View / Download 7.8 Mb

Sortase As a Tool in Biotechnology and Medicine

Joseph J. Bellucci III

Department of Biomedical Engineering Duke University

Date:______Approved:

______Ashutosh Chilkoti, Supervisor

______Charles A. Gersbach

______Kris Wood

______Fan Yuan

______Michael R. Zalutsky

Dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Biomedical Engineering in the Graduate School of Duke University

2016

ABSTRACT

Sortase As a Tool in Biotechnology and Medicine

Joseph J. Bellucci III

Department of Biomedical Engineering Duke University

Date:______Approved:

______Ashutosh Chilkoti, Supervisor

______Charles A. Gersbach

______Kris Wood

______Fan Yuan

______Michael R. Zalutsky

An abstract of a dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Biomedical Engineering in the Graduate School of Duke University

2016

Abstract

We have harnessed two reactions catalyzed by the enzyme sortase A and applied them to generate new methods for the purification and site-selective modification of recombinant protein therapeutics.

We utilized native peptide ligation —a well-known function of sortase A— to attach a small molecule drug specifically to the carboxy-terminus of a recombinant protein. By combining this reaction with the unique phase behavior of elastin-like polypeptides, we developed a protocol that produces homogenously-labeled protein- small molecule conjugates using only centrifugation. The same reaction can be used to produce unmodified therapeutic proteins simply by substituting a single reactant. The isolated proteins or protein-small molecule conjugates do not have any exogenous purification tags, eliminating the potential influence of these tags on bioactivity. Because both unmodified and modified proteins are produced by a general process that is the same for any protein of interest and does not require any chromatography, the time, effort, and cost associated with protein purification and modification is greatly reduced.

We also developed an innovative and unique method that attaches a tunable number of drug molecules to any recombinant protein of interest in a site-specific manner. Although the ability of sortase A to carry out native peptide ligation is widely used, we demonstrated that Sortase A is also capable of attaching small molecules to

proteins through an isopeptide bond at lysine side chains within a unique amino acid sequence. This reaction —isopeptide ligation— is a new site-specific conjugation method that is orthogonal to all available protein-small conjugation technologies and is the first site-specific conjugation method that attaches the payload to lysine residues. We show that isopeptide ligation can be applied broadly to peptides, proteins, and antibodies using a variety of small molecule cargoes to efficiently generate stable conjugates. We thoroughly assessed the site-selectivity of this reaction using a variety of analytical methods and showed that in many cases the reaction is site-specific for lysines in flexible, disordered regions of the substrate proteins. Finally, we showed that isopeptide ligation can be used to create clinically-relevant antibody-drug conjugates that have potent cytotoxicity towards cancerous cells.

Contents

Abstract ...... iv

List of Tables ...... xi

List of Figures ...... xii

Acknowledgements ...... xvii

1. Introduction ...... 1

1.1 Proteins and antibodies for cancer therapy ...... 1

1.1.1 Antibody-drug conjugates ...... 3

1.1.2 Protein-polymer conjugates ...... 4

1.2 Challenges associated with therapeutic protein production ...... 5

1.2.1 Column chromatography ...... 5

1.2.2 Purification tags ...... 8

1.2.3 Elastin-like polypeptides ...... 10

1.3 Challenges associated with protein modification ...... 11

1.3.1 Non-specific conjugation methods introduce heterogeneity ...... 11

1.3.2 Available methods for modifying proteins ...... 13

1.4 Sortase A catalyzes a unique reaction that can be applied to protein purification and modification ...... 20

2. Chromatography-free purification and site-specific labeling of recombinant proteins using sortase A ...... 22

2.1 Introduction ...... 22

2.2 Materials and methods ...... 23

2.2.1 Molecular biology used to construct fusion proteins ...... 23

2.2.2 Expression and purification of fusion proteins ...... 24

2.2.3 Sortase reactions ...... 26

2.2.4 Characterization of purified target proteins ...... 27

2.2.5 TNFα and TRAIL cytotoxicity assays ...... 29

2.3 Results and discussion ...... 30

2.3.1 Binary fusion reaction ...... 31

2.3.2 Ternary fusion reaction...... 39

2.3.3 Bioactivity of the purified proteins ...... 40

2.3.4 Simultaneous tag removal and site-specific conjugation of small molecules to the carboxy-termini of substrate proteins ...... 42

2.3.5 TRAIL-ELP fusions show potent in vitro and in vivo cytotoxicity towards TRAIL-sensitive cell lines ...... 48

2.4 Results and discussion ...... 54

3. A non-canonical function of sortase enables site-specific conjugation of small molecules to lysine residues in proteins ...... 55

3.1 Introduction ...... 55

3.2 Materials and Methods ...... 58

3.2.1 Protein expression and purification ...... 58

3.2.2 Protein-small molecule conjugation reactions ...... 58

3.2.3 LC-MS/MS and biotin:protein molar ratio ...... 59

3.2.4 Immunofluorescence microscopy ...... 60

3.3 Results and Discussion ...... 61

vii

3.3.1 Proof-of-concept of sortase-catalyzed isopeptide ligation using peptides ...... 61

3.3.2 Sortase-catalyzed protein polymerization through isopeptide bonds ...... 63

3.3.3 Conjugation of small molecules to proteins using sortase-catalyzed isopeptide ligation ...... 65

3.3.4 Isopeptide ligation of multiple small molecules to a single protein ...... 70

3.3.5 Analysis of the isopeptide ligation product by tandem mass spectrometry .... 78

3.3.6 Protein-small molecule conjugates retain bioactivity...... 84

3.4 Significance ...... 88

4. Generation of antibody-drug conjugates using sortase-mediated isopeptide ligation . 89

4.1 Introduction ...... 89

4.1.1 Potency requirements for ADC payloads ...... 91

4.1.2 Auristatins as ADC payloads ...... 92

4.2 Materials and Methods ...... 94

4.2.1 Antibody production by transient transfection ...... 94

4.2.2 Antibody purification by protein G chromatography ...... 94

4.2.3 Antibody-biotin conjugation using sortase ...... 95

4.2.4 Auristatin F-sortase recognition peptide coupling reaction ...... 96

4.2.5 Antibody-auristatin F conjugation using sortase ...... 97

4.2.6 In vitro testing of antibody conjugates ...... 97

4.3 Results and Discussion ...... 98

4.3.1 Procedure for cloning antibody genes from a hybridoma ...... 98

4.3.2 Antibody expression in mammalian cells using transient transfection...... 102

viii

4.3.3 Conjugation of biotin to an engineered monoclonal antibody ...... 116

4.3.4 Stability of isopeptide-linked antibody conjugates ...... 124

4.3.5 Synthesis of auristatin F linked to a sortase substrate peptide ...... 127

4.3.6 Conjugation of auristatin F to a model targeting protein ...... 131

4.3.7 Conjugation of auristatin F to an engineered anti-Her2 monoclonal antibody ...... 134

4.4 Significance ...... 138

5. Analysis of the sequence requirements for isopeptide ligation at the pilin domain ... 139

5.1 Introduction ...... 139

5.2 Materials and Methods ...... 140

5.2.1 Molecular biology ...... 140

5.2.2 Sortase reactions ...... 141

5.2.3 Electrophoresis, Western blotting, and mass spectrometry ...... 142

5.2.4 Pepspots array...... 142

5.3 Results and Discussion ...... 143

5.3.1 Amino acid sequence requirements for isopeptide ligation ...... 143

5.3.2 TRAIL is modified by isopeptide ligation at lysines in a flexible loop ...... 148

5.3.3 Other chemistries do not lead to TRAIL modification ...... 160

5.3.4 Native peptide ligation competes with isopeptide ligation if the N-terminus is accessible...... 178

5.3.5 Analysis of nucleophile requirements for N-terminal native peptide ligation by sortase A ...... 186

5.3.6 Analysis of native and isopeptide ligation in human antibodies ...... 191

5.4 Significance ...... 196

Appendix A...... 198

Appendix B ...... 200

Appendix C ...... 201

Appendix D...... 204

References ...... 207

Biography ...... 233

List of Tables

Table 1: Conversion, recovery, purity, and yield for proteins recovered from binary fusion sortase reactions...... 37

Table 2: Comparison of predicted and experimental molecular masses for TRAIL-Gly3 and TRAIL- CPT...... 46

Table 3: Predicted and experimental peptide masses for the conjugation of dabcyl to ELP-GLP1...... 70

Table 4: Primers used to clone anti-Her2 antibody 4D5 light and heavy chains from hybridoma CRL-10463 ...... 101

List of Figures

Figure 1. Process for purifying human erythropoietin produced in Pichia pastoris...... 7

Figure 2. Overview of ELP fusion protein purification...... 11

Figure 3. Conjugation of a cargo molecule containing an NHS-ester to a protein amine group (N-terminus or lysine side chain)...... 14

Figure 4. Conjugation of a cargo molecule containing a malemide to the thiol group of a cysteine side chain...... 15

Figure 5. Overview of the canonical reaction catalyzed by SrtA in S. aureus...... 21

Figure 6. ELP fusion proteins used in the binary and ternary fusion reactions...... 24

Figure 7. Thermal responsiveness of the SrtA-ELP fusion protein used in this study. .... 32

Figure 8. Scouting studies to identify conditions for binary fusion reactions...... 34

Figure 9. Conditions for reaction cleanup were selected to ensure removal of hydrophilic ELP fusion proteins...... 35

Figure 10. Protein purification by the binary fusion sortase-ELP reaction...... 36

Figure 11. HPLC analysis of target proteins purified by the binary fusion reaction...... 38

Figure 12. All-in-one protein purification by the ternary fusion reaction...... 40

Figure 13. Sortase reactions generate properly folded, bioactive TNFα...... 41

Figure 14. Protein-small molecule conjugation using the binary fusion reaction...... 44

Figure 15. Electrospray ionization mass spectrometry analysis of TRAIL-Gly3 and the TRAIL- camptothecin conjugate...... 45

Figure 16. TRAIL-camptothecin conjugates are bioactive and cytotoxic towards human breast adenocarcinoma cells in vitro...... 47

Figure 17. Different ELP fusion partners allow TRAIL to be delivered locally or systemically...... 49

xii

Figure 18. Purification conditions influence dimer formation and TRAIL cytotoxicity. .. 51

Figure 19. Intratumoral administration of coacervate-forming TRAIL-ELP in mice is well-tolerated, leads to tumor regression, and improves survival...... 53

Figure 20. Schematic of pilin assembly by sortases in Gram-positive bacteria...... 56

Figure 21. Reaction with peptide substrates demonstrates the isopeptide ligation reaction catalyzed by sortase A...... 62

Figure 22. Sortase A catalyzes isopeptide bond formation at the pilin domain lysine. .... 64

Figure 23. Sortase A catalyzes site-specific conjugation of dabcyl to GLP-ELP...... 67

Figure 24. MALDI-TOF mass spectrum of the tryptic digest of dabcyl-conjugated ELP- GLP1...... 69

Figure 25. Sortase A catalyzes site-specific conjugation of biotin to Fn3...... 74

Figure 26. Isopeptide ligation is a general in vitro function of SrtA...... 76

Figure 27. Site-specific conjugation of FITC to Fn3...... 77

Figure 28. Purification of biotinylated Fn3-PLN3-ELP using cation exchange chromatography...... 79

Figure 29. MS1 spectrum for the tryptic digest of biotinylated Fn3-PLN3-ELP...... 81

Figure 30. Tandem mass spectrometry (MS) of tryptic peptides of biotinylated Fn3- PLN3-ELP...... 82

Figure 31. Analysis of the biotin-to-protein molar ratio for biotinylated Fn3-PLN3-ELP. 84

Figure 32. Sortase A catalyzes site-specific conjugation of biotin to Fn3...... 86

Figure 33. Cellular uptake of anti-αvβ3 Fn3 proteins biotinylated by SrtA...... 87

Figure 34. Structure of auristatin F...... 93

Figure 35. Dot blot confirms that the CRL-10463 secretes antibody...... 100

xiii

Figure 36. Gel electrophoresis of PCR-amplified anti-Her2 light and heavy chain DNA...... 102

Figure 37. Antibody expression cassette...... 104

Figure 38. Timeline for transient transfection of Expi293 cells for antibody production...... 108

Figure 39. Antibody production using transient transfection...... 109

Figure 40. Predicted signal peptide cleavage for the murine anti-Her2 antibody light chain with the IL-2 signal peptide...... 112

Figure 41. Predicted signal peptide cleavage for the murine anti-Her2 antibody heavy chain with the IL-2 signal peptide...... 113

Figure 42. Predicted signal peptide cleavage for the murine anti-Her2 antibody light chain with the azurocidin signal peptide...... 114

Figure 43. Predicted signal peptide cleavage for the murine anti-Her2 antibody heavy chain with the azurocidin signal peptide...... 115

Figure 44. Sortase modifies a mouse antibody only at the pilin domain...... 117

Figure 45. Sortase-mediated isopeptide ligation can be used to modify monoclonal antibodies (mAbs)...... 119

Figure 46. Temperature dependence of the antibody modification reaction...... 120

Figure 47. Biotin-conjugated 4D5 anitbody retains antigen targeting...... 122

Figure 48. Quantification of the biotin-to-protein ratio in a monoclonal antibody modified by isopeptide ligation...... 123

Figure 49. Antibodies conjugated to small molecules using isopeptide ligation are stable in vitro...... 126

Figure 50. Reaction scheme for peptide coupling of auristatin F to the sortase recognition peptide GLPATGRAGG ...... 130

Figure 51. MALDI-TOF mass spectrum of auristatin F linked through a peptide bond to GLPATGRAGG peptide...... 131

xiv

Figure 52. Auristatin F can be conjugated to proteins using sortase-mediated isopeptide ligation...... 133

Figure 53. Schematic for the conjugation of auristatin F to the 4D5 monoclonal antibody...... 135

Figure 54. anti-Her2 conjugated to auristatin F using IPL is cytotoxic towards antigen- expressing cell lines in vitro...... 138

Figure 55. Procedure for mutagenesis by polymerase chain reaction...... 141

Figure 56. Three truncation mutants suggest that the pilin domain can be shortened. . 145

Figure 57. Analysis of the substrate sequence requirements for sortase-catalyzed isopeptide ligation...... 147

Figure 58. Structure of TRAIL monomer showing its lysine residues...... 149

Figure 59. Sortase-mediated isopeptide ligation reacts at sites other than the pilin domain in TRAIL...... 151

Figure 60. Mutagenesis confirms that the TRAIL AA” loop lysines are modified by sortase...... 153

Figure 61. TRAIL AA” loop lysines are modified by sortase while other lysines and the single cysteine in the protein are not modified...... 156

Figure 62. Lysines within secondary structural elements do not appear to be modified by isopeptide ligation...... 159

Figure 63. Proposed reaction scheme for the acylation of histidine by sortase...... 161

Figure 64. Proposed reaction scheme for the acylation of tyrosine...... 161

Figure 65. Structure of TRAIL monomer showing its histidine residues...... 163

Figure 66. Structure of TRAIL monomer showing its tyrosine residues...... 164

Figure 67. The modifications made to TRAIL by sortase are stable across a wide range of pH...... 165

Figure 68. Mutation of histidines in TRAIL does not reveal any reactive sites...... 166

Figure 69. The modifications that sortase makes to TRAIL are stable towards the strong nucleophile hydroxylamine...... 167

Figure 70. A solvent-exposed tyrosine in TRAIL is not modified by sortase...... 168

Figure 71. Proposed reaction scheme for the acylation of tryptophan...... 169

Figure 72. Structure of TRAIL monomer showing its tryptophan residues...... 170

Figure 73. Tryptophans in TRAIL are not modified by sortase...... 171

Figure 74. Structure of TRAIL monomer showing its methionine residues...... 173

Figure 75. Methionine in TRAIL appears to be modified by sortase...... 176

Figure 76. The methionine side chain is not modified by sortase...... 178

Figure 77. Cleavage of the TRAIL N-terminus does not conclusively indicate modification by sortase...... 180

Figure 78. Native peptide ligation at the N-terminus competes with isopeptide ligation within a protein...... 183

Figure 79. Native peptide ligation occurs with proteins that do not have an N-terminal glycine...... 185

Figure 80. All N-terminal amino acids are modified by sortase in a native peptide ligation reaction...... 190

Figure 81. Human antibodies are modified by sortase through native and isopeptide ligation...... 192

Figure 82. Human antibodies are modified by sortase through native and isopeptide ligation...... 195

xvi

Acknowledgements

I would like to thank my parents, Joe and Nancy Bellucci, for their constant support throughout my education. Without their hard work to provide my brothers and

I with a solid foundation in life, I would not have had any of the opportunities that have shaped who I am today.

My wife Marine Bellucci has been a constant source of motivation and a partner in the many ups and downs that I experienced during my PhD, particularly in the final year. I want to thank her for her understanding, unyielding support, and for sharing her life with me. She provides the balance in my life that makes putting in the hard work worth the effort. Thank you, Marine, for your love, patience, and compassion.

I want to thank the many friends that I made in graduate school for the memories we’ve shared. I would also like thank the past and present members of the

Chilkoti lab for making our group an engaging place to work. In particular, thank you

Dr. Mira Amiram for training me my first year in the group and Dr. Jayanta

Bhattacharryya for the friendship and advice that he provided over many cups of coffee.

Finally, thank you Dr. Ashutosh Chilkoti for giving me the opportunity to work in the group and for providing me with the flexibility to pursue projects that I found interesting. I am also grateful to my dissertation committee for their guidance.

xvii

1. Introduction

1.1 Proteins and antibodies for cancer therapy

Approximately 100 recombinant proteins and antibodies are approved for clinical use in the United States and Europe, with a 2010 market of over $100 billion annually.1 These therapeutics are used to treat myriad diseases, including but not limited to diabetes, hormone replacement therapy, antiviral therapy, and enzyme- replacement therapy. In cancer treatment, the majority of directly cytotoxic therapeutic proteins are monoclonal antibodies, though several non-cytotoxic proteins such as granulocyte-macrophage colony-stimulating factor (GM-CSF) and erythropoietin (EPO) are used to promote the expansion of white and red blood cells, respectively, both of which are commonly depleted by chemotherapy.

Antibodies are tetrameric proteins composed of two light chains (~25 kDa each) and two heavy chains (~50 kDa each). Naturally produced by B cells in humans, antibodies are one arm of the adaptive immune system that defends the body against

“non-self” antigens—bacteria, viruses, and mutated, cancerous cells that express proteins considered non-self, such as neonatal proteins that are normally expressed only during fetal development. B cells develop antibodies with extremely high affinity (KD <

10-10 M) for specific epitopes on the foreign antigen through a process of highly competitive, clonal mutation of their genomic DNA.2,3 By fusing antibody-producing B cells extracted from antigen-immunized mice with myeloma cells, an immortalized

fusion cell can be formed that produces the monoclonal antibody.4 Alternatively, the

DNA sequences of the antibody heavy and light chains can be cloned into an immortalized, heterologous cell line such as Chinese Hamster Ovary cells or used in cell-free expression systems to produce the recombinant antibody.5,6

Antibodies have been used for cancer treatment since the FDA’s approval of rituximab in 1997 for non-Hodgkin’s lymphoma. Antibodies can achieve highly selective, tight binding to their target, which has been exploited for the treatment of cancer in several ways.7 First, antibodies have been designed to block signaling molecules—such as vascular endothelial growth factor, which promotes the formation of new blood vessels—that are secreted by cancer cells to promote tumor growth.8,9,10

Second, antibodies have been designed that bind to receptors on the cancer cell that normally promote proliferation, so that when these receptors are bound by an antibody they are switched off, arresting cell division.11,12 Third, antibodies bound to surface antigens on cancer cells can trigger direct cell killing by the innate immune system, as high levels of surface-bound antibody can stimulate both complement-mediated cell lysis and antibody-dependent cell-mediated cytotoxicity, in which the cancer cells are eliminated by natural killer cells, macrophages, neutrophils, and eosinophils.13,14,15

Antibody-drug conjugates are the fourth, newest, and potentially most powerful class of antibody therapeutics developed for cancer therapy.

1.1.1 Antibody-drug conjugates

Drugs that are too toxic to administer on their own can be attached to antibodies that have high affinity for antigens or receptors on tumor cells and high specificity —a concomitantly low affinity for healthy cells— so that they localize specifically to cancerous cells, thereby exposing diseased tissue to lethal concentrations of the toxin while sparing healthy tissue. This approach utilizes the complementary advantages of the two ADC components: the antibody—which may not be effective in cell killing on its own—provides a means to harness the extreme toxicity of the small molecule by directing it only to diseased cells. However, available methods to produce these conjugates attach a variable number of drugs to a fraction of available positions on the antibody.16,17 Both the number of attached drugs and their location on the antibody dramatically affect the ADCs affinity, specificity, and stability, which can dilute their effectiveness as highly potent and selective tumor killing molecules.18 This motivates the need for site-specific methods of drug attachment that can easily produce ADCs with defined, tunable, and consistent properties.

Two ADCs are FDA approved: anti-CD30 coupled to monomethyl auristatin E for the treatment of classical Hodgkin’s lymphoma, and anti-Her2 linked to maytansine for the treatment of Her2+ breast cancer.19,20 Although these ADCs have achieved clinical success and extended the survival of patients, their development has been fraught with challenges, many of which derive from the complicated nature of the conjugation

methods used in their manufacturing. The primary limitation is the lack of site- specificity in drug attachment. In each batch of the approved ADCs, only the average number of attached drugs can be controlled by careful optimization and quality control of the manufacturing processes. However, small variations in the number of attached drugs or where the drugs are attached can result in dramatic changes in pharmacokinetics and efficacy.21,22 As a result, early preclinical studies required significant optimization of the production process to obtain material that was consistent and scalable through clinical trials and into bulk manufacturing.23,24,25 New methods to make site-specific ADCs with tightly controlled stoichiometry are therefore highly desirable.

1.1.2 Protein-polymer conjugates

Protein-polymer conjugates are another important class of therapeutics that is related to antibody drug conjugates in that both suffer from heterogeneity that is inherent to available processes for attaching polymers or drugs to proteins.26 Many non- antibody therapeutic proteins have a short circulating half-life that results from renal filtration, proteolytic digestion, and receptor-mediated lysosomal trafficking.27 A variety of modifications can be made to proteins in order to improve pharmacokinetics, though the most common involves attaching the synthetic polymer polyethylene glycol (PEG) to the polypeptide chain. PEGylation improves protein pharmacokinetics by two means.28

In many cases, the unmodified protein is small—its molecular weight is less than the ~60 kDa cutoff for kidney filtration—and as a result it is cleared from the blood while the protein-polymer conjugate is retained because of its bulkiness. Second, PEGylation can shield a protein from degradation by proteases or through low affinity interactions that mediate protein degradation in the liver.29 Monoclonal antibodies do not require

PEGylation due to their already large molecular weight and other biochemical factors

(i.e. FcRn recycling that captures endocytosed antibodies and returns them to the circulation) that give them a long circulating half-life. Nonetheless, some cancer therapeutics are marketed as PEGylated conjugates, including both erythropoietin (EPO) and granulocyte macrophage colony-stimulating factor (GM-CSF or filgrastim).30,31,32

1.2 Challenges associated with therapeutic protein production

1.2.1 Column chromatography

Recombinant proteins are important biomedical research tools and an established class of therapeutics that has fueled the rapid expansion of the biotechnology industry. However, the production of therapeutic proteins is limited by complicated, expensive purification processes that limit yield.

Recombinant proteins are produced using bacterial, insect, yeast, or mammalian cells and must be isolated from a complex mixture of host cell proteins.33,34 Production of a recombinant protein in a heterologous system no longer limits productivity, as

proteins are routinely produced in gram per liter quantities in modern expression systems; productivity is now in some cases even exceeding ten grams per liter.35,36

Purification is achieved over a series of many steps, and column chromatography is the primary mode of separation. It relies on a variety of functionalized polymer resins that allow selective separation of a protein of interest based on its physical properties. For instance, ion exchange chromatography uses conditions where (ideally) all proteins except the protein of interest are charged and bind to the resin, allowing the isolation of the target protein in the fraction of buffer that flows through the column. Other applications include affinity-based systems that have resins functionalized with small molecules or proteins that specifically capture the protein of interest from the mixture.

Figure 1. Process for purifying human erythropoietin produced in Pichia pastoris. Schematic showing the sequence of unit operations used to purify erythropoietin from Pichia pastoris, which includes 4 chromatography steps (boxed in red) and a PEGylation step (boxed in green).37 These steps are time consuming, expensive, and limit the yield of the protein due to chromatography resin cost and yield loss in PEGylation.

Regardless of type, all chromatography operations require customization for a given protein of interest, use expensive resins that have limited potential for re-use, and require significant time to optimize and execute. In addition, several different chromatography steps are typically required to purify a protein, since a subset of host cell proteins will invariably have one or more physical properties that are similar to the protein of interest. Because different modes of chromatography require different buffers, diafiltration steps required between each chromatography operation to exchange the product into an appropriate buffer for the following step, which adds additional cost,

time, and complexity to the overall purification process. This illustrated by Figure 1, which shows the process used to purify clinical-grade erythropoietin from a Pichia pastoris culture supernatant. The complexity of this process—driven by the 4 chromatography unit operations—as well as the associated column packing, testing, and cleaning, presents a major technical hurdle to successfully executing scaled-up manufacturing runs. In addition, the chromatography resin itself presents a financial bottleneck that limits the overall yield of the process. For example, protein A resin— used as the primary capture step for all antibodies—costs between $5,000 and $15,000 per kilogram of antibody ($100,000 to $300,000 per batch), depending on the antibody titer in the bioreactor, the number of cycles per batch (more cycles require longer process times), and the number of times the resin can be re-used (the costs listed above assume multiple batches for 120 total cycles).38,39 Importantly, re-use in multiple batches requires a significant investment in validation studies to prove that performance is consistent over time.

1.2.2 Purification tags

Purification tags are widely employed to ease the burden of column chromatography by providing a generalized set of conditions that can be applied to diverse proteins as long as they are engineered to contain the purification tag. For instance, hexahistidine tags chelate Ni2+ and Co2+ ions through the imidazole groups of

the amino acid side chains in the tag.40,41,42 Polymer resin with these metal ions embedded in it will bind hexahistidine-tagged proteins that can then be eluted by adding excess imidazole.

Another widely-used purification tag is Maltose-Binding Protein (MBP) from E. coli, which is also used to enhance the solubility of its fusion partner.43 MBP is normally involved in the binding and uptake of polysaccharides into E. coli, and as a result, MBP provides an efficient purification tag through its affinity for amylose, which can easily be incorporated into column chromatography resins.44,45,46 Binding to amylose is highly selective for the MBP fusion protein; most other host cell components will flow through the amylose column or bind only weakly and non-specifically, allowing highly-selective enrichment of the MBP fusion protein in a single capture step.

While these processes are convenient, in many cases the tag will interfere with the fused protein’s bioactivity and must be removed. This is of particular concern for

MBP fusion proteins, since the MBP component adds over 40 kDa to the protein molecular weight and in many cases (e.g. most cytokines or chemokines) this will be larger than the size of the bioactive protein itself. Purification tags are typically removed by adding an exogenous protease such as tobacco etch virus (TEV) protease, thrombin, or factor Xa, that cut specific recognition sequences that are engineered between the bioactive protein and the tag. However, after this reaction the advantage of the tag is lost, since the target protein must be purified from the mixture of the protease, the tag,

and undigested fusion protein. This requires a chromatography step that must be customized for the given protein of interest.

1.2.3 Elastin-like polypeptides

Elastin-like polypeptides are a type of purification tag that has been used extensively by the Chilkoti group for a variety of applications including peptide delivery47,48, small molecule drug delivery49,50,51,52,53, radionuclide delivery54,55, and protein purification56, among others. ELPs are a polymer composed of repeats of the pentapeptide subunit VPGXG, where X is any amino acid except proline. ELPs allow facile purification of genetically-fused proteins by a process termed inverse transition cycling.57 This procedure utilizes that fact that ELP fusion proteins are thermally- responsive and reversibly form insoluble aggregates when heated above a characteristic transition temperature. The transition temperature can be easily tuned to the application of interest, as it is a function of the identity of “X” in the VPGXG repeat, the number of repeats in the polymer, and environmental factors such as the concentration of the ELP and cosolutes. Importantly, reversible phase transition behavior is retained by ELPs when they are fused to a protein that normally does not display this behavior.58 ELP fusion proteins can therefore be purified by centrifugation rather than chromatography

(Figure 2). Nonetheless, ELPs suffer from the same limitation as any other purification

tag in that tag removal generates a mixture of proteins that requires chromatography to isolate the target protein.

Figure 2. Overview of ELP fusion protein purification. (a) Schematic of the inverse transition cycling procedure for ELP fusion protein purification. (b) Example of fractions from an inverse transition cycling purification of a tumor necrosis factor-related apoptosis-inducing ligand (TRAIL)-ELP fusion protein. The protocol is described in detail in the text. “Hot” centrifugation corresponds to room temperature with salt and selectively recovers insoluble ELP fusion in the pellet. “Cold” centrifugation corresponds to 4°C where the ELP fusion is soluble and is collected in the supernatant. CS1 = first cold spin. HS1 = first hot spin. CS2 = second cold spin.

1.3 Challenges associated with protein modification

1.3.1 Non-specific conjugation methods introduce heterogeneity

The effectiveness of both PEGylation and conjugation of drugs to antibodies depends critically on the site of attachment in the protein. For ADCs in particular, controlling the location of attachment is important because it has a dramatic impact on

the stability and aggregation of the ADC during storage as well as its circulation half-life upon injection59,60, cellular internalization of the drug61, and the rate of intracellular drug release62, all of which have a dramatic impact on both the efficacy and the unwanted side effects experienced by patients.63,64 In conventional methods for ADC synthesis, each antibody is attached to a stochastic number of drugs that are distributed across multiple reactive sites—typically cysteine or lysine side chains—on the protein scaffold.65,66,67

Thus, each batch of a conventional ADC is a heterogeneous mixture of conjugates; for example, if 4 drug molecules are non-specifically attached via amine-reactive chemistry to an antibody that contains 85 lysine residues, over 50 million possible variants can be generated. There are 1,680 combinations when 4 drug molecules are attached through cysteine-based chemistry, and although this is less than the number of possibilities with lysine-based chemistry it is still a large number of potential variants.

This heterogeneity is a problem because each variant has a different in vivo performance that is impossible to predict or evaluate a priori. This difference has been shown by a variety of studies with antibody-drug conjugates because aggregation, target binding, and interaction with other receptors is perturbed by stochastic attachment of a variable number of drugs across different sites.68,69,70 For example, modification of the antibody constant domain receptor-binding regions can lead to diminished antibody- dependent cell-mediated cytotoxicity (ADCC), which is responsible a significant amount of anti-tumor activity for certain mAbs.71

Drug loading is also a concern with methods for drug or polymer conjugation that are not site-specific. For ADCs, loading more drugs is desirable because efficacy generally improves with increasing drug loading as long as the stability of the antibody is not compromised.72 Stability can be compromised when hydrophobic drugs are attached to ADCs, though the use of hydrophilic linkers with these drugs has been shown to dramatically reduce aggregation and improve in vivo efficacy.73

Given the effects of conjugation site and conjugation number on bioavailability, new conjugation methods that tightly control the site and stoichiometry of attachment are highly desirable. A brief review of amino- and thiol-reactive reagents is provided below, along with a description of several available methods that control the conjugation site using enzymatic reactions and modification of unnatural amino acid side chains.

1.3.2 Available methods for modifying proteins

Amino-reactive crosslinking reagents

The ε-amino group of lysine is a commonly-used conjugation site because of its presence in many proteins and the availability of several reagents that provide efficient conjugation at primary amino groups. N-hydroxysuccinimide (NHS) esters are commonly used for lysine modification because they achieve conjugation under mild conditions (pH 7-9) and form a stable amide bond in the product.

Figure 3. Conjugation of a cargo molecule containing an NHS-ester to a protein amine group (N-terminus or lysine side chain). Conjugation is achieved under mild conditions (pH 7-9). Modification of the protein N- terminus is favored at lower pH while lysine modification is favored at higher pH, though there is significant reaction at both sites because modification depends on the solvent accessibility and the pKa of an individual amine group (which is affected by the local charge and hydrophobicity of the protein).

NHS esters will react with any accessible lysine ε-amino group as well as the α- amino group at the terminus of the protein. Because of the prevalence of lysine in proteins along with the reactivity of the amino-terminus, multiple sites are typically available for modification and the reaction will generate a mixture of products with different numbers of conjugated molecules distributed across all potential attachment sites.

Thiol-reactive crosslinking reagents

The low abundance of cysteine in proteins relative to lysine makes it a more specific target for conjugation. The most commonly-used crosslinking agent for cysteine modification is the malemide group, which reacts with a thiol on the protein to form a thioether linkage. The thioether linker is stable in solvent and can be further stabilized by succinimide ring hydrolysis, which can be improved by the use of certain linkers that lower the activation energy of the hydrolysis reaction.74 However, this stabilizing step is 14

typically not accomplished for the vast majority of commercially-available malemide- based conjugation reagents, limiting their effectiveness. In serum, these conjugates will undergo a slow but significant malemide exchange reaction with albumin, which contains a reactive free thiol group.75 In the case of antibody-drug conjugates that utilize extremely potent cytotoxic drugs, this non-specifically exposes cells to long-lived albumin-drug conjugates and can particularly affect the liver through further malemide exchange with free cysteine or glutathione in this organ. Though cysteine-based conjugation is more site-specific than lysine-based modifications, the off-target toxicity associated with malemide exchange is a significant drawback that requires careful implementation of this method.

Figure 4. Conjugation of a cargo molecule containing a malemide to the thiol group of a cysteine side chain. Conjugation of the malemide component to the thiol forms a thioether linkage between the protein and the cargo. Unless the succinimide ring undergoes subsequent hydrolysis, the thiol-malemide reaction can be reversed, which releases the cargo from the protein or transfers it to a thiol on albumin (through a disulfide bond).

In addition to malemide exchange, there are other problems associated with cysteine-directed conjugation. Cysteine residues are often present in proteins to stabilize tertiary or quaternary structure through disulfide bonds. Additionally, the presence of a reduced thiol in a protein suggests that this residue contributes to biological function, for instance, in tumor necrosis factor-related apoptosis-inducing ligand (TRAIL), the single free thiol is involved in chelating a zinc ion and stabilizing the homotrimeric quaternary structure of the protein76. Disulfide bonds can be reduced and the resulting thiols can be modified, but this has the potential to destabilize structure. For example, disulfides in an antibody hinge region are accessible and can be easily reduced, but complete modification of hinge cysteines will decouple the covalent bonds in the heavy chain dimer, and the quaternary, bivalent structure of the antibody will only be maintained by comparatively weak van der Waals, hydrogen bonding, and ionic interactions.

Site-directed mutagenesis can be used to incorporate a cysteine into a specific solvent-accessible site in a protein.77 In this case, conjugation via thiol-reactive crosslinkers is site-specific if no other cysteines are present in the protein, and can be selective if there are other cysteines in the protein and the mutagenized residue is positioned in a more reactive site, though this requires a significant protein engineering effort to ensure that the added cysteine does not destabilize structure or promote aggregation.78,79 This strategy has several disadvantages. Introducing a new cysteine by

mutagenesis or decoupling disulfide bonds can lead to the formation of mispaired disulfides or disulfide shuffling during folding and storage if other cysteines are present in the protein.80,81 Additionally, reducing agents are required during purification to mitigate intermolecular disulfide bond formation. This presents a problem for proteins with existing intramolecular disulfide bonds like antibodies, since reducing agents will break these bonds if they are solvent-accessible, making the resulting thiols available for additional conjugations.

Unnatural amino acids

A powerful strategy to generate site-specific protein-small molecule conjugates involves the incorporation of unnatural amino acids into the protein during expression.

For instance, trastuzumab has been conjugated to auristatin, a tubulin inhibitor, using this method.82 By introducing the amber stop codon into the heavy chain gene and producing the protein in an expression host bearing an orthogonal amber suppressor tRNA and aminoacyl-tRNA synthetase, a heavy chain variant was produced that had para-acetylphenylalanine at a single site. The replacement of the phenylalanine hydroxyl group with a keto group enabled site-specific conjugation of alkoyxamine-coupled auristatin, producing an oxime linkage between the drug and the heavy chain.

Despite the high specificity of this reaction, it suffers from several limitations.

First, though multiple unnatural amino acids have been incorporated into a single model protein, the payload has to date not been increased beyond 1 molecule per protein in a

manner that is efficient enough to be scaled up for therapeutic use, except in an elastin- like polypeptide that is not bioactive.83,84 More importantly, unnatural amino acids have not been widely applied for large-scale protein production. Overall expression yields are generally less than those for proteins containing only natural amino acids. For instance, para-acetylphenylalanine has been introduced into several mAbs in stirred-tank bioreactors at 1 g/L, though a maximum reactor volume of 5L was used, and it remains to be determined whether these processes can be scaled to match “routine” antibody production that generates several grams/liter of protein in thousands of liters of culture broth.85 Nonetheless, unnatural amino acids are a promising and rapidly-developing route to achieving site-specific protein-small molecule conjugates.

Formylglycine-generating enzyme

Several enzymes are available that attach compounds to specific recognition sequences in a protein substrate.86,87,88 For example, formylglycine-generating enzyme

(FGE) recognizes the “aldehyde tag” CXPXR and converts the cysteine side chain to an aldehyde group, an electrophile that can then react with compounds functionalized with aminooxy or hydrazide groups.89 This reaction is site-specific because it is bioorthogonal, meaning that no other protein or carbohydrate components will react to produce unwanted side products. Aldehyde formation can be accomplished as a posttranslational modification in either prokaryotic or eukaryotic cells by co-overexpressing the formylglycine-generating enzyme with the protein of interest.90,91,92 However,

introduction of a non-native cysteine into the protein primary structure can result in disulfide-linked dimers or misfolded monomers through improper intramolecular disulfide formation prior to aldehyde formation. Moreover, the efficiency of aldehyde formation has been shown to be dramatically impacted by the site of the aldehyde tag in the protein, with reported modification yields ranging from 60-100%.93 Thiols that are not converted to formylglycine have the same disadvantages as cysteine-based conjugation methods. Given these potential complications, the conjugation site must be carefully engineered to preserve protein structure and achieve properly-folded, high- yield product.

Bacterial transglutaminase and phosphopantetheinyl transferase (PPTase) have been used extensively to modify glutamine and serine, respectively, within specific recognition domains.94,95,96,97 In contrast to FGE, both transglutaminase and PPTase can directly install modified drugs onto antibodies. Transglutaminase will attach drugs that contain a lysine or cadaverine side chain to a LLQG recognition sequence in a flexible region of the antibody; a study has shown that when this sequence was engineered into

90 different sites in an anti-EGFR antibody, 12 of the positions demonstrated near quantitative modification and low levels of aggregation.98 Similarly, PPTase will attach a drug that is coupled to coenzyme A to recognition sequences containing 11 or 12 amino acids (DSLEFIASKLA, GDSLSWLLRLLN, and GDSLDMLEWSLM, where the underlined serines are modified). The PPTase modification sites have been successfully

engineered into 52 unique positions in human anti-Her2 (trastuzumab) with drug conjugation yields >90%, which represented about half of the total positional variants tested.99 Together, transglutaminase and PPTase demonstrate the effectiveness and site- specificity of enzyme-based approaches to make antibody-drug and other protein-small molecule conjugates.

1.4 Sortase A catalyzes a unique reaction that can be applied to protein purification and modification

Sortase A is an enzyme normally situated in the peptidoglycan of the Gram- positive bacterium Staphylococcus aureus. It catalyzes a transpeptidation reaction that the bacterium uses to attach secreted proteins bearing the enzyme recognition sequence to its exterior. A variety of sortase substrates have been identified, but the most notable of these is protein A (SpA), which binds the Fc region of human antibodies, orienting them opposite the direction normally required for phagocytosis and opsonization, and thereby shielding the bacteria from clearance by the host immune system during infection.100,101

SrtA recognizes the primary sequence LPXTG (where X is any amino acid) in a protein and cleaves the peptide bond between threonine and glycine, forming a stable intermediate that joins the catalytic thiol to the carboxyl group of threonine in a thioester bond. In S. aureus, this intermediate undergoes nucleophilic attack by the α-amino group of an oligoglycine branch in the peptidoglycan, generating a native peptide bond that 20

anchors the substrate protein to the cell wall (Figure 5).102 This reaction between LPXTG and (G)n (where n>1) has been used to attach a diverse array of compounds (such as triglycine-modified lipids, small molecules, peptides and proteins) to proteins by linking them to an oligoglycine or to the LPXTG peptide, making them suitable nucleophiles or

SrtA substrates, respectively.103,104,105,106,107

Figure 5. Overview of the canonical reaction catalyzed by SrtA in S. aureus. (a) As proteins bearing the LPXTG sortase recognition site are secreted through the cell membrane, SrtA breaks the peptide bond between the threonine and glycine residues and forms an enzyme-substrate intermediate through a thioester bond between the threonine residue and the catalytic cysteine. (b) The acyl-enzyme intermediate is stable but is resolved after nucleophilic attack by the α-amino group of a pentaglycine branch of the cell wall. (c) After the transpeptidation reaction is complete, the catalytic thiol of SrtA is regenerated and the substrate protein is anchored to the peptidoglycan through a native peptide bond.

We hypothesized that the transpeptidation reaction catalyzed by sortase A could be used both to improve both protein purification and to enable site-specific modification of the purified proteins. Using synthetic triglycine peptide as the

nucleophile in the reaction, sortase would enable the release of a purified target protein from its fusion tag, mimicking a proteolytic cleavage catalyzed by commonly-used enzymes such as thrombin, factor Xa, or tobacco etch virus protease.

We designed a strategy that offers two advantages over existing purification strategies. First, sortase was linked to the same purification tag as the target protein, which dramatically simplifies purification of the reaction product because the properties of the tag could be used to remove all undesired components after the reaction. The primary advantage of using sortase versus a protease is that sortase is a transpeptidase, and this activity of the enzyme can be used to both remove the purification tag and label the target protein at one of its termini. This reaction is extremely flexible and can accommodate a variety of payloads because sortase has been known to accept diverse molecules as long as they are linked to either an oligoglycine or an LPXTG peptide.

2. Chromatography-free purification and site-specific labeling of recombinant proteins using sortase A

2.1 Introduction

We have developed a general strategy that utilizes a fusion protein of an ELP and the transpeptidase sortase A to enzymatically release a target protein from an ELP fusion tag and allow it to be recovered at >95% purity by a single 15-minute centrifugation step. Purification of the reaction is enabled by the fact that all undesired

components remain bound to thermally-responsive ELP tags, allowing selective enrichment of bioactive, tag-free target protein. This system was successfully used to purify a variety of proteins including the model proteins thioredoxin and green fluorescent protein (GFP), as well as the therapeutic proteins tumor necrosis factor α

(TNF α) and tumor necrosis factor-related apoptosis-inducing ligand (TRAIL).

Additionally, the reaction was easily extended to allow conjugation of a small molecule to the target protein carboxy-terminus within the tag removal reaction, enabling the production of homogeneous, site-specific protein-small molecule conjugates.

2.2 Materials and methods

2.2.1 Molecular biology used to construct fusion proteins

The full-length SrtA gene was cloned from S. aureus (a gift from the McCafferty lab at Duke) and the amino-terminal 59 amino acids were removed by polymerase chain reaction. DNA coding for the ELPs (VPGVG)240, and (VPGXG)90, where X represents A,

G, and V in a molar ratio of 2:3:5 were constructed previously in our lab.

The genes for thioredoxin (TRX) and green fluorescent protein (GFP) were available from previous studies. DNA coding for soluble murine tumor necrosis factor α

(TNFα, amino acids 80-235) and soluble human tumor necrosis factor-related apoptosis- inducing ligand (TRAIL, amino acids 114-281) were designed with E. coli codon usage and purchased from Life Technologies (Carlsbad, CA).

For SrtA-ELP, pET24 (Merck KGaA, Darmstadt, Germany) was modified with synthetic oligonucleotides from Integrated DNA Technologies (Coralville, IA) to allow in-frame insertion of SrtA at the 5’ end of ELP (VPGVG)240 with a short intervening linker. Target protein-ELP fusions were generated in a modified pET25 vector by in- frame addition of target proteins to the 5’ end of ELP (VPGXG)90 [X=A:G:V=2:3:5

(mol/mol/mol)] with a linker coding for the LPETG enzyme recognition site. For ternary fusion constructs, the SrtA-ELP was transferred to pET25 (Merck KGaA). Target proteins were inserted in-frame at the 3’ end of ELP (VPGVG)240 with a linker containing the

LPGAG enzyme recognition site. The design of these fusion proteins is illustrated in

Figure 6.

Figure 6. ELP fusion proteins used in the binary and ternary fusion reactions. (a) Sortase A fused to ELP. (b) Ternary fusion proteins for GFP and TNFα. (c) Binary fusion proteins for TRX, GFP, TNFα, and TRAIL.

2.2.2 Expression and purification of fusion proteins

Expression vectors were transformed into E. coli strain BL21 (DE3), purchased as chemically-competent cells from EdgeBio (Gaithersburg, MD). Clones were screened by

colony PCR, sequenced, and subsequently maintained in 3.5% DMSO at -80°C. These stocks were used to inoculate a starter culture that was grown overnight at 37°C with orbital shaking at 250 rpm. All cultures were grown in Terrific Broth containing 45

µg/ml kanamycin or 100 µg/ml ampicillin for pET24 and pET25 constructs, respectively.

The starters were centrifuged, resuspended in fresh media, and 2% inoculums were added to 4L unbaffled Erlenmeyer shake flasks containing 1L of media with antibiotic.

Cultures were grown for 4-5 hours with 200 rpm orbital shaking at 25°C and protein expression was induced by the addition of isopropyl β-D-1-thiogalactopyranoside

(IPTG) to 0.5 mM final concentration. The culture temperature setpoint was changed to

16°C and induction was allowed to proceed overnight.

Cultures were centrifuged and resuspended in water at 1% of the original culture volume. Milli-Q water (EMD Millipore, Billerica, Massachusetts) was used throughout the purification of target protein-ELPs. Diothiothreitol was added to this buffer to a concentration of 5-10 mM in the purification of SrtA-ELP. 5 mM Tris-Cl, 5 mM ethylene glycol tetraacetic acid (EGTA), pH 7.5 was used throughout ternary fusion purifications.

Harvested cells were disrupted by sonication on ice using a Misonix Sonicator

3000 (Misonix, Inc., Farmingdale, NY). Polyethylenimine 10% (v/v) was added to the cell lysate to approximately 1% (v/v). Precipitated nucleic acids and insoluble host cell proteins were removed by centrifugation at 4°C. The supernatant was removed and the

ELP phase transition was triggered by addition of ammonium sulfate to 0.2-0.3M. ELP

fusion protein was pelleted by centrifugation, then resuspended in water or buffer.

