Site-Specific Conjugation to Native and Engineered Lysines in Human

Aug 18, 2017 - The use of microbial transglutaminase (MTG) to produce site-specific antibody–drug conjugates (ADCs) has thus far focused on the tran...
0 downloads 13 Views 799KB Size
Subscriber access provided by UNIVERSITY OF ADELAIDE LIBRARIES

Article

Site-Specific Conjugation to Native and Engineered Lysines in Human Immunoglobulins by Microbial Transglutaminase Jared Spidel, Benjamin Vaessen, Earl Albone, Xin Cheng, Arielle Verdi, and J. Bradford Kline Bioconjugate Chem., Just Accepted Manuscript • DOI: 10.1021/acs.bioconjchem.7b00439 • Publication Date (Web): 18 Aug 2017 Downloaded from http://pubs.acs.org on August 19, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Bioconjugate Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Bioconjugate Chemistry

Site-Specific Conjugation to Native and Engineered Lysines in Human Immunoglobulins by Microbial Transglutaminase Jared L. Spidel1, Benjamin Vaessen, Earl F. Albone, Xin Cheng, Arielle Verdi, and J. Bradford Kline Morphotek Inc., 210 Welsh Pool Road, Exton, PA, 19341 1

corresponding author: [email protected]

ACS Paragon Plus Environment

Bioconjugate Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ABSTRACT The use of microbial transglutaminase (MTG) to produce site-specific antibody-drug conjugates (ADCs) has thus far focused on transamidation of engineered acyl donor glutamine residues in an antibody based on the hypothesis that the lower specificity of MTG for acyl acceptor lysines may result in transamidation of multiple native lysine residues, thereby yielding heterogeneous products. We investigated the utilization of native IgG lysines as acyl acceptor sites for glutamine-based acyl donor substrates. Of the approximately 80 lysines in multiple recombinant IgG monoclonal antibodies (mAbs), none were transamidated. As recombinant mAbs lack the C-terminal Lys447 due to cleavage by carboxypeptidase B in the production cell host, we explored whether blocking the cleavage of Lys447 by the addition of a Cterminal amino acid could result in transamidation of Lys447 by a variety of acyl donor substrates. MTG efficiently transamidated Lys447 in the presence of any non-acidic, non-proline amino acid residue at position 448. Lysine scanning mutagenesis throughout the antibody further revealed several transamidation sites in both the heavy and light chain constant regions. Additionally, scanning mutagenesis of the hinge region in a Fab' fragment revealed sites of transamidation that were not reactive in the context of the full-length mAb. Here we demonstrate the utility of single lysine substitutions and the C-terminal Lys447 for engineering efficient acyl acceptor sites suitable for site-specific conjugation to a range of glutamine-based acyl donor substrates.

ACS Paragon Plus Environment

Page 2 of 27

Page 3 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Bioconjugate Chemistry

INTRODUCTION Production of homogeneous antibody drug-conjugate (ADC) products requires efficient conjugation to one or more specific amino acid residues and/or glycan structures. Traditional protein conjugation methods to lysines or cysteines typically result in heterogeneous drug products with varying drug-toantibody ratios (DARs) due to multiple lysines and cysteines existing throughout the antibody. Sitespecific conjugation methods can alleviate these complexities by using chemical or biochemical methods that are unique and specific to the desired target residue. Various methods for achieving site-specific conjugation have been previously described that include engineering of unpaired cysteines, incorporation of non-natural amino acids, and site-specific enzymatic modification. Thiol-based conjugation to an engineered unpaired cysteine results in conjugation to just the engineered cysteine due to all native IgG cysteines being involved in intra-and inter-chain disulfide bridges.1-5 Non-natural amino acids (NNAAs) are a powerful way to expand the genetic code whereby amino acids with unique chemically reactive groups (e.g., strain-promoted alkyne-azide cycloaddition [SPAAC] reactive groups) that can specifically react with the linker or cytotoxin are incorporated into an antibody.6 However, production of NNAAmodified antibodies requires a modified organism or cell line.7-9 Several site-specific enzymatic modifications have also been described, including SMAC-technology™,10-12 SMARTag™,13 and microbial transglutaminase (MTG).14-19 These approaches allow for the generation of site-specific conjugatable antibodies produced using standard cell lines and methods. The major disadvantage of using an enzymatic-based approach is a requirement for engineering a peptide tag specific for the enzyme into the antibody, thereby increasing the risk of immunogenicity, changing the biophysical nature of the antibody, and adding additional manufacturing cost of the enzyme during the production of the ADC. In developing an optimized site-specific conjugation method, we sought to incorporate the advantages of enzymatic-based conjugation technologies while minimizing the risks. MTG is a preferable and efficient mediator of amino acid conjugation due to its ability to form extremely stable covalent bonds and its familiarity with regulatory agencies due to its use within the food industry over the past few decades. Transglutaminases are a family of proteins that catalyze the formation of a stable isopeptide bond between the γ-carboxyamide group (acyl donor) of a glutamine and the ε-amino group (acyl acceptor) of a lysine (reviewed in 20-22). Members of the transglutaminase family, which are structurally and functionally related, are found in organisms ranging from bacteria to humans and are involved in a variety of cellular processes. MTG isolated from the bacterium Streptomyces mobaraensis has been used extensively to crosslink proteins together for various applications. It is an attractive conjugation enzyme due to its ability to function under a wide range of pH, salt, and temperature conditions, as well as its low cost. Recently several groups have explored utilizing MTG as a means to produce homogeneous ADCs. Due to the lower specificity for the acyl acceptor amine by MTG, research thus far has been focused on transamidation of antibody glutamine residues.14-19 Despite human IgG1 containing ~60 glutamines of which 80-90% are predicted to be solvent exposed, little or no transamidation of wild-type antibodies by an acyl-acceptor substrate could be demonstrated.15-17 However, deglycosylation of IgG1 either enzymatically by PNGase F or via the heavy chain point mutation N297Q enabled transamidation of Gln295 or both Gln295 and Gln297, respectively, by acyl acceptor substrates.15, 16, 18 It was further reported that the four amino acid acyl-donor peptide LLQG could be inserted at the N or C terminus as well as at various positions within an antibody to provide a suitable site for transamidation.17 Despite these successful demonstrations, several limitations exist with the current technology. First, aglycosylated antibodies typically undergo a conformational change that decreases their thermal stability and increases protease susceptibility and aggregation rates.23-28 Second, engineering an additional fouramino acid tag into an antibody may increase the risk of immunogenicity. Finally, the incorporation of a four-amino acid tag may alter the biophysical properties of an antibody by increasing hydrophobicity, or disruption of the structure when inserted at an internal site.

