Cheung, S.A. Beausoleil, X. Zhang, S. Sato, S.M. Schieferl, J.S. Wieler, J.G. CT-P13 in inflammatory bowel disease patients: a prospective observational co- Beaudet, R.K. Ramenani, L. Popova, M.J. Comb, J. Rush, R.D. Polakiewicz, A proteo- hort study, J. Crohns Colitis 10 (2016) 1287–1293. mics approach for the identification and cloning of monoclonal antibodies from [58] J. Brown, A. Kudrin, Switching to biosimilar infliximab (CT-P13): evidence of clin- serum, Nat. Biotechnol. 30 (2012) 447–452. ical safety, effectiveness and impact on public health, Biologicals 44 (2016) [35] S.T. Reddy, X. Ge, A.E. Miklos, R.A. Hughes, S.H. Kang, K.H. Hoi, C. Chrysostomou, 257–266. S.P. Hunicke-Smith, B.L. Iverson, P.W. Tucker, A.D. Ellington, G. Georgiou, Monoclo- [59] A.O. de Labry, E. Giménez, L. Lindner, L. García, J. Espín, J. Rovira, Biosimilars in the nal antibodies isolated without screening by analyzing the variable-gene reper- European market, Generics Biosimilars Initiative J. (GaBI J.) 2 (2013) 30–35. toire of plasma cells, Nat. Biotechnol. 28 (2010) 965–969. [60] E.A. Blackstone, P.F. Joseph, The economics of biosimilars, Am. Health Drug Benefits [36] U. Ravn, F. Gueneau, L. Baerlocher, M. Osteras, M. Desmurs, P. Malinge, G. 6(2013)469–478. Magistrelli, L. Farinelli, M.H. Kosco-Vilbois, N. Fischer, By-passing in vitro [61] P. Chames, M. Van Regenmortel, E. Weiss, D. Baty, Therapeutic antibodies: suc- screening–next generation sequencing technologies applied to antibody display cesses, limitations and hopes for the future, Br. J. Pharmacol. 157 (2009) 220–233. and in silico candidate selection, Nucleic Acids Res. 38 (2010), e193. [62] R. Niwa, M. Satoh, The current status and prospects of antibody engineering for [37] B.J. DeKosky, G.C. Ippolito, R.P. Deschner, J.J. Lavinder, Y. Wine, B.M. Rawlings, N. therapeutic use: focus on glycoengineering technology, J. Pharm. Sci. 104 (2015) Varadarajan, C. Giesecke, T. Dorner, S.F. Andrews, P.C. Wilson, S.P. Hunicke- 930–941. 16 Z. Elgundi et al. / Advanced Drug Delivery Reviews 122 (2017) 2–19

[63] J.B. Evans, B.A. Syed, From the analyst's couch: next-generation antibodies, Nat. [88] E. Ostermann, P. Garin-Chesa, K.H. Heider, M. Kalat, H. Lamche, C. Puri, D. Rev. Drug Discov. 13 (2014) 413–414. Kerjaschki, W.J. Rettig, G.R. Adolf, Effective immunoconjugate therapy in cancer [64] O. Shpilberg, C. Jackisch, Subcutaneous administration of rituximab (MabThera) models targeting a serine protease of tumor fibroblasts, Clin. Cancer Res. 14 and trastuzumab (Herceptin) using hyaluronidase, Br. J. Cancer 109 (2013) (2008) 4584–4592. 1556–1561. [89] D. Quail, J. Joyce, Microenvironmental regulation of tumor progression and metas- [65] S. Derer, C. Kellner, S. Berger, T. Valerius, M. Peipp, Fc engineering: design, expres- tasis, Nat. Med. 19 (2013) 1423–1437. sion, and functional characterization of antibody variants with improved effector [90] G.R. Pettit, Y. Kamano, C.L. Herald, A.A. Tuinman, F.E. Boettner, H. Kizu, J.M. function, Methods Mol. Biol. 907 (2012) 519–536. Schmidt, L. Baczynskyj, K.B. Tomer, R.J. Bontems, The isolation and structure of a [66] A. Beck, J.M. Reichert, Marketing approval of mogamulizumab: a triumph for glyco- remarkable marine animal antineoplastic constituent: dolastatin 10, J. Am. Chem. engineering, MAbs 4 (2012) 419–425. Soc. 109 (1987) 6883–6885. [67] M. Sachdeva, S. Dhingra, Obinutuzumab: a FDA approved monoclonal antibody in [91] W.M. Maiese, M.P. Lechevalier, H.A. Lechevalier, J. Korshalla, N. Kuck, A. Fantini, M.J. the treatment of untreated chronic lymphocytic leukemia, Int. J. Appl. Basic Med. Wildey, J. Thomas, M. Greenstein, Calicheamicins, a novel family of antitumor an- Res. 5 (2015) 54–57. tibiotics: taxonomy, fermentation and biological properties, J. Antibiot. 42 (1989) [68] M. Ratner, Genentech's glyco-engineered antibody to succeed Rituxan, Nat. 558–563. Biotechnol. 32 (2014) (6+). [92] J.M. Cassady, K.K. Chan, H.G. Floss, E. Leistner, Recent developments in the [69] J.M. Subramaniam, G. Whiteside, K. McKeage, J.C. Croxtall, Mogamulizumab: first maytansinoid antitumor agents, Chem. Pharm. Bull. 52 (2004) 1–26. global approval, Drugs 72 (2012) 1293–1298. [93] R.V. Chari, Targeted cancer therapy: conferring specificity to cytotoxic drugs, Acc. [70] A.K. Kakkar, S. Balakrishnan, Obinutuzumab for chronic lymphocytic leukemia: Chem. Res. 41 (2008) 98–107. promise of the first treatment approved with breakthrough therapy designation, [94] S.O. Doronina, B.A. Mendelsohn, T.D. Bovee, C.G. Cerveny, S.C. Alley, D.L. Meyer, E. J. Oncol. Pharm. Pract. 21 (2015) 358–363. Oflazoglu, B.E. Toki, R.J. Sanderson, R.F. Zabinski, A.F. Wahl, P.D. Senter, Enhanced [71] S.D. Liu, C. Chalouni, J.C. Young, T.T. Junttila, M.X. Sliwkowski, J.B. Lowe, activity of monomethylauristatin F through monoclonal antibody delivery: Afucosylated antibodies increase activation of FcgammaRIIIa-dependent signaling effects of linker technology on efficacy and toxicity, Bioconjug. Chem. 17 components to intensify processes promoting ADCC, Cancer Immunol. Res. 3 (2006) 114–124. (2015) 173–183. [95] T.V. Jerjian, A.E. Glode, L.A. Thompson, C.L. O'Bryant, Antibody-drug conjugates: a [72] T. Mizushima, H. Yagi, E. Takemoto, M. Shibata-Koyama, Y. Isoda, S. Iida, K. Masuda, clinical pharmacy perspective on an emerging cancer therapy, Pharmacotherapy M. Satoh, K. Kato, Structural basis for improved efficacy of therapeutic 36 (2016) 99–116. antibodies on defucosylation of their Fc glycans, Genes Cells 16 (2011) [96] R.C. Elgersma, R.G. Coumans, T. Huijbregts, W.M. Menge, J.A. Joosten, H.J. Spijker, 1071–1080. F.M. de Groot, M.M. van der Lee, R. Ubink, D.J. van den Dobbelsteen, D.F. Egging, [73] S. Preithner, S. Elm, S. Lippold, M. Locher, A. Wolf, A.J. da Silva, P.A. Baeuerle, N.S. W.H. Dokter, G.F. Verheijden, J.M. Lemmens, C.M. Timmers, P.H. Beusker, Design, Prang, High concentrations of therapeutic IgG1 antibodies are needed to compen- synthesis, and evaluation of linker-duocarmycin payloads: toward selection of sate for inhibition of antibody-dependent cellular cytotoxicity by excess endoge- HER2-targeting antibody-drug conjugate SYD985, Mol. Pharm. 12 (2015) nous immunoglobulin G, Mol. Immunol. 43 (2006) 1183–1193. 1813–1835. [74] N. Zhang, L. Liu, C. Dan Dumitru, N.R.H. Cummings, M. Cukan, Y. Jiang, Y. Li, F. Li, T. [97] J.A. Hartley, The development of pyrrolobenzodiazepines as antitumour agents, Ex- Mitchell, M.R. Mallem, Y. Ou, R.N. Patel, K. Vo, H. Wang, I. Burnina, B.-K. Choi, H.E. pert Opin. Investig. Drugs 20 (2011) 733–744. Huber, T.A. Stadheim, D. Zha, Glycoengineered Pichia produced anti-HER2 is com- [98] J.E. Koblinski, M. Ahram, B.F. Sloane, Unraveling the role of proteases in cancer, parable to trastuzumab in preclinical study, MAbs 3 (2011) 289–298. Clin. Chim. Acta 291 (2000) 113–135. [75] R.L. Shields, A.K. Namenuk, K. Hong, Y.G. Meng, J. Rae, J. Briggs, D. Xie, J. Lai, A. [99] S.O. Doronina, B.E. Toki, M.Y. Torgov, B.A. Mendelsohn, C.G. Cerveny, D.F. Chace, R.L. Stadlen, B. Li, J.A. Fox, L.G. Presta, High resolution mapping of the binding site on DeBlanc, R.P. Gearing, T.D. Bovee, C.B. Siegall, J.A. Francisco, A.F. Wahl, D.L. Meyer, human IgG1 for Fc gamma RI, Fc gamma RII, Fc gamma RIII, and FcRn and design P.D. Senter, Development of potent monoclonal antibody auristatin conjugates of IgG1 variants with improved binding to the Fc gamma R, J. Biol. Chem. 276 for cancer therapy, Nat. Biotechnol. 21 (2003) 778–784. (2001) 6591–6604. [100] C. Vaklavas, A. Forero-Torres, Safety and efficacy of brentuximab vedotin in pa- [76] G.A. Lazar, W. Dang, S. Karki, O. Vafa, J.S. Peng, L. Hyun, C. Chan, H.S. Chung, A. tients with Hodgkin lymphoma or systemic anaplastic large cell lymphoma, Eivazi, S.C. Yoder, J. Vielmetter, D.F. Carmichael, R.J. Hayes, B.I. Dahiyat, Engineered Ther. Adv. Hematol. 3 (2012) 209–225. antibody Fc variants with enhanced effector function, Proc. Natl. Acad. Sci. 103 [101] J. Bendell, M. Saleh, A.A. Rose, P.M. Siegel, L. Hart, S. Sirpal, S. Jones, J. Green, E. (2006) 4005–4010. Crowley, R. Simantov, T. Keler, T. Davis, L. Vahdat, Phase I/II study of the [77] J.B. Stavenhagen, S. Gorlatov, N. Tuaillon, C.T. Rankin, H. Li, S. Burke, L. Huang, S. antibody-drug conjugate glembatumumab vedotin in patients with locally ad- Vijh, S. Johnson, E. Bonvini, S. Koenig, Fc optimization of therapeutic antibodies vanced or metastatic breast cancer, J. Clin. Oncol. 32 (2014) 3619–3625. enhances their ability to kill tumor cells in vitro and controls tumor expansion [102] H.K. Erickson, P.U. Park, W.C. Widdison, Y.V. Kovtun, L.M. Garrett, K. Hoffman, R.J. in vivo via low-affinity activating Fcgamma receptors, Cancer Res. 67 (2007) Lutz, V.S. Goldmacher, W.A. Blattler, Antibody-maytansinoid conjugates are acti- 8882–8890. vated in targeted cancer cells by lysosomal degradation and linker-dependent in- [78] S.T. Jung, S.T. Reddy, T.H. Kang, M.J. Borrok, I. Sandlie, P.W. Tucker, G. Georgiou, tracellular processing, Cancer Res. 66 (2006) 4426–4433. Aglycosylated IgG variants expressed in bacteria that selectively bind FcgammaRI [103] S. Alley, X. Zhang, N. Okeley, M. Anderson, E. Oflazoglu, I. Stone, C.-L. Law, P. Senter, potentiate tumor cell killing by monocyte-dendritic cells, Proc. Natl. Acad. Sci. D. Benjamin, Effects of linker chemistry on tumor targeting by anti-CD70 antibody- 107 (2010) 604–609. drug conjugates, Cancer Res. 67 (2007) 916. [79] S.L. Sazinsky, R.G. Ott, N.W. Silver, B. Tidor, J.V. Ravetch, K.D. Wittrup, Aglycosylated [104] S. Girish, M. Gupta, B. Wang, D. Lu, I.E. Krop, C.L. Vogel, H.A. Burris Iii, P.M. LoRusso, immunoglobulin G1 variants productively engage activating Fc receptors, Proc. J.H. Yi, O. Saad, B. Tong, Y.W. Chu, S. Holden, A. Joshi, Clinical pharmacology of Natl. Acad. Sci. 105 (2008) 20167–20172. trastuzumab emtansine (T-DM1): an antibody-drug conjugate in development [80] M.J. Borrok, S.T. Jung, T.H. Kang, A.F. Monzingo, G. Georgiou, Revisiting the role of for the treatment of HER2-positive cancer, Cancer Chemother. Pharmacol. 69 glycosylation in the structure of human IgG Fc, ACS Chem. Biol. 7 (2012) (2012) 1229–1240. 1596–1602. [105] A.G. Polson, J. Calemine-Fenaux, P. Chan, W. Chang, E. Christensen, S. Clark, F.J. de [81] R. Stewart, G. Thom, M. Levens, G. Guler-Gane, R. Holgate, P.M. Rudd, C. Webster, L. Sauvage, D. Eaton, K. Elkins, J.M. Elliott, G. Frantz, R.N. Fuji, A. Gray, K. Harden, Jermutus, J. Lund, A variant human IgG1-Fc mediates improved ADCC, Protein Eng. G.S. Ingle, N.M. Kljavin, H. Koeppen, C. Nelson, S. Prabhu, H. Raab, S. Ross, D.S. Des. Sel. 24 (2011) 671–678. Slaga, J.P. Stephan, S.J. Scales, S.D. Spencer, R. Vandlen, B. Wranik, S.F. Yu, B. [82] V. Kayser, N. Chennamsetty, V. Voynov, K. Forrer, B. Helk, B.L. Trout, Glycosylation Zheng, A. Ebens, Antibody-drug conjugates for the treatment of non- influences on the aggregation propensity of therapeutic monoclonal antibodies, Hodgkin's lymphoma: target and linker-drug selection, Cancer Res. 69 Biotechnol. J. 6 (2011) 38–44. (2009) 2358–2364. [83] Y. Mimura, S. Church, R. Ghirlando, P.R. Ashton, S. Dong, M. Goodall, J. Lund, R. [106] H.K. Erickson, G.D. Lewis Phillips, D.D. Leipold, C.A. Provenzano, E. Mai, H.A. Jefferis, The influence of glycosylation on the thermal stability and effector func- Johnson, B. Gunter, C.A. Audette, M. Gupta, J. Pinkas, J. Tibbitts, The effect of differ- tion expression of human IgG1-Fc: properties of a series of truncated glycoforms, ent linkers on target cell catabolism and pharmacokinetics/pharmacodynamics of Mol. Immunol. 37 (2000) 697–706. trastuzumab maytansinoid conjugates, Mol. Cancer Ther. 11 (2012) 1133–1142. [84] J.L. Zaro, Mylotarg: revisiting its clinical potential post-withdrawal, in: J. Wang, W.- [107] R.P. Lyon, D.L. Meyer, J.R. Setter, P.D. Senter, Conjugation of anticancer drugs C. Shen, J.L. Zaro (Eds.), Antibody-Drug Conjugates, Springer International Publish- through endogenous monoclonal antibody cysteine residues, Methods Enzymol. ing 2015, pp. 179–190. 502 (2012) 123–138. [85] F.J. Burrows, P.E. Thorpe, Eradication of large solid tumors in mice with an [108] M.P. Brun, L. Gauzy-Lazo, Protocols for lysine conjugation, Methods Mol. Biol. 1045 immunotoxin directed against tumor vasculature, Proc. Natl. Acad. Sci. 90 (1993) (2013) 173–187. 8996–9000. [109] C.R. Behrens, E.H. Ha, L.L. Chinn, S. Bowers, G. Probst, M. Fitch-Bruhns, J. Monteon, [86] A. Palumbo, F. Hauler, P. Dziunycz, K. Schwager, A. Soltermann, F. Pretto, C. Alonso, A. Valdiosera, A. Bermudez, S. Liao-Chan, T. Wong, J. Melnick, J.W. Theunissen, M.R. G.F. Hofbauer, R.W. Boyle, D. Neri, A chemically modified antibody mediates com- Flory, D. Houser, K. Venstrom, Z. Levashova, P. Sauer, T.S. Migone, E.H. van der plete eradication of tumours by selective disruption of tumour blood vessels, Br. J. Horst, R.L. Halcomb, D.Y. Jackson, Antibody-drug conjugates (ADCs) derived from Cancer 104 (2011) 1106–1115. interchain cysteine cross-linking demonstrate improved homogeneity and other [87] A. Visintin, K. Knowlton, E. Tyminski, C.I. Lin, X. Zheng, K. Marquette, S. Jain, L. pharmacological properties over conventional heterogeneous ADCs, Mol. Pharm. Tchistiakova, D. Li, C.J. O'Donnell, A. Maderna, X. Cao, R. Dunn, W.B. Snyder, A.K. 12 (2015) 3986–3998. Abraham, M. Leal, S. Shetty, A. Barry, L. Zawel, A.J. Coyle, H.F. Dvorak, S.C. [110] J.R. Junutula, H. Raab, S. Clark, S. Bhakta, D.D. Leipold, S. Weir, Y. Chen, M. Simpson, Jaminet, Novel anti-TM4SF1 antibody-drug conjugates with activity against S.P. Tsai, M.S. Dennis, Y. Lu, Y.G. Meng, C. Ng, J. Yang, C.C. Lee, E. Duenas, J. Gorrell, tumor cells and tumor vasculature, Mol. Cancer Ther. 14 (2015) V. Katta, A. Kim, K. McDorman, K. Flagella, R. Venook, S. Ross, S.D. Spencer, W. Lee 1868–1876. Wong, H.B. Lowman, R. Vandlen, M.X. Sliwkowski, R.H. Scheller, P. Polakis, W. Z. Elgundi et al. / Advanced Drug Delivery Reviews 122 (2017) 2–19 17