Centrifugation at 4°C removed any remaining insoluble material and retained the soluble ELP-fusion in the supernatant. The cycle of centrifugation with salt followed by cold centrifugation of the resuspended pellet was repeated 4-5 times to achieve desired protein purity as assessed by SDS-PAGE. Phase transitions for protein-ELP fusions were characterized using a Cary 300 Bio UV-Vis spectrophotometer (Agilent Technologies,

Santa Clara, CA).

2.2.3 Sortase reactions

SrtA-ELP and target protein-ELP were combined in an enymze:substrate molar ratio of approximately 1:4. Protein concentrations were determined by the Beer-Lambert

Law using calculated extinction coefficients and the absorbance at 280 nm measured using a Nanodrop ND-1000 spectrophotometer (Thermo Scientific, Wilmington, DE).

Concentrated reaction buffer was added to a final concentration of 50 mM Tris-HCl, 150 mM NaCl, 10 mM CaCl2, pH 7.5. Synthetic triglycine peptide from Alfa Aesar (Ward

Hill, MA) was added to 10-20 molar excess over the target protein-ELP fusion and the reaction was allowed to proceed for 18-24 hours at 20°C. Products were analyzed by

SDS-PAGE.

For ternary fusion reactions, concentrated reaction buffer was added to 1x concentration and triglycine was added to 10-20 molar excess over the ternary fusion protein. The reaction was allowed to proceed 18-24 hours at 20°C.

Cleaved target proteins in both reaction designs were purified by centrifugation at 16.1 rcf in a fixed-angle benchtop centrifuge with temperature controlled at 40°C for

15 minutes. Prior to centrifugation, 1M sodium chloride was added to reactions where the target protein was originally fused to ELP (VPGXG)90 [X=A:G:V=2:3:5 (mol/mol/mol)] to ensure the transition temperatures of the free ELP and unreacted target protein-ELP fusions were lowered to below 40°C.

Labeling reactions were identical to the reaction protocol for target protein purification using SrtA-ELP, except that triglycine was replaced with camptothecin modified by the Duke University Small Molecule Synthesis Facility (Durham, NC) to contain triglycine covalently linked to the E-ring hydroxyl group.

2.2.4 Characterization of purified target proteins

SDS-PAGE was performed using 4-20% and 8-16% Tris-HCl gels purchased from

Bio-Rad Laboratories (Hercules, CA) with 25 mM Tris, 192 mM glycine, 0.1% SDS running buffer and Laemmli sample buffer (Bio-Rad). Native PAGE was run using 4-

20% Tris-glycine (TGX) gels from Bio-Rad with 25 mM Tris, 192 mM glycine running

buffer and native sample buffer (Bio-Rad). Gels were stained with copper chloride and/or coomassie blue according to standard protocols.

Purified TRX, GFP, TNFα, and TRAIL were analyzed by HPLC (Shimadzu North

America, Columbia, Maryland) using a Shodex KB-803 column (Showa Denko America,

New York, New York) with a Milli-Q water mobile phase and isocratic elution. TRAIL,

TRAIL-CPT, and Gly3-CPT were analyzed using an Aeris Widepore 3.6 µm XB-C18 column with a 70/30 (v/v) Milli-Q water/acetonitrile mobile phase and isocratic elution.

Proteins were detected by absorbance at 280 nm and camptothecin by absorbance at 365 nm using a SPD-10A UV-Vis detector (Shimadzu).

MALDI-TOF mass spectrometry was performed using an Applied Biosystems

Voyager-DE Pro system fitted with a nitrogen laser. All results were obtained using a sinapinic acid matrix in a 10:3 (v/v) ratio with the analyte.

ESI-MS was performed using an Agilent LC/MSD trap. Samples were prepared at 0.1 µM using Milli-Q water containing 0.3% formic acid and directly injected by infusion pump.

Static light scattering was performed using an ALV CGS-3 goniometer (ALV,

Langen, Germany). Samples were filtered through 20 nm filters with no concentration loss as measured by absorption at 280 nm. All testing was conducted at room temperature. Static measurements were made in triplicate for 15 seconds each from 30° to 150° in 5° increments.

2.2.5 TNFα and TRAIL cytotoxicity assays

L929 cells were obtained from the Duke University Health System Cell Culture

Facility (Durham, North Carolina) and maintained in RPMI 1640 supplemented with

10% fetal bovine serum, 1 mM sodium pyruvate, 5 mM HEPES, and 100 units penicillin/100 µg streptomycin per mL (all from Life Technologies, Grand Island, New

York). Cells were inoculated in 96-well flat-bottom culture dishes at a density of 50,000 cells/mL and treated with a range of concentrations of TNFα purified by cleavage of

TNFα-ELP, from the ternary fusion, or purchased from eBioscience (San Diego, CA).

Negative control groups were included that were untreated. Cultures were grown at

37°C under 5% CO2 for 36 hours.

MDA-MB-231 cells were obtained from the Duke University Health System Cell

Culture Facility and maintained in DMEM:F12 (1:1) supplemented with 10% fetal bovine serum and 100 units penicillin/100 µg streptomycin per mL (Life Technologies). Cells were inoculated into 96-well flat-bottom culture dishes at a density of 100,000 cells/mL and incubated for approximately 18 hours. The growth media was aspirated and replenished with serum-free DMEM:F12 (1:1) and TRAIL, Gly3-CPT, TRAIL-CPT conjugate, or non-conjugated CPT and TRAIL (0.8:1 mol/mol) was added at a range of concentrations. Cell viability relative to an untreated control group was assessed after 24 hour incubation with treatment.

Growth inhibition versus the untreated controls was assessed using a CellTiter

96 One Solution MTS/PMS viability assay purchased from Promega (Madison, WI) and used per the manufacturer’s instructions. All treatment concentrations were repeated in triplicate, and the means for each group were normalized to that of the untreated group.

All samples were further corrected for background absorbance by measuring the absorbance of the MTS/PMS reagent added to cell-free media. Dose-response data was fit to a 4-parameter logistic model using Wolfram Mathematica version 8.0 (Wolfram

Research, Inc., Champaign, IL).

For caspase activation assays, MDA-MB-231 cells were seeded in 96-well flat- bottom culture plates at 100,000 cells/mL and allowed to grow for 18 hours. The media was changed and TRAIL purified from cleavage of TRAIL-ELP was added at a range of concentrations. Caspase 3 and 7 activation was quantified after 5 hours of treatment using the Caspase-Glo 3/7 assay (Promega) per the manufacturer’s instructions. All treatment groups were normalized to a media-only control and caspase activation in

TRAIL-treated groups was reported relative to the untreated control.

2.3 Results and discussion

Combining the transpeptidase activity of SrtA from S. aureus with ELP purification tags, we developed a strategy to purify significant quantities of high-purity, bioactive recombinant proteins without column chromatography. The protocol is

scalable, requires minimal equipment, and can be applied to a variety of target proteins.

The ease with which our protocol is executed and the flexibility to perform an optional, one-step site-specific conjugation reaction that homogeneously labels the product provide a valuable, “three-in-one” set of tools for the production of therapeutic proteins and protein-small molecule conjugates.

2.3.1 Binary fusion reaction

We hypothesized that the transpeptidase activity of a SrtA-ELP fusion could cleave ELP fusion proteins, allowing recovery of the released target proteins at high purities by another round of phase transition-mediated purification without chromatography. We constructed a gene-level fusion of SrtA lacking its 59 amino- terminal residues – a truncation that has been shown to have no impact on its transpeptidase activity – to an ELP with the sequence (VPGVG)240.108 This fusion protein exhibited a reversible phase transition between 20-25°C, depending on its concentration

(Figure 7). A panel of target proteins including thioredoxin (TRX), green fluorescent protein (GFP), soluble murine tumor necrosis factor α (TNFα, amino acids 80-235), and soluble human tumor necrosis factor-related apoptosis-inducing ligand (TRAIL, amino acids 114-281) were fused at the gene level to a different ELP, (VPGXG)90, where X represents alanine (A), glycine (G), and valine (V) in a molar ratio of 2:3:5. These target proteins range from well behaved, single sub-unit proteins (TRX and GFP) to more

difficult to express, pharmaceutically relevant proteins that form stable quaternary structures (TNFα and TRAIL). The linker between all target proteins and their ELP fusion partners contained the LPETG sequence, which was previously demonstrated to be the optimal recognition sequence for SrtA.

Figure 7. Thermal responsiveness of the SrtA-ELP fusion protein used in this study. (a) Absorbance at 350 nm was measured for 100, 75, 50, and 25 µM SrtA-ELP concentrations as the temperature was ramped from 15°C-35°C at 1 ºC/min in sortase reaction buffer (50 mM Tris-HCl, 150 mM NaCl, 10 mM CaCl2, pH 8.5). The transition temperature is defined as the inflection point of the absorbance as a function of temperature. (b) Plotting the transition temperature versus SrtA-ELP concentration allows prediction of the SrtA-ELP transition temperature at a particular concentration by interpolation or extrapolation.

All fusion proteins were expressed in E. coli and were purified by inverse transition cycling (ITC), a method that we have previously developed. We designed an

ITC protocol that consisted of centrifugation of the cell lysate at 4°C in low-salt buffer where the ELP fusion was soluble (below its transition temperature), collection of the supernatant, centrifugation at 30°C in 0.3M ammonium sulfate where the ELP fusion was insoluble (above its transition temperature), followed by collection and solubilization of the pellet – containing the fusion protein – in cold, low-salt buffer.

Repeated cycles of cold and warm centrifugation led to enrichment of the ELP fusion protein by eliminating other E. coli proteins that did not exhibit reversible phase transition behavior.

Target protein-ELP fusions and SrtA-ELP were expressed and purified separately, then co-incubated in a buffer that contained Ca2+ (required for SrtA activity) and synthetic triglycine peptide. Based on scouting studies (Figure 8), we selected overnight incubation at 20°C and a 1:4 enzyme:target ratio (mol/mol).

Figure 8. Scouting studies to identify conditions for binary fusion reactions. (a) SrtA-ELP and TRX-ELP were reacted overnight at various enzyme:substrate molar ratios. (b) SrtA-ELP and TRX-ELP were reacted overnight at various temperatures. Product formation was evaluated using densitometry to evaluate the relative increase in the free TRX or free ELP bands or the relative decrease in the TRX-ELP band.

To isolate the target protein, the phase transitions of all other reaction products –

SrtA-ELP, uncleaved target protein-ELP fusions, and free ELP – were triggered by adding sodium chloride to 1M and heating to 40°C. These conditions were selected to ensure aggregation of the SrtA-ELP and unreacted TRX-ELP – one of our more hydrophilic fusions that would require the most salt and highest temperature to transition – so that a common protocol could be used for each target protein in our panel

(Figure 9). Importantly, the conditions used for cleanup of the reaction product can be changed depending on the application, for instance, by varying the type of salt used or its concentration. Dialysis or diafiltration of the supernatant removed the remaining triglycine from the purified target protein.

Figure 9. Conditions for reaction cleanup were selected to ensure removal of hydrophilic ELP fusion proteins. Conditions for target protein purification from cleavage reactions were selected to ensure aggregation of TRX-ELP, which was expected to have a high transition temperature (Tt) because TRX was one of the more hydrophilic target proteins tested and the concentration of TRX-ELP was low after the cleavage reaction. (a) Thermal response of TRX-ELP at 12.5 μM in cleavage reaction buffer with NaCl at a concentration of (●) 0M, (●) 0.25M, (●) 0.5M, (●) 0.75M, and (●) 1M. Absorbance was measured at 300nm during a 1ºC/min temperature ramp from 20-70°C. (b) The inverse transition temperature (Tt) of TRX-ELP shows a linear relationship with NaCl concentration over the range tested for fusion protein concentrations of (●) 12.5, (●) 25, and (●) 50 μM. The TRX-ELP Tt in reaction buffer was reduced approximately 20°C by the addition of 1M NaCl, which was sufficient to trigger the transition at 40°C for a fusion protein concentration as low as 10 µM. 35

Figure 10. Protein purification by the binary fusion sortase-ELP reaction. (a) Schematic of target protein purification by the binary fusion reaction. SrtA-ELP and target protein-ELP fusions were expressed and purified separately. Target proteins were cleaved by overnight reaction and the products centrifuged at 40°C for 15 minutes in the presence of 1M NaCl. (b) SDS-PAGE analysis of (left to right for each target protein): raw reaction product, centrifugation pellet, and centrifugation supernatant.TRX = thioredoxin, GFP = green fluorescent protein, TNFα = tumor necrosis factor α, TRAIL = TNF-related apoptosis-inducing ligand.

Figure 10 shows sodium dodecyl sulfate polyacrylamide gel electrophoresis

(SDS-PAGE) of the unpurified reaction product and centrifugation fractions for the purification of TRX, GFP, TNFα, and TRAIL. We observed excellent cleavage and target protein recovery (Table 1). The purity of each target protein was greater than 95% by densitometry and size exclusion high-performance liquid chromatography (Figure 11), and the molecular weight of each of the target proteins was confirmed by matrix- assisted laser desorption and ionization time of flight (MALDI-TOF) mass spectrometry

(not shown).

Table 1: Conversion, recovery, purity, and yield for proteins recovered from binary fusion sortase reactions.

Product Product yield Target Protein Reaction Product purity by (mg/L of conversion (%) recovery (%) HPLC (%) fermentation) Thioredoxin 88 100 98 35 GFP 85 100 100 28 TNFα 54 82 96 16 TRAIL 46 46 95 9

Figure 11. HPLC analysis of target proteins purified by the binary fusion reaction. Size exclusion HPLC elution profiles for (a) GFP, (b) TNFα, and (c) TRAIL purified from reactions of ELP fusion proteins with SrtA-ELP. (d) Comparison of elution profiles for unreacted TRAIL-ELP and recovered TRAIL product. All traces show the absorbance measured at 280 nm.

2.3.2 Ternary fusion reaction

We also developed a ternary fusion in which SrtA and the target protein were linked by the ELP (VPGVG)240 (sortase on the N-terminus and the target protein on the

C-terminus). This design allowed straightforward purification of the target protein without the need to add extraneous SrtA-ELP (Figure 12a). Use of the optimal LPETG enzyme recognition site produced no intact ternary fusion due to premature cleavage during expression. However, amino acid substitutions in the LPXTG motif have been demonstrated to lower the SrtA reaction rate.109 We identified a variant –LPGAG– that allowed expression of intact ternary fusion when the protein was expressed at low temperature and the purification buffer contained ethylene glycol tetraacetic acid

(EGTA) to scavenge the Ca2+ required for SrtA activity. The two mutations from the optimal LPETG sequence (E3G and T4A) had both been shown to individually lower the transpeptidation reaction rate, but had never been previously been shown to function together as a sortase substrate, suggesting that the enzyme may be more promiscuous than previously thought, an observation that would drive the studies described in a later section of this document. Figure 12b shows SDS-PAGE analysis for GFP and TNFα purified from ternary fusion reactions. As in the binary fusion system, purities of both target proteins were greater than 95% by densitometry.

Figure 12. All-in-one protein purification by the ternary fusion reaction. (a) Schematic of target protein purification by the ternary-fusion reaction. The SrtA-ELP- target protein fusion was expressed and purified under conditions that mitigated premature cleavage. The reaction was triggered, allowed to proceed overnight, then centrifuged at 40°C for 15 minutes. (b) SDS-PAGE analysis of (left to right for each target protein): unreacted ternary fusion, raw reaction product, centrifugation pellet, and centrifugation supernatant.

2.3.3 Bioactivity of the purified proteins

The fluorescence emission spectrum of GFP provided a simple indication that our methodology yielded properly folded, active protein. TNFα was investigated as a more complicated case, as homotrimer assembly is critical for TNFα signaling.110,111 Non- denaturing, native PAGE was used to compare the quaternary structure of commercially-available TNFα to that of protein purified from an ELP fusion using sortase. Protein from both sources ran similarly and appeared as three distinct bands on the gel, suggestive of monomers, dimers, and trimers (Figure 13b). The commercial preparation contains 0.5% bovine serum albumin as a stabilizer in the formulation (to

mitigate non-specific TNFα binding to the walls of the container), which accounts for the lower mobility bands at the top of the gel. Static light scattering on TNFα purified from the ternary fusion indicated an average molecular weight of 44.7 kDa, which is consistent with the average molecular weight of 49 kDa reported previously for TNFα.112

Figure 13. Sortase reactions generate properly folded, bioactive TNFα. (a) Viability of L929 cells after incubation with TNFα isolated from the binary fusion (●), from the ternary fusion (●), or from eBioscience (●) relative to an untreated control. EC50 values were 48 pM, 350 pM, and 57 pM, for TNFα isolated from cleavage of TNFα-ELP by SrtA-ELP, TNFα isolated from the ternary fusion, and commercial protein, respectively. Error bars indicate standard deviations. (b) Native PAGE of commercial TNFα (lane 1) and TNFα from the ternary fusion (lane 2) are comparable and suggest the presence of monomer, dimer, and trimer populations in the lower (most mobile) bands.

We examined the bioactivity of TNFα by measuring its effect on the L929 mouse fibrosarcoma cell line, which is known to be sensitive to cytolysis by TNFα at picomolar concentrations.113,114 Cell number relative to an untreated group was assessed for cells treated with SrtA-purified or commercial TNFα (Figure 13a). EC50 values were 48 pM,

350 pM, and 57 pM, for TNFα isolated from cleavage of TNFα-ELP by SrtA-ELP, TNFα isolated from the ternary fusion, and commercial protein, respectively. The similar bioactivity of the commercial and SrtA-purified TNFα further reinforced the proper folding and multimerization suggested by native PAGE and static light scattering.

2.3.4 Simultaneous tag removal and site-specific conjugation of small molecules to the carboxy-termini of substrate proteins

SrtA, unlike proteases or self-splicing inteins that simply cleave a target protein from a tag, also enables site-specific, covalent attachment of other molecules to the target protein. This reaction has been used to attach a diverse array of compounds to recombinant proteins by linking them to synthetic oligoglycine or LPXTG peptides, making them suitable nucleophiles or SrtA substrates, respectively.115,116,117,118 However, only the termini of a target protein can be modified using this reaction, making the method inflexible with regard to the conjugation site and limiting the number of molecules that can be attached to two. Our protein purification reactions were designed to mimic a typical protease digestion by using synthetic triglycine as a nucleophile.

However, the system is flexible in that it can be easily extended to accommodate alternative nucleophiles.

To demonstrate the flexibility of our system, we conjugated the chemotherapeutic camptothecin (CPT) to TRAIL, generating a hybrid anticancer agent.

We chose CPT for its potent antitumor activity and because it has an absorption peak at

365 nm that allows the molecule to be tracked using spectrophotometry. The drug was chemically modified such that the hydroxyl group on the E-ring was coupled to the carboxyl terminus of a triglycine peptide (denoted Gly3-CPT).

The conditions used for the TRAIL-CPT conjugation reaction and product recovery were the same as those used for the purification of unlabeled protein, except that Gly3-CPT was substituted for triglycine (Figure 14a). SDS-PAGE of parallel reactions with triglycine or Gly3-CPT (Figure 14b) indicated significant conversion in both cases, suggesting that Gly3-CPT was an effective nucleophile. Based on analysis of

SDS-PAGE images, we determined that the reaction conversion using Gly3-CPT was approximately 80% of that for a reaction using triglycine. Similarly, overall recoveries after reaction purification for Gly3-CPT reactions were consistently 65% of recoveries for reactions using Gly3. We attribute the reductions in yield and recovery to the reduced solubility of the CPT moiety, which likely lowered the concentration driving force for reaction and possibly contributed to some nonspecific aggregation of the product.

Electrospray ionization mass spectrometry confirmed that the m/z ratios of TRAIL and

the TRAIL-CPT conjugate differed by the mass of a single molecule of CPT (Figure 15 and Table 2).

Figure 14. Protein-small molecule conjugation using the binary fusion reaction. (a) Schematic of the reaction and purification used to make TRAIL-camptothecin (CPT) conjugates. Conditions are the same used in the binary fusion reaction for protein purification, except that here Gly3 is replaced by CPT-Gly3. (b) SDS-PAGE analysis of TRAIL-ELP reactions run in parallel with triglycine and triglycine-CPT. Shown left to right for each reaction are the raw product, centrifugation pellet, and centrifugation supernatant.

TRAIL-Gly3 charge state m/z calculated mass (Da) 30 686.5 20564.8 29 710.1 20563.7 28 735.4 20563.0 27 762.7 20565.7 26 791.9 20563.2 25 823.6 20564.8 24 857.8 20563.0 23 895.1 20564.1 22 935.8 20565.4 21 980.3 20565.1 20 1029.2 20563.8 19 1083.4 20565.4 18 1143.5 20564.9 17 1210.6 20563.1 Average = 20564.28484 b

TRAIL-CPT charge state m/z calculated mass (Da) 29 721.6 20897.2 28 747.3 20896.2 27 775 20897.8 26 804.7 20896.0 25 836.8 20894.8 24 871.7 20896.6 23 909.5 20895.3 22 950.8 20895.4 21 996 20894.8 20 1045.9 20897.8 19 1100.8 20896.0 Average = 20896.2

Figure 15. Electrospray ionization mass spectrometry analysis of TRAIL-Gly3 and the TRAIL- camptothecin conjugate. (a) Charge state deconvolution of TRAIL-Gly3 indicates that its average molecular mass is 20564.3 Da. (b) Charge state deconvolution of TRAIL- CPT conjugate indicates an average molecular mass of 20896.2 Da. Both masses are in agreement with the predicted values and differ by the mass of 1 molecule of camptothecin. 45

Table 2: Comparison of predicted and experimental molecular masses for TRAIL-Gly3 and TRAIL- CPT.

Species Predicted Mw (Da) ESI-MS Mw (Da)

TRAIL-CPT 20895.2 20896.2

TRAIL-Gly3 20563.9 20564.3 Difference = CPT 331.3 331.9

Bioactivity of the TRAIL-CPT conjugate was assessed by measuring apoptosis in the TRAIL-sensitive human breast adenocarcinoma cell line MDA-MB-231 (Figure 16).119

Viable cell number relative to an untreated control was assayed after incubation with

TRAIL, Gly3-CPT, TRAIL-CPT conjugate, or TRAIL and Gly3-CPT in a molar ratio equivalent to that of the conjugate. High concentrations of Gly3-CPT alone killed 40% of the cells, whereas TRAIL killed a maximum of 30% of the cells. Dosing TRAIL and CPT as a conjugate reduced the viable cell number by a maximum of 75%. Though improved cell killing was anticipated using a combination of CPT and TRAIL, it is noteworthy that the TRAIL-CPT conjugate had the same potency and efficacy as the combination of non- conjugated Gly3-CPT and TRAIL in an equivalent molar ratio, which suggested that each molecule within the conjugate retained its activity. Interestingly, both conjugated and non-conjugated drug combinations showed an additive effect. Though the particular combination of TRAIL and CPT does not appear to offer a synergistic advantage in our

cytotoxicity assays, it nonetheless provides a proof-of-concept example of the ease with which bioactive, site-specific conjugate therapeutics can be produced by this system.

Figure 16. TRAIL-camptothecin conjugates are bioactive and cytotoxic towards human breast adenocarcinoma cells in vitro. Cell death in MDA-MB-231 cells treated for 48 hours with TRAIL (●), Gly3-CPT (●), TRAIL-CPT conjugate (●), or a mixture of Gly3-CPT and TRAIL at a 0.8:1 molar ratio. (●). TRAIL-CPT conjugate kills cells with a similar dose response curve as that for a mixture of the drugs with an equal drug-to-protein ratio. Similarly, the dose response curves for TRAIL and Gly3-CPT alone are additive and their sum reproduces the dose response curve of the conjugate and drug mixture. This suggests that the TRAIL-CPT conjugate retains the bioactivity of both the TRAIL and CPT components, but these drugs are not synergistic on the MDA-MB-231 cell line as a conjugate or as a mixture.

2.3.5 TRAIL-ELP fusions show potent in vitro and in vivo cytotoxicity towards TRAIL-sensitive cell lines

Though it does not directly involve sortase, we found that TRAIL-ELP fusions had potent in vitro cytotoxicity towards TRAIL-sensitive cell lines that led to dramatic tumor regressions in mice.

We generated two versions of TRAIL ELP. For local intratumoral administration, we fused TRAIL to ELP (VPGXG)90 (X=A:G:V=2:3:5 mol/mol/mol), forming a fusion that would undergo its phase transition and form a coacervate (or gel) upon injection. For systemic administration, we fused TRAIL to ELP (VPGXG)80 (X=A:G=1:1 mol/mol), forming a fusion that would remain soluble upon injection. Absorbance versus temperature scans (Figure 17) showed that the coacervate-forming ELP did transition around body temperature. Extrapolation of the concentration dependence of the transition temperature indicated that a coacervate would form at 37°C for concentrations greater than 290 µM, making it suitable for intratumoral depot formation if injected above this concentration. Thermal scans of the TRAIL-ELP that was designed to be soluble on injection showed that this construct will not undergo a transition at body temperature, even when extrapolated to extremely high concentrations.

Figure 17. Different ELP fusion partners allow TRAIL to be delivered locally or systemically. TRAIL fused to a hydrophobic ELP with the composition (VPGXG)90, with X=A:G:V in a 2:3:5 molar ratio, transitions below body temperature and forms a coacervate when injected. (a) Absorbance at 350 nm as a function of temperature. (b) Inflection points from the A350 curves plotted against ELP concentration. Extrapolation of the fit suggests the TRAIL-ELP forms a coacervate at 37°C at concentrations of 290 μM or higher, making it suitable for local or depot delivery. In reality, this concentration will be lower because proteins and salts in the injection site will lower the Tt for all ELP concentrations. (c) and (d) Similar data for TRAIL fused to a hydrophilic ELP with the composition (VPGXG)80, with X=A:G in a 1:1 molar ratio. Analysis of the fit suggests that this construct does not transition at body temperature, making it suitable for intravenous administration. All data was obtained in buffer (0.5M arginine, 20 mM Tris, 300 µM ZnSO4, 5 mM DTT, pH 7.5) using a ramp rate of 1°C/minute.

We found the cytotoxicity of TRAIL-ELP to be highly dependent on the conditions used during purification of the protein. TRAIL, like TNFα, is a homotrimeric protein. Unlike TNFα, though, each TRAIL monomer contains a cysteine residue whose

side chain thiol chelates a Zn2+ ion in coordination with the cysteine side chains in the two other monomers. TRAIL activity depends on correct trimer formation: because the cysteines are spatially close in the trimer, they can form an intermolecular disulfide bond in the presence of chelating agents and oxidizing environments. The disulfide- linked species can still activate death receptor signaling, but the potency of this molecule is much lower than that of the correctly formed trimer.120 Indeed, we found that TRAIL-

ELP purified in buffer (20 mM Tris pH 7.5 or PBS) without zinc II or reducing agent

(dithiothreitol, DTT) was prone to aggregation and was much less cytotoxic towards the

TRAIL-sensitive COLO205 human colon adenocarcinoma cell line (Figure 18a). The EC50 for TRAIL-ELP purified in without Zn2+ and DTT was ~200 nM, whereas purification in

0.5M arginine, 20 mM Tris, 300 µM ZnSO4, pH 7.5 generated TRAIL with an EC50 that was 30-200 fold lower, depending on whether the ELP was hydrophobic (depot-forming at 37°C) or hydrophilic (soluble at 37°C). We used 0.5M arginine in our purification because TRAIL is known to be soluble in high osmolarity buffers, particularly for Tris,

NaCl, sodium acetate, and arginine.121 Importantly, all proteins were expressed the same way in BL21 E. coli: a 2.5% by volume inoculum was used to inoculate Terrific Broth containing 45 µg/ml kanamycin sulfate, the cells were grown for 4.5 hours at 25°C, then induced by adding IPTG to 0.1 mM and incubating the cells overnight at 16°C. We attribute the difference in cytotoxicities to the amount of disulfide-linked protein in the formulations, as approximately 50% of the TRAIL-ELP purified without supplemental

Zn2+ and DTT was found to have an intermolecular disulfide bond by SDS-PAGE (Figure

18b).

Figure 18. Purification conditions influence dimer formation and TRAIL cytotoxicity. Three different TRAIL-ELP fusion proteins were purified in Tris-based buffer containing diothiothreitol (DTT) and zinc. (a) Human colon adenocarcinoma cell line COLO 205 is sensitive to the TRAIL-ELP fusions, commercially-available TRAIL, and TRAIL-ELP purified without DTT and zinc, though the EC50 is improved over 2 orders of magnitude by purification in reducing conditions with zinc. Coacervate-forming TRAIL-ELP uses ELP (VPGXG)90, X=A:G:V=2:3:5 mol/mol/mol. Strong coacervate-forming TRAIL-ELP uses (VPGVG)60. Soluble TRAIL-ELP uses (VPGXG)80, X=A:G=1:1 mol/mol. (b) Non- reducing SDS-PAGE suggests that greater than 50% of TRAIL-ELP purified without DTT is disulfide-linked. This is in agreement with the much lower activity of this sample in the in vitro cytotoxicity assay.

The coacervate-forming TRAIL-ELP killed COLO205 cells in vitro with excellent potency, and we therefore sought to test this formulation in mice. We inoculated one million COLO205 cells subcutaneously into male athymic nude mice and allowed tumors to grow to ~75 mm3. A single dose of buffer, coacervate-forming TRAIL-ELP at

30 mg/kg, or coacervate forming TRAIL-ELP at 60 mg/kg were then injected into the tumors and the tumor volume, body weight, and survival were tracked over time. Both

TRAIL-ELP doses were well tolerated and the animals initially lost a small amount of body weight that was attributable to the anesthesia used during treatment. Tumors regressed in both TRAIL-ELP treatment groups, providing responses that were durable over the 5 weeks of the study (Figure 19). As a result, all mice in both TRAIL-ELP treatment groups were alive at the end of the study, while 75% of the vehicle-treated mice were euthanized as a result of high tumor burden. While this data is preliminary, it suggests that intratumoral administration of TRAIL-ELP coacervates has the potential to effectively treat TRAIL-sensitive tumors that are accessible to this route of administration.

Figure 19. Intratumoral administration of coacervate-forming TRAIL-ELP in mice is well-tolerated, leads to tumor regression, and improves survival. (a) Fold change in tumor volume after a single intratumoral injection of coacervate- forming TRAIL-ELP. Cultured COLO 205 cells were injected into the right flank of male nude athymic mice and allowed to form approximately 75 mm3 tumors prior to treatment. A single dose of vehicle or endotoxin-purified TRAIL-ELP solution (650 μM) at 30 mg/kg or 60 mg/kg was administered in an intratumoral injection at day 0. Widths and lengths of tumors were measured using calipers and volume was calculated according to (volume) = (1/6)*π*(width)*(length). (b) Mice treated with TRAIL-ELP at either dose did not show a loss of body mass compared to vehicle-treated mice, suggesting that TRAIL-ELP does not exhibit systemic toxicity at the doses used. (c) Survival curve. Mice were euthanized when the tumor volume reached 1000 mm3.

2.4 Results and discussion

The combination of SrtA and ELPs represents a powerful and flexible system for purification and site-specific chemical modification of proteins. As with inteins, our ternary fusion provides a straightforward, all-in-one system that is subject to premature cleavage during expression. However, our binary fusion system provides an alternative with excellent control over reactivity, high product yields, and virtually no increase in complexity.

Using our reaction strategies, significant quantities of high-purity, bioactive recombinant proteins can be purified without column chromatography by a protocol that is easy to execute and applicable to a variety of target proteins. The ease with which our protocol is executed and the flexibility to perform an optional, one-step, and site- specific conjugation reaction that homogeneously labels the product provide a valuable,

“three-in-one” set of tools for the production of therapeutic proteins and protein-small molecule conjugates.

3. A non-canonical function of sortase enables site- specific conjugation of small molecules to lysine residues in proteins

We show in this section that sortase catalyzes a transpeptidation reaction that joins two moieties through an isopeptide bond. Conjugation occurs at the ε-amino group of a lysine residue, and because this reactive site is positioned within a defined sequence element, our technique provides a new method to generate protein-small molecule conjugates that allows precise control of the number of conjugated molecules as well their location. We demonstrate the first application of isopeptide transpeptidation to modify recombinant proteins and show that SrtA from S. aureus catalyzes this reaction, uncovering a function of this widely-used enzyme that was previously unknown. This method is the only available procedure that site-specifically modifies lysine residues, thus making it a valuable, orthogonal addition to the protein conjugation toolbox.

3.1 Introduction

Certain Gram-positive bacteria utilize homologs of S. aureus SrtA to assemble pili—fibrous polymers of structural proteins—that extend from the bacterial surface and are implicated in adhesion and biofilm formation.122 The initial step for pilin formation is the same as that used to anchor proteins to the cell wall and generates a thioester intermediate upon cleavage of the LPXTG motif. Extension of a pilin polymer proceeds when the thioester undergoes nucleophilic attack by the ε-amino group of a lysine 55

residue in another sortase-linked pilin monomer (Figure 20). This process repeats and is stochastically terminated by a “housekeeping” sortase that anchors the pilin polymer to the cell wall through a native peptide bond at the α-amino group of oligoglycine. The pilin polymer is therefore composed of a series of branched monomers linked by isopeptide bonds, with each monomer retaining an unmodified α-amino group. In

Corynebacterium diphtheriae (C. diphtheriae) and Actinomyces naeslundii (A. naeslundii), the nucleophilic lysine used in polymer extension is contained in a “pilin domain” with the sequence WX3VXVYPKH, where X is a non-conserved residue.123

Figure 20. Schematic of pilin assembly by sortases in Gram-positive bacteria. (a) Secreted pilin monomer protein (SpaA) is cleaved by a pilin-specific sortase, forming a thioester-linked intermediate. (b) The pilin polymer extends after the intermediate undergoes nucleophilic attack by the ε-amino group of a lysine in the next monomer, forming an isopeptide bond. (c) After a stochastic number of extension steps, a housekeeping sortase terminates polymerization by anchoring the assembled pilin chain to a pentaglycine branch in the peptidoglycan through a native peptide bond.

We hypothesized that a compound bearing the sortase recognition sequence

(LPXTG) could be conjugated to a protein engineered to contain the pilin domain, and that these two sequence elements would together provide a means of highly selective attachment. Moreover, we hypothesized that recombinant SrtA would accept the pilin domain lysine as a nucleophile in vitro in the absence of its preferred nucleophile, the α- amino group of oligoglycine. We chose to explore the ability of SrtA to catalyze this reaction, rather than the sortases from C. diphtheriae because SrtA is expressed at high yield in E. coli, while the alternative sortase has proven difficult to express in heterologous systems. To test these hypotheses, we used the reaction to generate a branched protein polymer that demonstrated its feasibility and allowed optimization of reaction conditions. We then utilized the reaction mechanism to site-specifically conjugate model small molecules to an engineered fibronectin type 3 (Fn3) domain that targets the αvβ3 integrin. We show that the SrtA isopeptide ligation reaction displays an exquisite level of control over the site of conjugation and generates bioactive conjugates exclusively at one or more unique sites engineered into the protein. This conjugation method is distinct from existing methods, and provides a new tool for the synthesis of site-specific protein-drug and protein-polymer conjugates.

3.2 Materials and Methods

3.2.1 Protein expression and purification

SrtA-ELP, Fn3-ELP, and Fn3-PLN3-ELP fusion proteins were genetically assembled in pET24b vectors, expressed in BL21 (DE3) E. coli (EdgeBio, Gaithersburg,

MD), and purified from soluble bacterial lysate as described previously.

3.2.2 Protein-small molecule conjugation reactions

Fn3-ELP or Fn3-PLN3-ELP were reacted at 50 µM with SrtA-ELP at 100 µM and biotin-LPETGRAGG or FITC-LPETGRAGG peptide at 500 µM in reaction buffer (50 mM

Tris-HCl, 150 mM NaCl, 10 mM CaCl2). Pilin domain peptide was purchased from

Genscript and was reacted at 10 µM with SrtA-ELP at 100 µM and biotin-LPETGRAGG peptide at 500 µM in reaction buffer. All reactions were allowed to proceed overnight at

32°C, then isopeptide ligation was quenched by the addition of triglycine peptide to 10

µM.

Reaction product was run on SDS-PAGE on anykD stain-free TGX gels (Bio-Rad).

For Western blots, the SDS-PAGE product was transferred to a Trans-blot turbo PVDF membrane (Bio-Rad), blocked with 5% nonfat milk, and probed with 1 mg/ml streptavidin-Cy5 conjugate (Life Technologies) at a 1:20,000 dilution in Tris-buffered saline with 1% Tween-20 and 0.5% nonfat milk. Fluorescent bands were detected with a

Typhoon variable mode laser scanner (GE Healthcare Life Sciences, Pittsburgh, PA). The

unpurified reaction product was trypsinized and analyzed by MALDI-TOF using α-

Cyano-4-hydroxycinnamic acid matrix. The MS-Digest function of ProteinProspector v

5.10.14 (available at http://prospector.ucsf.edu/prospector/mshome.htm) was used to calculate expected tryptic peptide masses for recombinant protein sequences.

3.2.3 LC-MS/MS and biotin:protein molar ratio

Biotinylated Fn3-PLN3-ELP purified by cation exchange chromatography was digested with MS-grade trypsin overnight at 20°C at a trypsin:substrate ratio of approximately 1:20 by mass. Desalted peptides in 50 mM ammonium bicarbonate were separated by reverse-phase high pressure liquid chromatography using a Supelco

Ascentis (5cmx1mm, 3µm) C18 column over an elution gradient of 5-50% buffer B (5%

H2O, 95% acetonitrile). MS data was collected using an Agilent 1100 series quadrupole ion trap (Agilent Technologies, Santa Clara, CA). For fragmentation and MS2, parent ions were isolated in the trap and fragmented by collisions with helium. Parent and daughter ions were identified using the MS-Digest and MS-Product functions of

ProteinProspector v 5.10.14.

The biotin:protein ratio was assessed by comparison of the integrated areas of summed extracted ion chromatograms for ions containing either biotinylated or non- biotinylated pilin domain lysines. For Fn3-PLN3-ELP reactions, a colorimetric quantitation kit (Thermo Scientific) that measured the decrease in HABA 500 nm

absorbance when displaced from avidin by protein-bound biotin was also used according to the manufacturer’s instructions. Prior to analysis, biotinylated proteins were diafiltered against at least 30 diavolumes of PBS. Protein concentrations used in the assay were obtained using the bicinchoninic acid (BCA) assay. Measurements of biotin and protein concentrations were performed in triplicate for each sample and standard.

3.2.4 Immunofluorescence microscopy

SK-OV-3 cells were obtained from the Duke University Health System Cell

Culture Facility and were maintained in McCoy’s 5a medium with 10% heat inactivated fetal bovine serum, 5 units/ml penicillin, and 5 µg/ml streptomycin (Gibco). Cells were attached to glass coverslips overnight. Cells were washed in PBS, then fixed with 4% paraformaldehyde in PBS at room temperature for 15 minutes. Fixed cells were stained with a 1:200 dilution of 0.5 mg/ml anti-Her2 (with a pilin domain on each heavy chain) biotinylated by reaction with SrtA or a biotinylated murine IgG1 isotype control antibody, followed by a 1:200 dilution of 1 mg/ml streptavidin-FITC conjugate. Cells were analyzed with a Nikon TE2000-U inverted fluorescent microscope with a 60x

1.25NA oil immersion objective.

3.3 Results and Discussion

3.3.1 Proof-of-concept of sortase-catalyzed isopeptide ligation using peptides

To date, only a single paper has been published on isopeptide ligation catalyzed by SrtA from S. aureus, but this study did not investigate the site-specificity of the reaction. Moreover, until our recent work isopeptide ligation had not been applied to modify proteins (the previous study only used peptide substrates).124 We first confirmed the ability of SrtA to carry out isopeptide ligation by reacting a LPETGRAGG peptide containing an amino-terminal biotin with a pilin domain peptide

(WLQDVHVYPKHGGSGR). The enzyme used to catalyze this reaction, SrtA, was produced as a fusion protein with an elastin-like polypeptide (ELP).125,126 ELPs are inert

—they impart no new bioactivity beyond phase transition behavior to their fused peptide or protein partner— and provide efficient purification tags that allow easy recovery of ELP fusions without column chromatography by exploiting their phase transition behavior. After incubating these two peptides with the SrtA-ELP overnight, the SrtA-ELP was efficiently removed by its phase transition induced aggregation and centrifugation, and the reaction product was analyzed by matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometry, which indicated the formation of a species that corresponded to the isopeptide-linked biotin-LPET and the pilin domain peptide (Figure 21).

Figure 21. Reaction with peptide substrates demonstrates the isopeptide ligation reaction catalyzed by sortase A. (a) Scheme showing the sortase-mediated isopeptide ligation of a peptide containing the pilin domain with a peptide containing the sortase recognition sequence LPETG and an amino-terminal biotin. (b) MALDI-TOF mass spectrometry indicates the formation of a reaction product with a molecular weight corresponding to peptides linked through an isopeptide bond when SrtA-ELP is included in the reaction (upper panel), but not when the enzyme is omitted from the reaction mixture (lower panel).

3.3.2 Sortase-catalyzed protein polymerization through isopeptide bonds

To extend the isopeptide ligation reaction to a model polypeptide, glucagon-like peptide-1 (GLP-1) was selected because GLP-1 is a pharmaceutically-relevant peptide with two internal lysines that could serve as off-target sites for isopeptide bond formation. We fused GLP-1 to both the pilin domain and the sortase LPETG recognition sequence, generating a polypeptide that was expressed as a fusion protein with an ELP.

Because each ELP-GLP1 contained both the pilin domain and LPETG enzyme recognition sequence, it was polymerized by SrtA-ELP according to the reaction shown in Figure 22a. SDS-PAGE of the reaction product (Figure 22b) showed a ladder of polymers with different numbers of subunits. Polymerization was temperature- dependent and greater than 90% monomer consumption was achieved between 29°C and 34°C (Figure 22c). This temperature range also showed the greatest fraction of high molecular weight (>250 kDa) polymers, with this population comprising 50% of the reaction product by densitometry. We therefore selected 32°C for all subsequent reactions. pH had a minor effect on reaction efficiency between pH 5.5 and pH 10, with the highest qualitative yield at pH 8-8.5 (not shown).

Figure 22. Sortase A catalyzes isopeptide bond formation at the pilin domain lysine. (a) Schematic of the polymerization reaction model for isopeptide bond formation. Predicted tryptic peptides of the polymerization product are boxed. (b) SDS-PAGE shows polymerization of the 41 kDa ELP-GLP1 monomer at all temperatures tested except reactions stored at -20°C. (c) Maximum monomer consumption in the polymerization reaction was achieved between 29°C and 34°C. Conversions are reported as the average of 3 independent reactions carried out at each temperature that were analyzed by SDS-PAGE and densitometry. Error bars indicate standard deviations. (d) MALDI-TOF mass spectrometry of a tryptic digest of 33°C reaction product shows the isopeptide linkage at the pilin domain lysine (ion 5) along with other ions that correspond to unmodified peptides from the ELP-GLP1 and SrtA-ELP fusion proteins.

Assuming that nucleophilic attack by a primary amine resolves the thioester intermediate, it is possible that polymerization proceeds through the terminal α-amino group of the ELP-GLP1 fusion rather than the ε-amino group of the pilin domain lysine.

If the ε-amino group acts as the nucleophile, the junction of two monomers will be branched as shown in Figure 22a, whereas a linear junction would arise from head-to- tail polymerization through the terminal α-amino group. To determine the location of the polymer junction, we treated the unpurified reaction product with mass spectrometry-grade trypsin and analyzed the resulting peptides by MALDI-TOF mass spectrometry. Trypsin normally cleaves a peptide bond on the carboxy-terminal side of lysine or arginine residues, but cannot cleave at lysines with modified ε-amino groups.

The MALDI-TOF spectrum showed the presence of an ion at m/z 2809.4 from the branched junction of two monomers (peak 5 in Figure 22d). This ion corresponded to a peptide with a single missed digestion at the pilin domain lysine and an additional mass of 625.7 Da, which indicated that amino acids from the carboxy-terminus of another monomer had been joined to the pilin domain lysine through an isopeptide bond.

Notably, we did not observe any signal at the predicted m/z values for ions that would have arisen from modification of either of the two lysines in GLP-1 or at the amino- terminus of the protein (Appendix A), suggesting that SrtA-mediated conjugation is specific for the pilin domain lysine.

3.3.3 Conjugation of small molecules to proteins using sortase- catalyzed isopeptide ligation

The protein-protein ligation reaction in the previous section was valuable in that the effects of changing reaction conditions could be easily evaluated using only SDS- 65

PAGE and inexpensive protein reagents. However, we wished to extend the reaction to make protein-small molecule conjugates, as this would allow the method to be used to conjugate drugs and polymerization initiators to proteins, both of which are important reactions for clinically-used therapeutics. Moreover, recombinant protein modification would provide a stringent test of the site-specificity of the reaction due to the presence of multiple, potentially cross-reactive lysine residues in most proteins of interest.

We initially tested the ability of SrtA to conjugate the small molecule quencher dabcyl to our ELP-GLP1 protein, as a dabcyl-LPETG-edans reagent is commercially available. We analyzed the tryptic digest of the reaction product by MALDI-TOF mass spectrometry (Figure 23). Though both polymerization and dabcyl conjugation were observed (as expected), both reactions were site-specific for the ε-amino group of the pilin domain lysine.

Figure 23. Sortase A catalyzes site-specific conjugation of dabcyl to GLP-ELP. (a) Structure of dabcyl-LPETG-edans peptide. Dabcyl absorbs at the 337 nm wavelength of the N2 laser used in MALDI-TOF and reproducibly lost a 132 Da segment due to fragmentation (the proposed fragment is highlighted in blue). (b) Schematic for the conjugation of dabcyl to ELP-GLP1. The ELP-GLP1 is the same protein used in the polymerization model, therefore, both polymerization and dabcyl conjugation can occur through the pilin domain as shown in the schematic. (c) MALDI-TOF spectra for the unpurified reaction product. The corresponding peak table is available for reference in Appendix B. The spectrum shows the presence of both conjugated dabcyl (ion 1) and the branched polymer junction (ion 5) at the pilin domain lysine. For dabcyl attachment, another ion is present (ion 1 – 132) that corresponds to attachment at the pilin domain lysine but with the loss of 132 Da. This same loss is also observed for ions corresponding to the reaction side product, dabcyl-LPETGGG, as indicated on the spectrum. Both conjugation and polymerization reactions are specific for isopeptide-linked attachment at the pilin domain lysine ε-amino group, as all other ions in the spectra map to unmodified segments of SrtA-ELP or ELP-GLP1.

Because we directly digested the reaction product with trypsin, the presence of the SrtA-ELP fusion required the addition of excess triglycine peptide during trypsinization, as many of the peptides generated by trypsin contained amino-terminal

glycine residues and were able to resolve the SrtA-thioester intermediate through the well-characterized SrtA transpeptidation reaction. Because triglycine is a preferred nucleophile in this reaction, its presence mitigated the formation of these modified peptides. Omission of triglycine during trypsinization resulted in more complicated

MALDI-TOF spectra with peaks corresponding to peptides linked to LPET-bearing substrate molecules through an α-amino group (Figure 24 and Table 3). These ions were only observed for tryptic peptides that had an amino-terminal glycine residue and were not present when triglycine was included during trypsinization. Though more complicated, this spectrum also suggests the formation of isopeptide bonds at the pilin domain lysine based on the skipped tryptic digestion at this residue and the additional mass resulting from dabcyl-LPET conjugation. Interestingly, the importance of adding triglycine to the tryptic digest reaction is highlighted by the pilin domain peptide itself, as attachment of dabcyl can occur through both the pilin domain lysine ε-amino group

(ions 2 and 5) and the α-amino group of the amino-terminal glycine (ions 1 and 5) that is exposed when trypsin cleavage produces this peptide. In contrast, the artifact of conjugation at the amino-terminal glycine is not observed when excess triglycine is included in the reaction (ions 1 and 5 in Figure 24 are not present in Figure 23).

Figure 24. MALDI-TOF mass spectrum of the tryptic digest of dabcyl-conjugated ELP- GLP1. Isopeptide ligation is apparent based on the missed tryptic digests at the pilin domain lysine and the increase in peptide mass by 692.3 Da (corresponding to dabcyl-LPET). However, without excess triglycine, the tryptic digest produces peptides with amino- terminal glycine residues that can react with remnant SrtA thioester reaction intermediates, leading to the conjugation of dabcyl-LPET through native peptide bonds for these peptides (ions 1, 3, 5, and 7).

Table 3: Predicted and experimental peptide masses for the conjugation of dabcyl to ELP-GLP1.

Predicted Found Peak Source Peptide sequence mass mass (Da) (Da) Dabcyl-LPET + GSSGGWLQDVHVYPK 1 PLN 2322.6 2323.0 (through α-amino group) GSSGGWLQDVHVYPKHGGSGR + Dabcyl- 2 PLN 2874.2 2876.9 LPET (through lysine ε-amino group) Dabcyl-LPET + GAHGEGTFTSDVSSYLEEQAAK 3 GLP1 2977.2 2977.8 (through α-amino group)

4 SRT15-42 VAGYIEIPDADIKEPVYPGPATPEQLNR 3054.4 3055.8 GSSGGWLQDVHVYPKHGGSGR + 2 Dabcyl- 5 PLN LPET (through both lysine ε-amino group and α- 3566.0 3566.2 amino group) GVSFAEENESLDDQNISIAGHTFIDRPNYQFTN 6 SRT43-77 3972.3 3973.8 LK Dabcyl-LPET + 7 SRT43-77 GVSFAEENESLDDQNISIAGHTFIDRPNYQFTN 4664.1 4664.6 LK

3.3.4 Isopeptide ligation of multiple small molecules to a single protein

We next attempted to extend the isopeptide ligation reaction to site-specifically label a bioactive protein with small molecules. Additionally, by incorporating multiple pilin domains into the substrate protein, we hypothesized that a stoichiometry of more than one small molecule per protein could be achieved. This is desirable because antibody-drug conjugates contain between 2 and 8 drugs per antibody, with an average of ~4 drugs/mAb typically used to achieve therapeutic intracellular concentrations of the cargo drug.

The ability to attach multiple cargo molecules to a protein would also represent a significant advantage over the canonical, native peptide ligation reaction catalyzed by sortase, which only attaches one molecule to the carboxy-terminus of a polypeptide chain. Additionally, isopeptide ligation is in principle not restricted to the termini of the substrate protein, a feature of the native peptide transpeptidation reaction that is a major limitation of sortase for proteins in which the termini are important for biological function.

We initially used the fibronectin type III (Fn3) domain as a model substrate. Fn3 domains are of interest as an alternative to antibodies because they can be affinity matured against a target but do not contain the complex quaternary structure, disulfide bonds, or glycosylation found in antibodies.127,128 We generated a Fn3 domain fused to an

ELP with 3 intervening copies of the pilin domain (Fn3-PLN3-ELP), and a Fn3-ELP fusion lacking any pilin domains as a control. We incorporated multiple copies of the pilin domain into the Fn3 fusion protein between the Fn3 domain and the ELP to investigate whether sortase could attach a payload of several small molecules to a single protein molecule.

The Fn3-ELP and Fn3-PLN3-ELP fusions were incubated with SrtA-ELP and biotin-LPETGRAGG peptide overnight (Figure 25a) and the product was analyzed by

Western blot using a streptavidin-Cy5 conjugate to detect biotinylated protein. SDS-

PAGE and the corresponding Western blot are shown in Figure 25b. Only the reaction of

biotin-LPETGRAGG peptide with Fn3-PLN3-ELP resulted in biotinylated target protein, which suggested that biotinylation was specific for the pilin domain. No biotinylation of the Fn3-ELP without pilin domains was observed—no band was observed in lane 1 of the Western blot at the expected molecular weight of 36 kDa—despite the fact that this protein contains 3 lysine residues and a terminal primary amine that offer sites for off- target conjugation. Only SrtA-ELP was observed in this reaction (at approximately 120 kDa) because the thioester-linked SrtA-biotin intermediate is stable in SDS-PAGE. In contrast, biotinylated Fn3-PLN3-ELP appeared as a 43 kDa band in lane 2 of the Western blot, along with the thioester-linked SrtA-ELP intermediate at 120 kDa. Lanes 3 and 4 show control reactions containing Fn3-ELP and Fn3-PLN3-ELP, respectively, where the biotin-LPETGRAGG peptide was excluded. As expected, no bands were observed in the

Western blot for the reactions in lanes 3 and 4.

In order to determine the biotinylation site and the selectivity of the reaction for the pilin domain, the unpurified product was digested with trypsin and analyzed by

MALDI-TOF mass spectrometry. The spectrum indicated the presence of peptides with a missed digestion at the pilin domain lysine and an additional mass of 666.8 Da

(matching the mass of biotin-LPET) in the reaction of Fn3-PLN3-ELP with biotin-

LPETGRAGG (ions 1 and 2 in Figure 25c). These ions corresponded to m/z values of both

2763.5 and 3871.2 because the tryptic peptide of the amino-terminal copy of the pilin

domain contained some residues from the Fn3 sequence. These ions were not observed when biotin-LPETGRAGG peptide was not included in the reaction (Figure 25d). All other ions in the spectra mapped to expected, unmodified segments of the Fn3-PLN3-

ELP or SrtA-ELP fusion proteins. Importantly, no ions corresponding to biotinylation at any of the 3 lysines in the Fn3 domain or at the Fn3 amino-terminus were observed in the MALDI-TOF spectra.. A peak table for these MALDI-TOF spectra is available in

Appendix C.

Figure 25. Sortase A catalyzes site-specific conjugation of biotin to Fn3. (a) Overview of the reaction used to conjugate biotin to the pilin domain in the Fn3- PLN3-ELP fusion protein, along with the expected tryptic peptides of the reaction product (boxed). (b) SDS-PAGE of biotinylation reactions with Fn3-ELP (36 kDa) and Fn3-PLN3-ELP (43 kDa). (c) Western blot with streptavidin-Cy5 indicates that only the reaction containing both the pilin domain and biotin-modified LPETG peptide yields biotinylated target protein. (d) MALDI-TOF spectrum of the unpurified Fn3-PLN3-ELP biotinylation reaction product—corresponding to lane 2 of the Western blot in panel (c)—after trypsinization. Biotin is installed specifically at the pilin domain lysine (ions 1 and 2), and all other ions corresponding to unmodified peptides from the Fn3-PLN3-ELP and SrtA-ELP fusion proteins. (e) MALDI-TOF spectrum of tryptic peptides from Fn3- PLN3-ELP reacted without biotin-modified LPETG peptide, corresponding to lane 4 of

the Western blot in panel (c). The spectrum shows the same peaks as in panel (d) except those corresponding to ions 1 and 2, confirming that these ions arise from biotinylation of the pilin domain lysine when both the biotinylated LPETG peptide and the pilin domain are present in the reaction.

The same SrtA linked to a hexahistidine tag—rather than an ELP—and purified by immobilized metal affinity chromatography also biotinylated the Fn3-PLN3-ELP through isopeptide bonds at pilin domain lysines, as assessed by Western blotting and

MALDI-TOF (Figure 26). This rules out the potential influence of the ELP tag on the enzyme’s activity and suggests that protein modification by isopeptide ligation is a general in vitro function of SrtA. We also attached fluorescein isothiocyanate (FITC) coupled to an LPETGRAGG peptide to the Fn3-PLN3-ELP by isopeptide ligation (Figure

27), and confirmed conjugation to the pilin domain lysine by in-gel fluorescence detection and MALDI-TOF mass spectrometry. This demonstrates that the reaction is not restricted to biotin and can be used with bulkier, more drug-like small molecules.

Figure 26. Isopeptide ligation is a general in vitro function of SrtA. His-tagged SrtA achieves site-specific conjugation of biotin to the pilin domain lysine. (a) SDS-PAGE of reactions of H6-SrtA and Fn3-PLN3-ELP run overnight at a range of temperatures from 21-42°C with biotin-LPETGRAGG peptide. (b) Western blot of the gel in panel (a) using streptavidin-Cy5 to detect biotinylated protein. H6-SrtA appears as a biotinylated band at ~25 kDa in the reactions containing biotin-LPETGRAGG because the thioester intermediate of H6-SrtA and biotin-LPET is stable in SDS-PAGE. A disulfide-linked dimer of the enzyme also appears as a biotinylated protein in the blot due to the amino-terminal glycine of the enzyme, which can be biotinylated by the native peptide transpeptidation reaction at its α-amino group. (c) MALDI-TOF mass spectrum of tryptic peptides of the unpurified reaction product contains ions that correspond to the same biotinylated pilin domain peptides (ions 1 and 2) that appear in reactions run with SrtA-ELP, as well as a peptide corresponding to the amino-terminus of the H6-SrtA enzyme biotinylated through a native peptide bond at the terminal glycine residue.

Figure 27. Site-specific conjugation of FITC to Fn3. FITC can be conjugated to substrate proteins by sortase-mediated isopeptide ligation. (a) FITC-Ahx-LPETGRAGG peptide was reacted with Fn3-PLN3-ELP and SrtA- ELP overnight at 32°C in reaction buffer (50 mM Tris-HCl, 150 mM NaCl, 10 mM CaCl2). (b) The reaction product was analyzed by SDS-PAGE and total protein was visualized (upper panel). Scanning the gel with a Typhoon scanner with 488 nm laser excitation and a 520 BP 40 nm emission filter showed that Fn3-PLN3-ELP was conjugated to FITC (lower panel). SrtA-ELP is also visible in this image because the thioester reaction intermediate of SrtA and the FITC-LPET peptide is stable in SDS-PAGE. (c) The reaction product was trypsinized and the resulting peptides were analyzed by MALDI-TOF mass spectrometry. The MALDI-TOF spectrum showed the presence of two ions with m/z 77

3040.7 and 4148.7, which correspond to the predicted molecular weight of pilin domain tryptic peptides linked to FITC-Ahx-LPET through an isopeptide bond at the pilin domain lysine. As with the biotin conjugation reactions, these peptides contain a missed tryptic digestion at the pilin domain lysine because trypsin is unable to cleave at these lysines after isopeptide bond formation. (d) The chemical structure of the FITC-Ahx- LPET moiety attached to the pilin domain lysine.

3.3.5 Analysis of the isopeptide ligation product by tandem mass spectrometry

We purified the biotinylated Fn3-PLN3-ELP from the reaction mixture using cation exchange chromatography (Figure 28), then trypsinized the protein and analyzed the resulting peptides by liquid chromatography-electrospray ionization tandem mass spectrometry (LC-MS/MS). This achieved two aims: (1) conclusive identification of the isopeptide bond in the linker between biotin and the pilin domain and (2) quantitation of the average number of biotin molecules per protein by comparing the relative amounts of peptides with modified and unmodified pilin domain lysines.

Figure 28. Purification of biotinylated Fn3-PLN3-ELP using cation exchange chromatography. The biotinylation reaction of Fn3-PLN3-ELP was purified with cation exchange chromatography (CEX). (a) Absorbance at 280 nm, pH, and conductivity traces for the CEX run. The biotinylation reaction product was loaded on a HiTrap SP FF 5 ml column at 5 ml/min, washed in 20 mM Tris, pH 6.0, and eluted in a gradient to 100% 1M NaCl, 20 mM Tris, pH 6.0 over 50 minutes. Non-bound (NB) and elution (Elu) fractions were collected. (b) SDS-PAGE of the CEX load, non-bound, and elution fractions indicated that Fn3-PLN3-ELP was selectively recovered in the elution fraction. (c) A Western blot of the CEX load, non-bound (NB), and elution (Elu) fractions using streptavidin-Cy5 indicates successful recovery of biotinylated Fn3-PLN3-ELP in the elution.

The MS1 spectrum (Figure 29 and Appendix D) indicated the presence of biotinylated pilin domain peptides and other, unmodified peptides from the Fn3-PLN3-

ELP sequence, with biotinylation only observed for peptides containing the pilin domain lysine, in agreement with the results obtained using MALDI-TOF mass spectrometry. To provide direct evidence for the isopeptide conjugation of biotin to the pilin domain lysine, we fragmented the biotinylated pilin domain peptide ion by low-energy collisions with helium. The MS2 spectrum for the parent ion with z = +4 (Figure 30a) contains daughter ions formed by breaking peptide bonds in the protein backbone (the fragmentation pattern is outlined in Figure 30b). Several of these daughter ions correspond to peptides with an isopeptide junction between biotin-LPET and the pilin domain lysine (ions 1-4, with the chemical structures shown in Figure 30c). A similar fragmentation pattern was observed for the parent ion with z = +3 (not shown). The presence of ions 1-4 in the MS2 spectra for multiple charge states conclusively shows that the sortase-mediated reaction installs biotin-LPET at the pilin domain lysine through an isopeptide bond, as the daughter ions produced by tandem-MS provide an unambiguous signature that specifically confirms the identity of the biotinylated parent ion.

Figure 29. MS1 spectrum for the tryptic digest of biotinylated Fn3-PLN3-ELP. Peaks labeled in red are ions that contain biotin-LPET attached to the pilin domain lysine through an isopeptide bond. Peaks labeled in blue contain unmodified pilin domain lysines. Peaks labeled both red and blue correspond to multiple linked pilin domains (caused by missed trypsin cuts) that contain both modified and unmodified lysine residues. A peak table with detailed information about the labeled peaks can be found in Appendix D.

Figure 30. Tandem mass spectrometry (MS) of tryptic peptides of biotinylated Fn3-PLN3-ELP. LC-MS/MS confirms biotinylation at the pilin domain lysine through an isopeptide bond. (a) MS2 spectrum of daughter ions produced by isolating the +4 charge state (m/z

= 691) of a biotinylated pilin domain peptide. The spectrum shows four ions with multiple charge states whose masses confirm the expected chemical structures for biotin-LPET linked to the pilin domain lysine through an isopeptide bond (ions 1-4). These fragments correspond to y- and b-type ions produced by breaking peptide bonds along the pilin domain backbone (ions 1, 3, and 4) as well as within the LPET linker region (ions 1 and 2). (b) Outline of the observed fragmentation pattern and the nomenclature used to classify daughter ions, with colors corresponding to the daughter ion chemical structures in panel (c). Peptide bond cleavage results in a charged amino- terminal fragment (b ions) or a charged carboxy-terminal fragment (y ions). bIP and yIP correspond to daughter ions from fragmentation events within the isopeptide-linked LPET moiety.

Extracted ion chromatograms (EICs) from MS1 for peptides with biotinylated or unmodified pilin domain lysines track the intensity of these ions over time as they elute from the chromatography column. This allows the relative abundance of biotinylated versus unmodified lysines to be determined by comparing the integrated areas of their corresponding EICs. By summing the EIC intensities for all identified ions containing either biotinylated or unmodified pilin domain lysine residues, we determined that 75% of the pilin domain lysines were biotinylated, which corresponds to an average of 2.2 biotinylated pilin domains per Fn3-PLN3-ELP protein (Figure 31a). To verify this result, we utilized a colorimetric 4'-hydroxyazobenzene-2-carboxylic acid (HABA) biotin quantitation assay to determine the molar ratio of biotin to protein, which yielded a molar ratio of biotin to protein of 1.7±0.2, whereas unreacted Fn3-PLN3-ELP gave no signal above background (Figure 31b).

Figure 31. Analysis of the biotin-to-protein molar ratio for biotinylated Fn3- PLN3-ELP. (a) Extracted ion chromatograms (EICs) for individual ions were summed for peptides containing biotinylated (▬) or non-biotinylated (▬) pilin domain lysines. (b) Quantitation of the biotin:protein molar ratio using a colorimetric assay measuring HABA displacement from avidin. A chemically-crosslinked biotin-horseradish peroxidase (HRP) conjugate was analyzed as a positive control.

3.3.6 Protein-small molecule conjugates retain bioactivity

To demonstrate that our biotinylated fusion protein retained the targeting affinity of the Fn3 domain, we investigated its uptake by human umbilical vein endothelial cells (HUVECs), which express αvβ3 integrin according to flow cytometry.

Interestingly, the biotinylated FN3-PLN3-ELP could be used as a reagent in flow 84

cytometry. In this case, surface αvβ3 was detected with either a biotinylated antibody spanning the integrin dimer or the Fn3-PLN3-ELP, and streptavidin-FITC was used as a secondary stain. Fn3-PLN3-ELP biotinylated by reaction with SrtA showed fluorescence similar to that obtained with the antibody, whereas unreacted Fn3-PLN3-ELP did not give fluorescence signal above background (Figure 32)

We treated HUVECs with 100 nM biotinylated Fn3-PLN3-ELP for 30 minutes, then washed and immediately fixed, permeabilized, and stained with streptavidin-FITC to detect intracellular biotin. Cells showed significant uptake of Fn3-PLN3-ELP by fluorescence microscopy, as evidenced by punctate fluorescence in the FITC channel

(Figure 33a-c). Uptake of the biotinylated reagent appeared to be receptor-mediated, as it was dramatically reduced by saturating αvβ3 dimers with a 10-fold molar excess of unlabeled ligand during incubation of the cells with biotinylated Fn3-PLN3-ELP (Figure

33d-f).

Figure 32. Sortase A catalyzes site-specific conjugation of biotin to Fn3. Flow cytometry confirmed αvβ3 expression on HUVECs. (a) Cells were harvested using Hanks-based enzyme-free dissociation buffer and 2.5x105 cells were stained with a 1:100 dilution of 0.5 mg/ml biotinylated isotype control antibody (▬, filled histogram) or a 1:100 dilution of 0.5 mg/ml biotinylated anti-human CD51/CD61 (▬, open histogram) monoclonal antibody (eBioscience catalog number 13-0519), washed, and stained with a 1:200 dilution of 0.5 mg/ml streptavidin-FITC. (b) Flow cytometry for HUVECs stained with either biotinylated mouse IgG1 κ (▬, filled histogram), unreacted Fn3-PLN3-ELP (▬, open histogram), or biotinylated Fn3-PLN3-ELP (▬, open histogram) followed by secondary detection with a 1:200 dilution of 0.5 mg/ml streptavidin-FITC. Cells were analyzed immediately using a BD LSRII flow cytometer.

This data demonstrates that the SrtA isopeptide transpeptidation reaction can be used to produce bioactive protein-small molecule conjugates, as the reaction conditions and conjugation did not destabilize the Fn3 structure and allowed it to maintain affinity for its target antigen.

Figure 33. Cellular uptake of anti-αvβ3 Fn3 proteins biotinylated by SrtA. (a-c) HUVECs were incubated with 100 nM of purified, biotinylated Fn3-PLN3-ELP for 30 min and uptake was visualized by fluorescence microscopy. The merged image shows nuclei (blue), glycoproteins (for cell morphology, red), and intracellular biotin

(green). Punctate intracellular fluorescence in the green channel suggests that biotinylated Fn3-PLN3-ELP was internalized. (d-f) HUVECs incubated with biotinylated Fn3-PLN3-ELP and a 10-fold molar excess of unlabeled Fn3-ELP showed dramatically reduced uptake of the biotinylated ligand compared to the image in panel (b). This suggests that receptor-mediated endocytosis of biotinylated Fn3-PLN3-ELP was blocked by saturation of the αvβ3 integrin dimer with unlabeled ligand.

3.4 Significance

We have demonstrated that the heretofore unexplored isopeptide ligation function of S. aureus SrtA can conjugate multiple small molecules to a protein at internal and terminal sites when their sequences are modified to contain one or more copies of the pilin domain. This method overcomes two major limitations of the canonical native peptide ligation reaction catalyzed by SrtA, which can only conjugate a single moiety to one terminus of a polypeptide chain. The precise control over the attachment site and stoichiometry, as well as the fact that this site-specific ligation reaction targets lysine residues, provides a new bioconjugation technique that is orthogonal to available strategies and should have a variety of research and therapeutic applications. In particular, the site-specificity of the SrtA-mediated isopeptide ligation reaction could be used to make antibody-drug conjugates with improved homogeneity, which would address a significant problem for this burgeoning class of therapeutics. This application of sortase-mediated isopeptide ligation is explored in the next chapter of this document.

4. Generation of antibody-drug conjugates using sortase-mediated isopeptide ligation

4.1 Introduction

Drugs that are too toxic to administer on their own can be attached to antibodies that have high affinity for antigens or receptors on tumor cells and high specificity — and a concomitantly low affinity for healthy cells— so that they localize specifically to cancerous cells, thereby exposing diseased tissue to lethal concentrations of the toxin while sparing healthy tissue. This approach utilizes the complementary advantages of the two ADC components: the antibody —which may not be effective in cell killing on its own— provides a means to harness the extreme toxicity of the small molecule by directing it only to diseased cells.

There is enormous clinical need to develop new, improved ADCs because of their ability to specifically target and directly kill only cancer cells. Two ADCs have been approved in the United States: an anti-Her2 mAb linked to a maytansine for the treatment of Her2-positive breast cancer and an anti-CD30 mAb coupled to monomethyl auristatin E for the treatment of Hodgkin lymphoma and systemic anaplastic large cell lymphoma. Many others are in the preclinical pipeline and in clinical trials.129,130,131,132,133,134,135,136 The conjugation of highly potent cytotoxic agents to tumor selective antibodies has led to significant improvement in patient outcomes in various types of cancers, and this provides the rationale for the development of new and better

ADCs. For instance, patients unresponsive to unarmed trastuzumab with metastatic 89

disease survive nearly 25% longer when treated with toxin-conjugated trastuzumab

(Kadcyla), which amounts to a total survival time of 2.5 years.137 Similarly, the anti-CD30

ADC achieves overall response rates of 73% and 86% in Hodgkin lymphoma and systemic anaplastic large cell lymphoma (sALCL), respectively.138,139,140 In both groups of patients, the anti-CD30 ADC is used in the face of few other treatment options for these patients. Despite this, many have a complete response to the therapy, and in sALCL, the majority of patients respond. Though ADCs have demonstrated significant advantages over unarmed antibodies, they are a relatively new class of drugs with issues in their synthesis that currently limit their clinical efficacy

Available methods to produce ADCs attach a variable number of drugs to a fraction of available positions on the antibody.141,142 Both the number of attached drugs and their location on the antibody dramatically affect the ADCs affinity, specificity, and stability. If the number of attached drug molecules and their location on the mAb are not tightly controlled, it can greatly reduce the effectiveness of the ADC.143 Even small variations in the average number of attached dug molecules lead to dramatic changes in pharmacokinetics and efficacy of the ADC.144,145 As a result, early preclinical studies required significant optimization of the production process to obtain material that was consistent and scalable through clinical trials and into bulk manufacturing.146,147 New methods to make site-specific ADCs with tightly controlled stoichiometry are therefore

highly desirable, and we sought to apply sortase-mediated isopeptide ligation to this problem.

4.1.1 Potency requirements for ADC payloads

The type of drug conjugated to the antibody is an important parameter in ADC design. In one of the earliest ADCs that was developed in the late 1990’s, doxorubicin was conjugated to an antibody against the carbohydrate Lewis Y antigen that is expressed on a variety of human tumors, including 75% of breast carcinomas; however, this ADC failed to show significant efficacy in clinical trials.148,149 This failure led to a retrenchment of the ADC field, and a realization that ADCs are likely to be successful only if they carry a far more potent —and hence more systemically toxic— payload. This is because only 2-8 drugs can be loaded onto a mAb without compromising protein stability, so that achieving a therapeutically-relevant intracellular drug concentration with traditional chemotherapeutics is not feasible.150,151

The field of ADCs has enjoyed a renaissance in recent years with the discovery of improved ADC components: highly potent small molecule toxins, linkers that efficiently release active drug into targeted cells, and high affinity and high specificity mAbs that selectively target tumors but not healthy tissue. These new ADCs have been developed with drugs that have a EC50 values of < 1 nM, and are hence 100-1,000 times more potent than the traditional chemotherapeutics —such as doxorubicin, paclitaxel, and

gemcitabine— that were used in the previous generation of unsuccessful ADCs. These next-gen ADCs use highly potent drugs that belong to one of four general classes: DNA damaging agents (calicheamicins, duocarmycins, and pyrrolobenzodiazepines), tubulin inhibitors (auristatins, maytansines), RNA polymerase inhibitors (amatoxins such as α- amanitin), and protein synthesis inhibitors (bacterial and plant protein toxins, which are less commonly used due to their immunogenicity). This list of high-potency compounds is rapidly expanding as medicinal chemistry identifies new compounds and old molecules that were previously considered to be “failures” as a result of their non- specific toxicities are repurposed in ADCs.

4.1.2 Auristatins as ADC payloads

Auristatins are currently one of the gold-standard cytotoxic ADC payloads, and along with maytansines are used in the majority of ADCs in development and in the clinically approved brentuximab vedotin.152,153,154 Auristatins are a class of small molecule tubulin polymerization inhibitors derived from the marine natural product dolastatin

10, originally discovered in the 1980s.155 Extensive medicinal chemistry efforts have since improved the potency of dolastatin 10, generating a variety of derivatives such as auristatin E, auristatin F, and auristatin D, all of which kill rapidly dividing cells at sub- nanomolar concentrations, making them powerful anti-cancer agents when linked to antibodies that target tumors with high selectivity.156,157,158 Auristatin F is a peptide-like

molecule: the structure is partially natural peptide, but contains the two residues dolaisoleuine and dolaproine, which contain two additional carbons in the amino acid backbone compared to leucine and proline. Both dolaisoleuine and dolaproine contain additional hydrophobic groups compared to the natural amino acids. In dolaleuine, position C3 has a methyl ether side chain and a terminal methylated amine. In dolaproine, position C3’ contains a methyl ether side chain and position C2’ and N1 of the heterocycle are methylated (Figure 34). Additionally, the amino-terminus of auristatin F is dimethylated. These features are similar in other molecules of the auristatin class; for example, monomethyl auristatin E (used in the marketed anti-CD30

ADC) contains both dolaisoleuine and dolaproine and differs from auristatin F only in that it is monomethylated at its amino-terminus and contains a thiazole group instead of a terminal carboxylic acid group (converting the phenylalanine into a dolaphenine residue).

Figure 34. Structure of auristatin F. Auristatin F contains dolaisoleuine and dolaproine, which contain two additional carbons in the amino acid backbone compared to leucine and proline. In dolaleuine, position C3 has a methyl ether side chain and a terminal methylated amine. In dolaproine, position C3’ contains a methyl ether side chain and position C2’ and N1 of the heterocycle are methylated. The amino-terminus of auristatin F is also dimethylated.

4.2 Materials and Methods

4.2.1 Antibody production by transient transfection

Genes encoding the heavy and light chains of the murine IgG1 κ anti-human

Her2 antibody (clone 4D5) were cloned from a hybridoma obtained through the Duke

University Health System Cell Culture Facility. Unmodified antibodies and antibodies modified to contain isopeptide attachment sequences were expressed by transient transfection of HEK293 cells using the Expi293 system (Life Technologies). The procedures for cloning and expression of the antibody are described in detail in the

Results and Discussion section.

4.2.2 Antibody purification by protein G chromatography

Secreted antibodies were purified from the culture supernatant by protein G column chromatography. Protein G is expressed on the surface of group C and G

Streptococcal bacteria and functions like protein A from S. aureus, which binds to the Fc regions of antibodies to help the bacteria evade phagocytosis and complement activation.159 Protein G and A differ in their binding specificities across species, with protein G binding to a greater number of mouse IgG isotypes.

Transfected cell culture supernatant was dialyzed overnight at 4°C against PBS or diluted 1:1 in PBS. A protein G column (Pierce Protein Sciences, Thermo Fisher) was equilibrated with 5-10 column volumes of PBS and then loaded with approximately one-

third of the total process volume (16-20 ml for a 30 ml culture diluted in PBS). The other two-thirds were loaded later —in 2 separate column injections— to avoid overloading the resin. The column was then washed with 8-10 column volumes of PBS, and bound antibody was eluted with 5 column volumes of 0.1M glycine, pH 2.5. Elution was collected into a tube containing one-tenth of the elution volume of 1M Tris, pH 8.5 to immediately neutralize the low-pH elution buffer as it was collected, thereby minimizing the exposure of the antibody to low-pH.

4.2.3 Antibody-biotin conjugation using sortase

Purified antibodies at 5 µM were reacted with SrtA-ELP at 10 µM (2-fold excess over the mAb) and biotin-LPETGRAGG peptide at 500 µM (100-fold excess over the mAb) in 10 mM CaCl2, 20 mM Tris pH 7.4. After overnight incubation at room temperature, isopeptide ligation was stopped by the addition of triglycine peptide to 10

µM. Antibodies were purified from the reaction by adding NaCl to 1M, then centrifuging at 16.1 rcf for 20 minutes at 40°C to remove SrtA-ELP, followed by diafiltration against ~30 diavolumes of buffer to remove excess biotin-LPETGRAGG. The biotin:protein ratio was determined for purified antibodies using the fluorescent version of the HABA displacement and bicinchoninic acid assays (Pierce Protein Biology,

Thermo Scientific) according to the manufacturer’s instructions.

4.2.4 Auristatin F-sortase recognition peptide coupling reaction

Auristatin F (95% purity) was purchased from Concortis Biosystems, Inc.

GLPATGRAGG peptide (>90% purity) was synthesized by GenScript USA, Inc. N,N- diisopropylethylamine (DIEA or Hünig's base), 1-[Bis(dimethylamino)methylene]-1H-

1,2,3-triazolo[4,5-b]pyridinium 3-oxid hexafluorophosphate (HATU), anhydrous dimethylformamide (DMF) were all purchased from Sigma Aldrich.

1 molar equivalent of Auristatin F, 1 equivalent of HATU, and 4 equivalents of

DIEA were combined in 1 ml of DMF under argon with stirring at room temperature.

This was designed to activate the carboxylic acid of the drug without any

GLPATGRAGG peptide in the mixture to minimize the reaction of HATU with the unprotected carboxy-terminus of the peptide. After 30 minutes, GLPATGRAGG peptide in 0.5 ml of DMF was added to the reaction dropwise over the course of 1 hour. The coupling reaction was allowed to proceed for a total of 2 hours at room temperature, with stirring and under argon the entire time. The reaction was stopped by evaporating all volatile components under high vacuum. The product was washed with a 1:1 (v/v) methanol chloroform mixture, which was then evaporated under vacuum. Finally, the product was washed with water and lyophilized.

4.2.5 Antibody-auristatin F conjugation using sortase

Purified antibodies at 5 µM were reacted with SrtA-ELP at 10 µM (2-fold excess over the mAb) and AF-GLPATGRAGG peptide at ~500 µM (100-fold excess over the mAb) in 10 mM CaCl2, 20 mM Tris pH 7.4. After overnight incubation at room temperature, isopeptide ligation was quenched by the addition of triglycine peptide to

10 µM.

4.2.6 In vitro testing of antibody conjugates

SK-BR-3 human breast adenocarcinoma cells (high Her2 expression), SK-OV-3 human ovarian adenocarcinoma (high Her2 expression), MDA-MB-231 human breast adenocarcinoma (low Her2 expression), and MDA-MB-468 human breast adenocarcinoma (no Her2 expression) were obtained from the Duke University Health

System Cell Culture Facility. SK-BR-3 and SK-OV-3 were maintained in McCoy’s 5a medium containing 10% heat-inactivated fetal bovine serum and 100 units/ml penicillin and 100 µg/ml streptomycin. MDA-MB-231 and MDA-MB-468 were maintained in

Dulbecco’s Modified Eagle’s Medium/F12 (1:1) containing 10% heat-inactivated fetal bovine serum and 100 units/ml penicillin and 100 µg/ml streptomycin.

For viability assays, cells were plated in 96-well flat-bottom culture dishes at

2,250 cells/ml in 90 µl complete media and allowed to adhere for 4-6 hours, followed by treatment with auristatin F-GLPATGRAGG conjugate, 4D5 antibody, 4D5 conjugated to

auristatin F by reaction with sortase, or vehicle (20 mM Tris, pH 7.4). The plates were then incubated for 6 days at 37°C in a 5% CO2 atmosphere. The media was aspirated and replaced with 100 µl fresh media, and the viable cell number was immediately assessed by the addition of CellTiter 96 One Solution MTS/PMS viability reagent (Promega). The absorbance at 490 nm was measured after incubation at 37°C for 1-2 hours and corrected for background absorbance at 650 nm. Viable cell number was reported relative to the vehicle-treated positive control group and to a group of negative control wells containing only media.

4.3 Results and Discussion

4.3.1 Procedure for cloning antibody genes from a hybridoma

The addition of an isopeptide attachment sequence to an antibody requires modification of the antibody gene sequence, since no commercial antibodies are available that contain this sequence. To demonstrate that sortase-mediated isopeptide ligation could be used to modify an antibody, we initially chose the a monoclonal antibody raised against the extracellular domain of the human Her2 antigen, as this target is known to internalize bound antibody and has been exploited as a target by the

ADC trastuzumab-DM1 (marketed as Kadcyla).160,161 The 4D5 monoclonal antibody against human Her2 is the parental antibody of trastuzumab: the murine 4D5 was

humanized by Genentech in the early 1990s to generate the human mAb later marketed as Herceptin.162,163,164

We obtained a hybridoma producing the 4D5 mAb (ATCC CRL-10463) and cultured it in Dulbecco’s Modified Eagle’s Medium containing 10% heat-inactivated fetal bovine serum. To confirm that these cells expressed the murine IgG1 isotype antibody, a dot blot was performed on the concentrated culture supernatant. This procedure is similar to a Western blot but does not rely on separation of the proteins in the mixture by SDS-PAGE prior to blotting. Two microliters of concentrated hybridoma culture supernatant or a mouse IgG1 isotype standard were directly spotted onto a PVDF membrane and allowed to dry. Subsequently, the membrane was blocked by 1 hour incubation in 5% nonfat milk TBST, washed three times in TBST, stained with an anti- mouse IgG1-horseradish peroxidase (HRP) conjugate in 0.5% nonfat milk TBST for 1 hour at room temperature, then washed 3 times in TBST and once in TBS. Conversion of an ECL chemiluminescent substrate by the HRP indicated the presence of mouse IgG1 in the hybridoma supernatant and for the IgG1 control (Figure 35). The intensity of the spots differs as a result of a difference in antibody concentration in the hybridoma supernatant and the IgG1 standard.

Figure 35. Dot blot confirms that the CRL-10463 secretes antibody. Supernatant from a culture of hybridoma ATCC CRL-10463 (producing the mouse anti- human Her2 antibody 4D5) or a purified mouse IgG1 isotype control antibody were spotted (2 µl) on a PVDF membrane. The spots were allowed to dry, and the blot was blocked and stained with a polyclonal anti-mouse antibody linked to horseradish peroxidase. Bound antibody was visualized using ECL reagent and exposure to X-ray film. The three spots for each sample are replicates of each other. This assay is not quantitative but does confirm that the hybridoma secretes murine IgG1 antibody.

After several passages, 5 million cells were lysed by aspiration with a narrow- bore 21.5 gauge needle and RNA was purified using an RNeasy spin column from

Qiagen. cDNA corresponding to the spliced open reading frame for the heavy and light chain genes was generated using RevertAid reverse transcriptase (Thermo). The reverse transcription utilized gene-specific primers for the 4D5 heavy and light chains shown in

Table 4. From the cDNA, the heavy and light chain genes of 4D5 were amplified by PCR using the primers in Table 4 and Phusion high-fidelity DNA polymerase (New England

Biolabs). Reverse primers for the reverse transcription and the PCR were designed based on a Basic Local Alignment Search Tool (BLAST, from the National Center for

Biotechnology Information) result for the germline sequence murine IgG1 Fc domain.

100

Forward primers for the PCR reactions were based on data published on the humanization of the murine antibody.165 Restriction sites were added into the 5’ end of each primer to enable the PCR products to be cloned into a pET24 shuttle vector containing the bacterial T7 promoter and T7 terminator, which allowed complete sequencing of the antibody genes.

Table 4: Primers used to clone anti-Her2 antibody 4D5 light and heavy chains from hybridoma CRL-10463

Restriction Primer name site added Sequence (5’ to 3’) LC forward NdeI AATTTACATATGGACATCGTGATGACCCAGTCC LC reverse EcoRI TATATTGGATCCACACTCTCCCCTGTTGAAGCT HC forward NdeI AATTTACATATGCAGGTTCAGCTGCAGCAGTCT HC reverse HindIII ATTTATAAGCTTTTTACCCGGAGACAGGGAGAG

Success of this cloning procedure was initially assessed by analysis of the PCR reaction product. The amplification resulted in the production of light chain and heavy chain

DNA fragments of ~650 and ~1400 base pairs, respectively (Figure 36). These bands were gel purified, digested using the dual restriction enzyme sites that were incorporated into the ends of the fragments (Table 4), ligated into the pET24 shuttle vector, and sequenced using multiple bacterial clones and the purified PCR product.

101

Figure 36. Gel electrophoresis of PCR-amplified anti-Her2 light and heavy chain DNA. The anti-Her2 monoclonal antibody 4D5 light and heavy chains were cloned from the antibody-expressing hybridoma according to the procedure described in the text. DNA electrophoresis showed that the PCR amplification of the light and heavy chain genes matched the expected length of the sequences at ~650 and ~1400 base pairs, respectively.

4.3.2 Antibody expression in mammalian cells using transient transfection

With the 4D5 heavy and light chain genes in hand, we next required a method to use them to produce the antibody. Antibodies require eukaryotic cells for efficient production because they are glycosylated and have complex tertiary and quaternary structures that are maintained by 16 disulfide bonds. All previous proteins in our group had been produced in E. coli, and despite the fact that some reports have described the 102

production of full-length antibodies in bacteria, this was not an attractive route for antibody production because of the low productivity of these systems.166,167 The vast majority of recombinant antibodies are in Chinese Hamster Ovary (CHO) and human embryonic kidney (HEK 293) cells. Both of these types of mammalian cells can be transfected transiently to produce protein, which has the advantage of being very fast, allowing genetic-level changes to the transgene to be made in a just a matter of days. In contrast, stably-transfected cell lines take weeks or months to produce, as they require clonal expansion under growth conditions that are supported by a selectable marker (an antibiotic or an essential nutrient) and subsequent clonal screening for the efficiency of protein production. As we anticipated making many changes to the antibody sequences

(e.g. the attachment of different numbers of isopeptide attachment sequences), we chose transient HEK293 expression for our constructs. The choice between HEK293 and CHO was application-based, as we opted for the enhanced productivity of HEK293 over

CHO’s biosafety advantages (it is not an immortalized human cell line, like HEK293, which would be important if the protein was being produced for human testing).

103

Figure 37. Antibody expression cassette. pcDNA5/FRT vector from Invitrogen was modified to contain the woodchuck hepatitis virus posttranslational regulatory element (WHVPRE), and the interleukin-2 signal peptide. One of the antibody genes—light chain, heavy chain, or heavy chain with an isopeptide attachment sequence—are inserted between these two elements. The image was adapted from pcDNATM5/FRT Vector user guide, Life Technologies part number 25-0307, revision 12 September 2012.

Since transient transfection does not rely on a selectable marker or plasmid replication, we chose the pcDNA5/FRT vector from Invitrogen. Available from a previous study, this vector contains all the components required for transient transfection: a cytomegalovirus promoter and a bovine growth hormone polyadenylation signal flanking a multi-cloning site for transgene insertion (Figure 37).

104

The vector contains a pUC replication origin and an ampicillin resistance gene for propagation in bacteria, in addition to a flippase recognition target (FRT) recombination site and a hygromycin resistance gene, neither of which is used in our application.

We added two components to the pcDNA5/FRT vector to facilitate and enchance antibody production. First, the interleukin-2 (IL-2) signal peptide

MQLLSCIALSLALVTNS was added to the 5’ end of the multi-cloning site at NheI. This sequence was added during PCR of the transgene of interest, such that it was gapless and in-frame with the first codon of the transgene. A signal peptide sequence at the amino-terminus of a protein drives translation of the protein product into the endoplasmic reticulum (ER), enabling folding and glycosylation. The protein is eventually directed to the Golgi where it can undergo additional glycosylation and is then routed for export from the cell.168 Several signal peptides have been identified; though variable in sequence, they are generally composed of a 6-15 residue hydrophobic sequence flanked by a 3-5 residue polar region on the amino-terminus and another polar region on the carboxyl-terminal side that is cleaved in the ER, resulting in removal of the signal peptide from the protein. The IL-2 signal peptide is known to be a functional for protein secretion, as IL-2 is rapidly produced and secreted from T cells in response to antigenic stimulation. Though several improved signal peptides are reported in the literature, these enhance expression by only 2- or 3-fold.169,170 The body of evidence that these sequences are truly improved is limited, whereas the IL-2 signal peptide is

105

established and is used in several commercially-available mammalian expression vectors. We therefore proceeded with the wild-type IL-2 signal peptide to maximize our chances of successfully expressing the antibody.

The second element added to the expression cassette was the woodchuck hepatitis virus posttranscriptional regulatory element (WHVPRE). This 598 bp sequence is similar to a variety of sequences found in retroviruses. Most mammalian genes contain introns and it is thought that the splicing events needed to remove these introns from the mRNA promotes gene expression.171,172 In contrast, retroviral genes do not typically contain introns and are therefore expressed at low levels when transfected. It is believed that the use of posttranslational elements like the WHVPRE counters the loss of expression resulting from the lack of introns in the viral genes, likely by providing secondary structure that increases transcript termination and improves 3’ transcript cleavage and polyadenylation.173 This does not affect the half-life of the mRNA, but the improved termination and processing leads to an increased amount of transcribed mRNA in the cell, which is believed to directly drive increased translation.174 Notably, inclusion of the WHVPRE has been shown to increase the expression of transgenes by

>10-fold, which is a similar increase to that observed when an intron is included in the transgene. Importantly though, the WHVPRE and introns are not mutually exclusive, suggesting that inclusion of an intron in our antibody gene sequences could further improve expression levels, though this is an only an issue of yield.

106

With the plasmids constructed, we transfected the cells according to the procedure shown in Figure 38. The day before transfection, the cells are seeded at a density of 2 million cells/ml in 30 ml of fresh Expi293 media (an Invitrogen formulation designed to sustain the high densities achieved by the Expi293 cells). The next day, the cell density was diluted to 2.9 million cells/ml in a total culture volume of 25.5 ml. The transfection reagent—Expifectamine, which according to the manufacturer is optimized for transfection of high-density suspension cell cultures—was diluted 1:18 in Opti-MEM low serum media in a total volume of 1.5 ml and allowed to equilibrate for 5 minutes at room temperature. To this, 30 µg of DNA was added in a volume of 1.5 ml Opti-MEM, mixed gently, and allowed to complex with the transfection reagent for 20 minutes at room temperature. The DNA:Expifectamine complexes were then added to the cells. 20-

24 hours after transfection, Expi293 enhancers 1 and 2 were added to the culture. The formulation of these components is not disclosed, but we believe that one of the two may contain valproic acid, which is believed to act as a histone deacetylase inhibitor and has been shown to dramatically increase (by a factor of ~10) the levels of transgene mRNA and protein produced in transiently transfected CHO and HEK293 cells.175,176,177

The other enhancer is likely a nutritional supplement —a glucose and amino acid bolus— that would help support the high cell density culture for up to 1 week after transfection.

107

Figure 38. Timeline for transient transfection of Expi293 cells for antibody production.

Production of the antibody was assessed by SDS-PAGE of the culture supernatant (Figure 39). Under non-reducing conditions (no 2-mercaptoethanol in the loading buffer), the antibody should migrate slightly higher than 150 kDa (since it is a disulfide-linked tetramer, it does not run as a linear protein in SDS-PAGE, and the apparent molecular weight is overestimated). Under reducing conditions, the antibody breaks into its light and heavy chains. In the 4D5 mAb, both light and heavy chains are glycosylated at asparagines 67 and 293, respectively, and therefore run heavier than 25 kDa and 50 kDa molecular weights that would be predicted based on their amino acid sequence. Prior to transfection, no antibody bands appeared in the culture supernatant under non-reducing or reducing conditions; only background proteins were present. In contrast, 3 days after transfection the supernatant contained abundant antibody, evident from a ~150 kDa band under non-reducing conditions and ~50 kDa and ~25 kDa bands 108

under reducing conditions. The quantity of antibody is expected to increase as the culture is incubated further. Harvesting the protein at 5 to 7 days consistently recovered

30-50 mg per liter of fermentation after protein G purification.

Figure 39. Antibody production using transient transfection. SDS-PAGE analysis of the Expi293 culture supernatant before transfection did not show any antibody at ~150 kDa under non-reducing conditions, or at ~50 kDa and ~25 kDa under reducing conditions. 3 days after transfection, abundant antibody had been secreted into the culture media. The identity of these bands was confirmed by Western blot using an anti-mouse IgG1 polyclonal antibody raised against the heavy and light chains. Antibody was purified using protein G chromatography columns as described in the Materials and Methods section. The left two images are coomassie-stained gels, whereas the rightmost image is a stain-free gel where the band intensity is proportional to the number of tryptophans, therefore, the intensity of the light chain is less than that of the heavy chain.