Mallet, Site-specific conjugation of a cytotoxic drug to an antibody improves the [132] P.A. Moore, W. Zhang, G.J. Rainey, S. Burke, H. Li, L. Huang, S. Gorlatov, M.C. Veri, S. therapeutic index, Nat. Biotechnol. 26 (2008) 925–932. Aggarwal, Y. Yang, K. Shah, L. Jin, S. Zhang, L. He, T. Zhang, V. Ciccarone, S. Koenig, E. [111] J.Y. Axup, K.M. Bajjuri, M. Ritland, B.M. Hutchins, C.H. Kim, S.A. Kazane, R. Halder, Bonvini, S. Johnson, Application of dual affinity retargeting molecules to achieve J.S. Forsyth, A.F. Santidrian, K. Stafin, Y. Lu, H. Tran, A.J. Seller, S.L. Biroc, A. optimal redirected T-cell killing of B-cell lymphoma, Blood 117 (2011) 4542–4551. Szydlik, J.K. Pinkstaff, F. Tian, S.C. Sinha, B. Felding-Habermann, V.V. Smider, P.G. [133] S.M. Kipriyanov, G. Moldenhauer, J. Schuhmacher, B. Cochlovius, C.W. Von der Schultz, Synthesis of site-specific antibody-drug conjugates using unnatural Lieth, E.R. Matys, M. Little, Bispecific tandem diabody for tumor therapy with im- amino acids, Proc. Natl. Acad. Sci. 109 (2012) 16101–16106. proved antigen binding and pharmacokinetics, J. Mol. Biol. 293 (1999) 41–56. [112] E.S. Zimmerman, T.H. Heibeck, A. Gill, X. Li, C.J. Murray, M.R. Madlansacay, C. Tran, [134] F. McAleese, M. Eser, RECRUIT-TandAbs: harnessing the immune system to kill N.T. Uter, G. Yin, P.J. Rivers, A.Y. Yam, W.D. Wang, A.R. Steiner, S.U. Bajad, K. Penta, cancer cells, Future Oncol. 8 (2012) 687–695. W. Yang, T.J. Hallam, C.D. Thanos, A.K. Sato, Production of site-specific antibody- [135] N. Liddy, G. Bossi, K.J. Adams, A. Lissina, T.M. Mahon, N.J. Hassan, J. Gavarret, F.C. drug conjugates using optimized non-natural amino acids in a cell-free expression Bianchi, N.J. Pumphrey, K. Ladell, E. Gostick, A.K. Sewell, N.M. Lissin, N.E. system, Bioconjug. Chem. 25 (2014) 351–361. Harwood, P.E. Molloy, Y. Li, B.J. Cameron, M. Sami, E.E. Baston, P.T. Todorov, S.J. [113] A.M. Sochaj, K.W. Swiderska, J. Otlewski, Current methods for the synthesis of ho- Paston, R.E. Dennis, J.V. Harper, S.M. Dunn, R. Ashfield, A. Johnson, Y. McGrath, G. mogeneous antibody-drug conjugates, Biotechnol. Adv. 33 (2015) 775–784. Plesa, C.H. June, M. Kalos, D.A. Price, A. Vuidepot, D.D. Williams, D.H. Sutton, B.K. [114] R. van Geel, M.A. Wijdeven, R. Heesbeen, J.M. Verkade, A.A. Wasiel, S.S. van Berkel, Jakobsen, Monoclonal TCR-redirected tumor cell killing, Nat. Med. 18 (2012) F.L. van Delft, Chemoenzymatic conjugation of toxic payloads to the globally 980–987. conserved N-glycan of native mAbs provides homogeneous and highly [136] G. Schaefer, L. Haber, L.M. Crocker, S. Shia, L. Shao, D. Dowbenko, K. Totpal, A. efficacious antibody-drug conjugates, Bioconjug. Chem. 26 (2015) Wong, C.V. Lee, S. Stawicki, R. Clark, C. Fields, G.D. Lewis Phillips, R.A. Prell, D.M. 2233–2242. Danilenko, Y. Franke, J.P. Stephan, J. Hwang, Y. Wu, J. Bostrom, M.X. Sliwkowski, [115] A. Nisonoff, M.M. Rivers, Recombination of a mixture of univalent antibody frag- G. Fuh, C. Eigenbrot, A two-in-one antibody against HER3 and EGFR has superior ments of different specificity, Arch. Biochem. Biophys. 93 (1961) 460–462. inhibitory activity compared with monospecific antibodies, Cancer Cell 20 [116] U.D. Staerz, M.J. Bevan, Hybrid hybridoma producing a bispecific monoclonal anti- (2011) 472–486. body that can focus effector T-cell activity, Proc. Natl. Acad. Sci. 83 (1986) [137] C. Wu, H. Ying, C. Grinnell, S. Bryant, R. Miller, A. Clabbers, S. Bose, D. McCarthy, R.R. 1453–1457. Zhu, L. Santora, R. Davis-Taber, Y. Kunes, E. Fung, A. Schwartz, P. Sakorafas, J. Gu, E. [117] K. Gunasekaran, M. Pentony, M. Shen, L. Garrett, C. Forte, A. Woodward, S.B. Ng, T. Tarcsa, A. Murtaza, T. Ghayur, Simultaneous targeting of multiple disease media- Born, M. Retter, K. Manchulenko, H. Sweet, I.N. Foltz, M. Wittekind, W. Yan, En- tors by a dual-variable-domain immunoglobulin, Nat. Biotechnol. 25 (2007) hancing antibody Fc heterodimer formation through electrostatic steering effects: 1290–1297. applications to bispecific molecules and monovalent IgG, J. Biol. Chem. 285 (2010) [138] K.E. Conrath, M. Lauwereys, M. Galleni, A. Matagne, J.M. Frere, J. Kinne, L. Wyns, S. 19637–19646. Muyldermans, Beta-lactamase inhibitors derived from single-domain antibody [118] T.S. Von Kreudenstein, E. Escobar-Carbrera, P.I. Lario, I. D'Angelo, K. Brault, J. Kelly, fragments elicited in the camelidae, Antimicrob. Agents Chemother. 45 (2001) Y. Durocher, J. Baardsnes, R.J. Woods, M.H. Xie, P.A. Girod, M.D. Suits, M.J. 2807–2812. Boulanger, D.K. Poon, G.Y. Ng, S.B. Dixit, Improving biophysical properties of a [139] S.D. Jones, P. Seymour, H.L. Levine, CMC Activities for Development of MAbs, Con- bispecific antibody scaffold to aid developability: quality by molecular design, tract Pharma, 2010 60–63. MAbs 5 (2013) 646–654. [140] M. Chartrain, L. Chu, Development and production of commercial therapeutic [119] S.M. Lewis, X. Wu, A. Pustilnik, A. Sereno, F. Huang, H.L. Rick, G. Guntas, A. Leaver- monoclonal antibodies in Mammalian cell expression systems: an overview Fay, E.M. Smith, C. Ho, C. Hansen-Estruch, A.K. Chamberlain, S.M. Truhlar, E.M. of the current upstream technologies, Curr. Pharm. Biotechnol. 9 (2008) Conner, S. Atwell, B. Kuhlman, S.J. Demarest, Generation of bispecific IgG antibod- 447–467. ies by structure-based design of an orthogonal Fab interface, Nat. Biotechnol. 32 [141] M.P. Robinson, N. Ke, J. Lobstein, C. Peterson, A. Szkodny, T.J. Mansell, C. Tuckey, (2014) 191–198. P.D. Riggs, P.A. Colussi, C.J. Noren, C.H. Taron, M.P. DeLisa, M. Berkmen, Efficient ex- [120] J.B. Ridgway, L.G. Presta, P. Carter, ‘Knobs-into-holes’ engineering of antibody CH3 pression of full-length antibodies in the cytoplasm of engineered bacteria, Nat. domains for heavy chain heterodimerization, Protein Eng. Des. Sel. 9 (1996) Commun. 6 (2015) 8072. 617–621. [142] S.T. Jung, W. Kelton, T.H. Kang, D.T. Ng, J.T. Andersen, I. Sandlie, C.A. Sarkar, G. [121] W. Schaefer, J.T. Regula, M. Bahner, J. Schanzer, R. Croasdale, H. Durr, C. Gassner, G. Georgiou, Effective phagocytosis of low Her2 tumor cell lines with engineered, Georges, H. Kettenberger, S. Imhof-Jung, M. Schwaiger, K.G. Stubenrauch, C. aglycosylated IgG displaying high FcgammaRIIa affinity and selectivity, ACS Sustmann, M. Thomas, W. Scheuer, C. Klein, Immunoglobulin domain crossover Chem. Biol. 8 (2013) 368–375. as a generic approach for the production of bispecific IgG antibodies, Proc. Natl. [143] H. Li, N. Sethuraman, T.A. Stadheim, D. Zha, B. Prinz, N. Ballew, P. Bobrowicz, B.K. Acad. Sci. 108 (2011) 11187–11192. Choi, W.J. Cook, M. Cukan, N.R. Houston-Cummings, R. Davidson, B. Gong, S.R. [122] E.A. Rossi, D.M. Goldenberg, T.M. Cardillo, W.J. McBride, R.M. Sharkey, C.H. Chang, Hamilton, J.P. Hoopes, Y. Jiang, N. Kim, R. Mansfield, J.H. Nett, S. Rios, R. Stably tethered multifunctional structures of defined composition made by the Strawbridge, S. Wildt, T.U. Gerngross, Optimization of humanized IgGs in dock and lock method for use in cancer targeting, Proc. Natl. Acad. Sci. 103 glycoengineered Pichia pastoris, Nat. Biotechnol. 24 (2006) 210–215. (2006) 6841–6846. [144] T.I. Potgieter, S.D. Kersey, M.R. Mallem, A.C. Nylen, M. d'Anjou, Antibody expression [123] E.A. Rossi, D.M. Goldenberg, C.H. Chang, The dock-and-lock method combines re- kinetics in glycoengineered Pichia pastoris, Biotechnol. Bioeng. 106 (2010) combinant engineering with site-specific covalent conjugation to generate multi- 918–927. functional structures, Bioconjug. Chem. 23 (2012) 309–323. [145] J. Ye, J. Ly, K. Watts, A. Hsu, A. Walker, K. McLaughlin, M. Berdichevsky, B. Prinz, D. [124] J.H. Davis, C. Aperlo, Y. Li, E. Kurosawa, Y. Lan, K.M. Lo, J.S. Huston, SEED bodies: Sean Kersey, M. d'Anjou, D. Pollard, T. Potgieter, Optimization of a glycoengineered fusion proteins based on strand-exchange engineered domain (SEED) CH3 Pichia pastoris cultivation process for commercial antibody production, Biotechnol. heterodimers in an Fc analogue platform for asymmetric binders or Prog. 27 (2011) 1744–1750. immunofusions and bispecific antibodies, Protein Eng. Des. Sel. 23 (2010) [146] J.K. Ma, T. Lehner, P. Stabila, C.I. Fux, A. Hiatt, Assembly of monoclonal antibodies 195–202. with IgG1 and IgA heavy chain domains in transgenic tobacco plants, Eur. J. [125] R. Zeidler, J. Mysliwietz, M. Csanady, A. Walz, I. Ziegler, B. Schmitt, B. Wollenberg, Immunol. 24 (1994) 131–138. H. Lindhofer, The Fc-region of a new class of intact bispecific antibody mediates ac- [147] B. De Muynck, C. Navarre, M. Boutry, Production of antibodies in plants: status after tivation of accessory cells and NK cells and induces direct phagocytosis of tumour twenty years, Plant Biotechnol. J. 8 (2010) 529–563. cells, Br. J. Cancer 83 (2000) 261–266. [148] G. Moussavou, K. Ko, J.H. Lee, Y.K. Choo, Production of monoclonal antibodies in [126] M.G. Ott, F. Marme, G. Moldenhauer, H. Lindhofer, M. Hennig, R. Spannagl, M.M. plants for cancer immunotherapy, Biomed. Res. Int. 2015 (2015) 306164. Essing, R. Linke, D. Seimetz, Humoral response to catumaxomab correlates with [149] B.M. Grohs, Y. Niu, L.J. Veldhuis, S. Trabelsi, F. Garabagi, J.A. Hassell, M.D. McLean, clinical outcome: results of the pivotal phase II/III study in patients with malignant J.C. Hall, Plant-produced trastuzumab inhibits the growth of HER2 positive cancer ascites, Int. J. Cancer 130 (2012) 2195–2203. cells, J. Agric. Food Chem. 58 (2010) 10056–10063. [127] E. Wolf, R. Hofmeister, P. Kufer, B. Schlereth, P.A. Baeuerle, BiTEs: bispecific anti- [150] A.D. Bandaranayake, S.C. Almo, Recent advances in mammalian protein produc- body constructs with unique anti-tumor activity, Drug Discov. Today 10 (2005) tion, FEBS Lett. 588 (2014) 253–260. 1237–1244. [151] S.M. Browne, M. Al-Rubeai, Selection methods for high-producing mammalian cell [128] P.A. Baeuerle, P. Kufer, R. Bargou, BiTE: teaching antibodies to engage T-cells for lines, Trends Biotechnol. 25 (2007) 425–432. cancer therapy, Curr. Opin. Mol. Ther. 11 (2009) 22–30. [152] S. Shi, R.G. Condon, L. Deng, J. Saunders, F. Hung, Y.S. Tsao, Z. Liu, A high-through- [129] M.J. Newman, D.J. Benani, A review of blinatumomab, a novel immunotherapy, J. put automated platform for the development of manufacturing cell lines for pro- Oncol. Pharm. Pract. 22 (2016) 639–645. tein therapeutics, J. Vis. Exp. 55 (2011), e3010 http://dx.doi.org/10.3791/3010. [130] M.S. Topp, N. Gokbuget, A.S. Stein, G. Zugmaier, S. O'Brien, R.C. Bargou, H. Dombret, [153] L. Chu, D.K. Robinson, Industrial choices for protein production by large-scale cell A.K. Fielding, L. Heffner, R.A. Larson, S. Neumann, R. Foa, M. Litzow, J.M. Ribera, A. culture, Curr. Opin. Biotechnol. 12 (2001) 180–187. Rambaldi, G. Schiller, M. Bruggemann, H.A. Horst, C. Holland, C. Jia, T. Maniar, B. [154] H.F. Liu, J. Ma, C. Winter, R. Bayer, Recovery and purification process development Huber, D. Nagorsen, S.J. Forman, H.M. Kantarjian, Safety and activity of for monoclonal antibody production, MAbs 2 (2010) 480–499. blinatumomab for adult patients with relapsed or refractory B-precursor acute [155] J.R. Birch, A.J. Racher, Antibody production, Adv. Drug Deliv. Rev. 58 (2006) lymphoblastic leukaemia: a multicentre, single-arm, phase 2 study, Lancet Oncol. 671–685. 16 (2015) 57–66. [156] R.L. Fahrner, H.L. Knudsen, C.D. Basey, W. Galan, D. Feuerhelm, M. Vanderlaan, G.S. [131] S. Johnson, S. Burke, L. Huang, S. Gorlatov, H. Li, W. Wang, W. Zhang, N. Tuaillon, J. Blank, Industrial purification of pharmaceutical antibodies: development, opera- Rainey, B. Barat, Y. Yang, L. Jin, V. Ciccarone, P.A. Moore, S. Koenig, E. Bonvini, Effec- tion, and validation of chromatography processes, Biotechnol. Genet. Eng. Rev. 18 tor cell recruitment with novel Fv-based dual-affinity re-targeting protein leads to (2001) 301–327. potent tumor cytolysis and in vivo B-cell depletion, J. Mol. Biol. 399 (2010) [157] W. Schwart, D. Judd, M. Wysocki, L. Guerrier, E. Birck-Wilson, E. Boschetti, 436–449. Comparison of hydrophobic charge induction chromatography with affinity 18 Z. Elgundi et al. / Advanced Drug Delivery Reviews 122 (2017) 2–19