109

Use of the azurocidin signal peptide (MTRLTVLALLAGLLASSRA) improved yields further. Azurocidin is a protein that is secreted by neutrophils in response to bacterial infection that serves as a chemoattractant for inflammatory monocytes.178,179

This signal peptide is known to give high productivity of antibodies and proteins secreted from CHO and insect cells.180,181 Though this is not the only signal peptide that has been shown to function well for recombinant antibody secretion, we tested it because it was predicted to function much better with our antibodies than the IL-2 signal peptide. This is based on use of the SignalP 4.1 server, which uses a training set of sequences from proteins known to be secreted, non-secreted, or transmembrane to predict the efficiency of signal peptide cleavage.182,183,184 Eukaryotic signal peptides direct ribosomal translation to the endoplasmic reticulum (ER) membrane, causing the protein to be expressed into the ER and enabling folding, disulfide bond formation, glycosylation, and secretion via the Golgi apparatus.185 Secreted proteins are expressed into the ER as membrane-bound proteins that are anchored by the signal peptide. Thus, for the protein to be secreted, the signal peptide must be cleaved to generate the native

N-terminus of the protein. Signal peptide cleavage is dictated by its C-terminal region, which typically contains polar, helix-breaking residues (particularly G and P) and by the

N-terminus of the native protein, which typically contains an amino acid with a short side chain in the first position.186

110

The SignalP 4.1 software scores the likelihood of signal peptide cleavage for a given signal peptide/protein combination. The SignalP report for the murine antibody light and heavy chains with the IL-2 signal peptide is shown in Figures 40 and 41, respectively. The software predicts a relatively low cleavage score (0.5) for both the heavy and light chains, which suggests that the IL-2 signal peptide is not an efficient signal peptide for this antibody. Moreover, potential alternative cleavage sites with cleavage probabilities that are similar to the primary site are identified (3 in the light chain and 1 in the heavy chain). Conversely, the SignalP results for the azurocidin signal peptide predicts that this signal peptide will be removed with high efficiency (0.8) only at the correct site for both the light and heavy chains (Figures 42 and 43). Simply by replacing the IL-2 signal peptide with the azuorcidin signal peptide in our dual-vector transfection approach, we were able to increase the expression level of antibody by a factor of ~10 (from ~40 mg/ml to ~400 mg/ml).

111

Figure 40. Predicted signal peptide cleavage for the murine anti-Her2 antibody light chain with the IL-2 signal peptide. The SignalP 4.1 server was used to predict the signal site and efficiency of signal peptide cleavage. The “C-score” indicates the probability of cleavage at a particular site. The “S- score” indicates the probability that a given residue is in a signal peptide. The “Y-score” is the geometric mean of the C- and S-scores. For the IL2SP-mLC, SignalP predicts 4 possible signal peptide cleavage sites, all of which are cleaved inefficiently (Y-scores < 0.5).

112

Figure 41. Predicted signal peptide cleavage for the murine anti-Her2 antibody heavy chain with the IL-2 signal peptide. The SignalP 4.1 server was used to predict the signal site and efficiency of signal peptide cleavage. The “C-score” indicates the probability of cleavage at a particular site. The “S- score” indicates the probability that a given residue is in a signal peptide. The “Y-score” is the geometric mean of the C- and S-scores. For the IL2SP-mHC, SignalP predicts 2 possible signal peptide cleavage sites, both of which are cleaved inefficiently (Y-scores < 0.5).

113

Figure 42. Predicted signal peptide cleavage for the murine anti-Her2 antibody light chain with the azurocidin signal peptide. The SignalP 4.1 server was used to predict the signal site and efficiency of signal peptide cleavage. The “C-score” indicates the probability of cleavage at a particular site. The “S- score” indicates the probability that a given residue is in a signal peptide. The “Y-score” is the geometric mean of the C- and S-scores. For the AZSP-mLC, SignalP predicts one high efficiency signal peptide cleavage site (Y-score = 0.8).

114

Figure 43. Predicted signal peptide cleavage for the murine anti-Her2 antibody heavy chain with the azurocidin signal peptide. The SignalP 4.1 server was used to predict the signal site and efficiency of signal peptide cleavage. The “C-score” indicates the probability of cleavage at a particular site. The “S- score” indicates the probability that a given residue is in a signal peptide. The “Y-score” is the geometric mean of the C- and S-scores. For the IL2SP-mLC, SignalP predicts one high efficiency signal peptide cleavage site (Y-score = 0.7).

It is important to note that the productivity of the cells could also be increased by improving the expression vector design. The flippase recognition target (FRT) recombination site and the hygromycin resistance gene are both not necessary for transient protein expression and could be removed. This would increase the number of copies of the plasmid in each transfection, since the amount of DNA used is based on mass (18 µg of heavy chain plasmid and 12 µg of light chain plasmid per 30 mL culture).

This would likely improve antibody expression by increasing the number of plasmids 115

delivered to each cell. The ~400 mg/L expression achieved with the azurocidin signal peptide was sufficient for our applications, but it is worth noting that this could likely be further improved.

4.3.3 Conjugation of biotin to an engineered monoclonal antibody

Having characterized the isopeptide ligation reaction with the Fn3 domain, we next sought to generalize this reaction by modifying a more complex protein. To examine the feasibility of isopeptide ligation to generate antibody-drug conjugates, we conjugated biotin to the 4D5 monoclonal antibody genetically modified to contain a pilin domain at the carboxy-termini of its heavy chains. The recombinant antibody was incubated with SrtA-ELP and biotin-LPETGRAGG peptide overnight (Figure 44a), and the product was compared to a reaction that used the anti-Her2 antibody without a pilin domain. An anti-biotin Western blot indicated that only the heavy chain of the antibody that contained the pilin domain was modified in the reaction (Figure 44b).

116

Figure 44. Sortase modifies a mouse antibody only at the pilin domain. Two mouse IgG1 antibodies were tested in parallel: the anti-Her2 antibody 4D5 and the same antibody with a pilin domain at each heavy chain C-terminus. (a) Schematic of the biotinylation reaction catalyzed by SrtA-ELP. (b) Reaction products were analyzed by a dual-color Western blot using FITC-labeled polyclonal anti-mouse IgG and a Cy5- labeled streptavidin to detect biotinylated protein. Total antibody staining (left panel) shows the positions of the non-reduced antibody and the reduced heavy and light chains on the blot. The right panel shows that only the non-reduced antibody containing the pilin domains is biotinylated. This modification of the antibody maps only to the heavy chain when the sample is reduced, which suggests site-specificity for the pilin domain.

117

Site-selectivity for the pilin domain was also assessed using control reactions that contained a panel of murine antibodies. Based on an anti-biotin Western blot (Figure 45),

SrtA-ELP did not biotinylate any sites on any of 4 murine IgG isotype controls (IgG1,

IgG2a, IgG2b, or IgG3), or either class of murine light chain (kappa or lambda). As before, the anti-Her2 antibody was not biotinylated when it did not contain the pilin domain. Only the anti-Her2 heavy chain modified to contain the pilin domain was biotinylated by SrtA-ELP, which —consistent with our data for the Fn3 domain— strongly suggests that isopeptide ligation is specific for this amino acid sequence. The specificity of this reaction is remarkable, as the proteins tested herein contain many potential off-target sites for conjugation (86 lysines in the case of the anti-Her2 mAb).

118

Figure 45. Sortase-mediated isopeptide ligation can be used to modify monoclonal antibodies (mAbs). An anti-Her2 mAb genetically modified to contain the pilin domain at each heavy chain C-terminus was reacted with biotin-LPETGRAGG peptide and SrtA-ELP overnight and compared to other murine antibodies across all isotypes. (b) SDS-PAGE and an anti- biotin Western blot indicate that the pilin domain is required for biotinylation of antibodies. Non-reduced anti-Her2 containing the IPA sequence on its heavy chain is biotinylated in the reaction and this modification maps exclusively to the heavy chain when the antibody’s interchain disulfide bonds are reduced prior to electrophoresis. No biotinylation is observed for reactions on a panel of antibodies including anti-Her2 without a pilin domain, as well as all murine IgG isotypes and both kappa and lambda murine light chain variants.

119

To determine the temperature dependence of the antibody modification reaction, we ran 7 parallel reactions on the anti-Her2 substrate containing one pilin domain on each heavy chain C-terminus. An anti-biotin Western blot indicated that maximum biotinylation was achieved between at 37°C, with measurable biotinylation at all temperatures tested (Figure 46).

Figure 46. Temperature dependence of the antibody modification reaction. Anti-Her2 antibodies containing one copy of the pilin domain at each heavy chain C- terminus were reacted with SrtA-ELP and biotin-LPETGRAGG across a range of temperatures. SDS-PAGE was stained for total protein, showing that the protein load in each reaction was the same. An anti-biotin Western blot of the same SDS-PAGE gel showed maximum protein biotinylation at 37°C.

The biotinylated antibody also retained antigen targeting, as it showed efficient labeling of Her2 on SK-OV-3 human ovarian adenocarcinoma cells (Figure 47a-c) compared to a biotinylated isotype control antibody (Figure 47d-f) in immunofluorescence microscopy. Using a HABA displacement assay, we determined

120

that there were 1.8 ± 0.1 biotins conjugated per antibody out of 2 possible pilin domain attachment sites (Figure 48). This equates to a conversion of 90%, which is consistent with reported yields for other enzymatic reactions at the carboxy-termini of antibody heavy chains.187 This assay uses avidin that is covalently linked to Dylight 493. The

Dylight fluoresces weakly because in the absence of biotin the quencher molecule HABA is weakly bound to the avidin in the biotin-binding pockets. The proximity of HABA quenches the Dylight molecules through FRET. Adding protein that is conjugated to biotin to the assay displaces the HABA and eliminates FRET, which results in an increase in the fluorophores’ emission that can be measured and compared to a standard curve to determine the number of moles of biotin in the sample. Since the number of moles of protein in the same sample can be accurately determined using the BCA assay, this allows the biotin:protein molar ratio to be calculated.

121

Figure 47. Biotin-conjugated 4D5 anitbody retains antigen targeting. (a-c) Immunofluorescence of the Her2-overexpressing cell line SK-OV-3 using biotinylated anti-Her2 containing a pilin domain on its heavy chain followed by secondary staining with streptavidin-FITC conjugate. The level of green staining confirms the expected strong Her2 expression of these cells, and the ability of the biotinylated 4D5 antibody to bind Her2. (d-f) Immunofluorescence of SK-OV-3 using a biotinylated isotype control antibody in the primary staining step shows the level of non-specific background staining. Blue = nuclear stain, green = FITC, scale bars are 15 µm.

122

Figure 48. Quantification of the biotin-to-protein ratio in a monoclonal antibody modified by isopeptide ligation. Quantitation of the biotin-to-protein molar ratio for anti-Her2 with one pilin domain inserted at the carboxy-terminus of each heavy chain. Antibody was biotinylated by overnight reaction at 32°C with SrtA-ELP and biotin-LPETGRAGG peptide in 2- and 100-fold molar excess respectively. Purified antibody was dialyzed extensively (30 diavolumes) with PBS to remove unreacted biotin. Protein-bound biotin was quantified using a fluorescence biotin detection kit with a known concentration of biocytin as a standard, according to the manufacturer’s instructions. Protein concentration was determined by BCA assay using a bovine serum albumin standard per the manufacturer’s instructions. Biotin and protein content were evaluated for undiluted antibody as well as 4 dilutions of the conjugate (1:4, 1:6, 1:8, and 1:10). Plotting the concentration of biotin versus the concentration of protein for each sample gives a slope of 1.8 with a standard error of 0.1, which indicates that 90% of the pilin domains in the antibody sample were biotinylated in the isopeptide ligation reaction. All samples and standards were measured in triplicate and error bars indicate standard deviations.

123

4.3.4 Stability of isopeptide-linked antibody conjugates

To assess the stability of the isopeptide bond and the peptide linker between the antibody and the cargo molecule, we produced biotin-mAb conjugates and incubated them at 37°C for 10 days with either cells or complete media containing 5% fetal bovine serum. We assessed the biotin-to-antibody ratio over time by performing a dual-color

Western blot on the culture supernatant in which we simultaneously stained with an anti-murine IgG1 conjugated to Alexa 488 to quantify total mouse antibody and streptavidin-Cy5 to quantify the total biotinylated protein. a cell line expressing high levels of Her2 (SK-BR-3), there was a statistically-significant decline in the biotin-to- antibody ratio at days 6 and 10 (p<0.05, Figure 49a and 49c), whereas biotinylated antibody incubated in the complete media without cells did not show a substantial loss of biotin. This suggests that antigen-mediated internalization of the biotinylated antibody was required to release the cargo. This is consistent with the approach used for both clinically approved antibody drug conjugates and the majority of ADCs in preclinical development, which use chemically stable linkers consisting of either thioethers or peptide bonds that are cleaved specifically by cathepsin D. In both cases, active drug is released only when the linker or attachment site in the antibody is degraded by endosomal and lysosomal enzymes.188,189,190,191,192,193 Interestingly, when the same experiment was conducted with Her2-negative MDA-MB-468 cells, the biotin:antibody ratio did not decline over time and there was no difference between the

124

group incubated with cells and the group incubated in media alone at any of the time points tested (Figure 49b and 49d). This suggests that the linkage between the antibody and the cargo is highly stable in the context of antigen-negative cells.

125

Figure 49. Antibodies conjugated to small molecules using isopeptide ligation are stable in vitro. Anti-Her2 was biotinylated at the pilin domains on its heavy chain C-termini and was incubated at 37°C in either cell culture media containing 5% fetal bovine serum or the same media inoculated with cells. The biotin-to-antibody ratio in the culture supernatant was assessed by simultaneously staining for total biotinylated protein and total antibody in a dual-color Western blot. Representative Western blots for biotinylated antibody incubated with (a) Her2-positive SK-BR-3 cells or their complete media and (b) Her2-negative MDA-MB-231 cells or their media. Relative amounts of biotin and antibody at each time point were quantified by analyzing the band intensities by densitometry and the results were normalized to day 0. Quantified results for (c) SK- BR-3 and (d) MDA-MB-468. Error bars are based on triplicate measurements. * indicates statistical significance at p<0.05 by two-tailed t-test.

126

4.3.5 Synthesis of auristatin F linked to a sortase substrate peptide

Auristatin F was linked to a sortase recognition peptide to enable its use in the sortase isopeptide ligation reaction. We chose to use auristatin F as an ADC payload because of its commercial availability and because its structure allows for simple chemical modification. We attached a synthetic peptide containing the LPATG sortase recognition sequence to the carboxylic acid of auristatin F via a peptide coupling reaction. We changed the glutamic acid in the LPETG sequence to alanine because this simplifies the chemistry of the peptide coupling reaction by eliminating the potential side reaction at the carboxylic acid group present in the glutamic acid side chain.194

Importantly, the LPATG is known to be active in the sortase reaction.195

We followed a reaction scheme similar to that used in the manufacture of chemically-synthesized peptides. This mechanism has been used to attach a variety of two amino acid linkers to the carboxylic acid of auristatin F with yields ranging from 10-

60%, and to attach a primary amine in the maytansine DM-1 to a polyethylene glycol linker with yields of 75%.196,197 However, the auristatin F reactions involved peptide bond formation between two bulky residues—the phenylalanine on auristatin F and residues with large side chains such as methionine, isoleucine, and tyrosine on the peptides— which likely resulting in low reaction conversions because of a high degree of steric hindrance. We therefore designed our peptide with an amino-terminal glycine residue to minimize the steric hindrance around the nucleophilic α-amino group. The complete

127

peptide sequence was GLPATGRAGG; the GRAGG is cleaved off by sortase and the sequence of this segment is therefore somewhat arbitrary, though we designed it to be consistent with the peptide sequence used earlier with biotin and FITC, and hoped that the arginine would improve the water solubility of the otherwise hydrophobic peptide.

The scheme of the reaction used to couple auristatin F and the GLPATGRAGG peptide is shown in Figure 50. We reacted auristatin F at its carboxylic acid by reaction with 1 molar equivalent of 1-[Bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5- b]pyridinium 3-oxid hexafluorophosphate (HATU) for 30 minutes in anhydrous dimethylformamide (DMF) in the presence of 4 equivalents of N,N- diisopropylethylamine (DIEA or Hünig's base), a strong base that would deprotonate the auristatin F carboxylic acid. The reaction of the carboxylic acid with HATU converts it to an activated ester, which is then subject to nucleophilic attack by the amino- terminus of the GLPATGRAGG peptide in the next step.

Because the peptide has an unprotected carboxylic acid, it could in principle also be activated by HATU. We attempted to minimize this by pre-incubation of equimolar amounts of HATU and auristatin F, and added the peptide to the reaction dropwise over an hour to keep the auristatin F in as high an excess to the peptide as possible. In addition, only 0.9 equivalents of peptide were reacted, as a small amount of unreacted auristatin F would not affect the downstream sortase reaction and would be removed from the product of that reaction (along with any unreacted drug-peptide conjugate) by

128

diafiltration. The peptide coupling reaction was allowed to proceed for a total of two hours, the solvent was removed under vacuum, and the product was then reconstituted in water and lyophilized. From 10 mg of starting auristatin F and 9 mg of

GLPATGRAGG peptide, a total of 17 mg of product was recovered. We analyzed the product of the reaction by MALDI-TOF mass spectrometry (Figure 51). The spectrum confirmed that the product had a molecular mass that matched the expected value for

AF-GLPATGRAGG. Importantly, no unreacted peptide was present in the reaction product, which would inhibit the downstream sortase reaction by labeling substrate protein with GLPAT peptide with no attached drug molecule.

129

Figure 50. Reaction scheme for peptide coupling of auristatin F to the sortase recognition peptide GLPATGRAGG The carboxylic acid of auristatin F was activated by reaction with an equimolar amount of HATU in DMF for 30 minutes at room temperature in the presence of a 4-fold molar excess of the non-nucleophilic base DIEA. This step activates the carboxylic acid by conversion to a reactive ester with an electron-withdrawing leaving group. Adding the GLPATGRAGG peptide dropwise over 1 hour kept the activated auristatin F in as large an excess as possible, because the carboxylic acid on the peptide was not protected and could have in principle reacted with HATU and then itself, forming a peptide dimer. Once all the peptide was added, the reaction was allowed to proceed for 1 additional hour at room temperature before solvent evaporation under vacuum. 130

Figure 51. MALDI-TOF mass spectrum of auristatin F linked through a peptide bond to GLPATGRAGG peptide. The carboxylic acid of auristatin F and the amino-terminus of a GLPATGRAGG peptide were joined by a peptide coupling reaction. MALDI-TOF mass spectrometry indicated that the product of this reaction had molecular weight that matched that of the predicted product.

4.3.6 Conjugation of auristatin F to a model targeting protein

The tryptic digest of an unlabeled antibody is extremely complex and difficult to analyze by mass spectrometry, as it requires specialized equipment, software, and expertise. The situation becomes even more complex for antibody-drug conjugates, which give rise to additional tryptic peptides containing the covalently-bound small molecules.

We sought to quickly confirm that the AF-GLPATGRAGG reagent that we had produced was functional in the sortase reaction. This was not guaranteed, since

131

auristatin F had never been used in a sortase-based reaction and we needed to confirm that the LPATG sequence —rather than LPETG— was still functional in isopeptide ligation, though we expected that it would be given that the peptide sequence only affects the formation of the enzyme thioester intermediate, and would not affect nucleophilic attack on this intermediate.

To test the reaction with AF-GLPATGRAGG, we conjugated it to the Fn3-PLN3-

ELP protein that we had previously employed for biotin conjugation. We used the same conditions that were used for biotin conjugation: overnight incubation at 32°C and enzyme, Fn3-PLN3-ELP, and AF-GLPATGRAGG in a 2:1:10 molar ratio. We trypsinized the reaction product and analyzed the resulting peptides by MALDI-TOF mass spectrometry. The mass spectrum (Figure 52) indicated the presence of two ions at m/z

3264.7 and 4370.4 that correspond specifically to AF-GLPAT linked to the pilin domain lysine through an isopeptide bond (with a missed trypsin digestion at the pilin domain lysine). As with biotin, modification of the first copy of the pilin domain in the protein produces the ion at m/z 4370.4 and modification of the second and third copies of the pilin domain produce the same ion at m/z 3264.7, as a result of the fact that the tryptic peptide of the first of pilin domain retains some of the linker between itself and the Fn3 protein. The presence of these ions in the spectrum confirms that the AF-GLPATGRAGG reagent is functional in the sortase isopeptide ligation reaction, just as dabcyl-LPETG- edans, biotin-LPETGRAGG, and FITC-LPETGRAGG were.

132

Figure 52. Auristatin F can be conjugated to proteins using sortase-mediated isopeptide ligation. (a) Schematic of the reaction, which used the same conditions that were used in the biotin and FITC conjugation reactions. The Fn3-PLN3-ELP substrate protein was incubated with 2- and 10-fold molar excesses of SrtA-ELP and AF-GLPATGRAGG overnight at 32°C. (b) MALDI-TOF mass spectrum of the tryptic digest of the reaction product shows the modification of the pilin domain lysines with AF-GLPAT based on the appearance of the ions at m/z 3264.7 and 4370.4, which both contain a missed digestion at the modified pilin domain lysine. 133

4.3.7 Conjugation of auristatin F to an engineered anti-Her2 monoclonal antibody

With the knowledge that the AF-GLPATGRAGG reagent was functional in the isopeptide ligation reaction, we next attempted to use the reaction to conjugate auristatin F to the anti-Her2 monoclonal antibody 4D5 that had been engineered to contain one copy of the pilin domain at carboxy-terminus of each heavy chain (Figure

53). The reaction with auristatin F proceeded under the same conditions that were used to attach biotin to the antibody. Antibody, enzyme, and AF-GLPATGRAGG were mixed in a 1:2:100 molar ratio in 10 mM CaCl2 and 20 mM Tris, pH 7.4. After overnight reaction at room temperature, SrtA-ELP was removed by adding NaCl to 1M and centrifuging the reaction product for 20 minutes at 16.1 rcf and 40°C. Excess AF-GLPATGRAGG was then removed by diafiltration against 30 diavolumes of 20 mM Tris, pH 7.4.

134

Figure 53. Schematic for the conjugation of auristatin F to the 4D5 monoclonal antibody. Sortase was used to conjugate AF-GLPAT to the isopeptide attachment sequences at the termini of the 4D5 heavy chains. The reaction was incubated at 32°C for 24 hours and used antibody, enzyme, and AF-GLPATGRAGG in a 1:2:100 molar ratio. The reaction was quenched by adding triglycine to 10 mM. SrtA-ELP was removed by adding NaCl to 1M to trigger the ELP phase transition and centrifuging the reaction product at 16.1 rcf for 20 minutes. The supernatant was then collected and diafiltered to remove unreacted auristatin F.

We tested the cytotoxicity of the anti-Her2 ADCs made using isopeptide ligation on a panel of human breast adenocarcinoma cell lines expressing a range of Her2 levels:

SK-BR-3, which expresses high levels of Her2; MDA-MB-231, which expresses very low 135

but detectable Her2; and MDA-MB-468, which does not express detectable levels of

Her2.198,199 These cell lines were inoculated in 96-well plate format at 2,250 cells/ml, then treated with AF-GLPATGRAGG conjugate, 4D5 antibody, 4D5 conjugated to auristatin F by reaction with sortase, or vehicle (20 mM Tris, pH 7.4). After a 6 day incubation, the media was aspirated, replaced with fresh media, and the number of viable cells relative to relative to the vehicle-treated positive control group and a group of negative control wells containing only media was determined by the addition of by a colorimetric MTS assay.

For SK-BR-3 cells, which express a high level of Her2, the anti-Her2-AF ADC had an EC50 (defined as the concentration that kills 50% of the cells) of 0.5 nM (Figure 54).

This EC50 is similar to those obtained with ADCs targeting Her2 that use both Auristatin and Maytansine drug payloads.200,201,202 Additionally, the ADC’s EC50 of 0.5 nM was 36- fold lower than the EC50 for the AF-peptide (18 nM), which indicates that the cytotoxicity of the ADC was target-specific and was not due to non-specific uptake of conjugate by macropinocytosis. In cells with low antigen expression (MDA-MB-231), the ADC had an

EC50 of 23 nM that was similar to the EC50 obtained using the non-conjugated AF- peptide. The free drug and the ADC also had similar cytotoxicity towards Her2-negative

MDA-MB-468 cells, with EC50 values of 33 nM and 23 nM, respectively. This non-specific toxicity is typically observed for ADCs when they are applied to cells at high (10-1000 nM) concentrations in vitro. This non-specific killing in vitro is not a feature that is

136

unique to ADCs against Her2, as it has also been observed for a variety of targets including lymphocyte antigen 6 locus E (Ly6E)203, tumor-associated calcium signal transducer 2 (M1S1)204, CD70205, mesothelin206, and CD37.207 It is also not unique to the auristatin class of drugs, as it has also been observed in ADCs of other drugs such as duocarmycins208, maytansines209, and cryptophycins210.

137

Figure 54. anti-Her2 conjugated to auristatin F using IPL is cytotoxic towards antigen-expressing cell lines in vitro. Auristatin F was conjugated to the anti-Her2 mAb by SCIL and cells were treated with the purified reaction product or free drug for 6 days. The anti-Her2 mAb was engineered to contain one copy of the isopeptide ligation sequence at the carboxy- terminus of each heavy chain. (a) Her2-positive SK-BR-3 cells are killed by the ADC at concentrations 2 orders of magnitude lower then when the cells were treated with drug- GLPATGRAGG peptide that was not attached to the antibody. (b) MDA-MB-231 and (c) MDA-MB-468 (expressing low levels of Her2) are killed equally well by the ADC and drug-peptide. Together, these data indicate that the ADC maintains its antigen-targeting capability, kills antigen-positive cells efficiently, and kills cells expressing low levels of antigen through non-specific uptake.

4.4 Significance

We have extended the isopeptide ligation method to modify full-length monoclonal antibodies, which represents a significant increase in complexity over the peptide and protein substrates in our previous studies. We have demonstrated that

138

isopeptide ligation is site-specific for pilin domain sequences when engineered into an anti-Her2 antibody that contains 86 other lysines. Our data suggests that these antibody conjugates are stable in cell culture and are degraded only in the presence of antigen- positive cells. We have also applied this reaction to attach a clinically-relevant drug to the anti-Her2 antibody to create ADCs with potent cytotoxicity towards human cancer cell lines.

5. Analysis of the sequence requirements for isopeptide ligation at the pilin domain

5.1 Introduction

We have demonstrated that isopeptide ligation can be used to site-specifically modify only pilin domain lysine residues in the context of GLP-1, an anti-αvβ3 integrin

Fn3 domain, and an anti-human Her2 murine IgG1 κ. The pilin domain that we used— the 11 amino acid sequence WLQDVHVYPKH—was selected based on its homology in the pilin monomer proteins of C. diphtheriae and A. naeslundii. We next sought to determine whether this sequence could be shortened, since this would make it easier to engineer into accessible loops in protein structures (e.g. within the constant domains of antibodies). To identify the minimal sequence length of the pilin domain, we constructed

Fn3-PLN-ELP fusion proteins with a panel of PLN lengths. We also determined the site of reaction in TRAIL-ELP fusion proteins, which we found were efficiently labeled by sortase even without a pilin domain sequence. Finally, we determined that isopeptide 139

ligation competes with N-terminal native peptide ligation using a peptide array.

Interestingly, we found that native peptide ligation occurs at all N-terminal amino acids, whereas the literature suggests (without evidence) that only N-terminal glycine is functional in the sortase native peptide ligation reaction.

5.2 Materials and Methods

5.2.1 Molecular biology

We first constructed the gene coding for Fn3-PLN (one copy of the full-length pilin domain) using pRE-RDL in a pET24b vector. Deletion mutants were then constructed by polymerase chain reaction according to the workflow in Figure 55. Non- phosphorylated primers were designed flanking the sequence to be deleted that resulted in amplification of the rest of the Fn3-PLN sequence and the vector. After PCR with high-fidelity Phusion polymerase (New England Biolabs), the long dsDNA that was generated was gel-purified and the 5’ ends were phosphorylated with T4 polynucleotide kinase for 30 minutes at 37°C. The phosphorylated DNA was directly re-circularized using NEB quick-ligase under conditions that promote re-cicularization (high ligase concentration, polyethylene glycol in the ligation buffer, and <50 ng/ul DNA concentration). Colonies were then screened and the Fn3-mutant PLN was ligated to the gene coding for ELP (VPGVG)60.

140

Figure 55. Procedure for mutagenesis by polymerase chain reaction. A plasmid containing Fn3-PLN was amplified with primers flanking the site to be deleted. (a) Example of a deletion reaction where codons for the first 7 amino acids (WLQDVHYV) are deleted from the pilin domain. (b) Example of a substitution reaction where the codon for lysine is changed to a codon for methionine. The ATG codon is initially mismatched with the lysine codon but is extended in the PCR reaction, resulting in a substitution. The same reaction were used to generate the TRAIL variants described later in the text. PCR used Phusion polymerase from New England Biolabs. PNK (T4 polynucleotide kinase) and T4 ligase (quick reaction kit) were also from New England Biolabs.

5.2.2 Sortase reactions

Substrate proteins were reacted at 5 µM with SrtA-ELP at 10 µM and biotin-

LPETGRAGG or FITC-LPETGRAGG peptide at 500 µM in reaction buffer (0.5 mM Tris-

HCl, 10 mM CaCl2, pH 8.5). All reactions were allowed to proceed overnight at 37°C.

141

5.2.3 Electrophoresis, Western blotting, and mass spectrometry

SDS-PAGE was conducted using Bio-Rad anyKd stain-free gels. These gels were run under denaturing conditions (25 mM Tris, 192 mM glycine, 0.1% sodium dodecyl sulfate) with reducing (Laemmli buffer + 5% β-mercaptoethanol) or non-reducing

(Laemmli buffer without β-mercaptoethanol) sample buffers, depending on the protein being analyzed.

Western blotting was performed by transferring SDS-PAGE products to Bio-Rad

Trans-Blot Turbo PVDF membranes. Blots were blocked in 5% nonfat milk in Tris- buffered saline + 0.1% Tween-20 (TBST buffer) for 1-1.5 hours at room temperature with agitation, then washed 3 times for 5 minutes in TBST. Blots were then stained in 0.5% nonfat milk TBST containing a 1:20000 dilution of streptavidin-Cy5 conjugate (1 mg/ml stock).

MALDI-TOF mass spectrometry was performed using an Applied Biosystems

Voyager-DE Pro system fitted with a nitrogen laser. All results were obtained using α-

Cyano-4-hydroxycinnamic acid matrix in a 40:1 (v/v) ratio with the analytes.

5.2.4 Pepspots array

Peptide arrays were purchased from JPT Peptide Technologies GmbH (Berlin,

Germany). Peptides are synthesized on nitrocellulose membranes such that the C- terminus of each peptide is covalently linked to the membrane. The N-terminal 15 amino

142

acids of the heavy and light chains of trastuzumab (human anti-human Her2, IgG1 isotype) were synthesized, as well as variants of these peptides with all other amino acids except cysteine substituted at the N-terminal position. Each spot on the array therefore corresponds to a unique peptide. The array was rinsed with methanol, then washed 4x5 min with TBST buffer, then blocked in 5% nonfat milk TBST for 2 hours at room temperature with gentle agitation. The blocked membrane was then washed 3x5 min in TBST, then 3x5min with TBS and incubated overnight in a sortase reaction (5 mM

SrtA-ELP, 10 mM CaCl2, 500 µM biotin-LPETGRAGG) at 32°C with gentle agitation.

After the reaction, the blot was washed 4x in TBST then stained for 1 hour at room temperature in 0.5% nonfat milk TBST containing a 1:20,000 dilution of streptavidin-Cy5

(1 mg/ml stock). The blot was then washed 3x in TBST, once in TBS, and imaged using a

Typhoon variable mode laser scanner (GE Healthcare Life Sciences, Pittsburgh, PA).

5.3 Results and Discussion

5.3.1 Amino acid sequence requirements for isopeptide ligation

Because the majority of the pilin domain is conserved between C. diphtheriae and

A. naeslundii (WLQDVHVYPKH), we initially hypothesized that the majority of the pilin domain was required for isopeptide ligation.211 We therefore constructed 3 truncation mutants that removed non-conserved segments of the sequence: VHVYPKH (deletion of

WLQD), VYPKH (deletion of WLQDVH), and YPKH (deletion of WLQDVHV). Each of

143

these mutants was tested in overnight parallel reactions with SrtA-ELP and biotin-

LPETGRAGG at 37°C (Figure 56a). All three mutants were biotinylated through isopeptide bonds at the mutant pilin domain lysine according to MALDI-TOF mass spectrometry analysis of the trypsinized reaction product (Figure 56b). This is based on a missed digestion at the mutant PLN lysine (because the side chain is uncharged after isopeptide bond formation) and a gain of 666.8 Da corresponding to the addition of biotin-LPET through an isopeptide bond. Western blot analysis showed that all 3 mutant substrate proteins were biotinylated to a similar extent to the full-length pilin domain

(Figure 56c).

144

Figure 56. Three truncation mutants suggest that the pilin domain can be shortened. Fn3-PLN truncation mutants VHVYPKH, VYPKH, and YPKH function as sites isopeptide ligation in a manner similar to the full-length pilin domain. (a) Schematic of the reaction. (b) MALDI-TOF mass spectra showed that biotin-LPET was linked to all 3 pilin domain variants at their lysine residues. (c) Anti-biotin Western blot showed biotinylation of all the substrate proteins tested except Fn3 without any pilin domain. The amount of substrate biotinylation in all 3 pilin variants is similar to that of the full- length pilin domain.

Because no minimal motif was identified in the first 3 pilin domain truncations, we constructed 4 additional mutants (PKH, KH, WLQDVHVYPK, and K) that removed the remainder of the pilin sequence around the reactive lysine. To ensure that the linker 145

sequence flanking the reactive lysine was not influencing the reaction, we also constructed a mutant that did not contain the N-terminal linker (GGTSGSGSGGGSGG) and a mutant that replaced the flexible C-terminal GGSGR linker with a more rigid

SALVPR linker.

Substrate variants were incubated overnight at 37°C with SrtA-ELP and FITC-

LPETGRAGG and reaction products were analyzed by in-gel fluorescence (Figure 57).

This data indicated that only deletion of the pilin domain lysine (the “full deletion” sample) or exclusion of sortase prevented substrate protein biotinylation. This suggests that there is no sequence requirement for the pilin domain; instead, the reaction seems to depend on the flexibility and solvent accessibility of the reactive lysine side chain. To investigate this effect within the Fn3-PLN system, we generated an additional mutant that did not contain a pilin domain in the linker between the Fn3 and the ELP, but rather contained PKH in the solvent accessible, flexible FG loop in the Fn3 protein. Specifically, the FG loop was mutated from PRGDWNEGS to PKHGSPASS. The sequence C-terminal of the PKH returned the Fn3 sequence to its wild-type form, thereby reducing rigidity in the FG loop that had been incorporated into our Fn3 sequence by affinity maturation for

αvβ3 integrin binding.212,213 This mutant also was modified by sortase, further suggesting that isopeptide ligation depends on conformational flexibility and solvent accessibility rather than primary amino acid sequence.

146

Figure 57. Analysis of the substrate sequence requirements for sortase- catalyzed isopeptide ligation. A panel of deletion mutants was generated and tested for activity in an isopeptide ligation reaction with FITC-LPETGRAGG and SrtA-ELP. Deletion mutants are named

147

according to the section of the pilin domain that remains. N- and C-terminal linker deletions correspond to removal of glycine- and serine-rich flexible spacers with the sequences GGTSGSGSGGGSGG and GGSGR, N- and C-terminal to the pilin domain, respectively. We also produced a version of the Fn3 protein in which the FG loop was mutated from PRGDWNEGS to PKHGSPASS to include a PKH seqeunce for isopeptide ligation. (b) Fluorescence imaging of SDS-PAGE analysis of all substrate proteins tested showed fluorophore attachment unless the entire pilin domain was deleted (the “no sequence” lane). An image of the coomassie-stained gel is also shown to visualize the total protein in each lane. (c) Densitometry was used to quantify the fluorescence of each sample relative to the full pilin domain. This data confirms that the attachment site can be reduced to a solvent-accessible lysine with a local amino acid environment that enhances the nucleophilicity of the amino group of the lysine side chain. All analysis was repeated in triplicate and error bars indicate standard deviations.

5.3.2 TRAIL is modified by isopeptide ligation at lysines in a flexible loop

TRAIL was available from our previous studies that investigated the anticancer activity of TRAIL-camptothecin conjugates and depot-forming TRAIL-ELP fusions.

TRAIL is an interesting protein because it is contains 4 highly-flexible regions that contain lysines (Figure 58), and therefore represent potential substrates for isopeptide according the criteria identified with our Fn3-PLN mutants. TRAIL is composed of 10 β strands (A, A”, B’, B, C, D, E, F, G, and H).214,215,216 The loops connecting these strands are named for the two β strands that they bridge: according to this convention, the AA” loop connects the A and A” β strands. The AA”, CD, and EF loops are all highly flexible—the CD and EF loops crystallize with high B-factors (a measure of flexibility) whereas a large portion of the AA” loop has only been assigned in one crystal structure of TRAIL in the presence of the DR5 receptor (PDB ID: 1D4V).217 The AA” loop is

148

interesting in that is homologous to other tumor necrosis factor superfamily members

(e.g. TNFα, FasL, and CD40L) but contains a long segment of 16 amino acids that are not present in these other proteins. This flexible region only becomes ordered on receptor binding and is therefore responsible for a large fraction of the binding specificity of

TRAIL for its receptors. Notably, the N-terminal 8 amino acids of TRAIL are also disordered in all available crystal structures in addition to the AA”, CD, and EF loops.

Figure 58. Structure of TRAIL monomer showing its lysine residues. TRAIL monomer is shown with the two trimer interfaces arranged at the back right and left of the protein. The flexible AA” loop is shown in orange and adopts the conformation shown only when bound to death receptor 5. The N-terminus is shown in magenta, though 5 additional amino acids did not crystallize in this structure and are

149

not shown. The 11 lysine residues are labeled and are shown in green. Figure is based on PDB ID 1D4V. We constructed the fusion proteins TRAIL-PLN3-ELP and TRAIL-ELP (with no intervening linker) and tested them in the isopeptide ligation reaction to recapitulate the results seen with the Fn3 substrate proteins described earlier. Unlike with Fn3, however, both TRAIL and TRAIL-PLN3 were biotinylated to a similar extent according to Western blot analysis of the biotinylated reaction products (Figure 59a). After digestion with trypsin, MALDI-TOF mass spectrometry showed that the TRAIL-PLN3-ELP protein was biotinylated at the pilin domain lysine, which is in agreement with our earlier results.

However, analysis of the spectra for both TRAIL and TRAIL-PLN3 (Figure 59b) suggested two important features that we had not previously observed. First, we identified peaks that corresponded to modification of all three lysine residues in the

AA” loop. This was not surprising based on our mutagenesis study with Fn3 because the AA” loop is flexible and has high solvent exposure. It was surprising, however, that the spectra showed that the N-terminus of TRAIL was being modified, despite the fact that the terminal residue is a valine. All reports in the sortase literature suggest that only glycine can be used as a nucleophile in the native peptide transpeptidation reaction, because it is well-established that pentaglycine is the natural nucleophile for this reaction in S. aureus.218,219 Moreover, we did not observe any N-terminal ligation with our

Fn3 constructs that have an N-terminal valine. We therefore (perhaps incorrectly)

150

assumed that the peak corresponding to N-terminal modification of TRAIL was not assigned correctly and was actually a salt or matrix adduct with another ion.

Figure 59. Sortase-mediated isopeptide ligation reacts at sites other than the pilin domain in TRAIL. (a) An anti-biotin Western blot shows that both TRAIL-ELP and TRAIL-PLN3-ELP (3 copies of the pilin domain were used to mimic the design used with Fn3) are biotinylated by sortase. (b) MALDI-TOF mass spectrometry on a tryptic digest of the TRAIL-ELP reaction product suggested that TRAIL was modified at each of the 3 lysines 151

in the flexible AA” loop as well as its N-terminus. (c) MALDI-TOF mass spectrum of the tryptic digest of the TRAIL-PLN3-ELP reaction product also showed biotinylation at the 3 AA” loop lysines and N-terminus, along with expected isopeptide ligation at the pilin domain lysine.

We next sought to demonstrate these observations—that the N-terminus and the

AA” loop lysines were being modified—by a second method. We initially mutated

TRAIL by PCR such that the flexible N-terminal residues were removed, and the new N- terminus would begin with a valine residue that is in the bulk of the protein’s globular structure. All subsequent mutations were made using this framework. The 3 lysines in the AA” loop were changed to arginine to maintain the size and charge of the residues.

Similarly, the 3 lysines in the CD loop were mutated to arginine, and the single lysine in the EF loop was changed to alanine (this residue is buried in the trimerized form of

TRAIL, so the smaller Ala was substituted for Lys rather than Arg).

152

Figure 60. Mutagenesis confirms that the TRAIL AA” loop lysines are modified by sortase. We constructed TRAIL variants containing an 8 amino acid truncation of the N-terminus (ΔN8), mutation of the 3 lysines in the flexible AA” loop to arginine, mutation of the 3 lysines in the CD loop to arginine, and mutation of the single lysine in the EF loop to alanine. (a) These variants were tested in parallel in a reaction and biotinylated protein was determined by Western blot (total protein stain is also shown). (b) The relative biotin incorporation across the different TRAIL mutants in the Western blot was quantified and normalized to the amount of total protein in each lane.

Mutants ΔN8, AA”, CD, and EF were tested in parallel reactions with sortase.

Biotin incorporation into each substrate protein was assessed by Western blot (Figure

60a); the relative biotinylation of each substrate is compared in Figure 60b. All of the substrates were biotinylated in the reaction, however, AA” was the only mutant that had a significantly lower amount of biotin incorporation compared to wild-type TRAIL.

This is consistent with the mass spectra for TRAIL and TRAIL-PLN3, which showed modification of all three of the lysine residues in the AA” loop.

153

Although mutation of the 3 AA” lysines reduced the level of biotin incorporation, the remaining signal must have been due to modification of another site.

In addition to the seven lysines in the AA”, CD, and EF loops, TRAIL also contains 4 additional lysines. To investigate whether any of these were modified in the reaction, we mutated each additional lysine (K179R, K251R, K212A, and K224A) and tested the resulting protein in the sortase reaction. We also constructed combination mutants of the flexible TRAIL loops—AA”CD, AA”EF, CDEF, and AA”CDEF—with the hypothesis that the proximity of the CD and EF loops could have masked the individual contribution of each loop in the single-loop mutant proteins (i.e., mutation of one loop could have exposed the other to modification).

TRAIL also contains a single cysteine residue. Cysteine is widely used as a site of protein modification because it functions as a strong nucleophile that is compatible with a large number of protein labeling reagents. Although the cysteine in TRAIL is buried in the TRAIL trimer and is involved in chelating a zinc II ion, it was possible that a fraction of the TRAIL in our preparation was monomeric and could have reacted through the thiol side chain of the cysteine. To investigate this, we constructed a C280S mutant of

TRAIL.

Each of these new mutant proteins was tested alongside wild-type TRAIL and the single-loop mutants in parallel sortase reactions with FITC-LPETGRAGG peptide.

All mutant proteins were modified by sortase (Figure 61a), and all of the mutants had

154

the same fluorescence as wild-type TRAIL except those containing mutated AA” loop lysines, which had lower fluorescence that was the same for all variants with the AA” mutations. The fluorescence intensity of each band was quantified by densitometry

(Figure 61b). Comparing the average fluorescence intensity of 7 mutants containing lysines in the AA” loop (ΔN8, CD, EF, K179R, K251R, CDEF, and C280S) to the average intensity of 6 mutants containing lysine-to-arginine mutations in the AA” loop (AA”,

AA”CD, AA”EF, AA”CDEF, AA”K212A, and AA”K224A) indicated that mutation of the AA” loop lysines resulted in a statistically-significant reduction in fluorescence intensity at p = 0.005 (Figure 61c). This indicates that: (1) the cysteine in TRAIL is not modified by sortase and (2) no lysines in TRAIL other than the AA” loop lysines are modified.

155

Figure 61. TRAIL AA” loop lysines are modified by sortase while other lysines and the single cysteine in the protein are not modified. TRAIL variants were constructed in order to determine whether the amount of modified protein would decrease in response to the mutation of a particular amino acid. Single or multiple lysines within the same loop mutated to non-reactive amino acids. Similarly, the single cysteine in TRAIL was mutated to serine. (a) All variants were reacted in 156

parallel with SrtA-ELP and FITC-LPETGRAGG overnight and FITC attachment to the proteins was assessed by fluorescence scanning of SDS-PAGE gels. Coomassie stain of the same gel is shown to visualize the total protein in each lane. (b) All constructs showed similar levels of FITC incorporation when fluorescent band intensity was quantified by densitometry relative to wild-type TRAIL except for those with mutated AA” loop lysines, which were similar to each other. (c) Comparison of TRAIL variants with or without the AA” lysines mutated to arginine. The fluorescence intensity of 7 mutants containing lysines in the AA” loop (dN8, CD, EF, K179R, K251R, CDEF, and C280S) were averaged and compared to 6 mutants containing lysine-to-arginine mutations in the AA” loop (AA”, AA”CD, AA”EF, AA”CDEF, AA”K212A, and AA”K224A). The mutation of the AA” loop lysines resulted in a statistically-significant reduction in fluorescence intensity at p = 0.005 by a two-tailed t-test. All analysis was repeated in triplicate and error bars indicate standard deviations.

The fact that only the AA” loop lysines appear to be modified in TRAIL lead us to hypothesize that only lysines in flexible, unstructured regions of a substrate protein can be modified by isopeptide ligation. This would agree with data we obtained for

GLP-1, Fn3, monoclonal antibodies, and TRAIL. Another ELP substrate that Stefan

Roberts made for a different study in the Chilkoti group allowed us to further examine this hypothesis. He made two ELPs that contained the ELP sequence (VPGVG)15 linked to one of two C-terminal domains that form alpha helices with the sequences KA25A and

(KAAAA)5K (Figure 62a). We tested these two ELPs in parallel with a molecular weight

ELP-only control by incubating them with SrtA-ELP and FITC-LPETGRAGG peptide overnight at 32°C. A fluorescence image of the SDS-PAGE gel (Figure 62b) showed that all three substrates were modified to contain FITC, although this is expected because all three contain an N-terminal glycine residue that is a known substrate for sortase-

157

catalyzed native peptide ligation. However, when the trypsinized reaction product was analyzed by MALDI-TOF mass spectrometry, no isopeptide ligation was observed for any of the lysine residues in the helical elements (Figure 62c). This further suggests—in agreement with our earlier data on peptide and protein substrates—that isopeptide ligation occurs only at lysine residues that are located in flexible, disordered regions of the substrate protein.

158

Figure 62. Lysines within secondary structural elements do not appear to be modified by isopeptide ligation. (a) Sequence of three ELPs that were tested in the isopeptide ligation reaction with FITC. The KA25K and (KAAAA)5K domains form alpha helicies. (b) All 3 ELP constructs are modified to contain FITC in the reaction to a similar extent, though this is expected because glycine—a known nucleophile in the native peptide ligation reaction catalyzed by sortase—is the N-terminal residue in all 3 cases. (c) MALDI-TOF mass spectrometry does not show the presence of isopeptide ligation at any of the helical lysines in either construct, strongly suggesting that these residues are not modified in the reaction.

159

5.3.3 Other chemistries do not lead to TRAIL modification

Although TRAIL appears to be modified at the lysines in its flexible AA” loop, reactions with both biotin and FITC labels suggested that this was not the only site of modification in the TRAIL protein. We therefore sought to identify the reactive site(s) that accounted for the remaining modification(s).

We first investigated the stability of the isopeptide ligation product in order to obtain information that might help identify the type of amino acid that was being modified. We hypothesized that the modification could be occurring at histidine, tyrosine, tryptophan, or methionine, as these amino acids contain side chains that have the potential to be nucleophilic.

Histidine and tyrosine acylation has been demonstrated in the literature and these residues have the capacity to cross-react with some lysine-specific reagents.220,221,222

In the case of histidine, acylation occurs at either nitrogen in the imidazole ring (both can be protonated because of resonance in the ring structure). However, this modification is not stable at acidic pH, because the unmodified nitrogen will protonate, destabilizing the ring and leading to release of the acyl group (Figure 63).

160

Figure 63. Proposed reaction scheme for the acylation of histidine by sortase. Histidine can be acylated at either nitrogen due to resonant structures in the imidazole ring. Acylated histidine is unstable in acid because protonation of the non-modified amino group destabilizes and releases the acyl group from the neighboring amine.

Tyrosine is nucleophilic at its phenol hydroxyl group (Figure 64). Acylation of this position generates an ester that is subject to hydrolysis in acidic or basic conditions or by reaction with hydroxylamine, which is strongly nucleophilic and is known to attack the carbonyl of N-acylphenol.223,224

Figure 64. Proposed reaction scheme for the acylation of tyrosine. Tyrosine can be acylated at its side chain hydroxyl group if this group is initially deprotonated. The resulting ester bond is stable at neutral pH but can be hydrolyzed in acidic or basic conditions or by the strong nucleophile hydroxylamine.

161

TRAIL contains 6 histidines (Figure 65) and 11 tyrosines (Figure 66). To test whether histidine and tyrosine residues in TRAIL were being modified by sortase, we first ran a reaction on the wild-type TRAIL protein, then diafiltered the reaction product into pH 2.5, 7.5, or 8.5 buffers and incubated the solutions overnight at 37°C. We hypothesized that the low pH condition would lead to hydrolysis of any acylated histidine and tyrosine, whereas the high pH condition would have led to hydrolysis of only acylated tyrosine and neutral pH would not have favored hydrolysis of either species. However, an anti-biotin Western blot of the samples showed that the amount of biotinylated protein was similar in all 3 pH conditions (Figure 67).

162

Figure 65. Structure of TRAIL monomer showing its histidine residues. TRAIL monomer is shown with the two trimer interfaces arranged at the back right and left of the protein. The flexible AA” loop is shown in orange and adopts the conformation shown only when bound to death receptor 5. The N-terminus is shown in magenta, though 5 additional amino acids did not crystallize in this structure and are not shown. Histidines are labeled and are shown in green. Figure is based on PDB ID 1D4V.

163

Figure 66. Structure of TRAIL monomer showing its tyrosine residues. TRAIL monomer is shown with the two trimer interfaces arranged at the back right and left of the protein. The flexible AA” loop is shown in orange and adopts the conformation shown only when bound to death receptor 5. The N-terminus is shown in magenta, though 5 additional amino acids did not crystallize in this structure and are not shown. Tyrosines are labeled and are shown in green. Tyrosine 216 is in the top left of the structure. Figure is based on PDB ID 1D4V.

164

Figure 67. The modifications made to TRAIL by sortase are stable across a wide range of pH. TRAIL-ELP was biotinylated by reaction with sortase and aliquots of the reaction product were diafiltered into buffers with pH 7.5 (20 mM Tris), pH 2.5 (0.1M glycine), or pH 8.5 (20 mM Tris). The relative amount of biotinylated protein in each sample was assessed by Western blotting after overnight incubation at 37°C. No difference in the relative amount of biotinylated protein was observed when each lane of the Western was corrected for protein load.

We also mutated 4 solvent-exposed histidines in TRAIL (out of 6 total histidines) and tested these variants in a biotinylation reaction with sortase. All 4 variants showed similar biotinylation to wild-type TRAIL (Figure 68), which supports the conclusion that histidine is not modified by sortase.

165

Figure 68. Mutation of histidines in TRAIL does not reveal any reactive sites. TRAIL variants with individual histidines mutated to alanine (all containing the ΔN8 truncation) were biotinylated in parallel with wild-type TRAIL containing the ΔN8 truncation. All variants showed a similar level of biotinylation in an anti-biotin Western blot when corrected for total protein load.

To confirm that tyrosine modification was not occurring, we also biotinylated both TRAIL-PLN3 and TRAIL with sortase and incubated the reaction products with triglycine (to quench modification of the substrate proteins) and hydroxylamine under denaturing and non-denaturing conditions. We found that hydroxylamine did not reduce the biotinylation of either substrate protein under denaturing or native conditions, further confirming that tyrosine was not being modified in the reaction

(Figure 69).

166

Figure 69. The modifications that sortase makes to TRAIL are stable towards the strong nucleophile hydroxylamine. TRAIL-PLN3-ELP and TRAIL-ELP were biotinylated by overnight reaction with sortase, then incubated with hydroxylamine (HA), denaturant (sodium dodecyl sulfate, SDS), or hydroxylamine with denaturant. No reduction in the amount of biotinylated protein is evident after overnight incubation at 37°C, indicating that tyrosine residues in TRAIL are not acylated by sortase, as the ester bond in the acylated phenol is susceptible to cleavage by hydroxylamine, particularly under denaturing conditions which would reduce any steric hindrance around the modified site(s).

Tyrosine modification was also not supported by mutation of a highly solvent- exposed tyrosine (Y216) that is responsible for a large amount of contact area between

TRAIL and death receptor 5 (DR5) in a crystal structure of the TRAIL-DR5 complex.225

Because this residue was highly solvent exposed, we reasoned that it could be accessible for reaction with sortase. We compared a Y216A TRAIL mutant to a panel of other

167

TRAIL variants, but the Y216A mutant did not show different reactivity compared to these other proteins with the exception of the AA” lysine mutant that was already shown to be less reactive than wild-type TRAIL (Figure 70). Though there are 10 other tyrosines in TRAIL that are buried in the core of the protein and unlikely to react (Figure

66). The lack of a reaction at Y216 agrees with the fact that biotin was not removed by extremes of pH or hydroxylamine and leads us to conclude that tyrosines are not modified in the sortase reaction.

Figure 70. A solvent-exposed tyrosine in TRAIL is not modified by sortase. The side chain of tyrosine 216 in TRAIL is solvent-exposed and contributes significant contact area to the TRAIL-death receptor 5 complex. This residue was mutated to alanine and this variant is biotinylated to the same extent as a panel of other TRAIL variants by Western blotting.

168

Tryptophan is nucleophilic at all positions on its indole ring. Electrophiles typically favor the N1 position and these reactions require basic conditions that may not be achieved in aqueous media (Figure 71), but we reasoned that this type of reaction might be possible in the context of an enzymatic reaction, particularly because the sortase active site contains several hydrophobic amino acids that could undergo pi-pi stacking with the hydrophobic tryptophan side chain.

Figure 71. Proposed reaction scheme for the acylation of tryptophan. The tryptophan indole contains a nucleophilic nitrogen that could be acylated in reaction with sortase by the mechanism shown. This type of reaction typically involves better leaving groups (e.g. trifluoroacetic anhydride), and tryptophan is typically unreactive because it is hydrophobic and buried in the protein core. Nonetheless, because TRAIL contains only 2 tryptophans (one of which is in the flexible AA” loop), we mutated these residues to phenylalanine (to maintain hydrophobicity) and tested the mutants to determine whether they would be modified by sortase.

TRAIL contains two tryptophans, one of which is in the flexible AA” loop

(Figure 72). We individually mutated these residues to phenylalanine and tested them in parallel reactions with wild-type TRAIL. Both tryptophan mutants were as reactive as

169

the wild-type protein (Figure 73), which suggests that tryptophans are not modified in the sortase reaction.

Figure 72. Structure of TRAIL monomer showing its tryptophan residues. TRAIL monomer is shown with the two trimer interfaces arranged at the back right and left of the protein. The flexible AA” loop is shown in orange and adopts the conformation shown only when bound to death receptor 5. The N-terminus is shown in magenta, though 5 additional amino acids did not crystallize in this structure and are not shown. Tryptophans are labeled and are shown in green. Figure is based on PDB ID 1D4V.

170

Figure 73. Tryptophans in TRAIL are not modified by sortase. The 2 tryptophans in TRAIL were individually mutated to phenylalanine, then compared to wild-type TRAIL and TRAIL lacking its N-terminal 8 amino acids. Western blot indicated that all four variants had a similar amount of biotinylated protein after the reaction, suggesting that the tryptophans in TRAIL are not modified by sortase.

The methionine side chain contains a thioether group that is much less nucleophilic than the thiol group of cysteine, but methionine nonetheless undergoes some reactions that modify the sulfur atom. This group is readily oxidized, and this process is thought to relive oxidative stress in many proteins.226,227 Moreover, methionine is enzymatically converted into S-adenosylmethionine, which plays a critical role as a methyl donor in protein (particularly histone) and DNA methylation.228,229,230,231

Additionally, we reasoned that methionine modification could explain why

TRAIL was modified while Fn3 was not, because Fn3 does not contain any methionine

171

residues while TRAIL contains three (Figure 74). However, it is possible that one of the two proteins contained an N-terminal methionine because many examples have shown that recombinant proteins made in bacteria do not necessarily undergo complete removal of the N-terminal methionine that is encoded by the start codon. In bacteria this residue is initially added to the polypeptide chain as formylmethionine, which undergoes serial processing steps to remove the formyl group and then the methionine.

High-level expression of recombinant proteins can overwhelm the pool of bacterial methionine aminopeptidase (metAP) and low induction temperatures can also reduce the enzyme’s activity.232,233 Moreover, the amino acid sequence after the start codon has been shown to influence the efficiency of metAP.234,235,236 For these reasons, methionine removal is not guaranteed. Conversely, secreted proteins (e.g. antibodies) that are produced in mammalian cells do not have this issue because the entire N-terminal signal peptide is removed during expression.

172

Figure 74. Structure of TRAIL monomer showing its methionine residues. TRAIL monomer is shown with the two trimer interfaces arranged at the back right and left of the protein. The flexible AA” loop is shown in orange and adopts the conformation shown only when bound to death receptor 5. The N-terminus is shown in magenta, though 5 additional amino acids did not crystallize in this structure and are not shown. Methionines are labeled and are shown in green. Figure is based on PDB ID 1D4V.

To test whether methionine was undergoing sortase-mediated modification in

TRAIL, we oxidized the methionine side chains in several TRAIL variants by briefly exposing them to hydrogen peroxide, then quenched the oxidation reaction and

173

attached FITC to the proteins using isopeptide ligation. We borrowed the oxidation protocol from Stability of Proteins from Rates of Oxidation (SPROX) experiments, which is a mass spectrometry based method that uses the selective oxidation of methionine residues under a range of denaturant conditions to determine the structural stability of proteins in a sample.237,238

Wild-type TRAIL and three mutants were oxidized with 5% hydrogen peroxide for 10 minutes at room temperature, after which free hydrogen peroxide was quenched by adding L-methionine to 150 mM and incubating for 20 minutes at room temperature.

SrtA-ELP and FITC-LPETGRAGG were then added to the oxidized substrate proteins and the reactions were incubated at 37°C for ~3 hours. The reaction products were analyzed by SDS-PAGE (Figure 75a), which appeared to show that the reaction was completely inhibited in those TRAIL variants with oxidized methionines and mutated

AA” loop lysines. The amount of fluorescent protein was quantified by densitometry, which showed that less protein was modified for each of the 4 oxidized substrates compared to non-oxidized wild-type TRAIL (Figure 75b). However, the same data can also be normalized within each treatment group, that is, oxidized proteins relative to oxidized wild-type TRAIL and non-oxidized proteins to non-oxidized wild-type TRAIL.

This showed that hydrogen peroxide treatment inhibited the reaction by the same amount for all the TRAIL variants (Figure 75c). This could be explained by the fact that oxidation blocks the nucleophilicity of the methionine side chains. Alternatively, the

174

increased hydrophobicity of the oxidized proteins could also be to blame, as aggregation of these more hydrophobic substrates could have simply protected the reactive sites from modification by sortase. This experiment was therefore inconclusive, and we sought to test whether methionines were being modified by sortase using another technique.

175

Figure 75. Methionine in TRAIL appears to be modified by sortase. The methionine side chains in four TRAIL variants were oxidized with hydrogen peroxide, then reacted with SrtA-ELP and FITC-LPETGRAGG peptide. (a) Reaction products were analyzed by SDS-PAGE and the gel was imaged by fluorescence scanning. Because no signal was observed for variants with AA” loop lysine mutations, this suggested that the AA” loop lysines and methionines were sites of modification. (b) Fluorescent protein bands in the gel were quantified by densitometry and quantified relative to non-oxidized TRAIL. When normalized to non-oxidized wild-type TRAIL, samples that had been oxidized showed a reduction in fluorescence for all 4 variants. (c) However, when the same data was normalized to wild-type treated in the same manner (oxidized or non-oxidized), the results indicated that the oxidation step resulted in fluorescence that was lower by the same relative amount for each sample. This is likely due to the fact that oxidation makes each substrate more hydrophobic and caused precipitation that protected any reactive sites. This is supported by the fact that the 176

overall amount of fluorescence inversely correlates with the hydrophobicity of the samples: as the substrate becomes more hydrophobic (e.g. AA” 3KA is the most hydrophobic), the amount of protein modified by sortase decreases. This suggests that methionine is not a reactive site and the hydrophobicity of each substrate determines the extent of modification.

Because the Fn3 sequence does not contain any methionines, we generated a new version of this protein that contained a methionine in the flexible linker connecting it to its ELP fusion partner (Fn3-M-ELP). We then tested Fn3-M-ELP in the sortase reaction in parallel with a panel of other Fn3 variants from our earlier studies (Figure 76).

Consistent with our earlier results, Fn3 variants containing the full pilin domain, PKH, or only lysine in the linker were biotinylated by sortase, whereas Fn3-ELP did not react.

Fn3-M-ELP also did not react, indicating that the methionine side chain is not modified by sortase in the reaction. We therefore conclude that the TRAIL data showing the complete reduction in signal after oxidation of the AA” loop variant is the result of aggregation of the protein that prevents the modification of reactive sites. Combining the TRAIL and Fn3 results, we conclude that the methionine side chain cannot act as a nucleophile to react with the sortase acyl-enzyme intermediate.

177

Figure 76. The methionine side chain is not modified by sortase. The sequence of Fn3 does not contain any methionine residues. We therefore added a methionine to the linker between the Fn3 and the ELP (Fn3-M-ELP) and tested this protein in the sortase reaction with biotin. Whereas Fn3 containing the full pilin domain, PKH, or only lysine in the linker reacted in parallel reactions, Fn3-ELP and Fn3-M-ELP did not react. This indicates that the methionine side chain is not modified by sortase.

5.3.4 Native peptide ligation competes with isopeptide ligation if the N-terminus is accessible

We next examined whether the N-terminus of TRAIL was being modified in the reaction. We generated a mutant that had a Factor Xa cleavage site proximal to the N- terminus with the sequence VAIEGRAQRVAAH (bold = Factor Xa recognition site, italics = TRAIL ΔN8 sequence). The mutant with an N-terminal Factor Xa site was reacted in parallel with wild-type TRAIL and TRAIL with the ΔN8 deletion and AA”

178

loop lysines mutated to arginines. Factor Xa is known to cut off-target at GR sequences, particularly when these sites are in unstructured segments of the protein.239 TRAIL contains two of these sites (underlined and shown in bold):

VRERGPQRVAAHITGTRGRSNTLSSPNSKNEKALGRKINSW. The AA” lysine mutation removes this second site by changing the second site to GRR, which Factor Xa will not cut because it is does not cut GR sites that are followed by proline or arginine.240

After overnight biotinylation by sortase, the three reaction products were digested with

0, 1, or 2 units of Factor Xa for ~24 hours. Regardless of the version of TRAIL that was tested, the biotinylation signal was similar (~85%) to the same sample that was incubated without any Factor Xa (Figure 77). This result suggested that sortase was not modifying the N-terminus of the TRAIL mutants. Each mutant appeared as multiple bands on the Western after digestion with Factor Xa, which was expected given the fact that each contains one or more canonical or off-target Factor Xa sites. The fact that each of these bands was biotinylated suggests biotinylation at a site other than the N- terminus. However, SrtA-ELP is still present in this reaction and is potentially active.

The multiple biotinylated bands in the Western could therefore be explained by sortase modifying the new N-termini that are exposed after factor Xa cleavage, especially given that factor Xa side reactions typically occur at positions in the substrate protein with high flexibility and accessibility, which also seem to be preferred by sortase. This type of reaction should have been mitigated by the fact that we included 5 mM triglycine in the

179

factor Xa digestion, but it would explain the pattern of biotinylated protein observed in the Western blot. We therefore cannot convincingly conclude that the N-terminus of

TRAIL is being modified by sortase in this experiment, despite the fact that ions corresponding to biotinylated N-terminal peptides were observed in MALDI-TOF mass spectrometry.

Figure 77. Cleavage of the TRAIL N-terminus does not conclusively indicate modification by sortase. A version of TRAIL with a factor Xa cleavage site at the N-terminus VAIEGRAQRVAAH, factor Xa site in bold) was tested in a reaction with sortase in parallel with wild-type TRAIL and TRAIL ΔN8. The products of these reactions were incubated with factor Xa (0, 1, or 2 units) overnight at room temperature and the products analyzed by anti-biotin Western blotting. Though factor Xa cut each of the substrates (due to background sites in TRAIL), cleavage reduced the biotinylation of all three TRAIL variants to ~85% of the amount in samples incubated without any factor Xa. This suggests that there is N-terminal modification of all 3 TRAIL variants, though this is not statistically significant.

180

In light of the complications associated with the factor Xa digestion of TRAIL, we designed another assay to investigate whether native peptide ligation was occurring at

N-terminal residues other than glycine. We constructed an ELP fusion that had one pilin domain at its N-terminus (with a terminal serine) and a sortase recognition site at the C- terminus of the ELP. In this case, the fusion contained an arginine residue near the N- terminus that allowed us to distinguish isopeptide ligation from native peptide ligation

(Figure 78a). In the case of isopeptide ligation, trypsinization of the polymer would produce only short peptide fragments and the bulk of the polymer would be unaffected.

In the case of native peptide ligation at the N-terminus, the bulk of the polymer would disintegrate into monomer-sized pieces when trypsinized because each monomer would be linked through the segment containing the arginine. We reasoned that this system was improved over the earlier experiments that used Factor Xa to cleave TRAIL, since our earlier experience indicated that trypsin digestion is thorough and rapid (2-4 hours).

Furthermore, the PLN-ELP-LPXTG fusion is unstructured and should not have a high degree of steric hindrance at the trypsin cut site.

The PLN-ELP-LPXTG fusion was reacted overnight with SrtA-ELP 32°C and isopeptide ligation was quenched by adding triglycine to 5 mM. The product was then incubated with either buffer or a ~1:20 ratio (w/w) of mass spectrometry-grade acetylated trypsin at room temperature. After ~24 hours the product was analyzed by

SDS-PAGE. The results (Figure 78b) indicate that the PLN-ELP-LPXTG was polymerized

181

by sortase; however, trypsinization degraded the majority of the polymer. The polymer should not degrade during trypsinization if polymerization occurred through isopeptide bonds the pilin domain lysine. The gel shows that the majority of the high molecular weight polymer was destroyed and that a short peptide band appeared at the bottom of the gel, which corresponds to the GSSGGWLQDVHVYPK peptide. Monomer does not appear on the gel because it does not contain a tryptophan, and band visualization depends on chemical conversion of tryptophan side chains to a fluorescent product.

Together, these data suggest that polymerization primarily occurred through native peptide bond formation at the N-terminal serine residue with some isopeptide ligation at the side chain of the pilin domain lysine.

182

Figure 78. Native peptide ligation at the N-terminus competes with isopeptide ligation within a protein. PLN-ELP-LPXTG, a fusion protein containing the full length pilin domain at the N- terminus and the sortase recognition site LPLTG, will polymerize when incubated with sortase. (a) Schematic of the reaction, showing the structure of the polymer that results from native peptide ligation through the N-terminus or that which results from isopeptide ligation at the pilin domain lysine. In the case of native peptide ligation, the polymer will be cleaved by trypsin to yield monomer-sized fragments. In the case of isopeptide ligation, trypsin will only cleave small peptides off of the polymer and the high molecular weight backbone will remain intact. (b) SDS-PAGE analysis of the 183

reaction product shows that a significant fraction of the high molecular weight polymers disappear from the gel after incubation with trypsin. Though not visible with the stain- free gel chemistry, these polymers are cleaved to monomers based on the appearance of the tryptophan-containing peptide fragment at the bottom of the gel. Some dimer and trimer remain after trypsin digestion, which likely correspond to isopeptide-linked polymers whose backbone is unaffected by trypsin.

To further confirm these results, we generated a version of the same PLN-ELP-

LPXTG with the pilin domain lysine mutated to an alanine (Figure 79a). This version of the monomer cannot polymerize by isopeptide ligation, however, efficient polymerization was observed after overnight incubation at 32°C (Figure 79b). This suggests that polymerization was occurring through the N-terminal amino group and this conclusion was further confirmed by trypsinizing the product. Mostly monomer- like proteins were present after trypsinization. Two versions of the monomer were present that we attribute to missed trypsin digestions, though this does not affect the conclusion that polymerization is occurring through the N-terminal serine of the monomer protein.

184

Figure 79. Native peptide ligation occurs with proteins that do not have an N- terminal glycine. PLN(KA)-ELP-LPXTG, a fusion protein containing the pilin domain with its lysine mutated to alanine will polymerize when incubated with sortase only if native peptide ligation occurs at the N-terminal serine. (a) Schematic of the reaction, showing the structure of the polymer that results from native peptide ligation through the N- terminus. Isopeptide ligation is not possible with this substrate. In the case of native peptide ligation, the polymer will be cleaved by trypsin to yield monomer-sized fragments. (b) SDS-PAGE analysis of the sortase and trypsinization reaction products. Polymerization in the sortase reaction indicates that native peptide ligation is occurring at the N-terminal serine. Trypsinization degrades the polymer into two monomer-sized fragments (the upper band contains a missed digestion at a single site). 185

5.3.5 Analysis of nucleophile requirements for N-terminal native peptide ligation by sortase A

The polymerization and TRAIL data indicates that sortase will utilize amino acids other than glycine at the N-terminus of a protein as a nucleophile in the native peptide ligation reaction. This data contradicts the principle in the literature that only N- terminal glycine—preferably, oligoglycine—can be used as a nucleophile in the reaction.

Many reports have indeed confirmed that pentaglycine is indeed the nucleophile when the enzyme performs its natural function in S. aureus, which has led to the use of oligoglycine as the nucleophile for sortase. Many reports re-state this concept and use sortase only as a tool to connect one component bearing the LPXTG enzyme recognition site to another component with an N-terminal oligoglycine. While the reaction works well for this purpose, these reports have not considered testing the reaction with N- terminal nucleophiles other than glycine.

Only one report has attempted to explore the nucleophile selectivity of the sortase native peptide ligation reaction.241 This was a cursory investigation that reported that the three peptides AA, AG, and VG showed “no stimulation over the rate of the hydrolysis reaction.” This experiment was lacking for two reasons. First, it tested a very limited set of peptides in the sortase reaction, and while this provides very little information it is understandable considering the high cost of peptide synthesis. Second

—and more importantly— the experiment overlooks the fact that the peptides without

N-terminal glycine may in fact have reacted, albeit at a low rate. The authors concluded

186

that the peptides they tested did not stimulate the reaction over the rate of hydrolysis of the LPXTG enzyme substrate with a fluorescence resonance energy transfer pair (o- aminobenzoic acid [Abz] and 2,4-dinitrophenyl [Dnp]) at its termini. Thus, while the

AA, AG, and VG peptides may not have stimulated Abz-LPXTG-Dnp cleavage, the experiment provided no data on whether sortase was actually able to ligate any of the nucleophiles to the Abz-LPXT component.

The polymerization and TRAIL results suggested that native peptide ligation was possible at N-terminal serine and valine, respectively. We therefore sought to identify whether any other amino acids could be modified by sortase in a native peptide ligation reaction at the N-terminus. We designed and purchased a peptide array from

JPT Peptide Technologies that contained 38 peptides. We used the N-termini of the trastuzumab (human anti-human Her2 IgG1 kappa) heavy and light chains as model peptides because they are conserved in the majority of human IgG sequences and would therefore reflect antibodies that we would eventually use to make therapeutically- relevant antibody-drug conjugates. 19 of the peptides corresponded to the N-terminal 16 amino acids of the trastuzumab heavy chain. The other 19 corresponded to the N- terminal 16 amino acids of the trastuzumab light chain. Each of the 38 peptides contained a different N-terminal amino acid, so that all amino acids except cysteine were tested in both the heavy and light chain peptides. These peptides were synthesized

187

directly on a cellulose membrane such that their C-termini were covalently linked to the membrane (Figure 80a).

The peptide array was blocked in 5% nonfat milk TBST buffer then washed and incubated overnight at 32°C in a reaction containing 10 µM SrtA-ELP, 100 µM biotin-

LPETGRAGG peptide, and 0.1 mM CaCl2. After the reaction, the membrane was washed extensively with TBST, then stained with streptavidin-Cy5 to visualize biotin that had been covalently linked to the peptides. Fluorescence imaging of the array (Figure 80b) showed that all peptides were biotinylated in the reaction. We also analyzed a blot that was reacted in parallel that included the same peptide spots, except in this case each peptide was acetylated at its N-terminus (Figure 80c). We quantified the signal for each spot in both blots using densitometry and compared the results for the heavy and light chains (Figure 80c and 80d). This data indicated that sortase was able to modify the N- terminus for all N-terminal amino acids in both the light and the heavy chains.

Additionally, acetylation of the N-termini largely blocked biotinylation, though some residual signal was observed that we attributed to incomplete acetylation of spots with particular N-terminal residues (e.g. tryptophan). The quantitation of biotinylated peptides suggested that glycine was one of the most efficiently modified residues in the

N-terminal position, which agrees with the commonly accepted principle that one or more N-terminal glycines are necessary for native peptide ligation. Importantly, though, the relative amounts of biotinylation are not conclusive because we have not quantified

188

the amount of peptide in each spot. We also analyzed the Kabatman database of antibody sequences (approximately 2000 sequences) and found that 97% of the antibodies contained E, Q, or D at the heavy chain N-terminus (54%, 42%, and 1%, respectively). Combined with the peptide array data, this suggests that most human antibodies can be modified by sortase at their N-termini. However, this assumes that the

N-termini are accessible to the enzyme in the bulk structure of the antibody, which we sought to test in the next section.

189

Figure 80. All N-terminal amino acids are modified by sortase in a native peptide ligation reaction. A series of peptides were synthesized such that their C-termini were covalently linked to a cellulose membrane. The array contains peptides corresponding to the first 16 amino acids of either the light chain or the heavy chain of trastuzumab, a human IgG1 antibody. Each peptide spot in the array has a different N-terminal amino acid covering all 20 amino acids except cysteine. (a) The array was blocked with 5% nonfat milk, washed, and incubated overnight at 32°C in a reaction solution containing SrtA-ELP and biotin-LPETGRAGG peptide. After washing, biotinylated peptide spots were detected by streptavidin-Cy5 (SA-Cy5) staining. (b) Image of fluorescent spots after SA-Cy5 staining. Row 1 contains all heavy chain peptides and row 2 contains the light chain peptides. (c) In parallel, an array containing the same peptides with acetylated N-termini (which should eliminate their nucleophilicity) was reacted and analyzed. Relative biotinylation for both arrays was assessed by densitometry and is reported for the heavy chain (d) and the light chain (e). Both panels are normalized to their respective peptides with glycine in the N-terminal position. This data indicates that all peptides with N- terminal amino groups were biotinylated by sortase. Acetylation dramatically reduced

190

biotinylation for nearly all peptides, though we attribute some residual signal at certain spots to variations in acetylation efficiency for different N-terminal residues.

5.3.6 Analysis of native and isopeptide ligation in human antibodies

Having analyzed peptides corresponding to human antibody N-termini for their ability to undergo native peptide ligation, we next examined the reactivity of full-length human IgG antibodies against Her2 and CD30, a marker of resting B cells and activated

T cells that is upregulated in several types of hematological cancers.242 Both antibodies are the same sequences as those used in the clinically-approved antibody drug conjugates trastuzumab emtansine (against Her2) and brentuximab vedotin (against

CD30). The heavy chains of these antibodies are both IgG1 and their sequences are 88% identical, 8% similar, and 4% dissimilar (Figure 81a). The kappa light chains of these antibodies are 84% identical, 8% similar, and 8% dissimilar (Figure 81b).

191

Figure 81. Human antibodies are modified by sortase through native and isopeptide ligation. Alignment of the trastuzumab (anti-Her2) and brentuximab (anti-CD30) sequences. (a) The heavy chains are 88% identical, 8% similar, and 4% dissimilar. (b) The light chains are 84% identical, 8% similar, and 8% dissimilar. Identical residues are highlighted in dark gray. Amino acids that are not identical but which have similar hydrophobicities or charges are highlighted in light gray. Lysines are highlighted in yellow. Alignment was produced using the UniProt Align tool available at http://www.uniprot.org/align/.

192

We incubated these antibodies with SrtA-ELP and biotin-LPETGRAGG overnight at room temperature and analyzed the biotinylation of each protein chain by reducing SDS-PAGE and anti-biotin Western blotting. The Western blot indicated that both anti-Her2 and anti-CD30 antibodies were biotinylated on their heavy chains, but not on their light chains (Figure 82a and 82b). Though the peptide array data suggested that both chains were reactive at the N-termini, we believe that only the heavy chain is biotinylated in the context of an intact antibody because it is structurally accessible towards sortase while the light chain N-terminus is not. Interestingly, this result is different than that obtained with the murine version of the anti-Her2 antibody, which highlights the importance of testing the competition between isopeptide and native peptide ligation on a case-by-case basis for each protein of interest. It was possible that biotinylation occurred at the C-terminal lysine in the heavy chain, which is frequently removed by carboxypeptidases produced by the host cells during expression. However, we generated a version of trastuzumab where this lysine was genetically deleted, and this did not affect the biotinylation signal that was observed after reaction with sortase

(Figure 82c).

Our analysis of the Kabatman database (Figure 80f) previously found that 97% of the deposited antibodies sequences contained E, Q, or D at the heavy chain N-terminus

(54%, 42%, and 1%, respectively). We therefore mutated the N-terminal residue of trastuzumab from glutamic acid (E) to glutamine (Q) and aspartic acid (D). These

193

antibodies were also biotinylated at the heavy chain by sortase (Figure 82d). We also attempted to eliminate N-terminal ligation by removing residues from the N-terminus at the gene level, but we found that deletion of residues 1, 1 and 2, or 1-3 did not dramatically impact the level of heavy chain biotinylation (Figure 82e). Moreover, we found that the reaction was also unaffected when the Her2 variable regions were grafted onto a human IgG2 isotype (Figure 82f), indicating that this effect is not limited to human IgG1 isotypes. Non-specific binding of streptavidin was not the source of the signal in the anti-biotin reactions as trastuzumab or brentuximab that were reacted with

SrtA-ELP but without biotin-LPETGRAGG were not visualized in an anti-biotin Western blot (Figure 82g).

194

Figure 82. Human antibodies are modified by sortase through native and isopeptide ligation. (a) Sortase biotinylates trastuzumab (anti-Her2) with and without a pilin domain at its heavy chain C-terminus. (b) Sortase biotinylates brentuximab (anti-CD30) with and without a pilin domain at its heavy chain C-terminus. (c) Trastuzumab reactivity does not depend on whether the heavy chain C-terminal lysine is removed during expression, as genetic removal of this lysine from the antibody sequence does not affect biotinylation. (d) Trastuzumab with mutation of the heavy chain N-terminal glutamic acid to aspartic acid or glutamine does not affect reactivity, suggesting that the reaction will occur at a large fraction (97% of the Kabat database) of human antibody heavy chains. (e) Genetically deleting N-terminal residues 1, 1-2, or 1-3 does not affect trastuzumab reactivity. (f) The variable regions of trastuzumab were grafted onto the constant framework of a human IgG2, which was biotinylated by sortase and suggests that N-terminal modification is not isotype-dependent. (g) Signal in the anti-biotin Western blots is not due to non-specific streptavidin binding, as neither trastuzumab or brentuximab was visualized in the Western blot after reactions that did not include biotin-LPETGRAGG peptide. (h) Densitometry indicated that biotinylation of trastuzumab and brentuximab are increased 65-70% when the antibodies contain a pilin domain. 195

We found that all the versions of trastuzumab and brentuximab that we generated were biotinylated by sortase. Importantly, though, the level of biotinylation for both antibodies increased —by 72±30% for trastuzumab and 65±23% for brentuximab— when a pilin domain was included at each heavy chain C-terminus compared to the same antibodies without any pilin domains. These increases in biotinylation were found to be statistically significant by a two-tailed t test with p values equal to 0.05 and 0.04 for trastuzumab and brentuximab, respectively. This indicates that the antibodies with pilin domains are modified by both native peptide and isopeptide ligation, and that isopeptide ligation is able to modify the pilin domain with high efficiency. This is consistent with our previous yield estimates for isopeptide ligation, in which we found that pilin domains were modified at 57-75% for Fn3-PLN3-ELP and

~90% for murine anti-Her with one pilin domain at each heavy chain C-terminus. This data once again highlights the need to test sortase-mediated ligation on a case-by-case basis to evaluate the competition between isopeptide and native peptide ligation.

5.4 Significance and Future Directions

We have provided direct evidence—using TRAIL, an ELP, and a peptide array— that sortase is able to catalyze native peptide ligation at N-terminal residues other than glycine. The view of the literature is that N-terminal glycine (preferably, oligoglycine) is

196

necessary for sortase native peptide ligation, though no report has tested whether this reaction can occur at N-terminal residues that are not glycine. We have provided evidence that all amino acids (except cysteine, which was not tested) can be act as nucleophiles in the sortase native peptide ligation reaction. This data is convincing yet preliminary, and it would be interesting to evaluate a number of factors in future work.

For instance, kinetic parameters for the reaction could be evaluated and may differ for each of the different N-terminal amino acids. Additionally, it would be interesting to determine whether the identity of the second or third amino acids in the protein or peptide chain affect the kinetics of native peptide ligation. Importantly, this data does not contradict our earlier work with isopeptide ligation. We conclude that isopeptide ligation is able to modify lysine residues in flexible, solvent-accessible regions of a given protein and this reaction will compete with native peptide ligation at the N-terminus if it is also accessible. Isopeptide ligation can function alone (e.g. in Fn3 and murine antibodies) or in tandem with native peptide ligation (e.g. in TRAIL and human antibodies), and the competition between these two reactions must therefore be tested on a case-by-case basis. We believe this data has contributed to our understanding of the widely-used sortase A enzyme and has provided valuable bioconjugation approaches that can be applied to construct a variety of novel protein-small molecule conjugates.

197

Appendix A

Expected tryptic peptides for digestion of the unpurified polymerization reaction product containing SrtA-ELP and ELP-GLP1. The table shows the tryptic peptide sequences for peaks identified in the MALDI-TOF spectrum. Included are the predicted molecular weight, the experimentally measured molecular weight (calculated by subtracting the mass of H+ from the m/z value), and the difference between these two values. The complete sequences of the SrtA-ELP and ELP-GLP1 are also provided for reference.

SrtA-ELP amino acid sequence

MGQAKPQIPKDKSKVAGYIEIPDADIKEPVYPGPATPEQLNRGVSFAEENESLDDQNIS

IAGHTFIDRPNYQFTNLKAAKKGSMVYFKVGNETRKYKMTSIRDVKPTDVGVLDEQK

GKDKQLTLITCDDYNEKTGVWEKRKIFVATEVKALVTMGVG-(VPGVG)240

ELP-PLN-GLP-LPETG amino acid sequence

MSKGPGVG-(VPGVG)80-

VPGSGLVPRGSSGGWLQDVHVYPKHGGSGRGAHGEGTFTSDVSSYLEEQAAKEFIA

WLVKGAGLPETGRAGG

198

Peak table for tryptic peptides of SrtA-ELP reaction with ELP-PLN-GLP-LPETG

Peak Peptide sequence Predicted Found mass Difference Label mass (Da) [m/z-H+] (Da) (Da) 1 GSSGGWLQDVHVYPK 1630.8 1630.7 0.1 2 HGGSGR 570.6 568.5 2.1 3 GAHGEGTFTSDVSSYLE 2285.4 2285.8 -0.4 EQAAK 4 EFIAWLVK 1006.2 1006.3 -0.1 5 GSSGGWLQDVHVYPK 2182.4 + 625.7 2808.4 -0.3 HGGSGR + GAGLPET = 2808.1 (through lysine ε-amino group) 6 GAGLPETGGG + 2Na+ 814.8 + 44 = 858.3 0.5 858.8 S VAGYIEIPDADIKEPVY 3054.4 3054.1 0.3 RT15-42 PGP ATPEQLNR S GVSFAEENESLDDQNISIAG 3972.3 3972.7 -0.4 RT43-77 HTFIDRPNYQFTNLK GAGLPET + MSK (attachment 625.7 + 364.4 = Not found N/A to N-terminus through α-amino 990.1 group) GAGLPET + SK (attachment 625.7 + 233.2 = Not found N/A to N-terminus through α-amino 858.9 group, methionine loss) GAGLPET + GAHGEGTFTS 625.7 + 3272.6 Not found N/A DVSSYLEEQAAKEFIAWLVK = 3898.3 (attachment to lysine ε-amino group in GLP-1) GAGLPET + EFIAWLVKGAG 625.7 + 1802.1 Not found N/A LPETGGG = 2427.8 GAGLPET + EFIAWLVKGAG 625.7 + 2029.3 Not found N/A LPETGRAGG = 2655.0

199

Appendix B

Peak table for tryptic peptides of SrtA-ELP reaction with dabcyl-LPETG-edans and

ELP-PLN-GLP-LPETG

Peak Peptide sequence Predicted Found Difference Label mass (Da) mass [m/z (Da) -H+] (Da) 1 GSSGGWLQDVHVYPKH 2182.4 + 2873.0 1.2 GGSGR + Dabcyl-LPET 691.8 = 2874.2 1 – 132 GSSGGWLQDVHVYPKH 2874.2 – 132 2741.0 1.2 GGSGR + Dabcyl-LPET – 132 Da = 2742.2 (due to dabcyl fragmentation) 2 GSSGGWLQDVHVYPK 1630.8 1629.3 -0.5 3 GAHGEGTFTSDVSSYLEE 2285.4 2284.5 0.9 QAAK 4 EFIAWLVK 1007.2 1006.3 0.9 5 GSSGGWLQDVHVYPKHG 2808.1 2807.1 1.0 GSGR + GAGLPETG Dabcyl- Dabcyl-LPETGGG + 2Na+ 691.8 + 189 + 925.2 -0.2 LPETGGG 44 = 925.0 Dabcyl- Dabcyl-LPETGGG + 2Na+ 925.0 – 132 = 793.3 -0.3 LPETGGG - 132 Da (due to dabcyl 793.0 – 132 fragmentation) SRT15-42 VAGYIEIPDADIKEPVY 3054.4 3054.1 0.3 PGPATPEQLNR SRT43-77 GVSFAEENESLDDQNISI 3972.3 3971.1 1.2 AGHTFIDRPNYQFTNLK Dabcyl-LPET + MSK 691.8 + 364.4 Not found N/A (attachment to amino- = 1056.2 terminus through α-amino group) Dabcyl-LPET + SK 691.8 + 233.2 Not found N/A (attachment to N-terminus = 925.0 through α-amino group, methionine loss)

200

Appendix C

Expected peptides from tryptic digestion of the unpurified product of the SrtA-

ELP-catalyzed biotinylation of Fn3-PLN3-ELP. Included are the predicted molecular weight, the experimentally measured molecular weight (m/z value – H+ mass), and the difference between these two values. The complete sequences of SrtA-ELP and Fn3-

PLN3-ELP are provided for reference.

SrtA-ELP amino acid sequence

MGQAKPQIPKDKSKVAGYIEIPDADIKEPVYPGPATPEQLNRGVSFAEENESLDDQNIS

IAGHTFIDRPNYQFTNLKAAKKGSMVYFKVGNETRKYKMTSIRDVKPTDVGVLDEQK

GKDKQLTLITCDDYNEKTGVWEKRKIFVATEVKALVTMGVG-(VPGVG)240

Fn3-PLN3-ELP amino acid sequence

MVSDVPRDLEVVAATPTSLLISWDAPAVTVRYYRITYGETGGNSPVQEFTVPGSKSTA

TISGLKPGVDYTITVYAVTPRGDWNEGSKPISINYRTGGTSGGTSGSGSGGGSGGWLQ

DVHVYPKHGGSGRGSGGWLQDVHVYPKHGGSGRGSGGWLQDVHVYPKHGGSGRG

VGV-(VPGVG)60

Fn3-ELP amino acid sequence

201

MVSDVPRDLEVVAATPTSLLISWDAPAVTVRYYRITYGETGGNSPVQEFTVPGSKSTA

TISGLKPGVDYTITVYAVTPRGDWNEGSKPISINYRTGGTSGGTSGSGSGGGVG-

(VPGVG)60

Peak table for tryptic peptides of SrtA-ELP reaction with Fn3-PLN3-ELP and biotin-

LPETG

Peak Peptide sequence Predicted mass Found mass [m/z Differen Label (Da) -H+] (Da) ce (Da) 1 TGGTSGGTSGSGSGGG 3203.3 + 666.8 = 3870.2 -0.1 SGGWLQDVHVYPKHG 3870.1 GSGR + biotin-LPET 2 GSGGWLQDVHVYPKH 2095.3 + 666.8 = 2762.5 -0.4 GGSGR + biotin-LPET 2762.1 (attachment to lysine ε- amino group) 3 GSGGWLQDVHVYPK 1543.7 1543.5 0.2 4 TGGTSGGTSGSGSGGG 2651.8 2651.6 0.2 SGGWLQDVHVYPK Biotin- Biotin-LPETGGG + 2Na+ 666.8 + 189.2 + 44 = 900.6 -0.6 LPET 900.0 GGG SRT28- EPVYPGPATPEQLNR 1668.9 1669.0 -0.1 42 SRT15- VAGYIEIPDADIKEPVY 3054.4 3053.5 0.9 42 PGPATPEQLNR SRT121- QLTLITCDDYNEKTGV 3430.0 3430.8 -0.8 149 WEKRKIFVATEVK SRT43- GVSFAEENESLDDQNIS 3972.3 3972.3 0.0 77 IAGHTFIDRPNYQFTNL K Fn380-94 GDWNEGSKPISINYR 1736.9 1736.2 0.7 Fn335-55 ITYGETGGNSPVQEFTV 2169.4 2168.4 1.0 PGSK Fn380- Biotin-LPET + 666.8 + 1736.9 = 2403.7 0.0 94* GDWNEGSKPISINYR 2403.7

202

(through α-amino group) Fn356-79 STATISGLKPGVDYTIT 2511.9 2512.1 -0.2 VYAVTPR Fn38-31 DLEVVAATPTSLLISWD 2525.9 2526.2 -0.3 APAVTVR Biotin-LPET + 666.8 + 804.0 = Not found N/A MVSDVPR (attachment 1470.8 to amino-terminus through α-amino group) Biotin-LPET + VSDVPR 666.8 + 672.8 = Not found N/A (attachment to amino- 1339.6 terminus through α- amino group, methionine loss) STATISGLKPGVDYTIT 2511.9 + 666.8 = Not found N/A VYAVTPR + biotin-LPET 3178.7 (attachment to lysine ε- amino group in Fn3) ITYGETGGNSPVQEFTV 4662.2 + 666.8 = Not found N/A PGSKSTATISGLKPGVD 5329.0 YTITVYAVTPR + biotin- LPET (attachment to lysine ε-amino group in Fn3)

203

Appendix D

Annotated MS1 spectrum and peak table for LC-MS/MS of biotinylated Fn3-PLN3-ELP.

Peak Peptide sequence Charge Predicted m/z Found m/z label (u = lysine isopeptide- state linked to biotin-LPETG) 1 GSGGWLQDVHVYPK +3 514.9 514.7 2 GDWNEGSKPISINYR +3 579.3 579.0 3 VSDVPR (1Met loss) +1 672.4 671.1 4 GSGGWLQDVHVYPuH +4 690.8 690.6 GGSGR 5 ITYGETGGNSPVQEFT +9 709.3 709.0 VPGSKSTATISGLKPG VDYTITVYAVTPRGD WNEGSKPISINYR 6 ITYGETGGNSPVQEFT +3 723.4 722.9 VPGSK 7 GDWNEGSKPISINYRT +6 731.5 730.0 GGTSGGTSGSGSGGG SGGWLQDVHVYPK + NH4+ 8 DLEVVAATPTSLLISW +4 752.4 752.1 DAPAVTVRYYR 9 TGGTSGGTSGSGSGG +5 774.4 775.3 GSGGWLQDVHVYPuH GGSGR 10 GDWNEGSKPISINYRT +7 801.1 800.2 204

GGTSGGTSGSGSGGG SGGWLQDVHVYPuH GGSGR + NH4+ 11 MVSDVPR +1 819.4 818.5 + oxidation at M 12 TGGTSGGTSGSGSGG +8 827.0 826.6 GSGGWLQDVHVYPuH GGSGRGSGGWLQDV HVYPuHGGSGR 13 STATISGLKPGVDYTIT +3 837.5 837.1 VYAVTPR 14 STATISGLKPGVDYTIT +3 842.8 841.2 VYAVTPR + NH4+ 15 TGGTSGGTSGSGSGGG +11 855.6 855.0 SGGWLQDVHVYPuHG GSGRGSGGWLQDVHV YPuHGGSGRGSGGWL QDVHVYPuHGGSGR + 3NH4+ 16 GDWNEGSKPISINYR +2 868.3 868.4 17 TGGTSGGTSGSGSGG +3 884.1 885.2 GSGGWLQDVHVYPK 18 TGGTSGGTSGSGSGG +6 899.4 900.0 GSGGWLQDVHVYPuH GGSGRGSGGWLQDV HVYPK 19 GSGGWLQDVHVYPuH +3 920.4 919.1 GGSGR 20 GDWNEGSKPISINYRT +6 931.6 932.1 GGTSGGTSGSGSGGG SGGWLQDVHVYPuHG GSGR 21 MVSDVPRDLEVVAAT +4 952.5 954 PTSLLISWDAPAVTVR YYR 22 TGGTSGGTSGSGSGGG +6 991.3 991.2 SGGWLQDVHVYPuHG GSGRGSGGWLQDVHV YPKHGGSGR 23 DLEVVAATPTSLLISW +5 1031.9 1033.6 205

DAPAVTVRYYRITYG ETGGNSPVQEFTVPGSK 24 HGGSGRGSGGWLQDV +5 1076.8 1078.1 HVYPuHGGSGRGSGGW LQDVHVYPKHGGSGR 25 ITYGETGGNSPVQEFTV +8 1126.8 1127.8 PGSKSTATISGLKPGVD YTITVYAVTPRTGGTSG GTSGSGSGGGSGGWLQ DVHVYPK

206

References

1 Dimitrov DS. Therapeutic proteins. Methods Mol. Cell Biol. 2012, 899, 1-26.

2 Muramatsu M, Kinoshita K, Fagarasan S, Yamada S, Shinkai Y, Honjo T. Class switch recombination and hypermutation require activation-induced cytidine deaminase (AID), a potential RNA editing enzyme. Cell, 2000, 102, 553-563.