chromatography on protein A for harvest and purification of antibodies, J. [186] V.I. Razinkov, M.J. Treuheit, G.W. Becker, Accelerated formulation development of Chromatogr. A 908 (2001) 251–263. monoclonal antibodies (mAbs) and mAb-based modalities: review of methods [158] R. van Reis, S. Gadam, L.N. Frautschy, S. Orlando, E.M. Goodrich, S. Saksena, R. and tools, J. Biomol. Screen. 20 (2015) 468–483. Kuriyel, C.M. Simpson, S. Pearl, A.L. Zydney, High performance tangential flow fil- [187] H. Liu, G. Gaza-Bulseco, J. Sun, Characterization of the stability of a fully human tration, Biotechnol. Bioeng. 56 (1997) 71–82. monoclonal IgG after prolonged incubation at elevated temperature, J. [159] B. Lebreton, A. Brown, R. van Reis, Application of high-performance tangential Chromatogr. B 837 (2006) 35–43. flow filtration (HPTFF) to the purification of a human pharmaceutical antibody [188] J. Vlasak, R. Ionescu, Fragmentation of monoclonal antibodies, MAbs 3 (2011) fragment expressed in Escherichia coli, Biotechnol. Bioeng. 100 (2008) 253–263. 964–974. [189] R.J. Harris, B. Kabakoff, F.D. Macchi, F.J. Shen, M. Kwong, J.D. Andya, S.J. Shire, N. [160] T. Arakawa, M. Futatsumori-Sugai, K. Tsumoto, Y. Kita, H. Sato, D. Ejima, MEP Bjork, K. Totpal, A.B. Chen, Identification of multiple sources of charge heterogene- HyperCel chromatography II: binding, washing and elution, Protein Expr. Purif. ity in a recombinant antibody, J. Chromatogr. B Biomed. Sci. Appl. 752 (2001) 71 (2010) 168–173. 233–245. [161] O. Manzke, H. Tesch, V. Diehl, H. Bohlen, Single-step purification of bispecific [190] H. Liu, G. Gaza-Bulseco, C. Chumsae, Glutamine deamidation of a recombinant monoclonal antibodies for immunotherapeutic use by hydrophobic interaction monoclonal antibody, Rapid Commun. Mass Spectrom. 22 (2008) 4081–4088. chromatography, J. Immunol. Methods 208 (1997) 65–73. [191] Y.T. Zhang, J. Hu, A.L. Pace, R. Wong, Y.J. Wang, Y.-H. Kao, Characterization of aspar- [162] D. Nagorsen, P. Kufer, P.A. Baeuerle, R. Bargou, Blinatumomab: a historical perspec- agine 330 deamidation in an Fc-fragment of IgG1 using cation exchange chroma- tive, Pharmacol. Ther. 136 (2012) 334–342. tography and peptide mapping, J. Chromatogr. B 965 (2014) 65–71. [163] L. Ducry, Challenges in the development and manufacturing of antibody-drug con- [192] A.L. Pace, R.L. Wong, Y.T. Zhang, Y.-H. Kao, Y.J. Wang, Asparagine deamidation de- jugates, Methods Mol. Biol. 899 (2012) 489–497. pendence on buffer type, pH, and temperature, J. Pharm. Sci. 102 (2013) [164] R.A. Rader, Bioprocessing Challenges of Antibody–Drug Conjugates, BioProcess In- 1712–1723. ternational, 2014. [193] L. Yi, N. Beckley, B. Gikanga, J. Zhang, Y.J. Wang, H.-W. Chih, V.K. Sharma, Isomeri- [165] C. Scanlan, E. Gousseinov, A. Becerra-Artega, R. Waghmare, Challenges and Strate- zation of Asp–Asp motif in model peptides and a monoclonal antibody fab frag- gies for the Downstream Processing of BiSpecific Antibodies (BsAbs), Journal of ment, J. Pharm. Sci. 102 (2013) 947–959. Antibody-Drug Conjugates, InPress Media Group, 2014. [194] V. Timm, P. Gruber, M. Wasiliu, H. Lindhofer, D. Chelius, Identification and charac- [166] M.K. Joubert, M. Hokom, C. Eakin, L. Zhou, M. Deshpande, M.P. Baker, T.J. Goletz, terization of oxidation and deamidation sites in monoclonal rat/mouse hybrid an- B.A. Kerwin, N. Chirmule, L.O. Narhi, V. Jawa, Highly aggregated antibody therapeu- tibodies, J. Chromatogr. B 878 (2010) 777–784. tics can enhance the in vitro innate and late-stage T-cell immune responses, J. Biol. [195] L. Huang, J. Lu, V.J. Wroblewski, J.M. Beals, R.M. Riggin, In vivo deamidation charac- Chem. 287 (2012) 25266–25279. terization of monoclonal antibody by LC/MS/MS, Anal. Chem. 77 (2005) [167] S. Hermeling, D.J. Crommelin, H. Schellekens, W. Jiskoot, Structure-immunogenicity 1432–1439. relationships of therapeutic proteins, Pharm. Res. 21 (2004) 897–903. [196] A. Sreedhara, J. Yin, M. Joyce, K. Lau, A.T. Wecksler, G. Deperalta, L. Yi, Y. John Wang, [168] D.S. Rehder, D. Chelius, A. McAuley, T.M. Dillon, G. Xiao, J. Crouse-Zeineddini, L. B. Kabakoff, R.S.K. Kishore, Effect of ambient light on IgG1 monoclonal antibodies Vardanyan, N. Perico, V. Mukku, D.N. Brems, M. Matsumura, P.V. Bondarenko, during drug product processing and development, Eur. J. Pharm. Biopharm. 100 Isomerization of a single aspartyl residue of anti-epidermal growth factor receptor (2016) 38–46. immunoglobulin gamma2 antibody highlights the role avidity plays in antibody [197] A. Sreedhara, K. Lau, C. Li, B. Hosken, F. Macchi, D. Zhan, A. Shen, D. Steinmann, C. activity, Biochemistry 47 (2008) 2518–2530. Schöneich, Y. Lentz, Role of Surface Exposed Tryptophan as Substrate Generators for [169] A.L. Daugherty, R.J. Mrsny, Formulation and delivery issues for monoclonal anti- the Antibody Catalyzed Water Oxidation Pathway, Mol. Pharm. 10 (2013) 278–288. bodytherapeutics,Adv.DrugDeliv.Rev.58(2006)686–706. [198] R. Torosantucci, C. Schöneich, W. Jiskoot, Oxidation of therapeutic proteins and [170] Y. Yan, H. Wei, Y. Fu, S. Jusuf, M. Zeng, R. Ludwig, S.R. Krystek, G. Chen, L. Tao, T.K. peptides: structural and biological consequences, Pharm. Res. 31 (2014) 541–553. Das, Isomerization and oxidation in the complementarity-determining regions of a [199] W.Wang,J.Vlasak,Y.Li,P.Pristatsky,Y.Fang,T.Pittman,J.Roman,Y.Wang,T. monoclonal antibody: a study of the modification–structure–function correlations Prueksaritanont, R. Ionescu, Impact of methionine oxidation in human IgG1 Fc by hydrogen–deuterium exchange mass spectrometry, Anal. Chem. 88 (2016) on serum half-life of monoclonal antibodies, Mol. Immunol. 48 (2011) 2041–2050. 860–866. [171] B. Moorthy, B. Xie, E. Moussa, L. Iyer, S. Chandrasekhar, J. Panchal, E. Topp, Struc- [200] W. Wang, S. Singh, D.L. Zeng, K. King, S. Nema, Antibody structure, instability, and ture of monoclonal antibodies, in: A. Rosenberg, B. Demeule (Eds.), Biobetters, formulation, J. Pharm. Sci. 96 (2007) 1–26. Springer, New York 2015, pp. 81–89. [201] B. Yan, Z. Yates, A. Balland, G.R. Kleemann, Human IgG1 hinge fragmentation [172] N. Chennamsetty, V. Voynov, V. Kayser, B. Helk, B.L. Trout, Design of therapeutic as the result of H2O2-mediated radical cleavage, J. Biol. Chem. 284 (2009) proteins with enhanced stability, Proc. Natl. Acad. Sci. 106 (2009) 35390–35402. 11937–11942. [202] Z. Wei, J. Feng, H.-Y. Lin, S. Mullapudi, E. Bishop, G.I. Tous, J. Casas-Finet, F. Hakki, R. [173] N. Chennamsetty, V. Voynov, V. Kayser, B. Helk, B.L. Trout, Prediction of Strouse, M.A. Schenerman, Identification of a single tryptophan residue as critical aggregation prone regions of therapeutic proteins, J. Phys. Chem. B 114 for binding activity in a humanized monoclonal antibody against respiratory syn- (2010) 6614–6624. cytial virus, Anal. Chem. 79 (2007) 2797–2805. [174] F. Courtois, C.P. Schneider, N.J. Agrawal, B.L. Trout, Rational design of biobetters [203] C.J. Wilson, Rational protein design: developing next-generation biological thera- with enhanced stability, J. Pharm. Sci. 104 (2015) 2433–2440. peutics and nanobiotechnological tools, Wiley Interdiscip. Rev. Nanomed. [175] M. Ahmadi, C.J. Bryson, E.A. Cloake, K. Welch, V. Filipe, S. Romeijn, A. Hawe, W. Nanobiotechnol. 7 (2015) 330–341. Jiskoot, M.P. Baker, M.H. Fogg, Small amounts of sub-visible aggregates enhance [204] N. Cerda-Costa, A. Esteras-Chopo, F.X. Aviles, L. Serrano, V. Villegas, Early kinetics of the immunogenic potential of monoclonal antibody therapeutics, Pharm. Res. 32 amyloid fibril formation reveals conformational reorganisation of initial aggre- (2015) 1383–1394. gates, J. Mol. Biol. 366 (2007) 1351. [176] J. Bessa, S. Boeckle, H. Beck, T. Buckel, S. Schlicht, M. Ebeling, A. Kiialainen, A. [205] N. de Groot, I. Pallares, F. Aviles, J. Vendrell, S. Ventura, Prediction of “hot spots” of Koulov, B. Boll, T. Weiser, T. Singer, A.G. Rolink, A. Iglesias, The immunogenicity aggregation in disease-linked polypeptides, BMC Struct. Biol. 5 (2005) 18. of antibody aggregates in a novel transgenic mouse model, Pharm. Res. 32 [206] U. Ogmen, O. Keskin, S. Aytuna, R. Nussinoiv, A. Gursoy, PRISM: protein (2015) 2344–2359. interactions by structural matching, Nucleic Acids Res. 33 (2005) [177] M.C. Manning, D.K. Chou, B.M. Murphy, R.W. Payne, D.S. Katayama, Stability of pro- W331–W336. tein pharmaceuticals: an update, Pharm. Res. 27 (2010) 544–575. [207] G.G. Tartaglia, M. Vendruscolo, The Zyggregator method for predicting protein ag- [178] E. Sahin, A.O. Grillo, M.D. Perkins, C.J. Roberts, Comparative effects of pH and ionic gregation propensities, Chem. Soc. Rev. 37 (2008) 1395–1401. strength on protein–protein interactions, unfolding, and aggregation for IgG1 anti- [208] A. Trovato, F. Seno, S.C.E. Tosatto, The PASTA server for protein aggregation predic- bodies,J.Pharm.Sci.99(2010)4830–4848. tion, Protein Eng. Des. Sel. (2007), gzm042. [179] P. Arosio, S. Rima, M. Morbidelli, Aggregation mechanism of an IgG2 and two IgG1 [209] G.G. Tartaglia, A. Cavalli, R. Pellarin, A. Caflisch, Prediction of aggregation rate and monoclonal antibodies at low pH: from oligomers to larger aggregates, Pharm. Res. aggregation-prone segments in polypeptide sequences, Protein Sci. 14 (2005) 30 (2013) 641–654. 2723–2734. [180] S.N. Telikepalli, O.S. Kumru, C. Kalonia, R. Esfandiary, S.B. Joshi, C.R. Middaugh, D.B. [210] X. Wang, T.K. Das, S.K. Singh, S. Kumar, Potential aggregation prone regions in Volkin, Structural characterization of IgG1 mAb aggregates and particles generated biotherapeutics, MAbs 1 (2009) 254–267. under various stress conditions, J. Pharm. Sci. 103 (2014) 796–809. [211] V.E. Angarica, J. Sancho, Protein dynamics governed by interfaces of high polarity [181] F. Depreter, G. Pilcer, K. Amighi, Inhaled proteins: challenges and perspectives, Int. and low packing density, PLoS ONE 7 (2012), e48212. J. Pharm. 447 (2013) 251–280. [212] R. Zambrano, M. Jamroz, A. Szczasiuk, J. Pujols, S. Kmiecik, S. Ventura, [182] V. Kayser, N. Chennamsetty, V. Voynov, B. Helk, K. Forrer, B.L. Trout, Evaluation of a AGGRESCAN3D (A3D): server for prediction of aggregation properties of protein non-Arrhenius model for therapeutic monoclonal antibody aggregation, J. Pharm. structures, Nucleic Acids Res. 43 (2015) W306–W313. Sci. 100 (2011) 2526–2542. [213] D. Schweizer, T. Serno, A. Goepferich, Controlled release of therapeutic antibody [183] N. Kim, R.L. Remmele Jr., D. Liu, V.I. Razinkov, E.J. Fernandez, C.J. Roberts, Aggrega- formats, Eur. J. Pharm. Biopharm. 88 (2014) 291–309. tion of anti-streptavidin immunoglobulin gamma-1 involves Fab unfolding and [214] A.R. Karow, S. Bahrenburg, P. Garidel, Buffer capacity of biologics - from buffer salts competing growth pathways mediated by pH and salt concentration, Biophys. to buffering by antibodies, Biotechnol. Prog. 29 (2013) 480–492. Chem. 172 (2013) 26–36. [215] L. Chang, D. Shepherd, J. Sun, X. Tang, M.J. Pikal, Effect of sorbitol and residual mois- [184] H. Wu, R. Kroe-Barrett, S. Singh, A.S. Robinson, C.J. Roberts, Competing aggregation ture on the stability of lyophilized antibodies: Implications for the mechanism of pathways for monoclonal antibodies, FEBS Lett. 588 (2014) 936–941. protein stabilization in the solid state, J. Pharm. Sci. 94 (2005) 1445–1455. [185] S.B. Mehta, J.S. Bee, T.W. Randolph, J.F. Carpenter, Partial unfolding of a monoclonal [216] L. Platts, R.J. Falconer, Controlling protein stability: mechanisms revealed using for- antibody: role of a single domain in driving protein aggregation, Biochemistry 53 mulations of arginine, glycine and guanidinium HCl with three globular proteins, (2014) 3367–3377. Int. J. Pharm. 486 (2015) 131–135. Z. Elgundi et al. / Advanced Drug Delivery Reviews 122 (2017) 2–19 19

[217] S. Ohtake, Y. Kita, T. Arakawa, Interactions of formulation excipients with proteins [225] M. Smola, T. Vandamme, A. Sokolowski, Nanocarriers as pulmonary drug delivery in solution and in the dried state, Adv. Drug Deliv. Rev. 63 (2011) 1053–1073. systems to treat and to diagnose respiratory and non respiratory diseases, Int. J. [218] M.X. Yang, B. Shenoy, M. Disttler, R. Patel, M. McGrath, S. Pechenov, A.L. Margolin, Nanomedicine 3 (2008) 1–19. Crystalline monoclonal antibodies for subcutaneous delivery, Proc. Natl. Acad. Sci. [226] S.J. Koussoroplis, G. Paulissen, D. Tyteca, H. Goldansaz, J. Todoroff, C. Barilly, C. 100 (2003) 6934–6939. Uyttenhove, J. Van Snick, D. Cataldo, R. Vanbever, PEGylation of antibody frag- [219] D. Schweizer, K. Schonhammer, M. Jahn, A. Gopferich, Protein-polyanion interac- ments greatly increases their local residence time following delivery to the respi- tions for the controlled release of monoclonal antibodies, Biomacromolecules 14 ratory tract, J. Control. Release 187 (2014) 91–100. (2013) 75–83. [227] L. Dellamary, D.J. Smith, A. Bloom, S. Bot, G.R. Guo, H. Deshmuk, M. Costello, A. [220] D.W. Grainger, Controlled-release and local delivery of therapeutic antibodies, Ex- Bot, Rational design of solid aerosols for immunoglobulin delivery by modula- pert. Opin. Biol. Ther. 4 (2004) 1029–1044. tion of aerodynamic and release characteristics, J. Control. Release 95 (2004) [221] M. Shimizu, Y. Miwa, K. Hashimoto, A. Goto, Encapsulation of chicken egg yolk im- 489–500. munoglobulin G (IgY) by liposomes, Biosci. Biotechnol. Biochem. 57 (1993) [228] A.I. Bot, T.E. Tarara, D.J. Smith, S.R. Bot, C.M. Woods, J.G. Weers, Novel lipid-based 1445–1449. hollow-porous microparticles as a platform for immunoglobulin delivery to the re- [222] X.Y. Li, L.J. Jin, T.A. McAllister, K. Stanford, J.Y. Xu, Y.N. Lu, Y.H. Zhen, Y.X. Sun, Y.P. spiratory tract, Pharm. Res. 17 (2000) 275–283. Xu, Chitosan-alginate microcapsules for oral delivery of egg yolk immunoglobulin [229] I.M. El-Sherbiny, N.M. El-Baz, M.H. Yacoub, Inhaled nano- and microparticles for (IgY), J. Agric. Food Chem. 55 (2007) 2911–2917. drug delivery, Glob. Cardiol. Sci. Pract. 2015 (2015) 2. [223] Y. Yun, Y.W. Cho, K. Park, Nanoparticles for oral delivery: targeted nanoparticles [230] L. Detalle, T. Stohr, C. Palomo, P.A. Piedra, B.E. Gilbert, V. Mas, A. Millar, U.F. Power, with peptidic ligands for oral protein delivery, Adv. Drug Deliv. Rev. 65 (2013) C. Stortelers, K. Allosery, J.A. Melero, E. Depla, Generation and characterization of 822–832. ALX-0171, a potent novel therapeutic nanobody for the treatment of [224] M. Streit, Z. Beleznay, L.R. Braathen, Topical application of the tumour necrosis respiratory syncytial virus infection, Antimicrob. Agents Chemother. 60 factor-alpha antibody inﬂiximab improves healing of chronic wounds, Int. (2016) 6–13. Wound J. 3 (2006) 171–179. Journal of Visualized Experiments www.jove.com

Video Article Laboratory Scale Production and Purification of a Therapeutic Antibody

Zehra Elgundi1, Vicki Sifniotis1, Mouhamad Reslan1, Esteban Cruz1, Veysel Kayser1 1 Faculty of Pharmacy, University of Sydney

Correspondence to: Veysel Kayser at [email protected]

URL: http://www.jove.com/video/55153 DOI: doi:10.3791/55153 Keywords: Biochemistry, Issue 119, Biosimilar, antibodies, therapeutic antibody, antibody production, protein production, HEK-293, transfection, protein purification, affinity chromatography, FPLC Date Published: 1/24/2017 Citation: Elgundi, Z., Sifniotis, V., Reslan, M., Cruz, E., Kayser, V. Laboratory Scale Production and Purification of a Therapeutic Antibody. J. Vis. Exp. (119), e55153, doi:10.3791/55153 (2017).

Abstract

Ensuring the successful production of a therapeutic antibody begins early on in the development process. The first stage is vector expression of the antibody genes followed by stable transfection into a suitable cell line. The stable clones are subjected to screening in order to select those clones with desired production and growth characteristics. This is a critical albeit time-consuming step in the process. This protocol considers vector selection and sourcing of antibody sequences for the expression of a therapeutic antibody. The methods describe preparation of vector DNA for stable transfection of a suspension variant of human embryonic kidney 293 (HEK-293) cell line, using polyethylenimine (PEI). The cells are transfected as adherent cells in serum-containing media to maximize transfection efficiency, and afterwards adapted to serum-free conditions. Large scale production, setup as batch overgrow cultures is used to yield antibody protein that is purified by affinity chromatography using an automated fast protein liquid chromatography (FPLC) instrument. The antibody yields produced by this method can provide sufficient protein to begin initial characterization of the antibody. This may include in vitro assay development or physicochemical characterization to aid in the time-consuming task of clonal screening for lead candidates. This method can be transferable to the development of an expression system for the production of biosimilar antibodies.

Video Link

The video component of this article can be found at http://www.jove.com/video/55153/

Introduction

The success of therapeutic antibodies continues to drive substantial investment into antibody development as a wave of next generation therapeutics begins. The antibody market is expected to be reshaped by antibody fragments 1, antibody-drug conjugates 2, bispecific antibodies 3 and engineered antibodies with favorable properties 4. Another class gaining pharmaceutical interest are biosimilars. Biosimilar antibodies are 'highly similar' replicate products of a therapeutic antibody that has already received regulatory approval. A proposed biosimilar must be comparable with the originator antibody with respect to its structure, function, animal toxicity, clinical safety and effectiveness, human pharmacokinetics (PK), pharmacodynamics (PD) and immunogenicity 5,6.

The approval rates of biosimilar antibodies have been slow due to the strict constraints on the final quality of the product. The exact manufacturing processes such as specific cell lines and culturing conditions through to the final processing steps can remain proprietary. What is more, the manufacturing of antibodies inherently involves a degree of variability which can add to the challenge of producing a highly similar product. A comprehensive physiochemical and biophysical characterization and comparison is quite difficult, yet a number of studies demonstrating the characteristics of biosimilar antibodies are emerging in the literature 7,8,9.