3 Cantaert T, Schickel JN, Bannock JM, Ng YS, Massad C, Oe T, Wu R, Lavoie A, Walter JE, Notarangelo LD, Al-Herz W, Kilic SS, Ochs HD, Nonoyama S, Durandy A, Meffre E. Activation-Induced Cytidine Deaminase Expression in Human B Cell Precursors Is Essential for Central B Cell Tolerance. Immunity, 2015, 42, 884-895.

4 Köhler G, Milstein C. Continuous cultures of fused cells secreting antibody of predefined specificity. Nature, 1975, 256, 495-497.

5 Rajendra Y, Hougland MD, Alam R, Morehead TA, Barnard GC. A high cell density transient transfection system for therapeutic protein expression based on a CHO GS- knockout cell line: process development and product quality assessment. Biotechnol. Bioeng., 2015, 112, 977-986.

6 Xu Y, Lee J, Tran C, Heibeck TH, Wang WD, Yang J, Stafford RL, Steiner AR, Sato AK, Hallam TJ, Yin G. Production of bispecific antibodies in "knobs-into-holes" using a cell- free expression system. MAbs, 2015, 7, 231-242.

7 Scott AM, Wolchok JD, Old LJ. Antibody Therapy of Cancer. Nat. Rev. Cancer, 2012, 12, 278-287.

8 Heinemann V, von Weikersthal LF, Decker T, Kiani A, Vehling-Kaiser U, Al-Batran SE, Heintges T, Lerchenmüller C, Kahl C, Seipelt G, Kullmann F, Stauch M, Scheithauer W, Hielscher J, Scholz M, Müller S, Link H, Niederle N, Rost A, Höffkes HG, Moehler M, Lindig RU, Modest DP, Rossius L, Kirchner T, Jung A, Stintzing S. FOLFIRI plus cetuximab versus FOLFIRI plus bevacizumab as first-line treatment for patients with metastatic colorectal cancer (FIRE-3): a randomised, open-label, phase 3 trial. Lancet Oncol., 2014, 15, 1065-1075.

9 Loupakis F, Cremolini C, Masi G, Lonardi S, Zagonel V, Salvatore L, Cortesi E, Tomasello G, Ronzoni M, Spadi R, Zaniboni A, Tonini G, Buonadonna A, Amoroso D, Chiara S, Carlomagno C, Boni C, Allegrini G, Boni L, Falcone A. Initial therapy with

207

FOLFOXIRI and bevacizumab for metastatic colorectal cancer. N. Engl. J. Med., 2014, 37, 1609-1618.

10 Earl HM, Hiller L, Dunn JA, Blenkinsop C, Grybowicz L, Vallier AL, Abraham J, Thomas J, Provenzano E, Hughes-Davies L, Gounaris I, McAdam K, Chan S, Ahmad R, Hickish T, Houston S, Rea D, Bartlett J, Caldas C, Cameron DA, Hayward L; ARTemis Investigators. Efficacy of neoadjuvant bevacizumab added to docetaxel followed by fluorouracil, epirubicin, and cyclophosphamide, for women with HER2-negative early breast cancer (ARTemis): an open-label, randomised, phase 3 trial. Lancet Oncol., 2015, 16, 656-666.

11 Piccart-Gebhart MJ, Procter M, Leyland-Jones B, Goldhirsch A, Untch M, Smith I, Gianni L, Baselga J, Bell R, Jackisch C, Cameron D, Dowsett M, Barrios CH, Steger G, Huang CS, Andersson M, Inbar M, Lichinitser M, Láng I, Nitz U, Iwata H, Thomssen C, Lohrisch C, Suter TM, Rüschoff J, Suto T, Greatorex V, Ward C, Straehle C, McFadden E, Dolci MS, Gelber RD; Herceptin Adjuvant (HERA) Trial Study Team. Trastuzumab after adjuvant chemotherapy in HER2-positive breast cancer. N. Engl. J. Med., 2005, 353, 1659- 1672.

12 Swain SM, Baselga J, Kim SB, Ro J, Semiglazov V, Campone M, Ciruelos E, Ferrero JM, Schneeweiss A, Heeson S, Clark E, Ross G, Benyunes MC, Cortés J; CLEOPATRA Study Group. Pertuzumab, trastuzumab, and docetaxel in HER2-positive metastatic breast cancer. N. Engl. J. Med., 2015, 372, 724-734.

13 Iannello A, Ahmad A. Role of antibody-dependent cell-mediated cytotoxicity in the efficacy of therapeutic monoclonal antibodies. Cancer and Metastasis Reviews, 2005, 24, 487-499.

14 Gerdes CA, Nicolini VG, Herter S, van Puijenbroek E, Lang S, Roemmele M, Moessner E, Freytag O, Friess T, Ries CH, Bossenmaier B, Mueller HJ, Umaña P. GA201 (RG7160): a novel, humanized, glycoengineered anti-EGFR antibody with enhanced ADCC and superior in vivo efficacy compared with cetuximab. Clin. Cancer Res., 2013, 19, 1126-1138.

15 Romain G, Senyukov V, Rey-Villamizar N, Merouane A, Kelton W, Liadi I, Mahendra A, Charab W, Georgiou G, Roysam B, Lee DA, Varadarajan N. Antibody Fc engineering improves frequency and promotes kinetic boosting of serial killing mediated by NK cells. Blood, 2014, 124, 3241-3249.

16 Trail PA. Antibody-drug conjugates as cancer therapeutics. Antibodies, 2013, 2, 113-129.

208

17 Okeleley NM, Alley SC, Senter PD. Advancing antibody drug conjugation from the laboratory to an approved anticancer drug. Hematol. Oncol. Clin. N. Amer., 2008, 28, 13- 25.

18 Acchione M, Kwon H, Jochheim CM, Atkins WM. Impact of linker and conjugation chemistry on antigen binding, Fc receptor binding and thermal stability of model antibody-drug conjugates. mAbs, 2012, 4, 362-372.

19 Younes A, Bartlett NL, Leonard JP, Kennedy DA, Lynch CM, Sievers EL, Forero-Torres A. Brentuximab vedotin (SGN-35) for relapsed CD30-positive lymphomas. N. Engl. J. Med., 2010, 363, 1812-1821.

20 Verma S, Miles D, Gianni L, Krop IE, Welslau M, Baselga J, Pegram M, Oh DY, Diéras V, Guardino E, Fang L, Lu MW, Olsen S, Blackwell K; EMILIA Study Group. Trastuzumab emtansine for HER2-positive advanced breast cancer. N. Engl. J. Med., 2012, 367, 1783-1791.

21 Shen BQ, Xu K, Liu L, Raab H, Bhakta S, Kenrick M, Parsons-Reponte KL, Tien J, Yu SF, Mai E, Li D, Tibbitts J, Baudys J, Saad OM, Scales SJ, McDonald PJ, Hass PE, Eigenbrot C, Nguyen T, Solis WA, Fuji RN, Flagella KM, Patel D, Spencer SD, Khawli LA, Ebens A, Wong WL, Vandlen R, Kaur S, Sliwkowski MX, Scheller RH, Polakis P, Junutula JR. Conjugation site modulates the in vivo stability and therapeutic activity of antibody-drug conjugates. Nat. Biotechnol. 2012, 30, 184-189.

22 Hamblett KJ, Senter PD, Chace DF, Sun MM, Lenox J, Cerveny CG, Kissler KM, Bernhardt SX, Kopcha AK, Zabinski RF, Meyer DL, Francisco JA. Effects of drug loading on the antitumor activity of a monoclonal antibody drug conjugate. Clin. Cancer Res., 2004, 10, 7063-7070.

23 Panowski A, Chakta S, Raab H, Polakis P, Junutula JR. Site-specific antibody drug conjugates for cancer therapy. mAbs, 2014, 6, 34-45.

24 Behrens CR, Liu B. Methods for site-specific drug conjugation to antibodies. mAbs, 2014, 6, 46-53.

25 Kim MT, Chen Y, Marhoul J, Jacobson F. Statistical modeling of the drug load distribution on trastuzumab emtansine (Kadcyla), a lysine-linked antibody drug conjugate. Bioconjug. Chem., 2014, 25, 1223-1232.

209

26 Schulz JD, Patt M, Basler S, Kries H, Hilvert D, Gauthier MA, Leroux JC. Site-Specific Polymer Conjugation Stabilizes Therapeutic Enzymes in the Gastrointestinal Tract. doi: 10.1002/adma.201504797. Adv. Mater., 2015, [Epub ahead of print]

27 Mahmood, I Green, MD. Pharmacokinetic and pharmacodynamic considerations in the development of therapeutic proteins. Clin Pharmacokinet., 2005, 44, 331-347.

28 Harris JM, Chess RB. Effect of pegylation on pharmaceuticals. Nat. Rev. Drug Discov., 2003, 2, 214-221.

29 Bunn HF. New agents that stimulate erythropoiesis. Blood, 2007, 109, 868-873.

30 Liu L, Li H, Hamilton SR, Gomathinayagam S, Rayfield WJ, van Maanen M, Yin KC, Hong L, Prueksaritanont T. The impact of sialic acids on the pharmacokinetics of a PEGylated erythropoietin. J. Pharm. Sci., 2012, 101, 4414-4418.

31 Basser RL, Underhill C, Davis I, Green MD, Cebon J, Zalcberg J, MacMillan J, Cohen B, Marty J, Fox RM, Begley CG. Enhancement of platelet recovery after myelosuppressive chemotherapy by recombinant human megakaryocyte growth and development factor in patients with advanced cancer. J. Clin. Oncol., 2000, 18, 2852-2861.

32 Green MD, Koelbl H, Baselga J, Galid A, Guillem V, Gascon P, Siena S, Lalisang RI, Samonigg H, Clemens MR, Zani V, Liang BC, Renwick J, Piccart MJ; International Pegfilgrastim 749 Study Group. A randomized double-blind multicenter phase III study of fixed-dose single-administration pegfilgrastim versus daily filgrastim in patients receiving myelosuppressive chemotherapy. Ann. Oncol., 2003, 14, 29-35.

33 Georgiou G, Valax P. Expression of correctly folded proteins in Escherichia coli. Curr. Opin. In Biotech., 1996, 7, 190-197.

34 Bessette P, Aslund A, Beckwith J, Georgiou G. Efficient folding of proteins with multiple disulfide bonds in the Escherichia coli cytoplasm. Proc. Natl. Acad. Sci. USA, 1999, 96, 13703-13708.

210

36 Ye J, Alvin K, Latif H, Hsu A, Parikh V, Whitmer T, Tellers M, de la Cruz Edmonds MC, Ly J, Salmon P, Markusen JF. Rapid Protein Production Using CHO Stable Transfection Pools. Biotechnol. Prog. 2010, 26 1431-1437.

37 Ye J, Ly J, Watts K, Hsu A, Walker A, McLaughlin K, Berdichevsky M, Prinz B, Sean Kersey D, d'Anjou M, Pollard D, Potgieter T. Optimization of a Glycoengineered Pichia pastoris Cultivation Process for Commercial Antibody Production. Biotechnol. Prog. 2011, 27, 1744-1750.

38 Franzreb M, Muller E, Vajda J. Cost Estimation for Protein A Chromatography: an In Silico Approach to MAb purification strategy. Bioprocess Int. 2014, 12, 44-52.

39 Kobayashi S, Ueda Y. Comparing Protein A Resins for Monoclonal Antibody Purification. Biopharm. Int., 2013, 26.

40 Arnau J, Lauritzen C, Petersen GE, Pedersen J. Current strategies for the use of affinity tags and tag removal for the purification of recombinant proteins. Protein Expr. Purif., 2006, 48, 1-13.

41 Terpe K. Overview of tag protein fusions: from molecular and biochemical fundamentals to commercial systems. Appl. Microbiol. Biotechnol., 2003, 60, 525-533.

42 Jones C, Patel A, Griffin S, Martin J, Young P, O'Donnell K, Silverman C, Porter T, Chaiken I. Current trends in molecular recognition and bioseparation. J. Chromatogr. A, 1995, 707, 3-22.

43 Kapust RB, Waugh DS. Escherichia coli maltose-binding protein is uncommonly effective at promoting the solubility of polypeptides to which it is fused. Protein Science. 1999, 8, 1668-1674.

44 Pennati A, Deng J, Galipeau J. Maltose-binding protein fusion allows for high level bacterial expression and purification of bioactive mammalian cytokine derivatives. PLoS One, 2014, 9, e106724. doi: 10.1371/journal.pone.0106724

45 Jenny RJ, Mann KG, Lundblad RL. A critical review of the methods for cleavage of fusion proteins with thrombin and factor Xa. Protein Expr. Purif., 2003, 31, 1-11.

46 Smyth DR, Mrozkiewicz MK, McGrath WJ, Listwan P, Kobe B. Crystal structures of fusion proteins with large-affinity tags. Protein Sci., 2003, 12, 1313-1322.

211

47 Amiram M, Luginbuhl KM, Li X, Feinglos MN, Chilkoti A. A depot-forming glucagon- like peptide-1 fusion protein reduces blood glucose for five days with a single injection. J. Control. Release, 2013, 172, 144-151.

48 Amiram M, Luginbuhl KM, Li X, Feinglos MN, Chilkoti A. Injectable protease- operated depots of glucagon-like peptide-1 provide extended and tunable glucose control. Proc.Natl. Acad. Sci. USA, 2013, 110, 2792-2797.

50 Bhattacharyya J, Bellucci JJ, Weitzhandler I, McDaniel JR, Spasojevic I, Li X, Lin CC, Chi JT, Chilkoti A. A paclitaxel-loaded recombinant polypeptide nanoparticle outperforms Abraxane in multiple murine cancer models. Nat. Communications, 2015, 6, doi: 10.1038/ncomms8939

51 Mastria EM, Chen M, McDaniel JR, Li X, Hyun J, Dewhirst MW, Chilkoti A. Doxorubicin-conjugated polypeptide nanoparticles inhibit metastasis in two murine models of carcinoma. J. Control. Release, 2015, 208, 52-58.

52 McDaniel JR, Macewan SR, Dewhirst M, Chilkoti A. Doxorubicin-conjugated chimeric polypeptide nanoparticles that respond to mild hyperthermia. J. Control. Release, 2012, 159, 362-367.

53 MacKay JA, Chen M, McDaniel JR, Liu W, Simnick AJ, Chilkoti A. Self-assembling chimeric polypeptide-doxorubicin conjugate nanoparticles that abolish tumours after a single injection. Nat. Materials, 2009, 8, 993-999.

54 Liu W, MacKay JA, Dreher MR, Chen M, McDaniel JR, Simnick AJ, Callahan DJ, Zalutsky MR, Chilkoti A. Injectable intratumoral depot of thermally responsive polypeptide-radionuclide conjugates delays tumor progression in a mouse model. J. Control. Release, 2010, 144, 2-9.

55 Liu W, McDaniel J, Li X, Asai D, Quiroz FG, Schaal J, Park JS, Zalutsky M, Chilkoti A. Brachytherapy using injectable seeds that are self-assembled from genetically encoded polypeptides in situ. Cancer Res., 2012, 72, 5956-5965.

56 Meyer D, Chilkoti A. Purification of recombinant proteins by fusion with thermally- responsive polypeptides. Nat Biotechol., 1999, 17, 1112-1115. 212

57 Hassouneh W, Christensen T, and Chilkoti A. Elastin-like polypeptides as a purification tag for recombinant proteins. Curr Protein Sci., 2010, 61, 6.11.1–6.11.16.

58 MacEwan SR, Hassouneh W, Chilkoti A. Non-chromatographic purification of recombinant elastin-like polypeptides and their fusions with peptides and proteins from Escherichia coli. J. Vis. Exp., 2014, 88, doi: 10.3791/51583.

59 Boswell CA, Mundo EE, Zhang C, Bumbaca D, Valle NR, Kozak KR, Fourie A, Chuh J, Koppada N, Saad O, Gill H, Shen BQ, Rubinfeld B, Tibbitts J, Kaur S, Theil FP, Fielder PJ, Khawli LA, Lin K. Impact of drug conjugation on the pharmacokinetics and tissue distribution of anti-STEAP1 antibody-drug conjugates in rats. Bioconjug. Chem., 2011, 22, 1994-2004.

60 Shen BQ, Xu K, Liu L, Raab H, Bhakta S, Kenrick M, Parsons-Reponte KL, Tien J, Yu SF, Mai E, Li D, Tibbitts J, Baudys J, Saad OM, Scales SJ, McDonald PJ, Hass PE, Eigenbrot C, Nguyen T, Solis WA, Fuji RN, Flagella KM, Patel D, Spencer SD, Khawli LA, Ebens A, Wong WL, Vandlen R, Kaur S, Sliwkowski MX, Scheller RH, Polakis P, Junutula JR. Conjugation site modulates the in vivo stability and therapeutic activity of antibody-drug conjugates. Nat. Biotechnol. 2012, 30, 184-189.

61 Acchione M, Kwon H, Jochheim CM, Atkins WM. Impact of linker and conjugation chemistry on antigen binding, Fc receptor binding and thermal stability of model antibody-drug conjugates. mAbs. 2012, 4, 362-372.

62 Sanderson RJ, Hering MA, James SF, Sun MM, Doronina SO, Siadak AW, Senter PD, Wahl AF. In vivo drug-linker stability of an anti-CD30 dipeptide-linked auristatin immunoconjugate. Clin. Cancer Res., 2005, 11, 843-852.

63 Antczak C, Jaggi JS, LeFave CV, Curcio MJ, McDevitt MR, Scheinberg DA. Influence of the linker on the biodistribution and catabolism of actinium-225 self-immolative tumor- targeted isotope generators. Bioconjug. Chem., 2006, 17, 1551-1560.

64 Roberts SA, Andrews PA, Blanset D, Flagella KM, Gorovits B, Lynch CM, Martin PL, Kramer-Stickland K, Thibault S, Warner G. Considerations for the nonclinical safety evaluation of antibody drug conjugates for oncology. Regul. Toxicol. Pharmacol., 2013, 67, 382-391.

65 Kim MT, Chen Y, Marhoul J, Jacobson F. Statistical modeling of the drug load distribution on trastuzumab emtansine (Kadcyla), a lysine-linked antibody drug conjugate. Bioconjug. Chem., 2014, 25, 1223-1232. 213

66 Goldmacher VS, Amphlett G, Wang L, Lazar AC. Statistics of the Distribution of the Abundance of Molecules with Various Drug Loads in Maytansinoid Antibody-Drug Conjugates. Mol. Pharm., 2015, 12, 1738-1744.

67 Janin-Bussat MC, Dillenbourg M, Corvaia N, Beck A, Klinguer-Hamour C. Characterization of antibody drug conjugate positional isomers at cysteine residues by peptide mapping LC-MS analysis. J. Chromatogr. B. Analyt. Technol. Biomed. Life Sci., 2015, 15, 981-982.

68 Shen BQ, Xu K, Liu L, Raab H, Bhakta S, Kenrick M, Parsons-Reponte KL, Tien J, Yu SF, Mai E, Li D, Tibbitts J, Baudys J, Saad OM, Scales SJ, McDonald PJ, Hass PE, Eigenbrot C, Nguyen T, Solis WA, Fuji RN, Flagella KM, Patel D, Spencer SD, Khawli LA, Ebens A, Wong WL, Vandlen R, Kaur S, Sliwkowski MX, Scheller RH, Polakis P, Junutula JR. Conjugation site modulates the in vivo stability and therapeutic activity of antibody-drug conjugates. Nat. Biotechnol. 2012, 30, 184-189.

69 Hamblett KJ, Senter PD, Chace DF, Sun MM, Lenox J, Cerveny CG, Kissler KM, Bernhardt SX, Kopcha AK, Zabinski RF, Meyer DL, Francisco JA. Effects of drug loading on the antitumor activity of a monoclonal antibody drug conjugate. Clin. Cancer Res., 2004, 10, 7063-7070.

71 Daëron M. Fc receptor biology. Annu Rev Immunol. 1997, 15, 203-234.

72 Hamblett KJ, Senter PD, Chace DF, Sun MM, Lenox J, Cerveny CG, Kissler KM, Bernhardt SX, Kopcha AK, Zabinski RF, Meyer DL, Francisco JA. Effects of drug loading on the antitumor activity of a monoclonal antibody drug conjugate. Clin. Cancer Res., 2004, 10, 7063-7070.

73 Lyon RP, Bovee TD, Doronina SO, Burke PJ, Hunter JH, Neff-LaFord HD, Jonas M, Anderson ME, Setter JR, Senter PD. Reducing hydrophobicity of homogeneous antibody-drug conjugates improves pharmacokinetics and therapeutic index. Nat. Biotechnol. 2015, 33, 733-735. 214

74 Lyon RP, Setter JR, Bovee TD, Doronina SO, Hunter JH, Anderson ME, Balasubramanian CL, Duniho SM, Leiske CI, Li F, Senter PD. Self-hydrolyzing maleimides improve the stability and pharmacological properties of antibody-drug conjugates. Nat. Biotechnol. 2014, 32, 1059-1062.

75 B.Q. Shen, K. Xu, L. Liu, H. Raab, S. Bhakta, M. Kenrick, K.L. Parsons-Reponte, J. Tien, S.F. Yu, E. Mai, D. Li, J. Tibbitts, J. Baudys, O.M. Saad, S.J. Scales, P.J. McDonald, P.E. Hass, C. Eigenbrot, T. Nguyen, W.A. Solis, R.N. Fuji, K.M. Flagella, D. Patel, S.D. Spencer, L.A. Khawli, A. Ebens, W.L. Wong, R. Vandlen, S. Kaur, M.X. Sliwkowski, R.H. Scheller, P. Polakis, J.R. Junutula. Conjugation site modulates the in vivo stability and therapeutic activity of antibody-drug conjugates. Nat. Biotechnol. 2012, 30, 184-189.

76 Ashkenazi A, Holland P, Eckhardt SG. Ligand-based targeting of apoptosis in cancer: the potential of recombinant human apoptosis ligand 2/tumor necrosis factor-related apoptosis-inducing ligand (rhApo2/TRAIL). J. Clin. Oncol. 2008, 26, 3621-3630.

77 Shen BQ, Xu K, Liu L, Raab H, Bhakta S, Kenrick M, Parsons-Reponte KL, Tien J, Yu SF, Mai E, Li D, Tibbitts J, Baudys J, Saad OM, Scales SJ, McDonald PJ, Hass PE, Eigenbrot C, Nguyen T, Solis WA, Fuji RN, Flagella KM, Patel D, Spencer SD, Khawli LA, Ebens A, Wong WL, Vandlen R, Kaur S, Sliwkowski MX, Scheller RH, Polakis P, Junutula JR. Conjugation site modulates the in vivo stability and therapeutic activity of antibody-drug conjugates. Nat. Biotechnol. 2012, 30, 184-189.

78 Huh JH, White AJ, Brych SR, Franey H, Matsumura M. The identification of free cysteine residues within antibodies and a potential role for free cysteine residues in covalent aggregation because of agitation stress. J. Pharm. Sci., 2013, 102, 1701-1711.

79 Sakurai K, Nakahata R, Lee YH, Kardos J, Ikegami T, Goto Y. Effects of a reduced disulfide bond on aggregation properties of the human IgG1 CH3 domain. Biochim. Biophys. Acta., 2015, doi:10.1016/j.bbapap.2015.02.020 [Epub ahead of print].

80 Rispens T, Ooijevaar-de Heer P, Bende O, Aalberse RC. Mechanism of immunoglobulin G4 Fab-arm exchange. J. Am. Chem. Soc., 2011, 133, 10302-10311.

81 Liu YD, Chen X, Enk JZ, Plant M, Dillon TM, Flynn GC. Human IgG2 antibody disulfide rearrangement in vivo. J. Biol. Chem., 2008, 283, 29266-29272. PMCID: PMC2662016.

215

82 Axup JY, Bajjuri KM, Ritland M, Hutchins BM, Kim CH, Kazane SA, Halder R, Forsyth JS, Santidrian AF, Stafin K, Lu Y, Tran H, Seller AJ, Biroc SL, Szydlik A, Pinkstaff JK, Tian F, Sinha SC, Felding-Habermann B, Smider VV, Schultz PG. Synthesis of site- specific antibody-drug conjugates using unnatural amino acids. Proc. Natl. Acad. Sci. USA. 2012, 109, 16101-16106.

83 Xiao H, Chatterjee A, Choi SH, Bajjuri KM, Sinha SC, Schultz PG. Genetic incorporation of multiple unnatural amino acids into proteins in mammalian cells. Angew. Chemie Int. Ed. 2013, 52, 14080-14083.

84 Amiram M, Haimovich AD, Fan C, Wang YS, Aerni HR, Ntai I, Moonan DW, Ma NJ, Rovner AJ, Hong SH, Kelleher NL, Goodman AL, Jewett MC, Söll D, Rinehart J, Isaacs FJ. Evolution of translation machinery in recoded bacteria enables multi-site incorporation of nonstandard amino acids. Nat. Biotechnol., 2015, doi: 10.1038/nbt.3372. [Epub ahead of print].

85 Tian F, Lu Y, Manibusan A, Sellers A, Tran H, Sun Y, Phuong T, Barnett R, Hehli B, Song F, DeGuzman MJ, Ensari S, Pinkstaff JK, Sullivan LM, Biroc SL, Cho H, Schultz PG, DiJoseph J, Dougher M, Ma D, Dushin R, Leal M, Tchistiakova L, Feyfant E, Gerber HP, Sapra P. A general approach to site-specific antibody drug conjugates. Proc. Natl. Acad. Sci. USA. 2014, 111, 1766-1771.

86 Chen I, Howarth M, Lin W, Ting AY. Site-specific labeling of cell surface proteins with biophysical probes using biotin ligase. Nat. Methods. 2005, 2, 99-104.

87 Agarwal P, Bertozzi CR. Site-specific antibody-drug conjugates: the nexus of bioorthogonal chemistry, protein engineering, and drug development. 2015, 26, 176-192.

88 King M, Wagner A. Developments in the field of bioorthogonal bond forming reactions-past and present trends. Bioconjug. Chem., 2014, 25, 825-839.

89 Carrico IS, Carlson BL, Bertozzi CR. Introducing genetically encoded aldehydes into proteins. Nat. Chem. Biol. 2007, 3, 321-322.

90 Hudak JE, Barfield RM, de Hart GW, Grob P, Nogales E, Bertozzi CR, Rabuka D. Synthesis of heterobifunctional protein fusions using copper-free click chemistry and the aldehyde tag. Angew. Chemie Int. Ed. 2012, 51, 4161-4165.

216

91 Wu P, Shui W, Carlson BL, Hu N, Rabuka D, Lee J, Bertozzi CR. Site-specific chemical modification of recombinant proteins produced in mammalian cells by using the genetically encoded aldehyde tag. Proc. Natl. Acad. Sci. USA, 2009, 106, 3000-3005.

92 Drake PM, Albers AE, Baker J, Banas S, Barfield RM, Bhat AS, de Hart GW, Garofalo AW, Holder P, Jones LC, Kudirka R, McFarland J, Zmolek W, Rabuka D. Aldehyde Tag Coupled with HIPS Chemistry Enables the Production of ADCs Conjugated Site- Specifically to Different Antibody Regions with distinct in vivo efficacy and PK outcomes. Bioconjug. Chem., 2014, 25, 1331-1341.

93 Wu P, Shui W, Carlson BL, Hu N, Rabuka D, Lee J, Bertozzi CR. Site-specific chemical modification of recombinant proteins produced in mammalian cells by using the genetically encoded aldehyde tag. Proc. Natl. Acad. Sci. USA. 2009, 106, 3000-3005.

94 Strop P. Versatility of bacterial transglutaminase. Bioconjug. Chem., 2014, 25, 855-862.

95 Mero A, Spolaore B, Veronese FM, Fontana A. Transglutaminase-mediated PEGylation of proteins: direct identification of the sites of protein modification by mass spectrometry using a novel monodisperse PEG. Bioconjug. Chem. 2009, 20, 384-389.

96 Zhou Z, Cironi P, Lin AJ, Xu Y, Hrvatin S, Golan DE, Silver PA, Walsh CT, Yin J. Genetically encoded short peptide tags for orthogonal protein labeling by Sfp and AcpS phosphopantetheinyl transferases. ACS Chem. Biol., 2007, 2, 337-346.

97 Yin J, Straight PD, McLoughlin SM, Zhou Z, Lin AJ, Golan DE, Kelleher NL, Kolter R, Walsh CT. Genetically encoded short peptide tag for versatile protein labeling by Sfp phosphopantetheinyl transferase. Proc. Natl. Acad. Sci. USA, 2005, 102, 15815-15820.

98 Strop P, Liu SH, Dorywalska M, Delaria K, Dushin RG, Tran TT, Ho WH, Farias S, Casas MG, Abdiche Y, Zhou D, Chandrasekaran R, Samain C, Loo C, Rossi A, Rickert M, Krimm S, Wong T, Chin SM, Yu J, Dilley J, Chaparro-Riggers J, Filzen GF, O'Donnell CJ, Wang F, Myers JS, Pons J, Shelton DL, Rajpal A. Location matters: site of conjugation modulates stability and pharmacokinetics of antibody drug conjugates. Chem. Biol., 2013, 20, 161-167.

99 Grünewald J, Klock HE, Cellitti SE, Bursulaya B, McMullan D, Jones DH, Chiu HP, Wang X, Patterson P, Zhou H, Vance J, Nigoghossian E, Tong H, Daniel D, Mallet W, Ou W, Uno T, Brock A, Lesley SA, Geierstanger BH. Efficient Preparation of Site-Specific

217

Antibody-Drug Conjugates Using Phosphopantetheinyl Transferases. Bioconjug. Chem., 2015, 26, 2554-2562.

100 DeDent AC, McAdow M, Schneewind O. Distribution of protein A on the surface of Staphylococcus aureus. J. Bacteriol., 2007, 189, 4473-4484.

101 Pauli NT, Kim HK, Falugi F, Huang M, Dulac J, Henry Dunand C, Zheng NY, Kaur K, Andrews SF, Huang Y, DeDent A, Frank KM, Charnot-Katsikas A, Schneewind O, Wilson PC. Staphylococcus aureus infection induces protein A-mediated immune evasion in humans. J. Exp. Med., 2014, 211, 2331-2339.

102 Ton-That H, Liu G, Mazmanian SK, Faull KF, Schneewind O (1999) Purification and characterization of sortase, the transpeptidase that cleaves surface proteins of Staphylococcus aureus and the LPXTG motif. Proc Natl Acad Sci USA 96(22):12424-12429.

103 Mao H, Hart SA, Schink A, Pollok BA. Sortase-mediated protein ligation: a new method for protein engineering. J. Am. Chem. Soc. 2004, 126, 2670-2671.

104 Ling JJ, Policarpo RL, Rabideau AR, Liao X, Pentelute BL. Protein thioester synthesis enabled by sortase. J. Am. Chem. Soc. 2012, 134, 10749-10752.

105 Willamson DJ, Fascione MA, Webb ME, Turnbull WB. Efficient N-terminal labeling of proteins by use of sortase. Angew. Chem. Int. Ed. 2012, 51, 9377-9380.

106 Swee LK, Guimaraes CP, Sehrawat S, Spooner E, Barrasa MI, Ploegh HL. Sortase- mediated modification of αDEC205 affords optimization of antigen presentation and immunization against a set of viral epitopes. Proc. Natl. Acad. Sci. USA. 2013, 110, 1428- 1433.

107 Antos JM, Chew GL, Guimaraes CP, Yoder NC, Grotenbreg GM, Popp MW, Ploegh HL. Site-specific N- and C-terminal labeling of a single polypeptide using sortases of different specificity. J. Am. Chem. Soc. 2009, 131, 10800-10801.

108 Ilangovan U, Ton-That H, Iwahara J, Schneewind I, and Clubb RT. Structure of sortase, the transpeptidase that anchors proteins to the cell wall of Staphylococcus aureus. Proc. Natl. Acad. Sci. USA. 2001, 98, 6056-6061.

218

109 Kruger RG, Otvos B, Frankel BA, Bentley M, Dostal P, McCafferty DG. Analysis of the substrate specificity of the Staphylococcus aureus sortase transpeptidase SrtA. Biochemistry. 2004, 43, 1541-1551.

110 Ameloot P, Declercg W, Fiers W, Vandenabeele P, and Brouckaert P. Heterotrimers formed by tumor necrosis factors of different species or muteins. J. Biol. Chem. 2001, 276, 27098-27103.

111 Zhang XM, Weber I, Chen MJ. Site-directed mutational analysis of human tumor necrosis factor-alpha receptor binding site and structure-functional relationship. J. Biol. Chem. 1992, 267, 24069-24075.

112 Schoenfeld HJ, Poeschl B, Frey JR, Loetscher H, Hunziker W, Lustig A, Zulauf M. Efficient purification of recombinant human tumor necrosis factor β from Escherichia coli yields biologically active protein with a trimeric structure that binds to both tumor necrosis factor receptors. J. Biol. Chem. 1991, 266, 3863-3869.

113 Flick DA and Gifford GE. Comparison of in vitro cell cytotoxic assays for tumor necrosis factor. J. Immunol. Methods. 1984, 68, 167-175.

114 Liddil JD, Dorr RT, Scuderi P. Association of lysosomal activity with sensitivity and resistance to tumor necrosis factor in murine L929 cells. Cancer Res. 1989, 49, 2722-2728.

115 Mao H, Hart SA, Schink A, Pollok BA. Sortase-mediated protein ligation: a new method for protein engineering. J. Am. Chem. Soc. 2004, 126, 2670-2671.

116 Ling JJ, Policarpo RL, Rabideau AR, Liao X, Pentelute BL. Protein thioester synthesis enabled by sortase. J. Am. Chem. Soc. 2012, 134, 10749-10752.

117 Willamson DJ, Fascione MA, Webb ME, Turnbull WB. Efficient N-terminal labeling of proteins by use of sortase. Angew. Chem. Int. Ed. 2012, 51, 9377-9380.

118 Swee LK, Guimaraes CP, Sehrawat S, Spooner E, Barrasa MI, Ploegh HL. Proc. Natl. Acad. Sci. USA. 2013, 110, 1428-1433.

219

119 Chen JJ, Knudsen S, Mazin W, Dahlgaard J, Zhang B. A 71-gene signature of TRAIL sensitivity in cancer cells. Mol. Cancer Ther., 2012, 11, 34-44.

120 Bodmer JL, Meier P, Tschopp J, Schneider P. Cysteine 230 is essential for the structure and activity of the cytotoxic ligand TRAIL. J. Biol. Chem., 2000, 275, 20632-20637.

121 Lin TP, Shahrokh Z, Flores H, Pai R, Matthews TC. Apo2 ligand/TRAIL formulations. U.S. Patent 7,741,282 B2, June 22, 2010.

122 Ton-That H, Schneewind O. Assembly of pili on the surface of Corynebacterium diptheriae. Mol. Microbiol. 2003, 50, 1429-1438.

123 Maraffini LA, Dedent AC, Schneewind O. Sortases and the art of anchoring proteins to the envelopes of Gram-positive bacteria. Microbiol. Mol. Biol. Rev. 2006, 70, 192-221.

124 Dasgupta S, Samantaray S, Sahal D, Roy RP. Isopeptide ligation catalyzed by quintessential sortase A: mechanistic cues from cyclic and branched oligomers of indolicidin. J. Biol. Chem. 2011, 286, 23996-24006.

125 Meyer DR, Chilkoti A. Purification of recombinant proteins by fusion with thermally- responsive polypeptides. Nat. Biotechnol. 1999, 17, 1112-1115.

126 Bellucci JJ, Amiram M, Bhattacharyya J, McCafferty D, Chikoti A. Three-in-one chromatography-free purification, tag removal, and site-specific modification of recombinant fusion proteins using sortase A and elastin-like polypeptides. Angew. Chem. Int. Ed. 2013, 52, 3703-3708.

128 Duan J, Wu J, Valencia CA, Liu R. Fibronectin type III domain based monobody with high avidity. Biochemistry 2007, 46, 12656-12664.

129 Bendell J, Saleh M, Rose AA, Siegel PM, Hart L, Sirpal S, Jones S, Green J, Crowley E, Simantov R, Keler T, Davis T, Vahdat L. Phase I/II study of the antibody-drug conjugate glembatumumab vedotin in patients with locally advanced or metastatic breast cancer. J. Clin. Oncol., 2014, 32, 3619-3625.

220

130 Scales SJ, Gupta N, Pacheco G, Firestein R, French DM, Koeppen H, Rangell L, Barry- Hamilton V, Luis E, Chuh J, Zhang Y, Ingle GS, Fourie-O'Donohue A, Kozak KR, Ross S, Dennis MS, Spencer SD. Mol. Cancer Ther., 2014, 13, 2630-2640.

131 Fanale MA, Forero-Torres A, Rosenblatt JD, Advani RH, Franklin AR, Kennedy DA, Han TH, Sievers EL, Bartlett NL. A phase I weekly dosing study of brentuximab vedotin in patients with relapsed/refractory CD30-positive hematologic malignancies. Clin. Cancer Res., 2012, 18, 248-255.

132 Younes A, Bartlett NL, Leonard JP, Kennedy DA, Lynch CM, Sievers EL, Forero- Torres A. Brentuximab vedotin (SGN-35) for relapsed CD30-positive lymphomas. N. Engl. J. Med., 2010, 363, 1812-1821.

133 Oflazoglu E, Stone IJ, Gordon K, Wood CG, Repasky EA, Grewal ID, Law CL, Gerber HP. Potent anticarcinoma activity of the humanized anti-CD70 antibody h1F6 conjugated to the tubulin inhibitor auristatin via an uncleavable linker. Clin. Cancer Res., 2008, 14, 6171-6180.

134 Smith LM, Nesterova A, Alley SC, Torgov MY, Carter PJ. Potent cytotoxicity of an auristatin-containing antibody-drug conjugate targeting melanoma cells expressing melanotransferrin/p97. Mol. Cancer Ther., 2006, 5, 1474-1482.

135 Smith LM, Nesterova A, Ryan MC, Duniho S, Jonas M, Anderson M, Zabinski RF, Sutherland MK, Gerber HP, Van Orden KL, Moore PA, Ruben SM, Carter PJ. CD133/prominin-1 is a potential therapeutic target for antibody-drug conjugates in hepatocellular and gastric cancers. Br. J. Cancer, 2008, 99, 100-109.

136 Kim SY, Theunissen JW, Balibalos J, Liao-Chan S, Babcock MC, Wong T, Cairns B, Gonzalez D, van der Horst EH, Perez M, Levashova Z, Chinn L, D'Alessio JA, Flory M, Bermudez A, Jackson DY, Ha E, Monteon J, Bruhns MF, Chen G, Migone TS. A novel antibody-drug conjugate targeting SAIL for the treatment of hematologic malignancies. Blood Cancer J., 2015, 5:e316. doi: 10.1038/bcj.2015.39.

137 Verma S, Miles D, Gianni L, Krop IE, Welslau M, Baselga J, Pegram M, Oh DY, Diéras V, Guardino E, Fang L, Lu MW, Olsen S, Blackwell K; EMILIA Study Group. Trastuzumab emtansine for HER2-positive advanced breast cancer. N. Engl. J. Med., 2012, 367, 1783-1791.

138 Younes A, Gopal AK, Smith SE, Ansell SM, Rosenblatt JD, Savage KJ, Ramchandren R, Bartlett NL, Cheson BD, de Vos S, Forero-Torres A, Moskowitz CH, Connors JM,

221

Engert A, Larsen EK, Kennedy DA, Sievers EL, Chen R. Chen. Results of a pivotal phase II study of brentuximab vedotin for patients with relapsed or refractory Hodgkin's lymphoma. J. Clin. Oncol., 2012, 30, 2183-2189.

139 Fanale MA, Horwitz SM, Forero-Torres A, Bartlett NL, Advani RH, Pro B, Chen RW, Davies A, Illidge T, Huebner D, Kennedy DA, Shustov AR. Brentuximab vedotin in the front-line treatment of patients with CD30+ peripheral T-cell lymphomas: results of a phase I study. J. Clin. Oncol., 2014, 32, 3137-3143.

140 Pro B, Advani R, Brice P, Bartlett NL, Rosenblatt JD, Illidge T, Matous J, Ramchandren R, Fanale M, Connors JM, Yang Y, Sievers EL, Kennedy DA, Shustov A. Brentuximab vedotin (SGN-35) in patients with relapsed or refractory systemic anaplastic large-cell lymphoma: results of a phase II study. J. Clin. Oncol., 2012, 30, 2190-2196.

141 Trail PA. Antibody-drug conjugates as cancer therapeutics. Antibodies, 2013, 2, 113- 129.

142 Okeleley NM, Alley SC, Senter PD. Advancing antibody drug conjugation from the laboratory to an approved anticancer drug. Hematol. Oncol. Clin. N. Amer., 2008, 28, 13- 25.

143 Acchione M, Kwon H, Jochheim CM, Atkins WM. Impact of linker and conjugation chemistry on antigen binding, Fc receptor binding and thermal stability of model antibody-drug conjugates. mAbs, 2012, 4, 362-372.

144 Shen BQ, Xu K, Liu L, Raab H, Bhakta S, Kenrick M, Parsons-Reponte KL, Tien J, Yu SF, Mai E, Li D, Tibbitts J, Baudys J, Saad OM, Scales SJ, McDonald PJ, Hass PE, Eigenbrot C, Nguyen T, Solis WA, Fuji RN, Flagella KM, Patel D, Spencer SD, Khawli LA, Ebens A, Wong WL, Vandlen R, Kaur S, Sliwkowski MX, Scheller RH, Polakis P, Junutula JR. Conjugation site modulates the in vivo stability and therapeutic activity of antibody-drug conjugates. Nat. Biotechnol. 2012, 30, 184-189.

145 Hamblett KJ, Senter PD, Chace DF, Sun MM, Lenox J, Cerveny CG, Kissler KM, Bernhardt SX, Kopcha AK, Zabinski RF, Meyer DL, Francisco JA. Effects of drug loading on the antitumor activity of a monoclonal antibody drug conjugate. Clin. Cancer Res., 2004, 10, 7063-7070.

146 Panowski S, Chakta S, Raab H, Polakis P, Junutula JR. Site-specific antibody drug conjugates for cancer therapy. mAbs, 2014, 6, 34-45.

222

147 Behrens CR, Liu B. Methods for site-specific drug conjugation to antibodies. mAbs, 2014, 6, 46-53.

149 Tolcher AW, Sugarman S, Gelmon KA, Cohen R, Saleh M, Isaacs C, Young L, Healey D, Onetto N, Slichenmyer W. Randomized phase II study of BR96-doxorubicin conjugate in patients with metastatic breast cancer. J. Clin. Oncol., 1999, 17, 478-484.

150 Wang W, Wang EQ, Balthasar JP. Monoclonal antibody pharmacokinetics and pharmacodynamics. Clin. Pharmacol. Ther., 2008, 84, 548-558.

151 Thurber GM, Schmidt MM, Wittrup KD. Antibody tumor penetration: transport opposed by systemic and antigen-mediated clearance. Adv. Drug Deliv. Rev., 2008, 60, 1421-1434.

152 Katz J, Janik JE, Younes A. Brentuximab Vedotin (SGN-35). Clin. Cancer Res., 2011, 17, 6428-6436.

153 Younes A, Gopal AK, Smith SE, Ansell SM, Rosenblatt JD, Savage KJ, Ramchandren R, Bartlett NL, Cheson BD, de Vos S, Forero-Torres A, Moskowitz CH, Connors JM, Engert A, Larsen EK, Kennedy DA, Sievers EL, Chen R. Results of a pivotal phase II study of brentuximab vedotin for patients with relapsed or refractory Hodgkin's lymphoma. J. Clin. Oncol., 2012, 30, 2183-2189.

154 Pro B, Advani R, Brice P, Bartlett NL, Rosenblatt JD, Illidge T, Matous J, Ramchandren R, Fanale M, Connors JM, Yang Y, Sievers EL, Kennedy DA, Shustov A. Brentuximab vedotin (SGN-35) in patients with relapsed or refractory systemic anaplastic large-cell lymphoma: results of a phase II study. J. Clin. Oncol., 2012, 30, 2190-2196.

155 Pettit GR, Singh SB, Hogan F, Williams PL, Herald DL, Burkett DD, Clewlow PJ. The absolute configuration and synthesis of natural (-)-dolastatin 10. J. Am. Chem. Soc., 1989, 111, 5463-5465.

156 Francisco JA, Cerveny CG, Meyer DL, Mixan BJ, Klussman K, Chace DF, Rejniak SX, Gordon KA, DeBlanc R, Toki BE, Law CL, Doronina SO, Siegall CB, Senter PD, Wahl AF.

223

cAC10-vcMMAE, an anti-CD30-monomethyl auristatin E conjugate with potent and selective antitumor activity. Blood, 2003, 102, 1458-1465.

157 Bhatt S, Ashlock BM, Natkunam Y, Sujoy V, Chapman JR, Ramos JC, Mesri EA, Lossos IS. CD30 targeting with brentuximab vedotin: a novel therapeutic approach to primary effusion lymphoma. Blood, 2013, 122, 1233-1242.

158 Horwitz SM, Advani RH, Bartlett NL, Jacobsen ED, Sharman JP, O'Connor OA, Siddiqi T, Kennedy DA, Oki Y. Objective responses in relapsed T-cell lymphomas with single-agent brentuximab vedotin. Blood, 2014, 123, 3095-3100.

159 Bonifacino JS, Dell'Angelica EC, Springer TA. Immunoprecipitation. Curr. Protoc. Immunol., 2001, Chapter 8, Unit 8.3.

160 Austin CD, De Mazière AM, Pisacane PI, van Dijk SM, Eigenbrot C, Sliwkowski MX, Klumperman J, Scheller RH. Endocytosis and sorting of ErbB2 and the site of action of cancer therapeutics trastuzumab and gledanamycin. Molecular Biology of the Cell. 2004, 15, 5268-5282.

161 Rudnick SI, Lou J, Shaller CC, Tang Y, Klein-Szanto AJ, Weiner LM, Marks JD, Adams GP. Influence of affinity and antigen internalization on the uptake and penetration of anti-Her2 antibodies in solid tumors. Cancer Res. 2011, 71, 2550-2259.

162 Hudziak RM, Lewis GD, Winget M, Fendly BM, Shepard HM, Ullrich A. p185Her2 monoclonal antibody has antiproliferative effects in vitro and sensitizes human breast tumor cells to tumor necrosis factor. Mol. Cell Biol. 1989, 9, 1165-1172.

163 Lewis GD, Figari I, Fendly B, Wong WL, Carter P, Gorman C, Shepard HM. Differential responses of human tumor cell lines to anti-p185Her2 monoclonal antibodies. Cancer Immunol. Immunother. 1993, 37, 255-263.