Human IgG antibodies consist of two identical heavy chains and two identical light chains. Tightly regulated expression of both chains is essential for optimal production of heterologous IgG protein in mammalian cells 10. Intra- as well as inter-chain disulfide bonds have to be formed and a number of post-translational modifications have to be introduced during protein biosynthesis. A number of vectors are available that have been designed specifically to express antibody genes (refer to Table of Materials). These antibody-specific vectors usually express the constant regions for both heavy and light chains so only the variable regions of each chain require cloning.

endoplasmic reticulum. On the other hand, multi-cistronic vectors have internal ribosome entry site (IRES) elements incorporated that allow expression of multiple genes as a single mRNA transcript with translation permitted from internal regions of the mRNA 11. In this instance, the heavy and light chain-encoding genes are coupled in an arrangement to achieve co-expression of both antibody chains 10,12.

While transiently transfected cells yield sufficient protein to perform a limited number of experiments, stably transfected cell lines that have undergone selection for genome integration can deliver higher yields. Higher protein amounts allow for assay development relating to in vitro characterization and can provide an indication of antibody quality in consideration for downstream applications such as clonal cell line and lead candidate selection.

Protocol

NOTE: A suitable mammalian expression vector must be used for this protocol. Here, a single construct containing two expression cassettes is used (i.e. heavy and light chain expression is driven by separate promoters). Trastuzumab heavy and light chains were previously cloned into the vector. This vector was a gift from Andrew Beavil, obtained through a not-for-profit plasmid repository 13. 1. Recovery and Scale-up of Vector DNA

NOTE: Vector DNA was received as a soft agar stab culture in Escherichia coli XL-1 Blue strain; vector carries hygromycin resistance.

1. Insert a sterile inoculation stab or loop into the soft agar of the stab culture then streak a Luria Bertani (LB) agar plate prepared with 75 µg/ml hygromycin for isolated colonies. Incubate the plate at 37 °C for 18-24 hr. 2. Inoculate a single colony into 5 ml Terrific broth (TB) containing 75 µg/ml hygromycin. Incubate the culture at 37 °C for 18-24 hr with 225 rpm shaking. 3. Use the overnight culture to prepare a glycerol stock of vector DNA by gently mixing 800 µl of culture with 200 µl 80% glycerol in a cryovial and freeze at -80 °C. 1. Add 100 µl of the overnight culture to 100 ml TB containing 75 µg/ml hygromycin in a baffled shaker flask (i.e. 1/1,000 dilution). Incubate the culture at 37 °C for 18-24 hr with 225 rpm shaking.

4. After overnight culture, extract and purify the DNA according to manufacturer's instructions of midi/maxi preparation kit with the following exception; during the final step, elute or resuspend DNA with water (pH 7.0-8.5). 5. Check concentration and purity of DNA by absorbance readings at 260 and 280 nm then store DNA at -20 °C.

NOTE: Optional (highly recommended): Sequence DNA using vector-specific primers to confirm identity. 2. Stable Transfection of HEK-293 Cells

1. Grow and maintain HEK-293 cells (suspension cells) according to standard protocols in serum-free media supplemented with 0.1% non-ionic 5 surfactant in Erlenmeyer flasks. Maintain at 2 x 10 cells/ml and subculture every fourth day. Culture cells at 37 °C with 5% CO2 and 120 rpm rotation.

NOTE: Cells should have been in culture for no less than 4 days and no more than 4 weeks prior to transfection. HEK-293 cells grown as monolayers in serum-containing media can also be used for this procedure. 2. On the day prior to transfection, seed HEK-293 cells at 3 x 105 cells/ml in wells of a 12-well plate in 2 ml of Dulbecco's Modified Eagle Media (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS). On the day of transfection, check that cells have reached 80-90% confluency 3. Dilute DNA and polyethylenimine (PEI) separately in transfection media then mix together. 1. Dilute 1.25 µg vector DNA per well (15 µg for 12 wells) in 300 µl transfection media. Incubate at room temperature for 5 min. 2. Dilute 2.5 µl of 1 mg/ml PEI solution (30 µl for 12 wells; i.e. 1:2 ratio of DNA to PEI) in 300 µl transfection media. Incubate at room temperature for 5 min. 3. Add diluted DNA to diluted PEI, gently mix and incubate at room temperature for 15 min.

4. Add 50 µl of DNA/PEI mixture dropwise to each well of the plate. Rock the plate gently to distribute the transfection mixture then place the plate at 37 °C, 5% CO2 for 24 hr. 5. Add 50 µg/ml hygromycin B per well and return plate to 37 °C, 5% CO for 10 days. 2 NOTE: By Day 10, un-transfected cells will have died and detached from the surface of the plate, whereas stably-transfected cells will be viable and attached to the plate. 6. Replace the media in wells with 1/4 volume serum-free media and 3/4 volume DMEM supplemented with 10% FBS (i.e. 0.5 ml serum-free media + 1.5 ml DMEM = 7.5% final serum concentration) and incubate for 4 days. 7. Replace the media in wells with 1/2 volume serum-free media and 1/2 volume DMEM supplemented with 10% FBS (i.e. 1 ml serum-free media + 1 ml DMEM = 5% final serum concentration) and incubate for 4 days. 8. Replace the media in wells with 3/4 volume serum-free media and 1/4 volume DMEM supplemented with 10% FBS (i.e. 1.5 ml serum-free media + 0.5 ml DMEM = 2.5% final serum concentration) and incubate for 4 days.

9. Replace the media in wells with 2 ml of serum-free media supplemented with 0.1% non-ionic surfactant (i.e. 0% final serum concentration) and incubate for 4 days.

NOTE: Maintain 50 µg/ml hygromycin B selective pressure on cells during serum-free adaptation. Cells adapted to serum-free media detach from the surface of wells and may cluster in suspension. Refer to step 2.1 for culture conditions with the additional supplement of 50 µg/ml hygromycin B. 10. Using a pipette, gently mix the suspension cultures in the 12-well plate and pool cells into a separate tube. 1. Using a hemocytometer, enumerate pooled cells and seed in 30 ml serum-free media in a Erlenmeyer flask at 2 x 105 cells/ml. Culture cells at 37 °C with 5% CO2 and 120 rpm rotation. Subculture and expand every fourth day.

11. Continue to expand culture size to required cell density to obtain 10 vials at 1 x 107 cells/vial for cryopreservation in liquid nitrogen. Freeze the cells in serum-free media containing 10% dimethylsulfoxide (DMSO).

NOTE: Conditioned media containing antibody can be harvested at each subculture to confirm protein production or used to optimize purification conditions. To harvest conditioned media and maintain cells for subculture, centrifuge the culture at 300 x g for 5 min then filter- sterilize supernatant through a 0.22 µm filter. Filtered supernatant may be kept at 4 °C for 1-2 weeks or -20 °C for longer term storage. 3. Large Scale Antibody Production from Batch Overgrow Culture

NOTE: Antibody production can be followed on from step 2.11 where existing cells are expanded to the required cell density based on the batch overgrow culture volume to be setup. Otherwise, cells that have been thawed from cryopreservation begin at this stage once expanded to an appropriate cell density and volume. Various culture supplements may be used to optimize antibody production; the use of tryptone to increase antibody yield is demonstrated in this protocol.

1. Seed cells at 2 x 105 cells/ml in 100 ml serum-free media in two Erlenmeyer flasks (to compare antibody yields from un-supplemented and

nutrient-supplemented cultures). Culture at 37 °C, 5% CO2 and 120 rpm. 2. After 24 hr, add tryptone to a final concentration of 0.5% to nutrient-supplemented culture. Equalize volume of un-supplemented culture with serum-free media. 3. From day 8, perform cell counts of the cultures daily using a hemocytometer to monitor cell viability. Harvest the culture supernatants once cell viability is less than 80%. 1. Harvest supernatant by centrifugation of the cultures at 3,000 x g for 15 min then filter-sterilize the supernatant (containing antibody) through a 0.22 µm filter. Store the supernatants at 4 °C (short-term) or freeze at -20 °C (long-term).

4. Antibody Purification by Affinity Chromatography Using an Automated Fast Protein Liquid Chromatography (FPLC) System

NOTE: The following procedure can generally be applied to most automated systems. Purification can be performed at room temperature or at 4 °C (if FPLC system is kept in a cool room). A series of scouting tests can be performed to identify the optimal purification conditions including appropriate column matrix, binding buffer, elution buffer and pH to ensure maximum recovery of purified antibody from conditioned media (refer to Results section). The optimal conditions are dependent on the antibody or protein being purified. Purifications were performed on an automated FPLC system. Purifications were performed at room temperature using a 5 ml Protein A column.

1. Prepare the following buffers using ultrapure water and adjust to the recommended pH then filter through a 0.22 µm filter. 1. Prepare 1 L of phosphate buffered saline (PBS) pH 7.4 (binding buffer) by mixing the following: 0.14 M NaCl, 0.0027 M KCl, 0.01 M Na2HPO4 and 0.001 M KH2PO4. Adjust pH if necessary before filtering the buffer. 2. Prepare 500 ml of 0.1 M glycine-HCl pH 2.7 (elution buffer). Adjust pH before filtering the buffer. 3. Prepare 50 ml of 1 M Tris pH 9.0 (neutralization buffer).

2. Ensure all system power and communication connections with computer are made. Devices should be visible in the software 'System Control' module. Ensure UV cell is set at 280 nm wavelength. Optional: Calibrate pH meter if connected and to be used. 3. Immerse inlet tubes of A, B and sample pump in ultrapure water to wash the system. Purge the pumps with a syringe if there is air in the tubing or a suspicion of air in the system. Wash the system with water by manual operation via the system controller or via an automated method. Observe that pressure, conductivity, UV280 tracing and pH remain consistent and that the pressure limit for the column is not exceeded as per manufacturer's specification.

NOTE: Abnormalities may indicate air or blockage in the system that should be addressed before proceeding. 4. In the system controller module, start the flowrate at 1 ml/min manually via pump A in either load (bypassing sample loop) or inject (through the sample loop) position to begin connecting the column. Disconnect system tubing at the column position inlet and remove the stopper connected to the column inlet. 1. Allow water to flow from the system tubing dropwise on top of the column then attach the system tubing to the column. Having the column inlet and inlet fitting overflowing with drops of water ensures a connection free of air bubbles.

5. Attach the column outlet at the downstream outlet and verify all fittings are tightly fastened. With the column now attached to the system, do not exceed limits for maximum pressure limit and flowrate as outlined in column specifications. 6. Wash column with 5 column volumes (CV) of water. If necessary, continue column wash until UV280 tracing has stabilized. 7. Immerse inlet tubes of A in Binding buffer, B in Elution buffer and sample pump in Binding buffer to equilibrate the system in correct buffers. Fill the inlet tubes with buffer using PumpWash by manual operation. 8. In the 'Method Editor' module of the software, use the method wizard to setup an affinity chromatography method for the column that is intended to be used.

NOTE: Automated FPLC systems come with methods pre-filled with the recommended settings (i.e. flowrate and pressure limit) and run steps based on the column (manufacturer, matrix and size) selected for purification.

1. Use the following run steps for Protein A affinity chromatography: 1. Equilibrate system with 5 CV of Binding buffer and collect flowthrough into waste container. 2. Load sample onto the column (volume to be loaded is specified manually in the method) and collect flowthrough into a separate container. 3. Wash system with 5 CV of Binding buffer and collect flowthrough into a separate container. 4. Elute column with 5 CV isocratic fractionation using Elution buffer and collect purified antibody sample as fractions by fraction collector. 5. Equilibrate system with 5 CV of Binding buffer and collect flowthrough into waste container.

9. Once the method has been setup, specify the volume of conditioned media to be applied to the column and save the method.

NOTE: The specified volume in the method should be 5-10 ml less than the actual volume to avoid introduction of air during the sample run. The specified volume used in this protocol is 90-120 ml. 10. Submerge the sample pump tubing into the vessel containing the conditioned media. 11. Prepare a separate container to collect sample flowthrough via the specified outlet tubing.

NOTE: It is important to collect the flowthrough in the case an error occurs during the run or the binding capacity of the column is exceeded requiring the flowthrough to be reapplied to the column or the purification repeated. 12. Prepare collection tubes in the fraction collector. Add 100 µl of neutralization buffer per 1 ml fraction volume. The FPLC system will automatically elute the fractions into the collection tubes. 13. In the 'System Control' module of the software, open the method to be run.

NOTE: The method run is initiated in a series of pages that include checking the variables of the method, fraction collector setup and defining result file name and storage location. 14. Click START to initiate the run. The run can be monitored in the 'System Control' module. 15. At the completion of the purification run, check the resulting chromatogram in the 'Evaluation' module in the system software. 16. Combine all protein-containing fractions into a separate tube, buffer exchange and concentrate in PBS using a centrifugal filter device with 30 kDa molecular weight cut off. Perform centrifugation steps according to manufacturer's instructions. 17. Measure antibody concentration using bicinchoninic acid assay (BCA) according to manufacturer's instructions. 18. If another purification run is to be performed, prime the sample pump tubing in Binding buffer and repeat steps 4.9-4.14. 19. At the completion of purification runs, immerse inlet tubes of A, B and sample pump in ultrapure water and wash the system and column as performed in step 3. 20. Submerge A, B and sample pump tubing in 20% ethanol and repeat wash procedure for storage of the system and column. 21. Disconnect the column from the downstream outlet and replace column stopper, then disconnect column at the inlet and replace stopper, reconnect tubing to the system. Store the column at 4 °C.

Representative Results

Stable production of trastuzumab by transfected HEK-293 cells was confirmed using bio-layer interferometry (BLI) as presented in Figure 1. An IgG standard curve was generated by measuring the binding rate between an IgG antibody standard and Protein A biosensor (Figure 1A). The crude supernatant sample was similarly measured, then its concentration interpolated from the standard curve (Figure 1B). The supernatant concentration was measured to be approximately 25 µg/ml (sampled from 50 ml collected at subculture) two weeks after serum-free adaptation and prior to setting up cells for overgrow batch cultures.