164 Carter P, Presta L, Gorman CM, Ridgway JB, Henner D, Wong WL, Rowland AM, Kotts C, Carver ME, Shepard HM. Humanization of an anti-p185Her2 antibody for human cancer therapy. Proc. Natl. Acad. Sci. USA. 1992, 89, 4285-4289.

165 Carter P, Presta L, Gorman CM, Ridgway JB, Henner D, Wong WL, Rowland AM, Kotts C, Carver ME, Shepard HM. Humanization of an anti-p185Her2 antibody for human cancer therapy. Proc. Natl. Acad. Sci. USA. 1992, 89, 4285-4289.

224

166 Jung ST, Reddy ST, Kang TH, Borrok MJ, Sandlie I, Tucker PW, Georgiou G. Aglycosylated IgG variants expressed in bacteria that selectively bind FcgammaRI potentiate tumor cell killing by monocyte-dendritic cells. Proc. Natl. Acad. Sci. USA, 2010, 107, 604-609.

167 Mazor Y, Van Blarcom T, Mabry R, Iverson BL, Georgiou G. Isolation of engineered, full-length antibodies from libraries expressed in Escherichia coli. Nat. Biotechnol., 2007, 25, 563-565.

168 Stern B, Olsen LC, Trobe C, Ravenberg H, Pryme IF. Improving mammalian cell factories: the selection of signal peptide has a major impact on recombinant protein synthesis and secretion in mammalian cells. Trends in Cell & Molecular Biology. 2007, 2, 1- 17.

169 Zhang L, Leng Q, Mixson AJ. Alteration in the IL-2 signal peptide affects secretion of proteins in vitro and in vivo. J. Gene Med. 2005, 7, 354-365.

170 Kober L, Zehe C, Bode J. Optimized signal peptides for the development of high expressing CHO cell lines. Biotechnol. Bioeng. 2013, 110, 1164-1173.

171 Lucas BK, Giere LM, DeMarco RA, Shen A, Chisholm V, Crowley CW. High-level production of recombinant proteins in CHO cells using a dicistronic DHFR intron expression vector. Nucleic Acids Res., 1996, 24, 1774-1779.

172 Khamlichi AA, Rocca A, Cogné M. The effect of intron sequences on expression levels of Ig cDNAs. Gene, 1994, 150, 387-390.

174 Zufferey R, Donello JE, Trono D, Hope TJ. Woodchuck hepatitis virus posttranslational regulatory element enhances expression of transgenes delivered by retroviral vectors. J. Virol. 1999, 73, 2886-2892.

175 Wulhfard S, Baldi L, Hacker DL, Wurm F. Valproic acid enhances recombinant mRNA and protein levels in transiently transfected Chinese hamster ovary cells. J. Biotechnol. 2010, 148, 128-132.

225

176 Dalton AC, Barton WA. Over-expression of secreted proteins from mammalian cell lines. Prot. Sci. 2014, 23, 517-525.

177 Backliwal G, Hildinger M, Kuettel I, Delegrange F, Hacker DL, Wurm FM. Valproic acid: a viable alternative to sodium butyrate for enhancing protein expression in mammalian cell cultures. Biotechnol. Bioeng. 2008, 101, 182-189.

178 Soehnlein O, Zernecke A, Eriksson EE, Rothfuchs AG, Pham CT, Herwald H, Bidzhekov K, Rottenberg ME, Weber C, Lindbom L. Neutrophil secretion products pave the way for inflammatory monocytes. Blood. 2008, 112, 1461-1471.

179 Soehnlein O, Xie X, Ulbrich H, Kenne E, Rotzius P, Flodgaard H, Eriksson EE, Lindbom L. Neutrophil-derived heparin-binding protein (HBP/CAP37) deposited on endothelium enhances monocyte arrest under flow conditions. J. Immunol. 2005, 174, 6399-6405.

180 Kober L, Zehe C, Bode J. Optimized signal peptides for the development of high expressing CHO cell lines. Biotechnol. Bioeng. 2013, 110, 1164-1173.

181 Olczak M, Olczak T. Comparison of different signal peptides for protein secretion in nonlytic insect cell system. Anal. Biochem. 2006, 359, 45-53.

182 Petersen TN, Brunak S, von Heijne G, Nielsen H. SignalP 4.0: discriminating signal peptides from transmembrane regions. Nat. Methods. 2011, 28, 785-786.

183 Emanuelsson O, Brunak S, von Heijne G, Nielsen H. Locating proteins in the cell using TargetP, SignalP and related tools. Nat. Protocols. 2007, 2, 953-971.

184 Bendtsen JD, Nielsen H, von Heijne G, Brunak S. Improved prediction of signal peptides: SignalP 3.0. J. Mol. Bio. 2004, 340, 783-795.

185 Beate Stern B, Olsen LC, Trobe C, Ravneberg H, Pryme IF. Improving mammalian cell factories: The selection of signal peptide has a major impact on recombinant protein synthesis and secretion in mammalian cells. Trends Cell Mol. Biol. 2007, 2, 1-17.

186 Martoglio B, Dobberstein B. Signal sequences: more than just greasy peptides. Trends Cell Biol. 1998, 8, 410-415.

187 Drake PM, Albers AE, Baker J, Banas S, Barfield RM, Bhat AS, de Hart GW, Garofalo AW, Holder P, Jones LC, Kudirka R, McFarland J, Zmolek W, Rabuka D. Aldehyde tag

226

coupled with HIPS chemistry enables the production of ADCs conjugated site- specifically to different antibody regions with distinct in vivo efficacy and PK outcomes. Bioconjug. Chem. 2014, 25, 1331-1341.

188 Erickson HK, Park PU, Widdison WC, Kovtun YV, Garrett LM, Hoffman K, Lutz RJ, Goldmacher VS, Blättler WA. Antibody-maytansinoid conjugates are activated in targeted cancer cells by lysosomal degradation and linker-dependent intracellular processing. Cancer Res., 2006, 66, 4426-4433.

189 Uppal H, Doudement E, Mahapatra K, Darbonne WC, Bumbaca D, Shen BQ, Du X, Saad O, Bowles K, Olsen S, Lewis Phillips GS, Hartley D, Sliwkowski MX, Girish S, Dambach D, Ramakrishnan V. Potential mechanisms for thrombocytopenia development with trastuzumab emtansine (T-DM1). Clin. Cancer Res., 2015, 21, 123-133.

190 Austin CD, Wen X, Gazzard L, Nelson C, Scheller RH, Scales SJ. Oxidizing potential of endosomes and lysosomes limits intracellular cleavage of disulfide-based antibody- drug conjugates. Proc. Natl. Acad. Sci. USA, 2005, 102, 17987-17992.

191 Smith LM, Nesterova A, Alley SC, Torgov MY, Carter PJ. Potent cytotoxicity of an auristatin-containing antibody-drug conjugate targeting melanoma cells expressing melanotransferrin/p97. Mol. Cancer Ther., 2006, 5, 1474-1482.

192 Sutherland MS, Sanderson RJ, Gordon KA, Andreyka J, Cerveny CG, Yu C, Lewis TS, Meyer DL, Zabinski RF, Doronina SO, Senter PD, Law CL, Wahl AF. Lysosomal trafficking and cysteine protease metabolism confer target-specific cytotoxicity by peptide-linked anti-CD30-auristatin conjugates. J. Biol. Chem., 2006, 281, 10540-10547.

193 Francisco JA, Cerveny CG, Meyer DL, Mixan BJ, Klussman K, Chace DF, Rejniak SX, Gordon KA, DeBlanc R, Toki BE, Law CL, Doronina SO, Siegall CB, Senter PD, Wahl AF. cAC10-vcMMAE, an anti-CD30-monomethyl auristatin E conjugate with potent and selective antitumor activity. Blood, 2003, 102, 1458-1465.

194 Kruger RG, Otvos B, Frankel BA, Bentley M, Dostal P, McCafferty DG. Analysis of the substrate specificity of the Staphylococcus aureus sortase transpeptidase SrtA. Biochemistry, 2004, 43, 1541-1551.

195 Kruger RG, Otvos B, Frankel BA, Bentley M, Dostal P, McCafferty DG. Analysis of the substrate specificity of the Staphylococcus aureus sortase transpeptidase SrtA. Biochemistry, 2004, 1541-1551.

227

196 Doronina SO, Bovee TD, Meyer DW, Miyamoto JB, Anderson ME, Morris-Tilden CA, Senter PD. Novel peptide linkers for highly potent antibody-auristatin conjugate. Bioconjug. Chem. 2008, 19, 1960-1963.

197 Albers AE, Garofalo AW, Drake PM, Kudirka R, de Hart GW, Barfield RM, Baker J, Banas S, Rabuka D. Exploring the effects of linker composition on site-specifically modified antibody-drug conjugates. European Journal of Medicinal Chemistry, 2014, 17, 3-9.

199 Hudziak RM, Lewis GD, Winget M, Fendly BM, Shepard HM, Ullrich A. p185HER2 monoclonal antibody has antiproliferative effects in vitro and sensitizes human breast tumor cells to tumor necrosis factor. Mol. Cell Biol., 1989, 9, 1165-1172.

200 Bryant P, Pabst M, Badescu G, Bird M, McDowell W, Jamieson E, Swierkosz J, Jurlewicz K, Tommasi R, Henseleit K, Sheng X, Camper N, Manin A, Kozakowska K, Peciak K, Laurine E, Grygorash R, Kyle A, Morris D, Parekh V, Abhilash A, Choi JW, Edwards J, Frigerio M, Baker MP, Godwin A. In Vitro and In Vivo Evaluation of Cysteine Rebridged Trastuzumab-MMAE Antibody Drug Conjugates With Defined DARs. Mol. Pharm., 2015, 12, 1872-1879.

201 Albers AE, Garofalo AW, Drake PM, Kudirka R, de Hart GW, Barfield RM, Baker J, Banas S, Rabuka D. Exploring the effects of linker composition on site-specifically modified antibody-drug conjugates. Eur. J. Med. Chem., 2014, 88, 3-9.

202 Dennler P, Chiotellis A, Fischer E, Brégeon D, Belmant C, Gauthier L, Lhospice F, Romagne F, Schibli R. A Transglutaminase-Based Chemo-Enzymatic Conjugation Approach Yields Homogeneous Antibody-Drug Conjugates. Bioconjug. Chem., 2014, 25, 569-578.

203 Asundi J, Crocker L, Tremayne J, Chang P, Sakanaka C, Tanguay J, Spencer S, Chalasani S, Luis E, Gascoigne K, Desai R, Raja R, Friedman BA, Haverty PM, Polakis P, Firestein R. An antibody drug conjugate (ADC) targeting the Lymphocyte Antigen 6 Complex, Locus E (LY6E) provides robust tumor killing in a wide range of solid tumor malignancies. Clin. Cancer Res., 2015, 21, 3252-3262.

204 Dorywalska M, Strop P, Melton-Witt JA, Hasa-Moreno A, Farias SE, Galindo Casas M, Delaria K, Lui V, Poulsen K, Loo C, Krimm S, Bolton G, Moine L, Dushin R, Tran TT, Liu

228

SH, Rickert M, Foletti D, Shelton DL, Pons J, Rajpal A. Effect of attachment site on stability of cleavable antibody drug conjugates. Bioconjug. Chem., 2015, 26, 650-659.

205 Jeffrey SC, Burke PJ, Lyon RP, Meyer DW, Sussman D, Anderson M, Hunter JH, Leiske CI, Miyamoto JB, Nicholas ND, Okeley NM, Sanderson RJ, Stone IJ, Zeng W, Gregson SJ, Masterson L, Tiberghien AC, Howard PW, Thurston DE, Law CL, Senter PD. A Potent Anti-CD70 Antibody−Drug Conjugate Combining a Dimeric Pyrrolobenzodiazepine Drug with Site-Specific Conjugation Technology. Bioconjug. Chem., 2013, 24, 1256-1263.

206 Golfier S, Kopitz C, Kahnert A, Heisler I, Schatz CA, Stelte-Ludwig B, Mayer- Bartschmid A, Unterschemmann K, Bruder S, Linden L, Harrenga A, Hauff P, Scholle FD, Müller-Tiemann B, Kreft B, Ziegelbauer K. Anetumab ravtansine: a novel mesothelin-targeting antibody-drug conjugate cures tumors with heterogeneous target expression favored by bystander effect. Mol. Cancer Ther., 2014, 13, 1537-1538.

207 Deckert J, Park PU, Chicklas S, Yi Y, Li M, Lai KC, Mayo MF, Carrigan CN, Erickson HK, Pinkas J, Lutz RJ, Chittenden T, Lambert JM. A novel anti-CD37 antibody-drug conjugate with multiple anti-tumor mechanisms for the treatment of B-cell malignancies. Blood, 2013, 122, 3500-3510.

208 Elgersma RC, Coumans RG, Huijbregts T, Menge WM, Joosten JA, Spijker HJ, de Groot FM, van der Lee MM, Ubink R, van den Dobbelsteen DJ, Egging DF, Dokter WH, Verheijden GF, Lemmens JM, Timmers CM, Beusker PH. Design, Synthesis, and Evaluation of Linker-Duocarmycin Payloads: Toward Selection of HER2-Targeting Antibody-Drug Conjugate SYD985. Mol. Pharm., 2015, 12, 1813-1835.

209 Drake PM1, Albers AE, Baker J, Banas S, Barfield RM, Bhat AS, de Hart GW, Garofalo AW, Holder P, Jones LC, Kudirka R, McFarland J, Zmolek W, Rabuka D. Aldehyde Tag Coupled with HIPS Chemistry Enables the Production of ADCs Conjugated Site- Specifically to Different Antibody Regions with distinct in vivo efficacy and PK outcomes. Bioconjug. Chem., 2014, 25, 1331-1341.

210 Verma VA, Pillow TH, DePalatis L, Li G, Phillips GL, Polson AG, Raab HE, Spencer S, Zheng B. The cyptophycins as potent payloads for antibody drug conjugates. Bioorg. Med. Chem. Lett., 2015, 25, 864-868.

211 Maraffini LA, Dedent AC, Schneewind O. Sortases and the art of anchoring proteins to the envelopes of Gram-positive bacteria. Microbiol. Mol. Biol. Rev., 2006, 70, 192-221.

229

212 Richards J, Miller M, Abend J, Koide A, Koide S, Dewhurst S. J. Mol. Biol. 2003, 326, 1475-1488.

213 Duan J, Wi J, Valencia CA, Liu R, Biochemistry 2007, 46, 12656-12664.

214 Mongkolsapaya J, Grimes JM, Chen N, Xu XN, Stuart DI, Jones EY, Screaton GR. Structure of the TRAIL-DR5 complex reveals mechanisms conferring specificity in apoptotic initiation. Nat. Struct. Biol., 1999, 6, 1048-1053.

215 Cha SS, Kim MS, Choi YH, Sung BJ, Shin NK, Shin HC, Sung YC, Oh BH. 2.8 A resolution crystal structure of human TRAIL, a cytokine with selective antitumor activity. Immunity, 1999, 11, 253-261.

216 Cha SS, Sung BJ, Kim YA, Song YL, Kim HJ, Kim S, Lee MS, Oh BH. Crystal structure of TRAIL-DR5 complex identifies a critical role of the unique frame insertion in conferring recognition specificity. J. Biol. Chem., 2000, 275, 31171-31177.

217 Mongkolsapaya J, Grimes JM, Chen N, Xu XN, Stuart DI, Jones EY, Screaton GR. Structure of the TRAIL-DR5 complex reveals mechanisms conferring specificity in apoptotic initiation. Nat. Struct. Biol., 1999, 6, 1048-1053.

219 Perry AM, Ton-That H, Mazmanian SK, Schneewind O. Anchoring of surface proteins to the cell wall of Staphylococcus aureus. III. Lipid II is an in vivo peptidoglycan substrate for sortase-catalyzed surface protein anchoring. J. Biol. Chem., 2002, 277, 16241- 16248.

220 Dodds AW, Ren XD, Willis AC, Law SK. The reaction mechanism of the internal thioester in the human complement component C4. Nature, 1996, 379, 177-179.

221 Hnízda A, Santrůcek J, Sanda M, Strohalm M, Kodícek M. Reactivity of histidine and lysine side-chains with diethylpyrocarbonate -- a method to identify surface exposed residues in proteins. J. Biochem. Biophys.Methods, 2008, 70, 1091-1097.

222 Miles EW. Modification of histidyl residues in proteins by diethylpyrocarbonate. Methods Enzymol., 1977, 47, 431-432.

230

223 Allen G. Protein, Volume 2, Elsevier, 1999, Google Books, Web.

224 Lundblad RL. Chemical Modification of Biological Polymers, CRC Press, 2011, Google Books, Web.

225 Mongkolsapaya J, Grimes JM, Chen N, Xu XN, Stuart DI, Jones EY, Screaton GR. Structure of the TRAIL-DR5 complex reveals mechanisms conferring specificity in apoptotic initiation. Nat. Struct. Biol., 1999, 6, 1048-1053

226 Kim G, Weiss SJ, Levine RL. Methionine oxidation and reduction in proteins. Biochim. Biophys. Acta, 2014, 1840, 901-905.

227 Stadtman ER, Moskovitz J, Levine RL. Oxidation of methionine residues of proteins: biological consequences. Antioxid. Redox Signal., 2003, 5, 577-582.

228 Bauerle MR, Schwalm EL, Booker SJ. Mechanistic diversity of radical S- adenosylmethionine (SAM)-dependent methylation. J. Biol. Chem., 2015, 290, 3995-4002.

229 Mehta AP, Abdelwahed SH, Mahanta N, Fedoseyenko D, Philmus B, Cooper LE, Liu Y, Jhulki I, Ealick SE, Begley TP. Radical S-adenosylmethionine (SAM) enzymes in cofactor biosynthesis: a treasure trove of complex organic radical rearrangement reactions. J. Biol. Chem., 2015, 290, 3980-3986.

230 Kouzarides T. Chromatin modifications and their function. Cell, 2007, 128, 693-705.

231 Cheng X. DNA modification by methyltransferases. Curr. Opin. Struct. Biol., 1995, 5, 4- 10.

232 Rose K, Savoy LA, Simona MG, Offord RE, Wingfield PT, Mattaliano RJ, Thatcher DR. The state of the N-terminus of recombinant proteins: determination of N-terminal methionine (formylated, acetylated, or free). Anal. Biochem., 1987, 165, 59-69.

233 Liao YD, Jeng JC, Wang CF, Wang SC, Chang ST. Removal of N-terminal methionine from recombinant proteins by engineered E. coli methionine aminopeptidase. Protein Sci., 2004, 13, 1802-1810.

234 Hirel PH, Schmitter MJ, Dessen P, Fayat G, Blanquet S. Extent of N-terminal methionine excision from Escherichia coli proteins is governed by the side-chain length of the penultimate amino acid. Proc. Natl. Acad. Sci. USA, 1989, 86, 8247-8251.

231

235 Frottin F, Martinez A, Peynot P, Mitra S, Holz RC, Giglione C, Meinnel T. The proteomics of N-terminal methionine cleavage. Mol. Cell. Proteomics, 2006, 5, 2336-2349.

236 Xiao Q, Zhang F, Nacev BA, Liu JO, Pei D. Protein N-terminal processing: substrate specificity of Escherichia coli and human methionine aminopeptidases. Biochemistry, 2010, 49, 5588-5599.

237 Strickland EC, Geer MA, Tran DT, Adhikari J, West GM, DeArmond PD, Xu Y, Fitzgerald MC. Thermodynamic analysis of protein-ligand binding interactions in complex biological mixtures using the stability of proteins from rates of oxidation. Nat. Protoc., 2013, 8, 148-161.

238 Tran DT, Adhikari J, Fitzgerald MC. Stable Isotope Labeling with Amino Acids in Cell Culture (SILAC)-based strategy for proteome-wide thermodynamic analysis of protein- ligand binding interactions. Mol. Cell. Proteomics, 2014, 13, 1800-1813.

239 Jenny RJ, Mann KG, Lundblad RL. A critical review of the methods for cleavage of fusion proteins with thrombin and factor Xa. Protein Expr. Purif., 2003, 31, 1-11.

240 Terpe K. Overview of tag protein fusions: from molecular and biochemical fundamentals to commercial systems. Appl. Microbiol. Biotechnol., 2003, 60, 523-533.

241 Huang X, Aulabaugh A, Ding W, Kapoor B, Alksne L, Tabei K, Ellestad G. Biochemistry, 2003, 42, 11307-11315.

242 Horie R, Watanabe T. CD30: expression and function in health and disease. Semin. Immunol. 1998, 10, 457-470.

232

Biography

Joseph (Joe) Bellucci was born in Somers Point, New Jersey on June 22, 1986. He is the son of Joseph and Nancy Bellucci and has two brothers, Lou and Tom. In October

2015 he married his wife, Marine Bellucci. Joe graduated from Cornell University in May

2008 with a B.S. in Chemical & Biomolecular Engineering. In 2010, Joe began his work in the Chilkoti Lab at Duke. At Duke, Joe has received a University Scholars Fellowship, a

James B. Duke fellowship, the Medtronic Third Year Biomedical Engineering

Fellowship, and an NIH F31 Predoctoral Fellowship. He has authored the following manuscripts:

1. Bhattacharyya J, Bellucci JJ, Weitzhandler I, McDaniel JR, Spasojevic I, Lin CC, Chi JT, Chilkoti A. A paclitaxel-loaded recombinant polypeptide nanoparticle outperforms Abraxane in multiple murine cancer models. Nature Communications, 2015, 6:7939. doi: 10.1038/ncomms8939

2. Bellucci JJ, Bhattacharyya J, Chilkoti A. A noncanonical function of sortase enables site-specific conjugation of small molecules to lysine residues in proteins. Angew. Chemie Int. Ed., 2015, 54, 441-445.

3. Bellucci JJ, Amiram M, Bhattacharyya J, Chilkoti A. Three-in-one chromatography- free purification, tag removal, and site-specific modification of recombinant fusion proteins using sortase A and elastin-like polypeptides. Angew. Chemie Int. Ed., 2013, 52, 3703-3708.

4. Bellucci JJ and Hamaker KH. Evaluation of oxygen transfer rates in stirred-tank bioreactors for clinical manufacturing. Biotechnol. Prog., 2011, 27, 368-376.

5. Hamaker KH, Johnson DC, Bellucci JJ, Apgar KR, Soslow S, Gercke JC, Menzo DJ, Ton C. Design of a novel automated methanol feed system for pilot-scale fermentation of Pichia pastoris. Biotechnol. Prog., 2011, 27, 657-667.

233

1 D.S. Dimitrov. Therapeutic proteins. Methods Mol. Cell Biol. 2012, 899, 1-26. 2 Muramatsu M, Kinoshita K, Fagarasan S, Yamada S, Shinkai Y, Honjo T. Class switch recombination and hypermutation require activation-induced cytidine deaminase (AID), a potential RNA editing enzyme. Cell, 2000, 102, 553-563. 3 Cantaert T, Schickel JN, Bannock JM, Ng YS, Massad C, Oe T, Wu R, Lavoie A, Walter JE, Notarangelo LD, Al-Herz W, Kilic SS, Ochs HD, Nonoyama S, Durandy A, Meffre E. Activation-Induced Cytidine Deaminase Expression in Human B Cell Precursors Is Essential for Central B Cell Tolerance. Immunity, 2015, 42, 884-895. 4 Köhler G, Milstein C. Continuous cultures of fused cells secreting antibody of predefined specificity. Nature, 1975, 256, 495-497. 5 Rajendra Y, Hougland MD, Alam R, Morehead TA, Barnard GC. A high cell density transient transfection system for therapeutic protein expression based on a CHO GS- knockout cell line: process development and product quality assessment. Biotechnol. Bioeng., 2015, 112, 977-986. 6 Xu Y, Lee J, Tran C, Heibeck TH, Wang WD, Yang J, Stafford RL, Steiner AR, Sato AK, Hallam TJ, Yin G. Production of bispecific antibodies in "knobs-into-holes" using a cell- free expression system. MAbs, 2015, 7, 231-242. 7 A.M. Scott, J.D. Wolchok, L.J. Old. Antibody Therapy of Cancer. Nat. Rev. Cancer, 2012, 12 278-287. 8 Heinemann V, von Weikersthal LF, Decker T, Kiani A, Vehling-Kaiser U, Al-Batran SE, Heintges T, Lerchenmüller C, Kahl C, Seipelt G, Kullmann F, Stauch M, Scheithauer W, Hielscher J, Scholz M, Müller S, Link H, Niederle N, Rost A, Höffkes HG, Moehler M, Lindig RU, Modest DP, Rossius L, Kirchner T, Jung A, Stintzing S. FOLFIRI plus cetuximab versus FOLFIRI plus bevacizumab as first-line treatment for patients with metastatic colorectal cancer (FIRE-3): a randomised, open-label, phase 3 trial. Lancet Oncol., 2014, 15, 1065-1075. 9 Loupakis F, Cremolini C, Masi G, Lonardi S, Zagonel V, Salvatore L, Cortesi E, Tomasello G, Ronzoni M, Spadi R, Zaniboni A, Tonini G, Buonadonna A, Amoroso D, Chiara S, Carlomagno C, Boni C, Allegrini G, Boni L, Falcone A. Initial therapy with FOLFOXIRI and bevacizumab for metastatic colorectal cancer. N. Engl. J. Med., 2014, 37, 1609-1618. 10 Earl HM, Hiller L, Dunn JA, Blenkinsop C, Grybowicz L, Vallier AL, Abraham J, Thomas J, Provenzano E, Hughes-Davies L, Gounaris I, McAdam K, Chan S, Ahmad R, Hickish T, Houston S, Rea D, Bartlett J, Caldas C, Cameron DA, Hayward L; ARTemis Investigators. Efficacy of neoadjuvant bevacizumab added to docetaxel followed by fluorouracil, epirubicin, and cyclophosphamide, for women with HER2-negative early

234

breast cancer (ARTemis): an open-label, randomised, phase 3 trial. Lancet Oncol., 2015, 16, 656-666. 11 Piccart-Gebhart MJ, Procter M, Leyland-Jones B, Goldhirsch A, Untch M, Smith I, Gianni L, Baselga J, Bell R, Jackisch C, Cameron D, Dowsett M, Barrios CH, Steger G, Huang CS, Andersson M, Inbar M, Lichinitser M, Láng I, Nitz U, Iwata H, Thomssen C, Lohrisch C, Suter TM, Rüschoff J, Suto T, Greatorex V, Ward C, Straehle C, McFadden E, Dolci MS, Gelber RD; Herceptin Adjuvant (HERA) Trial Study Team. Trastuzumab after adjuvant chemotherapy in HER2-positive breast cancer. N. Engl. J. Med., 2005, 353, 1659- 1672. 12 Swain SM, Baselga J, Kim SB, Ro J, Semiglazov V, Campone M, Ciruelos E, Ferrero JM, Schneeweiss A, Heeson S, Clark E, Ross G, Benyunes MC, Cortés J; CLEOPATRA Study Group. Pertuzumab, trastuzumab, and docetaxel in HER2-positive metastatic breast cancer. N. Engl. J. Med., 2015, 372, 724-734. 13 A. Iannello, A. Ahmad. Role of antibody-dependent cell-mediated cytotoxicity in the efficacy of therapeutic monoclonal antibodies. Cancer and Metastasis Reviews, 2005, 24, 487-499. 14 Gerdes CA, Nicolini VG, Herter S, van Puijenbroek E, Lang S, Roemmele M, Moessner E, Freytag O, Friess T, Ries CH, Bossenmaier B, Mueller HJ, Umaña P. GA201 (RG7160): a novel, humanized, glycoengineered anti-EGFR antibody with enhanced ADCC and superior in vivo efficacy compared with cetuximab. Clin. Cancer Res., 2013, 19, 1126-1138. 15 Romain G, Senyukov V, Rey-Villamizar N, Merouane A, Kelton W, Liadi I, Mahendra A, Charab W, Georgiou G, Roysam B, Lee DA, Varadarajan N. Antibody Fc engineering improves frequency and promotes kinetic boosting of serial killing mediated by NK cells. Blood, 2014, 124, 3241-3249. 16 P.A. Trail. Antibody-drug conjugates as cancer therapeutics. Antibodies, 2013, 2, 113- 129. 17 N.M. Okeleley, S.C. Alley, P.D. Senter. Advancing antibody drug conjugation from the laboratory to an approved anticancer drug. Hematol. Oncol. Clin. N. Amer., 2008, 28, 13- 25. 18 M. Acchione, H. Kwon, C.M. Jochheim, W.M. Atkins. Impact of linker and conjugation chemistry on antigen binding, Fc receptor binding and thermal stability of model antibody-drug conjugates. mAbs, 2012, 4, 362-372. 19 Younes A, Bartlett NL, Leonard JP, Kennedy DA, Lynch CM, Sievers EL, Forero-Torres A. Brentuximab vedotin (SGN-35) for relapsed CD30-positive lymphomas. N. Engl. J. Med., 2010, 363, 1812-1821. 20 Verma S, Miles D, Gianni L, Krop IE, Welslau M, Baselga J, Pegram M, Oh DY, Diéras V, Guardino E, Fang L, Lu MW, Olsen S, Blackwell K; EMILIA Study Group.

235

Trastuzumab emtansine for HER2-positive advanced breast cancer. N. Engl. J. Med., 2012, 367, 1783-1791. 21 B.Q. Shen, et al, Conjugation site modulates the in vivo stability and therapeutic activity of antibody-drug conjugates. Nat. Biotechnol. 2012, 30, 184-189. 22 K.J. Hamblett, et al. Effects of drug loading on the antitumor activity of a monoclonal antibody drug conjugate. Clin. Cancer Res., 2004, 10, 7063-7070. 23 S. Panowski, S. Chakta, H. Raab, P. Polakis, J.R. Junutula. Site-specific antibody drug conjugates for cancer therapy. mAbs, 2014, 6, 34-45. 24 C.R. Behrens, B. Liu. Methods for site-specific drug conjugation to antibodies. mAbs, 2014, 6, 46-53. 25 M.T. Kim, Y. Chen, J. Marhoul, F. Jacobson. Statistical modeling of the drug load distribution on trastuzumab emtansine (Kadcyla), a lysine-linked antibody drug conjugate. Bioconjug. Chem., 2014, 25, 1223-1232. 26 Schulz JD, Patt M, Basler S, Kries H, Hilvert D, Gauthier MA, Leroux JC. Site-Specific Polymer Conjugation Stabilizes Therapeutic Enzymes in the Gastrointestinal Tract. doi: 10.1002/adma.201504797. Adv. Mater., 2015, [Epub ahead of print] 27 Mahmood, I Green, MD (2005) Pharmacokinetic and pharmacodynamic considerations in the development of therapeutic proteins. Clin Pharmacokinet 44(4):331-347. 28 Harris JM, Chess RB. Effect of pegylation on pharmaceuticals. Nat. Rev. Drug Discov., 2003, 2, 214-221. 29 Bunn HF. New agents that stimulate erythropoiesis. Blood, 2007, 109, 868-873. 30 Liu L, Li H, Hamilton SR, Gomathinayagam S, Rayfield WJ, van Maanen M, Yin KC, Hong L, Prueksaritanont T. The impact of sialic acids on the pharmacokinetics of a PEGylated erythropoietin. J. Pharm. Sci., 2012, 101, 4414-4418. 31 Basser RL, Underhill C, Davis I, Green MD, Cebon J, Zalcberg J, MacMillan J, Cohen B, Marty J, Fox RM, Begley CG. Enhancement of platelet recovery after myelosuppressive chemotherapy by recombinant human megakaryocyte growth and development factor in patients with advanced cancer. J. Clin. Oncol., 2000, 18, 2852-2861. 32 Green MD, Koelbl H, Baselga J, Galid A, Guillem V, Gascon P, Siena S, Lalisang RI, Samonigg H, Clemens MR, Zani V, Liang BC, Renwick J, Piccart MJ; International Pegfilgrastim 749 Study Group. A randomized double-blind multicenter phase III study of fixed-dose single-administration pegfilgrastim versus daily filgrastim in patients receiving myelosuppressive chemotherapy. Ann. Oncol., 2003, 14, 29-35. 33 G. Georgiou, P. Valax. Expression of correctly folded proteins in Escherichia coli. Curr. Opin. In Biotech., 1996, 7, 190-197. 34 P. Bessette, A. Aslund, J. Beckwith, G. Georgiou. Efficient folding of proteins with multiple disulfide bonds in the Escherichia coli cytoplasm. Proc. Natl. Acad. Sci. USA, 1999, 96, 13703-13708. 236

35 Staub JM, Garcia B, Graves J, Hajdukiewicz PT, Hunter P, Nehra N, Paradkar V, Schlittler M, Carroll JA, Spatola L, Ward D, Ye G, Russell DA. High-yield production of a human therapeutic protein in tobacco chloroplasts. Nat. Biotechnol. 2000, 18, 333-338. 36 Ye J, Alvin K, Latif H, Hsu A, Parikh V, Whitmer T, Tellers M, de la Cruz Edmonds MC, Ly J, Salmon P, Markusen JF. Rapid Protein Production Using CHO Stable Transfection Pools. Biotechnol. Prog. 2010, 26 1431-1437. 37 Ye J, Ly J, Watts K, Hsu A, Walker A, McLaughlin K, Berdichevsky M, Prinz B, Sean Kersey D, d'Anjou M, Pollard D, Potgieter T. Optimization of a Glycoengineered Pichia pastoris Cultivation Process for Commercial Antibody Production. Biotechnol. Prog. 2011, 27, 1744-1750. 38 Franzreb M, Muller E, Vajda J. Cost Estimation for Protein A Chromatography: an In Silico Approach to MAb purification strategy. Bioprocess Int. 2014, 12, 44-52. 39 Kobayashi S, Ueda Y. Comparing Protein A Resins for Monoclonal Antibody Purification. Biopharm. Int., 2013, 26. 40 Arnau J, Lauritzen C, Petersen GE, Pedersen J. Current strategies for the use of affinity tags and tag removal for the purification of recombinant proteins. Protein Expr. Purif., 2006, 48, 1-13. 41 Terpe K. Overview of tag protein fusions: from molecular and biochemical fundamentals to commercial systems. Appl. Microbiol. Biotechnol., 2003, 60, 525-533. 42 Jones C, Patel A, Griffin S, Martin J, Young P, O'Donnell K, Silverman C, Porter T, Chaiken I. Current trends in molecular recognition and bioseparation. J. Chromatogr. A, 1995, 707, 3-22. 43 R.B. Kapust, D.S. Waugh. Escherichia coli maltose-binding protein is uncommonly effective at promoting the solubility of polypeptides to which it is fused. Protein Science. 1999, 8, 1668-1674. 44 Pennati A, Deng J, Galipeau J. Maltose-binding protein fusion allows for high level bacterial expression and purification of bioactive mammalian cytokine derivatives. PLoS One, 2014, 9, e106724. doi: 10.1371/journal.pone.0106724 45 Jenny RJ, Mann KG, Lundblad RL. A critical review of the methods for cleavage of fusion proteins with thrombin and factor Xa. Protein Expr. Purif., 2003, 31, 1-11. 46 Smyth DR, Mrozkiewicz MK, McGrath WJ, Listwan P, Kobe B. Crystal structures of fusion proteins with large-affinity tags. Protein Sci., 2003, 12, 1313-1322. 47 Amiram M, Luginbuhl KM, Li X, Feinglos MN, Chilkoti A. A depot-forming glucagon- like peptide-1 fusion protein reduces blood glucose for five days with a single injection. J. Control. Release, 2013, 172, 144-151. 48 Amiram M, Luginbuhl KM, Li X, Feinglos MN, Chilkoti A. Injectable protease- operated depots of glucagon-like peptide-1 provide extended and tunable glucose control. Proc.Natl. Acad. Sci. USA, 2013, 110, 2792-2797.

237

49 McDaniel JR, Bhattacharyya J, Vargo KB, Hassouneh W, Hammer DA, Chilkoti A. Self- assembly of thermally responsive nanoparticles of a genetically encoded peptide polymer by drug conjugation. Angew. Chemie Int. Ed., 2013, 52, 1683-1687. 50 Bhattacharyya J, Bellucci JJ, Weitzhandler I, McDaniel JR, Spasojevic I, Li X, Lin CC, Chi JT, Chilkoti A. A paclitaxel-loaded recombinant polypeptide nanoparticle outperforms Abraxane in multiple murine cancer models. Nat. Communications, 2015, 6, doi: 10.1038/ncomms8939 51 Mastria EM, Chen M, McDaniel JR, Li X, Hyun J, Dewhirst MW, Chilkoti A. Doxorubicin-conjugated polypeptide nanoparticles inhibit metastasis in two murine models of carcinoma. J. Control. Release, 2015, 208, 52-58. 52 McDaniel JR, Macewan SR, Dewhirst M, Chilkoti A. Doxorubicin-conjugated chimeric polypeptide nanoparticles that respond to mild hyperthermia. J. Control. Release, 2012, 159, 362-367. 53 MacKay JA, Chen M, McDaniel JR, Liu W, Simnick AJ, Chilkoti A. Self-assembling chimeric polypeptide-doxorubicin conjugate nanoparticles that abolish tumours after a single injection. Nat. Materials, 2009, 8, 993-999. 54 Liu W, MacKay JA, Dreher MR, Chen M, McDaniel JR, Simnick AJ, Callahan DJ, Zalutsky MR, Chilkoti A. Injectable intratumoral depot of thermally responsive polypeptide-radionuclide conjugates delays tumor progression in a mouse model. J. Control. Release, 2010, 144, 2-9. 55 Liu W, McDaniel J, Li X, Asai D, Quiroz FG, Schaal J, Park JS, Zalutsky M, Chilkoti A. Brachytherapy using injectable seeds that are self-assembled from genetically encoded polypeptides in situ. Cancer Res., 2012, 72, 5956-5965. 56 D. Meyer and A. Chilkoti (1999) Purification of recombinant proteins by fusion with thermally-responsive polypeptides. Nat Biotech 17(11): 1112-1115. 57 Hassouneh W, Christensen T, and Chilkoti A (2010) Elastin-like polypeptides as a purification tag for recombinant proteins. Curr Protein Sci 61:6.11.1–6.11.16. 58 MacEwan SR, Hassouneh W, Chilkoti A. Non-chromatographic purification of recombinant elastin-like polypeptides and their fusions with peptides and proteins from Escherichia coli. J. Vis. Exp., 2014, 88, doi: 10.3791/51583. 59 C.A. Boswell, E.E. Mundo, C. Zhang, D. Bumbaca, N.R. Valle, K.R. Kozak, A. Fourie, J. Chuh, N. Koppada, O. Saad, H. Gill, B.Q. Shen, B. Rubinfeld, J. Tibbitts, S. Kaur, F.P. Theil, P.J. Fielder, L.A. Khawli, K. Lin. Impact of drug conjugation on the pharmacokinetics and tissue distribution of anti-STEAP1 antibody-drug conjugates in rats. Bioconjug. Chem., 2011, 22, 1994-2004. 60 B.Q. Shen, K. Xu, L. Liu, H. Raab, S. Bhakta, M. Kenrick, K.L. Parsons-Reponte, J. Tien, S.F. Yu, E. Mai, D. Li, J. Tibbitts, J. Baudys, O.M. Saad, S.J. Scales, P.J. McDonald, P.E. Hass, C. Eigenbrot, T. Nguyen, W.A. Solis, R.N. Fuji, K.M. Flagella, D. Patel, S.D. 238

Spencer, L.A. Khawli, A. Ebens, W.L. Wong, R. Vandlen, S. Kaur, M.X. Sliwkowski, R.H. Scheller, P. Polakis, J.R. Junutula. Conjugation site modulates the in vivo stability and therapeutic activity of antibody-drug conjugates. Nat. Biotechnol. 2012, 30, 184-189. 61 M. Acchione, H. Kwon, C.M. Jochheim, W.M. Atkins. Impact of linker and conjugation chemistry on antigen binding, Fc receptor binding and thermal stability of model antibody-drug conjugates. mAbs. 2012, 4, 362-372. PMCID: PMC3355488. 62 R.J. Sanderson, M.A. Hering, S.F. James, M.M. Sun, S.O. Doronina, A.W. Siadak, P.D. Senter, A.F. Wahl. In vivo drug-linker stability of an anti-CD30 dipeptide-linked auristatin immunoconjugate. Clin. Cancer Res., 2005, 11, 843-852. 63 C. Antczak, J.S. Jaggi, C.V. LeFave, M.J. Curcio, M.R. McDevitt, D.A. Scheinberg. Influence of the linker on the biodistribution and catabolism of actinium-225 self- immolative tumor-targeted isotope generators. Bioconjug. Chem., 2006, 17, 1551-1560. PMCID: PMC2570787. 64 S.A. Roberts, P.A. Andrews, D. Blanset, K.M. Flagella, B. Gorovits, C.M. Lynch, P.L. Martin, K. Kramer-Stickland, S. Thibault, G. Warner. Considerations for the nonclinical safety evaluation of antibody drug conjugates for oncology. Regul. Toxicol. Pharmacol., 2013, 67, 382-391. 65 M.T. Kim, Y. Chen, J. Marhoul, F. Jacobson. Statistical modeling of the drug load distribution on trastuzumab emtansine (Kadcyla), a lysine-linked antibody drug conjugate. Bioconjug. Chem., 2014, 25, 1223-1232. 66 V.S. Goldmacher, G. Amphlett, L. Wang, A.C. Lazar. Statistics of the Distribution of the Abundance of Molecules with Various Drug Loads in Maytansinoid Antibody-Drug Conjugates. Mol. Pharm., 2015 [Epub ahead of print]. 67 M.C. Janin-Bussat, M. Dillenbourg, N. Corvaia, A. Beck, C. Klinguer-Hamour. Characterization of antibody drug conjugate positional isomers at cysteine residues by peptide mapping LC-MS analysis. J. Chromatogr. B. Analyt. Technol. Biomed. Life Sci., 2015, 15, 981-982. 68 B.Q. Shen, K. Xu, L. Liu, H. Raab, S. Bhakta, M. Kenrick, K.L. Parsons-Reponte, J. Tien, S.F. Yu, E. Mai, D. Li, J. Tibbitts, J. Baudys, O.M. Saad, S.J. Scales, P.J. McDonald, P.E. Hass, C. Eigenbrot, T. Nguyen, W.A. Solis, R.N. Fuji, K.M. Flagella, D. Patel, S.D. Spencer, L.A. Khawli, A. Ebens, W.L. Wong, R. Vandlen, S. Kaur, M.X. Sliwkowski, R.H. Scheller, P. Polakis, J.R. Junutula. Conjugation site modulates the in vivo stability and therapeutic activity of antibody-drug conjugates. Nat. Biotechnol. 2012, 30, 184-189. 69 K.J. Hamblett, P.D. Senter, D.F. Chace, M.M. Sun, J. Lenox, C.G. Cerveny, K.M. Kissler, S.X. Bernhardt, A.K. Kopcha, R.F. Zabinski, D.L. Meyer, J.A. Francisco. Effects of drug loading on the antitumor activity of a monoclonal antibody drug conjugate. Clin. Cancer Res., 2004, 10, 7063-7070.

239

70 Junutula JR, Raab H, Clark S, Bhakta S, Leipold DD, Weir S, Chen Y, Simpson M, Tsai SP, Dennis MS, Lu Y, Meng YG, Ng C, Yang J, Lee CC, Duenas E, Gorrell J, Katta V, Kim A, McDorman K, Flagella K, Venook R, Ross S, Spencer SD, Lee Wong W, Lowman HB, Vandlen R, Sliwkowski MX, Scheller RH, Polakis P, Mallet W. Site-specific conjugation of a cytotoxic drug to an antibody improves the therapeutic index. Nat. Biotechnol. 2008, 26, 925-932. 71 Daëron M. Fc receptor biology. Annu Rev Immunol. 1997, 15, 203-234. 72 K.J. Hamblett, P.D. Senter, D.F. Chace, M.M. Sun, J. Lenox, C.G. Cerveny, K.M. Kissler, S.X. Bernhardt, A.K. Kopcha, R.F. Zabinski, D.L. Meyer, J.A. Francisco. Effects of drug loading on the antitumor activity of a monoclonal antibody drug conjugate. Clin. Cancer Res., 2004, 10, 7063-7070. 73 Lyon RP, Bovee TD, Doronina SO, Burke PJ, Hunter JH, Neff-LaFord HD, Jonas M, Anderson ME, Setter JR, Senter PD. Reducing hydrophobicity of homogeneous antibody-drug conjugates improves pharmacokinetics and therapeutic index. Nat. Biotechnol. 2015, 33, 733-735. 74 Lyon RP, Setter JR, Bovee TD, Doronina SO, Hunter JH, Anderson ME, Balasubramanian CL, Duniho SM, Leiske CI, Li F, Senter PD. Self-hydrolyzing maleimides improve the stability and pharmacological properties of antibody-drug conjugates. Nat. Biotechnol. 2014, 32, 1059-1062. 75 B.Q. Shen, K. Xu, L. Liu, H. Raab, S. Bhakta, M. Kenrick, K.L. Parsons-Reponte, J. Tien, S.F. Yu, E. Mai, D. Li, J. Tibbitts, J. Baudys, O.M. Saad, S.J. Scales, P.J. McDonald, P.E. Hass, C. Eigenbrot, T. Nguyen, W.A. Solis, R.N. Fuji, K.M. Flagella, D. Patel, S.D. Spencer, L.A. Khawli, A. Ebens, W.L. Wong, R. Vandlen, S. Kaur, M.X. Sliwkowski, R.H. Scheller, P. Polakis, J.R. Junutula. Conjugation site modulates the in vivo stability and therapeutic activity of antibody-drug conjugates. Nat. Biotechnol. 2012, 30, 184-189. 76 Ashkenazi A, Holland P, Eckhardt SG. Ligand-based targeting of apoptosis in cancer: the potential of recombinant human apoptosis ligand 2/tumor necrosis factor-related apoptosis-inducing ligand (rhApo2/TRAIL). J. Clin. Oncol. 2008, 26, 3621-3630. 77 B.Q. Shen, K. Xu, L. Liu, H. Raab, S. Bhakta, M. Kenrick, K.L. Parsons-Reponte, J. Tien, S.F. Yu, E. Mai, D. Li, J. Tibbitts, J. Baudys, O.M. Saad, S.J. Scales, P.J. McDonald, P.E. Hass, C. Eigenbrot, T. Nguyen, W.A. Solis, R.N. Fuji, K.M. Flagella, D. Patel, S.D. Spencer, L.A. Khawli, A. Ebens, W.L. Wong, R. Vandlen, S. Kaur, M.X. Sliwkowski, R.H. Scheller, P. Polakis, J.R. Junutula. Conjugation site modulates the in vivo stability and therapeutic activity of antibody-drug conjugates. Nat. Biotechnol. 2012, 30, 184-189.