Supernatant collected at each subculture after serum-free adaptation was pooled and used to optimize purification conditions; scouting of binding and elution buffers was performed as shown in Figure 2. The steps of purification are depicted in Figure 2A based on UV280 signal. Once the Protein A column equilibrates, an increase in UV280 signal represents sample loading; this increase is due to supernatant flowthrough passing through the column (containing unbound proteins which absorb UV light at this wavelength), while antibodies have been captured by the column. Once the sample has finished loading, the UV280 signal returns to baseline as any remaining unbound proteins are washed away. Elution occurs as pH decreases to the optimal pH to release the antibodies captured by the column, depicted by an increase in UV280 signal with eluted antibodies fractionated and collected by the fraction collector. Throughout the sample run, conductivity changes based on the salt concentration in the buffers whereas system pressure should remain relatively constant. Optimal elution, pH and buffer was scouted first (Figure 2B); it was observed that antibody eluted more efficiently off the column at lower pH using either 0.1 M glycine HCl or citric acid. However, below pH 2.7 there was no further improvement in the elution profile. In addition, elution peaks with 0.1 M glycine HCl appeared less broadened when compared with 0.1 M citric acid at similar pH (Figure 2B). Subsequently, 0.1 M glycine-HCl pH 2.7 buffer was used to optimize binding buffer conditions (Figure 2A). The effect of different binding buffers on elution was also compared (Figure 2A). It was observed that PBS pH 7.4 and 20 mM sodium phosphate pH 7 had comparable elution profiles, while the addition of 3 M NaCl to sodium phosphate buffer (to enhance antibody binding to the column) was not favorable. Due to the ease of PBS buffer preparation, PBS pH 7.4 was selected as the binding buffer for future purifications.

The purification chromatograms of batch overgrow cultures are shown in Figure 3; tryptone-supplemented culture is compared to un- supplemented culture. It was noted that the addition of tryptone to the supplemented culture resulted in an increased UV280 signal during the sample loading step, compared to the un-supplemented culture. The tryptone-supplemented culture yielded 3.8 mg and the un-supplemented culture 1.7 mg of trastuzumab, based on protein recovered after buffer exchange and concentration. Quality control of the purified antibody was confirmed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) as shown in Figure 4. Trastuzumab grown in un- supplemented and tryptone-supplemented conditions results in similar antibody profiles under non-reducing and reducing conditions. As expected, a prominent band at approximately 150 kDa confirms correctly folded antibody; other bands represent fragmented forms of the antibody, a result of disulfide bond breakage and a method-induced artefact. Likewise, two bands at 50 and approximately 25 kDa confirm the presence of antibody heavy and light chains, respectively.

Figure 1: Confirmation of protein production by stably transfected HEK-293 cells using bio-layer interferometry (BLI). (A) Standard curve was generated by measuring binding rate of IgG antibody standard to protein A biosensor over 120 seconds (grey) followed by crude supernatant sample of PEI-transfected trastuzumab (PEI-Tmab; black). (B) Concentration of antibody protein in crude supernatant (open black circle) was interpolated from standard curve (closed black circles) by plotting IgG antibody standard concentration versus binding rate. Please click here to view a larger version of this figure.

Figure 2: Purification chromatograms of elution and binding buffer scouting experiments. (A) The steps of purification are depicted according to UV280 signal (blue, green or black lines); sample loading onto the column followed by column wash to wash away unbound proteins then elution of bound protein from the column. Measurement of pH (red line), conductivity (brown line) and system pressure (grey line) is monitored throughout the run. Three buffers were tested for optimal binding of trastuzumab to the column and washing away of unbound protein to maximize elution yield. (B) 0.1 M glycine-HCl (pH 2.5-3.3) and 0.1 M citric acid (pH 2.5-3.5) buffers were tested for optimal elution of trastuzumab from protein A column. Please click here to view a larger version of this figure.

Figure 3: Purification chromatograms of trastuzumab (TmAb) batch overgrow cultures grown with no supplement or supplemented 5 with tryptone. Stably transfected HEK-293 cells were setup at 2 x 10 cells/ml and cultured for 8 days either un-supplemented (120 ml; black line) or supplemented with 0.5% tryptone (90 ml; green line). After harvesting, supernatants were purified using PBS pH 7.4 (binding buffer) and 0.1 M glycine-HCl pH 2.7 (elution buffer). Please click here to view a larger version of this figure.

Figure 4: Analysis of purified trastuzumab by SDS-PAGE. Approximately 2 µg of trastuzumab purified from un-supplemented (-) or tryptone- supplemented (+) cultures was sampled by 10% SDS-PAGE under non-reduced or reduced conditions. The gel is stained with coomassie R-250. Please click here to view a larger version of this figure.

Discussion

This protocol details the transfection, stable expression and purification of a therapeutic antibody in HEK-293 cells. Stable expression of antibody genes is the first step in generating an antibody-producing cell line for the development and manufacture of a therapeutic antibody. While Chinese hamster ovary (CHO) cells remain the expression platform of choice for therapeutic proteins, the HEK-293 cell line is gaining prominence with the realization that proteins produced in these cells are a closer match to naturally occurring human proteins, in terms of post- translational modifications and function 14,15.

Mammalian cell lines including CHO (e.g., CHO-DG44 and CHO-K1) as well as non-immunoglobulin secreting murine B cell lines NS0 and SP2/0 are predominantly used in biopharmaceutical production. Expression systems based on these cell lines are based on selection processes whereby high producing clones are induced and selected through incremental addition of a specific selective drug 16,17. The selection process for a stable clone can therefore be arduous and time-consuming. In comparison to these existing methods, the HEK-293 cell line robustly transfects with high efficiency. The selection process is simplified and very easily adapts to serum-free suspension culture, making it an ideal expression system for laboratory scale production of proteins 18.

A limitation of stable transfection is time in which a practical amount of protein can be produced and utilized in experiments. To try and overcome this, the current protocol describes a culturing step that can achieve substantial amounts of protein within 3-6 weeks of transfection. The batch overgrow cultures (large scale production) provide the opportunity to produce substantial amounts of antibody for the setup of in vitro characterization experiments in the lead up to selection of a clonal cell line and lead candidates. This has clear advantages over transient transfection, which can require higher amounts of DNA and reagents. In addition, the protein yield is not reproducible from batch to batch since expression is only temporary and determined by transfection efficiency.

The addition of tryptone in this protocol provided an increased yield in protein expression. Addition of tryptone to transfection cultures has previously been shown to improve protein synthesis 19,20. The tryptone supplemented culture resulted in improved trastuzumab antibody yields of 40 mg/L versus the un-supplemented culture of 14 mg/L (Figure 3). SDS-PAGE verified that: (1) the antibody was being produced correctly; and (2) the addition of tryptone did not alter structure based on comparison of the antibody bands under non-reducing and reducing conditions (Figure 4). The addition of tryptone to the cultures is optional and is particularly used to achieve higher protein yields. Other supplements have been considered to enhance protein expression such as sodium butyrate and valproic acid 21,22,23.

The first checkpoint of the protocol is the confirmation that protein is being produced by the transfected cell line. This step may be performed any time after the two-week period of selective pressure used to select stable transformants. A number of methods may be used to confirm protein production including bio-layer interferometry (Figure 1) or the similar approach of an IgG ELISA. Alternatively, a western blot may be performed using an anti-IgG detection antibody to identify antibody protein in the crude supernatant or purified sample.

Other researchers have shown that higher transfection efficiency is achieved using adherent cells in serum-containing media 24. The presence of supplements from animal origin is undesirable for biopharmaceutical production. Therefore, the sequential adaptation to serum-free media is a critical step in the protocol requiring patience and careful monitoring of cell viability. The 4-day turnover period suggested in this protocol is a guide and so if problems are encountered at a certain serum concentration, it is recommended that the cells be subcultured 2-3 times in the previous ratio of serum-containing to serum-free media before continuing with the next ratio. Prior to setting up batch cultures and cryopreservation of cells, consider that most cell lines are deemed fully adapted after three subcultures in 100% serum-free media.

One of the most crucial steps at the purification stage is the scouting for optimal conditions and buffers to ensure efficient binding of the antibody to the column and then complete elution off the column (Figure 2). The conditioned media collected at each subculture is useful for this purpose. In this case, the elution buffer of citric acid pH 3.5 generally recommended for protein A affinity chromatography was unsuitable for elution of trastuzumab from the column. The scouting experiment using two different elution buffers at various pH clearly showed that even within the narrow range of pH tested, the degree of elution was affected (Figure 2B).

by other protein quantification methods including measurement of the absorbance signal at 280 nm with concentration deduced from Beer Lambert's Law based on the extinction coefficient of the antibody.

This protocol has successfully produced the antibody protein of trastuzumab by stable transfection of HEK-293 cells. The antibody was purified and characterized to confirm integrity. The steps in this protocol detailing DNA preparation, stable transfection followed by serum-free adaptation, large scale production and automated purification can be transferred to producing other proteins.

Disclosures

The authors declare that they have no competing financial interests.

Acknowledgements

The research was supported by the University of Sydney. pVITRO1-Trastuzumab-IgG1/κ was a gift from Andrew Beavil (Addgene plasmid # 61883). We thank Tihomir S. Dodev for useful discussions regarding pVITRO1-Trastuzumab-IgG1/κ.

References

1. Nelson, A. L., & Reichert, J. M. Development trends for therapeutic antibody fragments. Nat Biotechnol. 27 (4), 331-337 (2009). 2. Drake, P. M., & Rabuka, D. An emerging playbook for antibody-drug conjugates: lessons from the laboratory and clinic suggest a strategy for improving efficacy and safety. Curr Opin Chem Biol. 28 174-180 (2015). 3. Kontermann, R. E., & Brinkmann, U. Bispecific antibodies. Drug Discov Today. 20 (7), 838-847 (2015). 4. Beck, A., Wurch, T., Bailly, C., & Corvaia, N. Strategies and challenges for the next generation of therapeutic antibodies. Nat Rev Immunol. 10 (5), 345-352 (2010). 5. Food and Drug Administration. (eds U.S. Department of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and Research (CDER), & Center for Biologics Evaluation and Research (CBER)) 22 (2015). 6. European Medicines Agency. (eds European Medicines Agency (EMA) & Committee for Medicinal Products for Human Use (CHMP)) 13 (2014). 7. Lopez-Morales, C. A. et al. Physicochemical and biological characterization of a biosimilar trastuzumab. Biomed Res Int. 2015 427235 (2015). 8. Jung, S. K. et al. Physicochemical characterization of Remsima. MAbs. 6 (5), 1163-1177 (2014). 9. Visser, J. et al. Physicochemical and functional comparability between the proposed biosimilar rituximab GP2013 and originator rituximab. BioDrugs. 27 (5), 495-507 (2013). 10. Li, J. et al. A comparative study of different vector designs for the mammalian expression of recombinant IgG antibodies. J Immunol Methods. 318 (1-2), 113-124 (2007). 11. Fitzgerald, K. D., & Semler, B. L. Bridging IRES elements in mRNAs to the eukaryotic translation apparatus. Biochim Biophys Acta. 1789 (9-10), 518-528 (2009). 12. Ho, S. C. et al. IRES-mediated Tricistronic vectors for enhancing generation of high monoclonal antibody expressing CHO cell lines. J Biotechnol. 157 (1), 130-139 (2012). 13. Dodev, T. S. et al. A tool kit for rapid cloning and expression of recombinant antibodies. Sci Rep. 4 5885 (2014). 14. Walsh, G. Post-translational modifications of protein biopharmaceuticals. Drug Discov Today. 15 (17-18), 773-780 (2010). 15. Backliwal, G. et al. Rational vector design and multi-pathway modulation of HEK 293E cells yield recombinant antibody titers exceeding 1 g/l by transient transfection under serum-free conditions. Nucleic Acids Res. 36 (15), e96 (2008). 16. Bebbington, C. R. et al. High-level expression of a recombinant antibody from myeloma cells using a glutamine synthetase gene as an amplifiable selectable marker. Biotechnology (N Y). 10 (2), 169-175 (1992). 17. Kaufman, R. J. et al. Coamplification and coexpression of human tissue-type plasminogen activator and murine dihydrofolate reductase sequences in Chinese hamster ovary cells. Mol Cell Biol. 5 (7), 1750-1759 (1985). 18. Dalton, A. C., & Barton, W. A. Over-expression of secreted proteins from mammalian cell lines. Protein Sci. 23 (5), 517-525 (2014). 19. Pham, P. L. et al. Transient gene expression in HEK293 cells: peptone addition posttransfection improves recombinant protein synthesis. Biotechnol Bioeng. 90 (3), 332-344 (2005). 20. Pham, P. L. et al. Large-scale transient transfection of serum-free suspension-growing HEK293 EBNA1 cells: peptone additives improve cell growth and transfection efficiency. Biotechnol Bioeng. 84 (3), 332-342 (2003). 21. Yang, W. C. et al. Addition of valproic acid to CHO cell fed-batch cultures improves monoclonal antibody titers. Molecular Biotechnology. 56 (5), 421-428 (2014). 22. Backliwal, G. et al. Valproic acid: a viable alternative to sodium butyrate for enhancing protein expression in mammalian cell cultures. Biotechnol Bioeng. 101 (1), 182-189 (2008). 23. Jiang, Z., & Sharfstein, S. T. Sodium butyrate stimulates monoclonal antibody over-expression in CHO cells by improving gene accessibility. Biotechnol Bioeng. 100 (1), 189-194 (2008). 24. Jordan, M., & Wurm, F. Transfection of adherent and suspended cells by calcium phosphate. Methods. 33 (2), 136-143 (2004).

ISSN: 1475-6366 (Print) 1475-6374 (Online) Journal homepage: https://www.tandfonline.com/loi/ienz20

Benzimidazolium-based novel silver N-heterocyclic carbene complexes: synthesis, characterisation and in vitro antimicrobial activity

Yakup Sarı, Senem Akkoç, Yetkin Gök, Vicki Sifniotis, İlknur Özdemir, Selami Günal & Veysel Kayser

To cite this article: Yakup Sarı, Senem Akkoç, Yetkin Gök, Vicki Sifniotis, İlknur Özdemir, Selami Günal & Veysel Kayser (2016) Benzimidazolium-based novel silver N-heterocyclic carbene complexes: synthesis, characterisation and invitro antimicrobial activity, Journal of Enzyme Inhibition and Medicinal Chemistry, 31:6, 1527-1530, DOI: 10.3109/14756366.2016.1156102 To link to this article: https://doi.org/10.3109/14756366.2016.1156102

Published online: 17 Mar 2016.

Submit your article to this journal

Article views: 566

View Crossmark data

Citing articles: 4 View citing articles

Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=ienz20 www.tandfonline.com/ienz ISSN: 1475-6366 (print), 1475-6374 (electronic)

J Enzyme Inhib Med Chem, 2016; 31(6): 1527–1530 ! 2016 Informa UK Limited, trading as Taylor & Francis Group. DOI: 10.3109/14756366.2016.1156102

RESEARCH ARTICLE Benzimidazolium-based novel silver N-heterocyclic carbene complexes: synthesis, characterisation and in vitro antimicrobial activity

Yakup Sarı1, Senem Akkoc¸2,3, Yetkin Go¨k1, Vicki Sifniotis3, _Ilknur O¨ zdemir1, Selami Gu¨nal4, and Veysel Kayser3

1Department of Chemistry, Faculty of Science and Arts, Ino¨nu¨ University, Malatya, Turkey, 2Department of Chemistry, Faculty of Science, Erciyes University, Kayseri, Turkey, 3Faculty of Pharmacy, The University of Sydney, Sydney, Australia, and 4Department of Microbiology, Faculty of Medicine, Ino¨nu¨ University, Malatya, Turkey

Abstract Keywords This study reports the synthesis, characterisation and antimicrobial activity of five novel silver Antimicrobial activity, benzimidazole, N-het- N-heterocyclic carbene (Ag–NHC) complexes obtained by N-propylphthalimide and erocyclic carbene, silver complex N-methyldioxane substituted benzimidazolium salts with silver oxide. The reactions were performed at room temperature for 24 h in the absence of light. The obtained complexes were History identified and characterised by 1H and 13C NMR, FT-IR and elemental analysis techniques. The minimum inhibitory concentration (MIC) of the complexes was determined for E. coli, Received 2 December 2015 P. aeruginosa, E. faecalis, S. aureus, C. tropicalis and C. albicans in vitro through agar and broth Revised 4 February 2016 dilution. The results indicated that these complexes exhibit antimicrobial activity. In particular, Accepted 4 February 2016 complex 3 presented the significant broad spectrum antimicrobial activity. Published online 16 March 2016

Introduction The aim of this study was to synthesise novel benzimidazolium-based Ag–NHC complexes (2–6) in the pursuit of finding It has been well established that silver cations (Ag+)have biologically active compounds that can be further investigated for antimicrobial activity against a broad range of microorganisms1–4. their antimicrobial activity. The complexes (2–6) were charac- Ag+ in low concentrations is non-toxic to humans, therefore a terised by IR, elemental analysis, 1H and 13C NMR. The compound that releases Ag+ to the environment in a steady rate antimicrobial activities of the compounds (2–6) were determined make an effective antimicrobial agent2–10. The effectiveness of a against a panel of non-fastidious bacteria and yeasts. Complexes 3 silver complex as an antimicrobial agent depends on the and 6 were particularly effective against gram negative bacteria bioavailability of Ag+ when the compound is solubilised. Many and yeast, with compound 3 having the highest antimicrobial factors can influence this bioavailability of Ag+ including the activity across the panel of microorganisms in comparison to the solubility and ionisation of the compound, the delivery route and other complexes. the presence of anions in the environment such as chloride, bromide and sulphide1–3,7. It has also been well established that silver N-heterocyclic Materials and methods carbenes (Ag–NHCs) are a class of organometallic compounds Chemicals and equipment that are stable in the presence of air or moisture; the release of Ag+ to the environment is slow and consistent4,5,7,8. As Ag–NHC All reactions were made under argon gas using Standard Schlenk complexes can be easily and efficiently synthesised, the charac- techniques. All primary chemicals were purchased commercially terisation of this class of compounds has generated considerable from the following suppliers: Merck (Darmstadt, Germany), interest in order to determine their effectiveness as antimicrobial Sigma-Aldrich (Aldrich and Fluka, St. Louis, MO), Introgen agents. From investigations to date, many groups of researchers (Carlo-Erba, Val de Reuil, France and Acros Organics, Geel, have synthesised and determined the biological activity of a broad Belgium) and ThermoFisher Scientific (Alfa-Aesar, Lancashire, range of Ag–NHC complexes10–12. Specifically, compounds UK). The purchased chemicals include: o-phenylenediamine, derived from benzimidazole-based NHC ligands have shown formic acid, dimethylformamide (DMF), hexane, silver oxide, 1- promising antimicrobial activity9,11,13–20 and various studies have (chloromethyl)naphthalene, benzyl chloride, 3-methylbenzyl extended their research to anticancer activity6,9,13,14. chloride, 4-methylbenzyl chloride, 2,4,6-trimethylbenzyl chloride, diethyl ether, dichloromethane and ethyl alcohol. Prior to use, the solvents were further purified through conventional methods21: Address for correspondence: Senem Akkoc¸, Department of Chemistry, diethyl ether and hexane by distillation over Na and dichlor- Faculty of Science, Erciyes University, Kayseri 38039, Turkey. Tel: +90 omethane by distillation over P4O10. 352 437 52 62/+61 02 9351 2330. Fax: +90 352 437 49 33. E-mail: Mueller-Hinton Broth was purchased from HiMedia [email protected]; [email protected] Vicki Sifniotis, Faculty of Pharmacy, The University of Sydney, Sydney Laboratories Pvt. Ltd. (Mumbai, India) and RPMI 1640 broth 2006, Australia. Tel: +61 02 9351 6834. E-mail: vsif0221@ was purchased from Sigma-Aldrich (Chemie GmbH, Taufkirchen, uni.sydney.edu.au Germany). 1528 Y. Sarı et al. J Enzyme Inhib Med Chem, 2016; 31(6): 1527–1530

R' R'

N Ag O, CH Cl N - 2 2 2 X AgX Antimicrobial activity N N

R R 1 2-6

Scheme 1. Synthesis of N-propylphthalimide and N-methyldioxane substituted Ag–NHC complexes.

The melting points (m.p.) of compounds were performed on an Chloro [1-(N-propylphthalimide)-3-naphthalen-1-yl- Electrothermal 9200 melting point apparatus. Elemental analyses methylbenzimidazol-2-ylidene] silver (I), 3 of 2–6 were conducted on a LECO CHNS-932 Elemental Compound 3 was synthesised as described above from Analyser. FT-IR analyses were done on a Mattson 1000 spectro- 1-(N-propylphthalimide)-3-naphthalen-1-yl-methylbenzimidazo- photometer. The NMR spectra of the Ag–NHC complexes were lium chloride (2 mmol) and Ag O (1 mmol). The reaction performed on a Bruker AC300P FT spectrometer. Chemical shifts 2 produced complex 3 (0.51 g, 83%) as a solid. M.p.: 195–196 C. () were given in ppm relative to tetramethylsilane. 1 IR (CN) ¼ 1430.5 cm . Anal. calcd. for C29H23AgClN3O2 (588.83 g/mol): C, 59.15%; H, 3.94%; N, 7.14%. Found C, Synthesis and characterisation of Ag–NHC complexes 59.03%; H, 4.08%; N, 7.15%. 1 (2–6) H NMR (300.13 MHz, DMSO-d6), ; 2.10 (p, 2H, The benzimidazolium salts were prepared as follows: KOH CH2CH2CH2N, J ¼ 7.2 Hz); 3.56 and 4.46 (t, 4H, CH2CH2CH2N, J ¼ 7.2 Hz); 6.10 (s, 2H, CH2C10H7); 6.89– was added to benzimidazole solution in ethyl alcohol and alkyl 13 halide was added to this solution after 1 h. The solution was 8.08 (m, 15H, Ar-H). C NMR (75 MHz, DMSO-d6), ; 29.4 refluxed for 6 h then it was allowed to cool to room temperature. (CH2CH2CH2N); 35.5 and 47.5 (CH2CH2CH2N); 50.4 The precipitated KCl was removed by filtering. The obtained N- (CH2C10H7); 112.8, 123.3, 124.7, 124.9, 125.1, 125.9, 126.7, alkylbenzimidazole was purified by crystallization. Subsequently, 127.2, 128.9, 129.2, 130.7, 132.2, 133.6, 133.8, 134.2, 134.6 alkyl halide was added to the DMF solution of synthesised and 134.9 (Ar–C); 168.5 (C ¼ O). N-alkylbenzimidazole and it was stirred for 24 h at 80 C. Benzimidazolium salts were synthesised and purified by crystal- Bromo [1,3-bis(N-propylphthalimide)benzimidazol-2-yli- lisation in a mixture of ethyl alcohol–diethylether (2:1)17,22–24. dene] silver (I), 4 The Ag(I)-NHC complexes (2–6) were synthesised Compound 4 was synthesised as described above from 1-(N- from N-propylphthalimide and N-methyldioxane substituted propylphthalimide)-3-benzylbenzimidazolium chloride (2 mmol) benzimidazolium salts (1) (2 mmol) and Ag2O (1 mmol) in and Ag2O (1 mmol). The reaction produced complex 4 (0.51 g, 20 mL dry dichloromethane (Scheme 1). All reactions were 86%) as a solid. M.p.: 246–247 C. IR (CN) ¼ 1434.8 cm1. conducted in a Schlenk tube that was wrapped with alumin- Anal. calcd. for C29H24AgBrN3O2 (634.29 g/mol): C, ium foil for 24 h at room temperature. After the completion of 54.91%; H, 3.81%; N, 6.62%. Found C, 54.71%; H, 4.00%; the reactions, the dichloromethane was removed by vacuum to N, 6.82%. 1 obtain the crude products. The products were washed twice H NMR (300 MHz, DMSO-d6), ; 2.51 (p, 4H, with diethyl ether, filtered and recrystallised from dichlor- CH2CH2CH2N, J ¼ 7.5 Hz); 3.64 and 4.42 (t, 8H, 13 omethane–diethylether (2:1) or chloroform–diethylether (2:1) CH2CH2CH2N, J ¼ 7.5 Hz); 7.47–7.85 (m, 12H, Ar–H). C mixtures at room temperature. The structures of the com- NMR (75 MHz, DMSO-d6), ; 29.3 (CH2CH2CH2N); 35.1 and 1 pounds were confirmed by elemental analysis, FT-IR, H and 45.2 (CH2CH2CH2N); 112.6, 123.1, 124.6, 131.8, 133.4 and 134.4 13 C NMR. (Ar–C); 168.2 (C ¼ O).

Chloro [1-(N-propylphthalimide)-3-benzylbenzimidazol-2- Chloro [1-(2-methyl-1,4-benzodioxane)-3-benzylbenzimi- ylidene] silver(I), 2 dazol-2-ylidene] silver(I), 5 Compound 2 was synthesised as described above from 1-(N- Compound 5 was synthesised as described above from 1-(2- propylphthalimide)-3-benzylbenzimidazolium chloride (2 mmol) methyl-1,4-benzodioxane)-3-benzylbenzimidazolium chloride and Ag2O (1 mmol). The reaction provided complex 2 (0.45 g, (2 mmol) and Ag2O (1 mmol). The reaction produced complex 72%) as a solid. M.p.: 212–213 C. IR (CN) ¼ 1431.2 cm1. 5 (0.51 g, 80%) as a solid. M.p.: 166–167 C. IR 1 Anal. calcd. for C25H21AgClN3O2 (538.77 g/mol) was determined (CN) ¼ 1454.1 cm . Anal. calcd. for C23H20AgClN2O2 as C, 55.73%; H, 3.93%; N, 7.80%. Found C, 55.61%; H, 4.08%; (499.74 g/mol): C, 55.28%; H, 4.03%; N, 5.61%. Found C, N, 7.75%. 55.14%; H, 4.21%; N, 5.56%. 1 1 H NMR (300 MHz, DMSO-d6), ; 2.40 (p, 2H, H NMR (300 MHz, DMSO-d6), ; 4.06 and 4.22 (dd, 2H, CH2CH2CH2N, J ¼ 7.2 Hz); 3.84 and 4.61 (t, 4H, OCH2CH(CH2)O, J ¼ 9.9 Hz); 4.88 (m, 1H, OCH2CHO); 4.71 CH2CH2CH2N, J ¼ 7.2 Hz); 5.66 (s, 2H, CH2C6H5); 7.26–7.87 and 4.79 (dd, 2H, OCH2CH(CH2)O, J ¼ 9.9 Hz); 5.74 (s, 2H, 13 13 (m, 13H, Ar–H). C NMR (75 MHz, DMSO-d6), ; 29.3 CH2C6H5); 6.68–7.90 (m, 13H, Ar–H). C NMR (75 MHz, (CH2CH2CH2N); 35.4 and 45.5 (CH2CH2CH2N); 53.4 DMSO-d6), ; 49.0 (OCH2CH(CH2)O); 55.6 (OCH2CH(CH2)O); (CH2C6H5); 111.3, 112.3, 123.5, 124.3, 127.2, 128.5, 129.1, 65.1 (OCH2CH(CH2)O); 71.8 (CH2C6H5); 112.8, 112.9, 117.6, 131.9, 133.8, 134.1 and 134.9 (Ar–C); 168.2 (C¼O). 117.7, 121.9, 122.0, 122.2, 124.6, 124.7, 127.8, 128.5, 129.2, 133.6, 134.4, 136.7, 142.3 and 143.2 (Ar–C). DOI: 10.3109/14756366.2016.1156102 Benzimidazolium-based novel silver N-heterocyclic carbene complexes 1529

Table 1. MIC values of the Ag–NHC complexes (2–6) against a panel of non-fastidious microorganisms.