240

78 J.H. Huh, A.J. White, S.R. Brych, H. Franey, M. Matsumura. The identification of free cysteine residues within antibodies and a potential role for free cysteine residues in covalent aggregation because of agitation stress. J. Pharm. Sci., 2013, 102, 1701-1711. 79 K. Sakurai, R. Nakahata, Y.H. Lee, J. Kardos, T. Ikegami, Y. Goto. Effects of a reduced disulfide bond on aggregation properties of the human IgG1 CH3 domain. Biochim. Biophys. Acta., 2015, doi:10.1016/j.bbapap.2015.02.020 [Epub ahead of print]. 80 T. Rispens, P. Ooijevaar-de Heer, O. Bende, R.C. Aalberse. Mechanism of immunoglobulin G4 Fab-arm exchange. J. Am. Chem. Soc., 2011, 133, 10302-10311. 81 Y.D. Liu, X. Chen, J.Z. Enk, M. Plant, T.M. Dillon, G.C. Flynn. Human IgG2 antibody disulfide rearrangement in vivo. J. Biol. Chem., 2008, 283, 29266-29272. PMCID: PMC2662016. 82 Axup JY, Bajjuri KM, Ritland M, Hutchins BM, Kim CH, Kazane SA, Halder R, Forsyth JS, Santidrian AF, Stafin K, Lu Y, Tran H, Seller AJ, Biroc SL, Szydlik A, Pinkstaff JK, Tian F, Sinha SC, Felding-Habermann B, Smider VV, Schultz PG. Synthesis of site- specific antibody-drug conjugates using unnatural amino acids. Proc. Natl. Acad. Sci. USA. 2012, 109, 16101-16106. 83 Xiao H, Chatterjee A, Choi SH, Bajjuri KM, Sinha SC, Schultz PG. Genetic incorporation of multiple unnatural amino acids into proteins in mammalian cells. Angew. Chemie Int. Ed. 2013, 52, 14080-14083. 84 Amiram M, Haimovich AD, Fan C, Wang YS, Aerni HR, Ntai I, Moonan DW, Ma NJ, Rovner AJ, Hong SH, Kelleher NL, Goodman AL, Jewett MC, Söll D, Rinehart J, Isaacs FJ. Evolution of translation machinery in recoded bacteria enables multi-site incorporation of nonstandard amino acids. Nat. Biotechnol., 2015, doi: 10.1038/nbt.3372. [Epub ahead of print]. 85 Tian F, Lu Y, Manibusan A, Sellers A, Tran H, Sun Y, Phuong T, Barnett R, Hehli B, Song F, DeGuzman MJ, Ensari S, Pinkstaff JK, Sullivan LM, Biroc SL, Cho H, Schultz PG, DiJoseph J, Dougher M, Ma D, Dushin R, Leal M, Tchistiakova L, Feyfant E, Gerber HP, Sapra P. A general approach to site-specific antibody drug conjugates. Proc. Natl. Acad. Sci. USA. 2014, 111, 1766-1771. 86 Chen I, Howarth M, Lin W, Ting AY. Site-specific labeling of cell surface proteins with biophysical probes using biotin ligase. Nat. Methods. 2005, 2, 99-104. 87 Agarwal P, Bertozzi CR. Site-specific antibody-drug conjugates: the nexus of bioorthogonal chemistry, protein engineering, and drug development. 2015, 26, 176-192. 88 King M, Wagner A. Developments in the field of bioorthogonal bond forming reactions-past and present trends. Bioconjug. Chem., 2014, 25, 825-839.

241

89 Carrico IS, Carlson BL, Bertozzi CR. Introducing genetically encoded aldehydes into proteins. Nat. Chem. Biol. 2007, 3, 321-322. 90 J.E. Hudak, R.M. Barfield, G.W. de Hart, P. Grob, E. Nogales, C.R. Bertozzi, D. Rabuka. Synthesis of heterobifunctional protein fusions using copper-free click chemistry and the aldehyde tag. Angew. Chemie Int. Ed. 2012, 51, 4161-4165. 91 P. Wu, W. Shui, B.L. Carlson, N. Hu, D. Rabuka, J. Lee, C.R. Bertozzi. Site-specific chemical modification of recombinant proteins produced in mammalian cells by using the genetically encoded aldehyde tag. Proc. Natl. Acad. Sci. USA, 2009, 106, 3000-3005. 92 P.M. Drake, A.E. Albers, J. Baker, S. Banas, R.M. Barfield, A.S. Bhat, G.W. de Hart, A.W. Garofalo, P. Holder, L.C. Jones, R. Kudirka, J. McFarland, W. Zmolek, D. Rabuka. Aldehyde Tag Coupled with HIPS Chemistry Enables the Production of ADCs Conjugated Site-Specifically to Different Antibody Regions with distinct in vivo efficacy and PK outcomes. Bioconjug. Chem., 2014, 25, 1331-1341. PMCID: PMC4215875. 93 Wu P, Shui W, Carlson BL, Hu N, Rabuka D, Lee J, Bertozzi CR. Site-specific chemical modification of recombinant proteins produced in mammalian cells by using the genetically encoded aldehyde tag. Proc. Natl. Acad. Sci. USA. 2009, 106, 3000-3005. 94 Strop P. Versatility of bacterial transglutaminase. Bioconjug. Chem., 2014, 25, 855-862. 95 Mero A, Spolaore B, Veronese FM, Fontana A. Transglutaminase-mediated PEGylation of proteins: direct identification of the sites of protein modification by mass spectrometry using a novel monodisperse PEG. Bioconjug. Chem. 2009, 20, 384-389. 96 Zhou Z, Cironi P, Lin AJ, Xu Y, Hrvatin S, Golan DE, Silver PA, Walsh CT, Yin J. Genetically encoded short peptide tags for orthogonal protein labeling by Sfp and AcpS phosphopantetheinyl transferases. ACS Chem. Biol., 2007, 2, 337-346. 97 Yin J, Straight PD, McLoughlin SM, Zhou Z, Lin AJ, Golan DE, Kelleher NL, Kolter R, Walsh CT. Genetically encoded short peptide tag for versatile protein labeling by Sfp phosphopantetheinyl transferase. Proc. Natl. Acad. Sci. USA, 2005, 102, 15815-15820. 98 Strop P, Liu SH, Dorywalska M, Delaria K, Dushin RG, Tran TT, Ho WH, Farias S, Casas MG, Abdiche Y, Zhou D, Chandrasekaran R, Samain C, Loo C, Rossi A, Rickert M, Krimm S, Wong T, Chin SM, Yu J, Dilley J, Chaparro-Riggers J, Filzen GF, O'Donnell CJ, Wang F, Myers JS, Pons J, Shelton DL, Rajpal A. Location matters: site of conjugation modulates stability and pharmacokinetics of antibody drug conjugates. Chem. Biol., 2013, 20, 161-167. 99 Grünewald J, Klock HE, Cellitti SE, Bursulaya B, McMullan D, Jones DH, Chiu HP, Wang X, Patterson P, Zhou H, Vance J, Nigoghossian E, Tong H, Daniel D, Mallet W, Ou W, Uno T, Brock A, Lesley SA, Geierstanger BH. Efficient Preparation of Site-Specific Antibody-Drug Conjugates Using Phosphopantetheinyl Transferases. Bioconjug. Chem., 2015, 26, 2554-2562. 242

100 DeDent AC, McAdow M, Schneewind O. Distribution of protein A on the surface of Staphylococcus aureus. J. Bacteriol., 2007, 189, 4473-4484. 101 Pauli NT, Kim HK, Falugi F, Huang M, Dulac J, Henry Dunand C, Zheng NY, Kaur K, Andrews SF, Huang Y, DeDent A, Frank KM, Charnot-Katsikas A, Schneewind O, Wilson PC. Staphylococcus aureus infection induces protein A-mediated immune evasion in humans. J. Exp. Med., 2014, 211, 2331-2339. 102 Ton-That H, Liu G, Mazmanian SK, Faull KF, Schneewind O (1999) Purification and characterization of sortase, the transpeptidase that cleaves surface proteins of Staphylococcus aureus and the LPXTG motif. Proc Natl Acad Sci USA 96(22):12424-12429. 103 H. Mao, S.A. Hart, A. Schink, B.A. Pollok. Sortase-mediated protein ligation: a new method for protein engineering. J. Am. Chem. Soc. 2004, 126, 2670-2671. 104 J.J. Ling, R.L. Policarpo, A.R. Rabideau, X. Liao, B.L. Pentelute. Protein thioester synthesis enabled by sortase. J. Am. Chem. Soc. 2012, 134, 10749-10752. 105 D.J. Willamson, M.A. Fascione, M.E. Webb, W.B. Turnbull. Efficient N-terminal labeling of proteins by use of sortase. Angew. Chem. Int. Ed. 2012, 51, 9377-9380. 106 Swee LK, Guimaraes CP, Sehrawat S, Spooner E, Barrasa MI, Ploegh HL. Sortase- mediated modification of αDEC205 affords optimization of antigen presentation and immunization against a set of viral epitopes. Proc. Natl. Acad. Sci. USA. 2013, 110, 1428- 1433. 107 Antos JM, Chew GL, Guimaraes CP, Yoder NC, Grotenbreg GM, Popp MW, Ploegh HL. Site-specific N- and C-terminal labeling of a single polypeptide using sortases of different specificity. J. Am. Chem. Soc. 2009, 131, 10800-10801. 108 Ilangovan U, Ton-That H, Iwahara J, Schneewind I, and Clubb RT. Structure of sortase, the transpeptidase that anchors proteins to the cell wall of Staphylococcus aureus. Proc. Natl. Acad. Sci. USA. 2001, 98, 6056-6061. 109 Kruger RG, Otvos B, Frankel BA, Bentley M, Dostal P, McCafferty DG. Analysis of the substrate specificity of the Staphylococcus aureus sortase transpeptidase SrtA. Biochemistry. 2004, 43, 1541-1551. 110 Ameloot P, Declercg W, Fiers W, Vandenabeele P, and Brouckaert P. Heterotrimers formed by tumor necrosis factors of different species or muteins. J. Biol. Chem. 2001, 276, 27098-27103. 111 Zhang XM, Weber I, Chen MJ. Site-directed mutational analysis of human tumor necrosis factor-alpha receptor binding site and structure-functional relationship. J. Biol. Chem. 1992, 267, 24069-24075. 112 Schoenfeld HJ, Poeschl B, Frey JR, Loetscher H, Hunziker W, Lustig A, Zulauf M. Efficient purification of recombinant human tumor necrosis factor β from Escherichia coli 243

yields biologically active protein with a trimeric structure that binds to both tumor necrosis factor receptors. J. Biol. Chem. 1991, 266, 3863-3869. 113 Flick DA and Gifford GE. Comparison of in vitro cell cytotoxic assays for tumor necrosis factor. J. Immunol. Methods. 1984, 68, 167-175. 114 Liddil JD, Dorr RT, Scuderi P. Association of lysosomal activity with sensitivity and resistance to tumor necrosis factor in murine L929 cells. Cancer Res. 1989, 49, 2722-2728. 115 Mao H, Hart SA, Schink A, Pollok BA. Sortase-mediated protein ligation: a new method for protein engineering. J. Am. Chem. Soc. 2004, 126, 2670-2671. 116 Ling JJ, Policarpo RL, Rabideau AR, Liao X, Pentelute BL. Protein thioester synthesis enabled by sortase. J. Am. Chem. Soc. 2012, 134, 10749-10752. 117 Willamson DJ, Fascione MA, Webb ME, Turnbull WB. Efficient N-terminal labeling of proteins by use of sortase. Angew. Chem. Int. Ed. 2012, 51, 9377-9380. 118 Swee LK, Guimaraes CP, Sehrawat S, Spooner E, Barrasa MI, Ploegh HL. Proc. Natl. Acad. Sci. USA. 2013, 110, 1428-1433. 119 Chen JJ, Knudsen S, Mazin W, Dahlgaard J, Zhang B. A 71-gene signature of TRAIL sensitivity in cancer cells. Mol. Cancer Ther., 2012, 11, 34-44. 120 Bodmer JL, Meier P, Tschopp J, Schneider P. Cysteine 230 is essential for the structure and activity of the cytotoxic ligand TRAIL. J. Biol. Chem., 2000, 275, 20632-20637. 121 Lin TP, Shahrokh Z, Flores H, Pai R, Matthews TC. Apo2 ligand/TRAIL formulations. U.S. Patent 7,741,282 B2, June 22, 2010. 122 Ton-That H, Schneewind O. Assembly of pili on the surface of Corynebacterium diptheriae. Mol. Microbiol. 2003, 50, 1429-1438. 123 Maraffini LA, Dedent AC, Schneewind O. Sortases and the art of anchoring proteins to the envelopes of Gram-positive bacteria. Microbiol. Mol. Biol. Rev. 2006, 70, 192-221. 124 S. Dasgupta, S. Samantaray, D. Sahal, R.P. Roy. Isopeptide ligation catalyzed by quintessential sortase A: mechanistic cues from cyclic and branched oligomers of indolicidin. J. Biol. Chem. 2011, 286, 23996-24006. 125 D.R. Meyer, A. Chilkoti. Purification of recombinant proteins by fusion with thermally-responsive polypeptides. Nat. Biotechnol. 1999, 17, 1112-1115. 126 J.J. Bellucci, M. Amiram, J. Bhattacharyya, D. McCafferty, A. Chikoti. Three-in-one chromatography-free purification, tag removal, and site-specific modification of recombinant fusion proteins using sortase A and elastin-like polypeptides. Angew. Chem. Int. Ed. 2013, 52, 3703-3708.

244

127 Richards J, Miller M, Abend J, Koide A, Koide S, Dewhurst S. Engineered fibronectin type III domain with a RGDWXE sequence binds with enhanced affinity and specificity to human alphavbeta3 integrin. J. Mol. Biol. 2003, 326, 1475-1488. 128 Duan J, Wu J, Valencia CA, Liu R. Fibronectin type III domain based monobody with high avidity. Biochemistry 2007, 46, 12656-12664. 129 J. Bendell, M. Saleh, A.A. Rose, P.M. Siegel, L. Hart, S. Sirpal, S. Jones, J. Green, E. Crowley, R. Simantov, T. Keler, T. Davis, L. Vahdat. Phase I/II study of the antibody- drug conjugate glembatumumab vedotin in patients with locally advanced or metastatic breast cancer. J. Clin. Oncol., 2014, 32, 3619-3625. 130 S.J. Scales, N. Gupta, G. Pacheco, R. Firestein, D.M. French, H. Koeppen, L. Rangell, V. Barry-Hamilton, E. Luis, J. Chuh, Y. Zhang, G.S. Ingle, A. Fourie-O'Donohue, K.R. Kozak, S. Ross, M.S. Dennis, S.D. Spencer. Mol. Cancer Ther., 2014, 13, 2630-2640. 131 M.A. Fanale, A. Forero-Torres, J.D. Rosenblatt, R.H. Advani, A.R. Franklin, D.A. Kennedy, T.H. Han, E.L. Sievers, N.L. Bartlett. A phase I weekly dosing study of brentuximab vedotin in patients with relapsed/refractory CD30-positive hematologic malignancies. Clin. Cancer Res., 2012, 18, 248-255. 132 A. Younes, N.L. Bartlett, J.P. Leonard, D.A. Kennedy, C.M. Lynch, E.L. Sievers, A. Forero-Torres. Brentuximab vedotin (SGN-35) for relapsed CD30-positive lymphomas. N. Engl. J. Med., 2010, 363, 1812-1821. 133 E. Oflazoglu, I.J. Stone, K. Gordon, C.G. Wood, E.A. Repasky, I.D. Grewal, C.L. Law, H.P. Gerber. Potent anticarcinoma activity of the humanized anti-CD70 antibody h1F6 conjugated to the tubulin inhibitor auristatin via an uncleavable linker. Clin. Cancer Res., 2008, 14, 6171-6180. 134 L.M. Smith, A. Nesterova, S.C. Alley, M.Y. Torgov, P.J. Carter. Potent cytotoxicity of an auristatin-containing antibody-drug conjugate targeting melanoma cells expressing melanotransferrin/p97. Mol. Cancer Ther., 2006, 5, 1474-1482. 135 L.M. Smith, A. Nesterova, M.C. Ryan, S. Duniho, M. Jonas, M. Anderson, R.F. Zabinski, M.K. Sutherland, H.P. Gerber, K.L. Van Orden, P.A. Moore, S.M. Ruben, P.J. Carter. CD133/prominin-1 is a potential therapeutic target for antibody-drug conjugates in hepatocellular and gastric cancers. Br. J. Cancer, 2008, 99, 100-109. PMCID: PMC2453027. 136 S.Y. Kim, J.W. Theunissen, J. Balibalos, S. Liao-Chan, M.C. Babcock, T. Wong, B. Cairns, D. Gonzalez, E.H. van der Horst, M. Perez, Z. Levashova, L. Chinn, J.A. D'Alessio, M. Flory, A. Bermudez, D.Y. Jackson, E. Ha, J. Monteon, M.F. Bruhns, G. Chen, T.S. Migone. A novel antibody-drug conjugate targeting SAIL for the treatment of hematologic malignancies. Blood Cancer J., 2015, 5:e316. doi: 10.1038/bcj.2015.39. 137 S. Verma, D. Miles, L. Gianni, I.E. Krop, M. Welslau, J. Baselga, M. Pegram, D.Y. Oh, V. Diéras, E. Guardino, L. Fang, M.W. Lu, S. Olsen, K. Blackwell. Trastuzumab

245

emtansine for HER2-positive advanced breast cancer. N. Engl. J. Med., 2012, 367, 1783- 1791. 138 A. Younes, A.K. Gopal, S.E. Smith, S.M. Ansell, J.D. Rosenblatt, K.J. Savage, R. Ramchandren, N.L. Bartlett, B.D. Cheson, S. de Vos, A. Forero-Torres, C.H. Moskowitz, J.M. Connors, A. Engert, E.K. Larsen, D.A. Kennedy, E.L. Sievers, R. Chen. Results of a pivotal phase II study of brentuximab vedotin for patients with relapsed or refractory Hodgkin's lymphoma. J. Clin. Oncol., 2012, 30, 2183-2189. PMCID: PMC3646316. 139 M.A. Fanale, S.M. Horwitz, A. Forero-Torres, N.L. Bartlett, R.H. Advani, B. Pro, R.W. Chen, A. Davies, T. Illidge, D. Huebner, D.A. Kennedy, A.R. Shustov. Brentuximab vedotin in the front-line treatment of patients with CD30+ peripheral T-cell lymphomas: results of a phase I study. J. Clin. Oncol., 2014, 32, 3137-3143. PMCID: PMC4171358. 140 B. Pro, R. Advani, P. Brice, N.L. Bartlett, J.D. Rosenblatt, T. Illidge, J. Matous, R. Ramchandren, M. Fanale, J.M. Connors, Y. Yang, E.L. Sievers, D.A. Kennedy, A. Shustov. Brentuximab vedotin (SGN-35) in patients with relapsed or refractory systemic anaplastic large-cell lymphoma: results of a phase II study. J. Clin. Oncol., 2012, 30, 2190- 2196. 141 P.A. Trail. Antibody-drug conjugates as cancer therapeutics. Antibodies, 2013, 2, 113- 129. 142 N.M. Okeleley, S.C. Alley, P.D. Senter. Advancing antibody drug conjugation from the laboratory to an approved anticancer drug. Hematol. Oncol. Clin. N. Amer., 2008, 28, 13-25. 143 M. Acchione, H. Kwon, C.M. Jochheim, W.M. Atkins. Impact of linker and conjugation chemistry on antigen binding, Fc receptor binding and thermal stability of model antibody-drug conjugates. mAbs, 2012, 4, 362-372. PMCID: PMC3355488. 144 B.Q. Shen, K. Xu, L. Liu, H. Raab, S. Bhakta, M. Kenrick, K.L. Parsons-Reponte, J. Tien, S.F. Yu, E. Mai, D. Li, J. Tibbitts, J. Baudys, O.M. Saad, S.J. Scales, P.J. McDonald, P.E. Hass, C. Eigenbrot, T. Nguyen, W.A. Solis, R.N. Fuji, K.M. Flagella, D. Patel, S.D. Spencer, L.A. Khawli, A. Ebens, W.L. Wong, R. Vandlen, S. Kaur, M.X. Sliwkowski, R.H. Scheller, P. Polakis, J.R. Junutula. Conjugation site modulates the in vivo stability and therapeutic activity of antibody-drug conjugates. Nat. Biotechnol. 2012, 30, 184-189. 145 K.J. Hamblett, P.D. Senter, D.F. Chace, M.M. Sun, J. Lenox, C.G. Cerveny, K.M. Kissler, S.X. Bernhardt, A.K. Kopcha, R.F. Zabinski, D.L. Meyer, J.A. Francisco. Effects of drug loading on the antitumor activity of a monoclonal antibody drug conjugate. Clin. Cancer Res., 2004, 10, 7063-7070. 146 S. Panowski, S. Chakta, H. Raab, P. Polakis, J.R. Junutula. Site-specific antibody drug conjugates for cancer therapy. mAbs, 2014, 6, 34-45. PMCID: PMC3929453. 147 C.R. Behrens, B. Liu. Methods for site-specific drug conjugation to antibodies. mAbs, 2014, 6, 46-53. PMCID: PMC3929454.

246

148 Sjögren HO, Isaksson M, Willner D, Hellström I, Hellström KE, Trail PA. Antitumor activity of carcinoma-reactive BR96-doxorubicin conjugate against human carcinomas in athymic mice and rats and syngeneic rat carcinomas in immunocompetent rats. Cancer Res., 1997, 57, 4530-4536. 149 Tolcher AW, Sugarman S, Gelmon KA, Cohen R, Saleh M, Isaacs C, Young L, Healey D, Onetto N, Slichenmyer W. Randomized phase II study of BR96-doxorubicin conjugate in patients with metastatic breast cancer. J. Clin. Oncol., 1999, 17, 478-484. 150 W. Wang, E.Q. Wang, J.P. Balthasar. Monoclonal antibody pharmacokinetics and pharmacodynamics. Clin. Pharmacol. Ther., 2008, 84, 548-558. 151 G.M. Thurber, M.M. Schmidt, K.D. Wittrup. Antibody tumor penetration: transport opposed by systemic and antigen-mediated clearance. Adv. Drug Deliv. Rev., 2008, 60, 1421-1434. 152 J. Katz, J.E. Janik, A. Younes. Brentuximab Vedotin (SGN-35). Clin. Cancer Res., 2011, 17, 6428-6436. 153 A. Younes, et al. Results of a pivotal phase II study of brentuximab vedotin for patients with relapsed or refractory Hodgkin's lymphoma. J. Clin. Oncol., 2012, 30, 2183- 2189. 154 Pro B, Advani R, Brice P, Bartlett NL, Rosenblatt JD, Illidge T, Matous J, Ramchandren R, Fanale M, Connors JM, Yang Y, Sievers EL, Kennedy DA, Shustov A. Brentuximab vedotin (SGN-35) in patients with relapsed or refractory systemic anaplastic large-cell lymphoma: results of a phase II study. J. Clin. Oncol., 2012, 30, 2190-2196. 155 G.R. Pettit, S.B. Singh, F. Hogan, P.L. Williams, D.L. Herald, D.D. Burkett, P.J. Clewlow. The absolute configuration and synthesis of natural (-)-dolastatin 10. J. Am. Chem. Soc., 1989, 111, 5463-5465. 156 Francisco JA, Cerveny CG, Meyer DL, Mixan BJ, Klussman K, Chace DF, Rejniak SX, Gordon KA, DeBlanc R, Toki BE, Law CL, Doronina SO, Siegall CB, Senter PD, Wahl AF. cAC10-vcMMAE, an anti-CD30-monomethyl auristatin E conjugate with potent and selective antitumor activity. Blood, 2003, 102, 1458-1465. 157 Bhatt S, Ashlock BM, Natkunam Y, Sujoy V, Chapman JR, Ramos JC, Mesri EA, Lossos IS. CD30 targeting with brentuximab vedotin: a novel therapeutic approach to primary effusion lymphoma. Blood, 2013, 122, 1233-1242. 158 Horwitz SM, Advani RH, Bartlett NL, Jacobsen ED, Sharman JP, O'Connor OA, Siddiqi T, Kennedy DA, Oki Y. Objective responses in relapsed T-cell lymphomas with single-agent brentuximab vedotin. Blood, 2014, 123, 3095-3100. 159 Bonifacino JS, Dell'Angelica EC, Springer TA. Immunoprecipitation. Curr. Protoc. Immunol., 2001, Chapter 8, Unit 8.3. 160 Austin CD, De Mazière AM, Pisacane PI, van Dijk SM, Eigenbrot C, Sliwkowski MX, Klumperman J, Scheller RH. Endocytosis and sorting of ErbB2 and the site of action of

247

cancer therapeutics trastuzumab and gledanamycin. Molecular Biology of the Cell. 2004, 15, 5268-5282. 161 Rudnick SI, Lou J, Shaller CC, Tang Y, Klein-Szanto AJ, Weiner LM, Marks JD, Adams GP. Influence of affinity and antigen internalization on the uptake and penetration of anti-Her2 antibodies in solid tumors. Cancer Res. 2011, 71, 2550-2259. 162 Hudziak RM, Lewis GD, Winget M, Fendly BM, Shepard HM, Ullrich A. p185Her2 monoclonal antibody has antiproliferative effects in vitro and sensitizes human breast tumor cells to tumor necrosis factor. Mol. Cell Biol. 1989, 9, 1165-1172. 163 Lewis GD, Figari I, Fendly B, Wong WL, Carter P, Gorman C, Shepard HM. Differential responses of human tumor cell lines to anti-p185Her2 monoclonal antibodies. Cancer Immunol. Immunother. 1993, 37, 255-263. 164 Carter P, Presta L, Gorman CM, Ridgway JB, Henner D, Wong WL, Rowland AM, Kotts C, Carver ME, Shepard HM. Humanization of an anti-p185Her2 antibody for human cancer therapy. Proc. Natl. Acad. Sci. USA. 1992, 89, 4285-4289. 165 Carter P, Presta L, Gorman CM, Ridgway JB, Henner D, Wong WL, Rowland AM, Kotts C, Carver ME, Shepard HM. Humanization of an anti-p185Her2 antibody for human cancer therapy. Proc. Natl. Acad. Sci. USA. 1992, 89, 4285-4289. 166 Jung ST, Reddy ST, Kang TH, Borrok MJ, Sandlie I, Tucker PW, Georgiou G. Aglycosylated IgG variants expressed in bacteria that selectively bind FcgammaRI potentiate tumor cell killing by monocyte-dendritic cells. Proc. Natl. Acad. Sci. USA, 2010, 107, 604-609. 167 Mazor Y, Van Blarcom T, Mabry R, Iverson BL, Georgiou G. Isolation of engineered, full-length antibodies from libraries expressed in Escherichia coli. Nat. Biotechnol., 2007, 25, 563-565. 168 B. Stern, L.C. Olsen, C. Trobe, H. Ravenberg, I.F. Pryme. Improving mammalian cell factories: the selection of signal peptide has a major impact on recombinant protein synthesis and secretion in mammalian cells. Trends in Cell & Molecular Biology. 2007, 2, 1- 17. 169 L.Zhang, Q. Leng, A.J. Mixson. Alteration in the IL-2 signal peptide affects secretion of proteins in vitro and in vivo. J. Gene Med. 2005, 7, 354-365. 170 L. Kober, C. Zehe, J. Bode. Optimized signal peptides for the development of high expressing CHO cell lines. Biotechnol. Bioeng. 2013, 110, 1164-1173. 171 Lucas BK, Giere LM, DeMarco RA, Shen A, Chisholm V, Crowley CW. High-level production of recombinant proteins in CHO cells using a dicistronic DHFR intron expression vector. Nucleic Acids Res., 1996, 24, 1774-1779. 172 Khamlichi AA, Rocca A, Cogné M. The effect of intron sequences on expression levels of Ig cDNAs. Gene, 1994, 150, 387-390.

248

173 Higashimoto T, Urbinati F, Perumbeti A, Jiang G, Zarzuela A, Chang LJ, Kohn DB, Malik P. The woodchuck hepatitis virus post-translational regulatory element reduces readthrough transcription from retroviral vectors. Gene Therapy. 2007, 14, 1298-1304. 174 R. Zufferey, J.E. Donello, D. Trono, T.J. Hope. Woodchuck hepatitis virus posttranslational regulatory element enhances expression of transgenes delivered by retroviral vectors. J. Virol. 1999, 73, 2886-2892. 175 S. Wulhfard, L. Baldi, D.L. Hacker, F. Wurm. Valproic acid enhances recombinant mRNA and protein levels in transiently transfected Chinese hamster ovary cells. J. Biotechnol. 2010, 148, 128-132. 176 A.C. Dalton, W.A. Barton. Over-expression of secreted proteins from mammalian cell lines. Prot. Sci. 2014, 23, 517-525. 177 Backliwal G, Hildinger M, Kuettel I, Delegrange F, Hacker DL, Wurm FM. Valproic acid: a viable alternative to sodium butyrate for enhancing protein expression in mammalian cell cultures. Biotechnol. Bioeng. 2008, 101, 182-189. 178 Soehnlein O, Zernecke A, Eriksson EE, Rothfuchs AG, Pham CT, Herwald H, Bidzhekov K, Rottenberg ME, Weber C, Lindbom L. Neutrophil secretion products pave the way for inflammatory monocytes. Blood. 2008, 112, 1461-1471. 179 Soehnlein O, Xie X, Ulbrich H, Kenne E, Rotzius P, Flodgaard H, Eriksson EE, Lindbom L. Neutrophil-derived heparin-binding protein (HBP/CAP37) deposited on endothelium enhances monocyte arrest under flow conditions. J. Immunol. 2005, 174, 6399-6405. 180 Kober L, Zehe C, Bode J. Optimized signal peptides for the development of high expressing CHO cell lines. Biotechnol. Bioeng. 2013, 110, 1164-1173. 181 Olczak M, Olczak T. Comparison of different signal peptides for protein secretion in nonlytic insect cell system. Anal. Biochem. 2006, 359, 45-53. 182 Petersen TN, Brunak S, von Heijne G, Nielsen H. SignalP 4.0: discriminating signal peptides from transmembrane regions. Nat. Methods. 2011, 28, 785-786. 183 Emanuelsson O, Brunak S, von Heijne G, Nielsen H. Locating proteins in the cell using TargetP, SignalP and related tools. Nat. Protocols. 2007, 2, 953-971. 184 Bendtsen JD, Nielsen H, von Heijne G, Brunak S. Improved prediction of signal peptides: SignalP 3.0. J. Mol. Bio. 2004, 340, 783-795. 185 Beate Stern B, Olsen LC, Trobe C, Ravneberg H, Pryme IF. Improving mammalian cell factories: The selection of signal peptide has a major impact on recombinant protein synthesis and secretion in mammalian cells. Trends Cell Mol. Biol. 2007, 2, 1-17. 186 Martoglio B, Dobberstein B. Signal sequences: more than just greasy peptides. Trends Cell Biol. 1998, 8, 410-415.

249

187 P.M. Drake, et al. Aldehyde tag coupled with HIPS chemistry enables the production of ADCs conjugated site-specifically to different antibody regions with distinct in vivo efficacy and PK outcomes. Bioconjug. Chem. 2014, 25, 1331-1341. 188 H.K. Erickson, P.U. Park, W.C. Widdison, Y.V. Kovtun, L.M. Garrett, K. Hoffman, R.J. Lutz, V.S. Goldmacher, W.A. Blättler. Antibody-maytansinoid conjugates are activated in targeted cancer cells by lysosomal degradation and linker-dependent intracellular processing. Cancer Res., 2006, 66, 4426-4433. 189 H. Uppal, E. Doudement, K. Mahapatra, W.C. Darbonne, D. Bumbaca, B.Q. Shen, X. Du, O. Saad, K. Bowles, S. Olsen, G.D. Lewis Phillips, D. Hartley, M.X. Sliwkowski, S. Girish, D. Dambach, V. Ramakrishnan. Potential mechanisms for thrombocytopenia development with trastuzumab emtansine (T-DM1). Clin. Cancer Res., 2015, 21, 123-133. 190 C.D. Austin, X. Wen, L. Gazzard, C. Nelson, R.H. Scheller, S.J. Scales. Oxidizing potential of endosomes and lysosomes limits intracellular cleavage of disulfide-based antibody-drug conjugates. Proc. Natl. Acad. Sci. USA, 2005, 102, 17987-17992. PMCID: PMC1298180. 191 L.M. Smith, A. Nesterova, S.C. Alley, M.Y. Torgov, P.J. Carter. Potent cytotoxicity of an auristatin-containing antibody-drug conjugate targeting melanoma cells expressing melanotransferrin/p97. Mol. Cancer Ther., 2006, 5, 1474-1482. 192 M.S. Sutherland, R.J. Sanderson, K.A. Gordon, J. Andreyka, C.G. Cerveny, C. Yu, T.S. Lewis, D.L. Meyer, R.F. Zabinski, S.O. Doronina, P.D. Senter, C.L. Law, A.F. Wahl. Lysosomal trafficking and cysteine protease metabolism confer target-specific cytotoxicity by peptide-linked anti-CD30-auristatin conjugates. J. Biol. Chem., 2006, 281, 10540-10547. 193 J.A. Francisco, C.G. Cerveny, D.L. Meyer, B.J. Mixan, K. Klussman, D.F. Chace, S.X. Rejniak, K.A. Gordon, R. DeBlanc, B.E. Toki, C.L. Law, S.O. Doronina, C.B. Siegall, P.D. Senter, A.F. Wahl. cAC10-vcMMAE, an anti-CD30-monomethyl auristatin E conjugate with potent and selective antitumor activity. Blood, 2003, 102, 1458-1465. 194 Kruger RG, et al (2004) Analysis of the substrate specificity of the Staphylococcus aureus sortase transpeptidase SrtA. Biochemistry 43(6): 1541-1551. 195 Kruger RG, Otvos B, Frankel BA, Bentley M, Dostal P, McCafferty DG. Analysis of the substrate specificity of the Staphylococcus aureus sortase transpeptidase SrtA. Biochemistry, 2004, 1541-1551. 196 S.O. Doronina, et al. Novel peptide linkers for highly potent antibody-auristatin conjugate. Bioconjug. Chem. 2008, 19, 1960-1963. 197 A.E. Albers, et al. Exploring the effects of linker composition on site-specifically modified antibody-drug conjugates. European Journal of Medicinal Chemistry, 2014, 17, 3-9.

250

198 Lewis GD, Figari I, Fendly B, Wong WL, Carter P, Gorman C, Shepard HM. Differential responses of human tumor cell lines to anti-p185HER2 monoclonal antibodies. Cancer Immunol. Immunother., 1993, 37, 255-263. 199 Hudziak RM, Lewis GD, Winget M, Fendly BM, Shepard HM, Ullrich A. p185HER2 monoclonal antibody has antiproliferative effects in vitro and sensitizes human breast tumor cells to tumor necrosis factor. Mol. Cell Biol., 1989, 9, 1165-1172. 200 A. Godwin, P. Bryant, M. Pabst, G. Badescu, M. Bird, W. McDowell, E. Jamieson, J. Swierkosz, K. Jurlewicz, R. Tommasi, K. Henseleit, X. Sheng, N. Camper, A. Manin, K. Kozakowska, K. Peciak, E. Laurine, R. Grygorash, A. Kyle, D. Morris, V. Parekh, A. Abhilash, J.W. Choi, J. Edwards, M. Frigerio, M.P. Baker. In Vitro and In Vivo Evaluation of Cysteine Rebridged Trastuzumab-MMAE Antibody Drug Conjugates With Defined DARs. Mol. Pharm., 2015 [Epub ahead of print] 201 A.E. Albers, A.W. Garofalo, P.M. Drake, R. Kudirka, G.W. de Hart, R.M. Barfield, J. Baker, S. Banas, D. Rabuka D. Exploring the effects of linker composition on site- specifically modified antibody-drug conjugates. Eur. J. Med. Chem., 2014, 88, 3-9. 202 P. Dennler, A. Chiotellis, E. Fischer, D. Brégeon, C. Belmant, L. Gauthier, F. Lhospice, F. Romagne, R. Schibli. A Transglutaminase-Based Chemo-Enzymatic Conjugation Approach Yields Homogeneous Antibody-Drug Conjugates. Bioconjug. Chem., 2014, 25, 569-578. 203 J. Asundi, L. Crocker, J. Tremayne, P. Chang, C. Sakanaka, J. Tanguay, S. Spencer, S. Chalasani, E. Luis, K. Gascoigne, R. Desai, R. Raja, B. Friedman, P.M. Haverty, P. Polakis, R. Firestein. An antibody drug conjugate (ADC) targeting the Lymphocyte Antigen 6 Complex, Locus E (LY6E) provides robust tumor killing in a wide range of solid tumor malignancies. Clin. Cancer Res., 2015, pii: clincanres.0156.2015. [Epub ahead of print]. 204 M. Dorywalska, P. Strop, J.A. Melton-Witt, A. Hasa-Moreno, S.E. Farias, M. Galindo Casas, K. Delaria, V. Lui, K. Poulsen, C. Loo, S. Krimm, G. Bolton, L. Moine, R. Dushin, T.T. Tran, S.H. Liu, M. Rickert, D. Foletti, D.L. Shelton, J. Pons, A. Rajpal. Bioconjug. Chem., 2015, 26, 650-659. 205 S.C. Jeffrey, P.J. Burke, R.P. Lyon, D.W. Meyer, D. Sussman, M. Anderson, J.H. Hunter, C. Leiske, J.B. Miyamoto, N.D. Nicholas, N.M. Okeley, R.J. Sanderson , I.J. Stone, W. Zeng, S.J. Gregson, L. Masterson, A.C. Tiberghien, P.W. Howard, D.E. Thurston, C.L. Law, P.D. Senter. A Potent Anti-CD70 Antibody−Drug Conjugate Combining a Dimeric Pyrrolobenzodiazepine Drug with Site-Specific Conjugation Technology. Bioconjug. Chem., 2013, 24, 1256-1263. 206 S. Golfier, C. Kopitz, A. Kahnert, I. Heisler, C.A. Schatz, B. Stelte-Ludwig, A. Mayer- Bartschmid, K. Unterschemmann, S. Bruder, L. Linden, A. Harrenga, P. Hauff, F.D. Scholle, B. Müller-Tiemann, B. Kreft, K. Ziegelbauer. Anetumab ravtansine: a novel

251

mesothelin-targeting antibody-drug conjugate cures tumors with heterogeneous target expression favored by bystander effect. Mol. Cancer Ther., 2014, 13, 1537-1538. 207 J. Deckert, P.U. Park, S. Chicklas, Y. Yi, M. Li, K.C. Lai, M.F. Mayo, C.N. Carrigan, H.K. Erickson, J. Pinkas, R.J. Lutz, T. Chittenden, J.M. Lambert. A novel anti-CD37 antibody-drug conjugate with multiple anti-tumor mechanisms for the treatment of B- cell malignancies. Blood, 2013, 122, 3500-3510. 208 R.C. Elgersma, R.G. Coumans, T. Huijbregts, W.M. Menge, J.A. Joosten, H.J. Spijker, F.M. de Groot, M.M. van der Lee, R. Ubink, D.J. van den Dobbelsteen, D.F. Egging, W.H. Dokter, G.F. Verheijden, J.M. Lemmens, C.M. Timmers, P.H. Beusker. Design, Synthesis, and Evaluation of Linker-Duocarmycin Payloads: Toward Selection of HER2-Targeting Antibody-Drug Conjugate SYD985. Mol. Pharm., 2015, DOI: 10.1021/mp500781a [Epub ahead of print] 209 P.M. Drake, A.E. Albers, J. Baker, S. Banas, R.M. Barfield, A.S. Bhat, G.W. de Hart, A.W. Garofalo, P. Holder, L.C. Jones, R. Kudirka, J. McFarland, W. Zmolek, D. Rabuka. Aldehyde Tag Coupled with HIPS Chemistry Enables the Production of ADCs Conjugated Site-Specifically to Different Antibody Regions with distinct in vivo efficacy and PK outcomes. Bioconjug. Chem., 2014, 25, 1331-1341. PMCID: PMC4215875. 210 V.A. Verma, T.H. Pillow, L. DePalatis, G. Li, G.L. Phillips, A.G. Polson, H.E. Raab, S. Spencer, B. Zheng. The cyptophycins as potent payloads for antibody drug conjugates. Bioorg. Med. Chem. Lett., 2015, 25, 864-868. 211 Maraffini LA, Dedent AC, Schneewind O (2006) Sortases and the art of anchoring proteins to the envelopes of Gram-positive bacteria. Microbiol Mol Biol Rev 70(1):192-221. 212 J. Richards, et al, J. Mol. Biol. 2003, 326, 1475-1488. 213 J. Duan, J. Wi, C.A. Valencia, R. Liu, Biochemistry 2007, 46, 12656-12664. 214 Mongkolsapaya J, Grimes JM, Chen N, Xu XN, Stuart DI, Jones EY, Screaton GR. Structure of the TRAIL-DR5 complex reveals mechanisms conferring specificity in apoptotic initiation. Nat. Struct. Biol., 1999, 6, 1048-1053. 215 Cha SS, Kim MS, Choi YH, Sung BJ, Shin NK, Shin HC, Sung YC, Oh BH. 2.8 A resolution crystal structure of human TRAIL, a cytokine with selective antitumor activity. Immunity, 1999, 11, 253-261. 216 Cha SS, Sung BJ, Kim YA, Song YL, Kim HJ, Kim S, Lee MS, Oh BH. Crystal structure of TRAIL-DR5 complex identifies a critical role of the unique frame insertion in conferring recognition specificity. J. Biol. Chem., 2000, 275, 31171-31177. 217 Mongkolsapaya J, Grimes JM, Chen N, Xu XN, Stuart DI, Jones EY, Screaton GR. Structure of the TRAIL-DR5 complex reveals mechanisms conferring specificity in apoptotic initiation. Nat. Struct. Biol., 1999, 6, 1048-1053.

252

218 Mazmanian SK, Ton-That H, Su K, Schneewind O. An iron-regulated sortase anchors a class of surface protein during Staphylococcus aureus pathogenesis. Proc. Natl. Acad. Sci. USA, 2002, 99, 2293-2298. 219 Perry AM, Ton-That H, Mazmanian SK, Schneewind O. Anchoring of surface proteins to the cell wall of Staphylococcus aureus. III. Lipid II is an in vivo peptidoglycan substrate for sortase-catalyzed surface protein anchoring. J. Biol. Chem., 2002, 277, 16241- 16248. 220 Dodds AW, Ren XD, Willis AC, Law SK. The reaction mechanism of the internal thioester in the human complement component C4. Nature, 1996, 379, 177-179. 221 Hnízda A, Santrůcek J, Sanda M, Strohalm M, Kodícek M. Reactivity of histidine and lysine side-chains with diethylpyrocarbonate -- a method to identify surface exposed residues in proteins. J. Biochem. Biophys.Methods, 2008, 70, 1091-1097. 222 Miles EW. Modification of histidyl residues in proteins by diethylpyrocarbonate. Methods Enzymol., 1977, 47, 431-432. 223 Allen G. Protein, Volume 2, Elsevier, 1999, Google Books, Web. 224 Lundblad RL. Chemical Modification of Biological Polymers, CRC Press, 2011, Google Books, Web. 225 Mongkolsapaya J, Grimes JM, Chen N, Xu XN, Stuart DI, Jones EY, Screaton GR. Structure of the TRAIL-DR5 complex reveals mechanisms conferring specificity in apoptotic initiation. Nat. Struct. Biol., 1999, 6, 1048-1053 226 Kim G, Weiss SJ, Levine RL. Methionine oxidation and reduction in proteins. Biochim. Biophys. Acta, 2014, 1840, 901-905. 227 Stadtman ER, Moskovitz J, Levine RL. Oxidation of methionine residues of proteins: biological consequences. Antioxid. Redox Signal., 2003, 5, 577-582. 228 Bauerle MR, Schwalm EL, Booker SJ. Mechanistic diversity of radical S- adenosylmethionine (SAM)-dependent methylation. J. Biol. Chem., 2015, 290, 3995-4002. 229 Mehta AP, Abdelwahed SH, Mahanta N, Fedoseyenko D, Philmus B, Cooper LE, Liu Y, Jhulki I, Ealick SE, Begley TP. Radical S-adenosylmethionine (SAM) enzymes in cofactor biosynthesis: a treasure trove of complex organic radical rearrangement reactions. J. Biol. Chem., 2015, 290, 3980-3986. 230 Kouzarides T. Chromatin modifications and their function. Cell, 2007, 128, 693-705. 231 Cheng X. DNA modification by methyltransferases. Curr. Opin. Struct. Biol., 1995, 5, 4- 10. 232 Rose K, Savoy LA, Simona MG, Offord RE, Wingfield PT, Mattaliano RJ, Thatcher DR. The state of the N-terminus of recombinant proteins: determination of N-terminal methionine (formylated, acetylated, or free). Anal. Biochem., 1987, 165, 59-69.

253

233 Liao YD, Jeng JC, Wang CF, Wang SC, Chang ST. Removal of N-terminal methionine from recombinant proteins by engineered E. coli methionine aminopeptidase. Protein Sci., 2004, 13, 1802-1810. 234 Hirel PH, Schmitter MJ, Dessen P, Fayat G, Blanquet S. Extent of N-terminal methionine excision from Escherichia coli proteins is governed by the side-chain length of the penultimate amino acid. Proc. Natl. Acad. Sci. USA, 1989, 86, 8247-8251. 235 Frottin F, Martinez A, Peynot P, Mitra S, Holz RC, Giglione C, Meinnel T. The proteomics of N-terminal methionine cleavage. Mol. Cell. Proteomics, 2006, 5, 2336-2349. 236 Xiao Q, Zhang F, Nacev BA, Liu JO, Pei D. Protein N-terminal processing: substrate specificity of Escherichia coli and human methionine aminopeptidases. Biochemistry, 2010, 49, 5588-5599. 237 Strickland EC, Geer MA, Tran DT, Adhikari J, West GM, DeArmond PD, Xu Y, Fitzgerald MC. Thermodynamic analysis of protein-ligand binding interactions in complex biological mixtures using the stability of proteins from rates of oxidation. Nat. Protoc., 2013, 8, 148-161. 238 Tran DT, Adhikari J, Fitzgerald MC. Stable Isotope Labeling with Amino Acids in Cell Culture (SILAC)-based strategy for proteome-wide thermodynamic analysis of protein- ligand binding interactions. Mol. Cell. Proteomics, 2014, 13, 1800-1813. 239 Jenny RJ, Mann KG, Lundblad RL. A critical review of the methods for cleavage of fusion proteins with thrombin and factor Xa. Protein Expr. Purif., 2003, 31, 1-11. 240 Terpe K. Overview of tag protein fusions: from molecular and biochemical fundamentals to commercial systems. Appl. Microbiol. Biotechnol., 2003, 60, 523-533. 241 Huang X, Aulabaugh A, Ding W, Kapoor B, Alksne L, Tabei K, Ellestad G. Biochemistry, 2003, 42, 11307-11315. 242 Horie R, Watanabe T. CD30: expression and function in health and disease. Semin. Immunol. 1998, 10, 457-470.

254