MIC (mg/mL) Gram negative bacteria Gram positive bacteria Yeasts Compound E. coli P. aeruginosa S. aureus E. faecalis C. albicans C. tropicalis 2 200 200 200 200 200 200 3 50 50 50 50 25 25 4 200 200 200 200 200 200 5 100 100 100 100 100 100 6 50 50 100 100 50 50 Ampicillin 3.12 – 3.12 1.56 – – Ciprofloxacin 1.56 3.12 0.39 0.78 – – Fluconazole – – – – 3.12 3.12

Chloro [1-(2-methyl-1,4-benzodioxane)-3-(4-methylben- In preparation for inoculating antimicrobial supplemented zyl)benzimidazol-2-ylidene]silver(I), 6 media, the cultures were prepared as a suspension to approximately 106 CFU/mL. The turbidity of the suspension was Compound 6 was synthesised as described above from 1-(2- standardised against a McFarlane Number 0.5 turbidity standard. methyl-1,4-benzodioxane)-3-(4-methylbenzyl)benzimidazolium Antimicrobial supplemented media was inoculated from the cell chloride (2 mmol) and Ag O (1 mmol). The reaction produced 2 suspensions with sterile 10 mL inoculating loops. After inocula- complex 6 (0.49 g, 77%) as a solid. M.p.: 172–173 C. IR tion, cultures were incubated at 35 C in static, aerobic conditions, (CN) ¼ 1443.2 cm1. Anal. calcd. for C H AgClN O 24 22 2 2 for 16–20 h for bacterial cultures and 48 h for yeast cultures. (513.76 g/mol): C, 56.11%; H, 4.32%; N, 5.45%. Found C, 56.03%; H, 4.52%; N, 5.46%. 1H NMR (300 MHz, DMSO), ; 4.07 and 4.20 (dd, 2H, Results and discussion OCH2CH(CH2)O, J ¼ 9.9 Hz); 4.93 (m, 1H, OCH2CHO); 4.75 Synthesis and characterisation and 4.81 (dd, 2H, OCH2CH(CH2)O, J ¼ 10.8 Hz); 2.30 (s, 3H, CH2C6H4(CH3)–4); 5.67 (s, 2H, CH2C6H4(CH3)-4); 6.69–7.87 (m, The Ag–NHC complexes can remain stable in the presence of air 13 12H, Ar–H). C NMR (75 MHz, DMSO), ; 21.1 (CH2C6H4 and moisture at room temperature, without evidence of decom- (CH3)-4); 49.0 (OCH2CH(CH2)O); 52.2 (OCH2CH(CH2)O); 65.3 position. Product yields were obtained between 72 and 86%. (OCH2CH(CH2)O); 71.9 (CH2C6H4(CH3)-4); 112.8, 112.9, 117.6, Generally, there is a typical proton signal of NCHN in the 1 117.7, 121.9, 122.2, 124.5, 127.8, 129.7, 133.6, 133.7, 134.4, range at 9–11 for benzimidazolium salts in the H NMR spectra. 137.8, 142.3 and 143.2 (Ar–C); 191.4 (C–Ag). However, we did not observed this single signal peak for 2–6. This result verified the synthesis of the Ag–NHC complexes. The pentet proton of N CH2CH2CH2C6H5 resonated at 2.40, 2.10 In vitro antimicrobial activity testing by agar and broth and 2.51 for 2–4, respectively. The triplet proton of N dilution CH2CH2CH2C6H5 resonated at 3.84 and 4.61 (for 2), 3.56 and 4.46 (for 3), 3.64 and 4.42 (for 4). The aromatic protons of silver The minimum inhibitory concentrations (MICs) for each of the compounds were seen at between 6.68 and 8.08 ppm in the 1H Ag–NHC complexes were determined on a variety of non- NMR. The carbene carbon chemical shift value belonging to fastidious bacteria and yeast strains. The Clinical and Laboratory compound 6 was observed at 191.4 ppm in the 13C NMR Standards Institute methods were followed: the agar dilution spectra. This single signal carbene peak was not observed for protocol was conducted on bacterial strains25 and the broth other synthesised new compounds (2–5). Carbonyl carbon was dilution protocol was conducted on yeast strains26. resonated at low area for example at 168.2, 168.5 and 168.2 for The following strains were used in this study: gram negative 2–4, respectively. The IR band was obtained at 1431.2, bacteria – Escherichia coli ATCC 25922TM and Pseudomonas (CN) 1430.5, 1434.8, 1454.1 and 1443.2 for 2–6, respectively. aeruginosa ATCC 27853TM; gram positive bacteria – Enterococcus faecalis ATCC 29212TM and Staphylococcus Antimicrobial activity aureus ATCC 29213TM; yeasts – Candida tropicalis and Candida albicans. All bacterial strains were sourced from the Protocols from the Clinical and Laboratory Standards Institute American Type Culture Collection (Rockville, MD) and all yeast were conducted to determine the MICs of 2–6 for E. coli, strains were kindly donated from the Department of P. aeruginosa, S. aureus, E. faecalis, C. albicans and C. Microbiology, Faculty of Medicine, Ege University (Izmir, tropicalis. The MIC was determined as the lowest concentration Turkey). at which the compound prevented visible growth for each All bacteria were subcultured on Mueller–Hinton Agar and all organism. yeasts were subcultured in RPMI 1640. For antimicrobial activity The results of the MIC values determined for complexes 2–6 tests, media was supplemented with Ag–NHC complexes and are provided in Table 1. The complexes showed varied antimicro- antimicrobial controls were prepared in serial dilution as bial activity against tested microorganisms. In comparison to the described in the CLSI protocols of the agar25 and broth26 dilution antibiotic controls, all complexes had a higher MIC value: methods. however complexes 3 and 6 were more effective than the other Stocks of the compounds 2–6 were prepared in dimethyl complexes. Complex 3, in particular, exhibited the most anti- sulfoxide (DMSO) and two-fold serial dilutions of the stocks were microbial activity across the whole panel compared to the other prepared in sterile distilled water. The Ag–NHC complexes were complexes. Compound 3 was more effective against the yeast prepared as an eight-dilution series from 800 mg/mL to strains C. albicans and C. tropicalis than against the bacterial 6.25 mg/mL. Ciprofloxacin, ampicillin and fluconazole were strains. The antimicrobial activity of complexes 2 and 4 was the used as positive control drugs to compare against 2–6. least effective, having similar activities across all microorganisms. 1530 Y. Sarı et al. J Enzyme Inhib Med Chem, 2016; 31(6): 1527–1530

Complex 5 again exhibited same activities across all microorgan- 10. Liu W, Gust R. Metal N-heterocyclic carbene complexes as potential isms, although it presented a moderate MIC value. Complexes 2 antitumor metallodrugs. Chem Soc Rev 2013;42:755–73. and 5 both contain a benzyl substituent, although 5 contains a 11. Deblock MC, Panzner MJ, Tessier CA, et al. Chapter 4 biologically active N-heterocyclic carbene–metal complexes. In: N-heterocyclic benzodioxane group and had lower MIC value than 2, which carbenes: from laboratory curiosities to efficient synthetic tools. contains a propyl phthalimide group. This suggests that the Royal Soc Chem; 2011:119–33. benzodioxane has an effect on the antimicrobial activity of 5. 12. Teyssot M-L, Jarrousse A-S, Manin M, et al. Metal–NHC Complex 3 includes a naphthalene-1-ylmethyl group and its complexes: a survey of anti-cancer properties. Dalton Trans performance is consistent with a previous work which indicated 2009;35:6894–902. that Ag–NHC with naphthalene-1-ylmethyl substituents has 13. Haque RA, Choo SY, Budagumpi S, et al. Synthesis, crystal higher antimicrobial activity, especially antifungal activity16. structures, characterization and biological studies of nitrile-functionalized silver(I) N-heterocyclic carbene complexes. Inorg Chim Complex 6 includes a 4-methylbenzyl group and its performance 27 Acta 2015;433:35–44. is markedly improved compared to a previous work , in which 14. Haque RA, Ghdhayeb MZ, Budagumpi S, et al. Non-symmetrically the authors demonstrated that their complex which included a 4- substituted N-heterocyclic carbene–Ag(I) complexes of benzimida- methylbenzyl group had lower antimicrobial activity. zol-2-ylidenes: synthesis, crystal structures, anticancer activity and transmetallation studies. Inorg Chim Acta 2013;394:519–25. 15. AkkoçS,Go¨kY,O¨ zdemir I,_ et al. N-Heterocyclic carbene silver Conclusion complexes: synthesis, characterization and in vitro antimicrobial This study describes the synthesis of five novel benzimidazolium studies. J Chin Adv Mater Soc 2014;2:20–30. ¨ ¨ ¨ based Ag–NHC complexes (2–6) which were characterised by IR 16. Gok Y, AkkoçS,Ç elikal OO, et al. In vitro antimicrobial studies of 1 13 naphthalen-1-ylmethyl substituted silver N-heterocyclic carbene and elemental analysis, H and C NMR. The complexes were complexes. Arab J Chem 2015. [Epub ahead of print]. doi: evaluated for biological activity against a panel of non-fastidious 10.1016/j.arabjc.2015.04.019. bacteria and yeasts. Complexes 3 and 6 were particularly effective 17. Go¨k Y, Akkoç S, Albayrak S, et al. N-Phenyl-substituted carbene against gram negative bacteria and yeast, with complex 3 having precursors and their silver complexes: synthesis, characterization the highest antimicrobial activity across the panel of microorgan- and antimicrobial activities. Appl Organomet Chem 2014;28: isms in comparison to the other complexes. 244–51. 18. Muskawar PN, Karthikeyan P, Aswar SA, et al. NHC–metal complexes based on benzimidazolium moiety for chemical trans- Declaration of interest formation. Arab J Chem 2012. [Epub ahead of print]. doi: 10.1016/ j.arabjc.2012.04.040. The authors report no declarations of interest. 19. Go¨k Y, AkkoçS,O¨ zerog˘Lu ÇelIkal_ O¨ , et al. N-functionalized We would like to thank the Inonu University Research Fund benzimidazol-2-ylidene silver complexes: synthesis, characteriza- (BAP 2011/25) for financial support. tion, and antimicrobial studies. Turkish J Chem 2013;37:1007–13. 20. Go¨k Y, Akkoç S, Erdog˘an H, et al. In vitro antimicrobial studies of new benzimidazolium salts and silver N-heterocyclic carbene References complexes. J Enzyme Inhib Med Chem 2016. [Epub ahead of 1. Maillard J-Y, Hartemann P. Silver as an antimicrobial: facts and gaps print]. doi: 10.3109/14756366.2015.1132210. in knowledge. Crit Rev Microbiol 2012;39:373–83. 21. Armarego WLF, Chai CLL. Chapter 4 – purification of organic 2. Kascatan-Nebioglu A, Panzner MJ, Tessier CA, et al. N- chemicals. In: Armarego WLF, Chai CLL, eds. Purification of Heterocyclic carbene–silver complexes: a new class of antibiotics. laboratory chemicals. 6th ed. Oxford: Butterworth-Heinemann; Coord Chem Rev 2007;251:884–95. 2009:88–444. 3. Asekunowo PO, Haque RA. Counterion-induced modulation in 22. AkkoçS,Go¨k Y. Synthesis and characterization of 1-phenyl-3- biochemical properties of nitrile functionalized silver(I)-N-hetero- alkylbenzimidazol-2-ylidene salts and their catalytic activities in the cyclic carbene complexes. J Coord Chem 2014;67:3649–63. Heck and Suzuki cross-coupling reactions. J Coord Chem 2013;66: 4. Hindi KM, Panzner MJ, Tessier CA, et al. The medicinal applica- 1396–404. tions of imidazolium carbene-metal complexes. Chem Rev 2009; 23. AkkoçS,Go¨kY,Ilhan_ IO_ ¨ , et al. In situ generation of efficient 109:3859–84. palladium N-heterocyclic carbene catalysts using benzimidazolium 5. Haque RA, Salman AW, Budagumpi S, et al. Silver(I)-N-heterocyc- salts for the Suzuki–Miyaura cross-coupling reaction. Curr Org lic carbene complexes of bis-imidazol-2-ylidenes having different Synth 2016. [Epub ahead of print]. doi: 10.2174/ aromatic-spacers: synthesis, crystal structure, and in vitro anti- 1570179413666151218200334. microbial and anticancer studies. App Organomet Chem 2013;27: 24. AkkoçS,Go¨kY,Ilhan_ IO_ ¨ , et al. N-Methylphthalimide-substituted 465–73. benzimidazolium salts and PEPPSI Pd–NHC complexes: synthesis, 6. Gautier A, Cisnetti F. Advances in metal–carbene complexes as characterization and catalytic activity in carbon–carbon bond- potent anti-cancer agents. Metallomics 2012;4:23–32. forming reactions. Beilstein J Org Chem 2016;12:81–8. 7. Garrison JC, Youngs WJ. Ag(I) N-heterocyclic carbene complexes: 25. Clinical and Laboratory Standards Institute. M07-A10: methods for synthesis, structure, and application. Chem Rev 2005;105: dilution antimicrobial susceptibility tests for bacteria that grow 3978–4008. aerobically; Approved Standard, 10th ed. Vol. M07-A10. Wayne, 8. Budagumpi S, Haque RA, Endud S, et al. Biologically relevant PA: Clinical and Laboratory Standards Institute; 2015. silver(I)-N-heterocyclic carbene complexes: synthesis, structure, 26. Clinical and Laboratory Standards Institute. M27-A3: reference intramolecular interactions, and applications. Eur J Inorg Chem method for broth dilution antifungal susceptibility testing of yeasts; 2013;2013:4367–88. Approved Standard, 3rd ed. Vol. M27-A3. Wayne, PA: Clinical and 9. Haque RA, Choo SY, Budagumpi S, et al. Silver(I) complexes of Laboratory Standards Institute; 2015. mono- and bidentate N-heterocyclic carbene ligands: synthesis, 27. Go¨k Y, Sarı Y, Akkoç S, et al. Antimicrobial studies of crystal structures, and in vitro antibacterial and anticancer studies. N-heterocyclic carbene silver complexes containing benzimidazol- Eur J Med Chem 2015;90:82–92. 2-ylidene ligand. Inter J Inorg Chem 2014;2014:1–6.