ACS Paragon Plus Environment

Bioconjugate Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

To mitigate the disadvantage of engineering an enzyme-recognition tag into the antibody and enhance the utility of MTG-mediated bioconjugation, we investigated the ability to utilize potential acyl acceptor sites in native human IgG. While it has been speculated that wild-type IgG1 antibodies may contain several native acyl acceptor sites on the surface of an IgG14, 16 due to the lower specificity of MTG for acyl acceptor sites,29-31 transamidation of IgG lysines has not been investigated. We, therefore, explored the ability to utilize native lysines or single lysine substitutions as acyl acceptor sites. Human IgG antibodies contain an average of 80 lysines, of which 80-90% are predicted to be solvent exposed.32 Additionally, the C-terminal codon of IgG is a lysine,33-35 however, serum-derived IgGs lack this lysine.36-39 Recombinant expression of IgG1 in HEK293 and CHO cells have also been found to produce antibodies lacking the C-terminal Lys447 (EU numbering),35, 40-42 presumably due to cleavage by a carboxypeptidase B specific for C-terminal lysine and arginine residues, as the addition of a non-basic C-terminal amino acid blocks removal of Lys447.43 Here we show that none of the native lysines in recombinantly expressed human IgG are efficiently modified by MTG for transamidation. Interestingly, lysine scanning mutagenesis revealed several positions throughout all IgG isotypes that enabled efficient transamidation by MTG. Moreover, modified lysine residues located in the middle and lower hinge regions were efficient acyl acceptors in the context of a Fab' format, but not in the context of a full-length antibody. Finally, our engineering efforts found that when the native Lys447 was protected from cleavage by a single C-terminal amino acid extension, MTG was able to utilize native Lys447 as a highly efficient acyl acceptor. RESULTS Natural IgG1 antibodies have no acyl acceptor sites for MTG-mediated transamidation Despite over two decades of research, the substrate specificity of MTG has not been clearly defined. In general, glutamines or lysines on exposed flexible loops with hydrophobic or positively charged adjacent residues are more likely to be substrates for MTG transamidation.44-49 Potential native acyl acceptor sites in IgG1 were identified by analyzing the crystal structures of human Fab (both kappa and lambda light chains) and Fc fragments for solvent exposed lysines in loop structures. Of the ~80 lysines in a human IgG1 antibody, approximately 50-60% were found to be both solvent exposed and located within loop regions (data not shown). To determine the ability of MTG to transamidate one or more of these native lysine residues several monoclonal antibodies (mAbs), including mouse-human chimeric (mAb1), humanized (mAb2, 3, 6), and fully human (mAb4, 5) mAbs containing both kappa (mAb1, 2, 3, 4, 6) and lambda (mAb5) light chains (LCs), were incubated with carboxybenzyl-glutamine-glycine-cadaverinebiotin (ZQG-biotin) and MTG. Transamidation was analyzed by determining the masses of the LC, Fd, and Fc fragments using electrospray ionization-mass spectrometry (ESI-MS) following digestion of the samples with IdeS and reduction with dithiothreitol (DTT). Since the mAbs were not deglycosylated, two mass peaks corresponding to the G0F (+1445 Da) and G1F (+1608 Da) glycoforms were observed for each Fc (Figure 1). Mab4 contains an N-linked glycosylation site in the variable heavy region, and two glycan species, G2FS and G2FS2 were observed. All samples lacked the C-terminal lysine 447 (128 Da) as evidenced by the -130 to -132 Da difference between the observed and theoretical mass for the Fc. Although these antibodies contained 42-50 solvent exposed lysines within loops, neither the heavy chain (HC) nor the LC was found to be modified by the acyl donor substrate.

ACS Paragon Plus Environment

Page 4 of 27

Page 5 of 27

A. 100

23213

mAb1

25069

100

23750

mAb2

100

23478

mAb3

25071

%

26097

%

%

25198 25359

25258

25198

25420

25360

0 20000

100

22500

25000

27500

mass 30000

23530

mAb4

0 20000 100

22500

25000

27500

mass 30000

22653

mAb5

0 20000 100

22500

25000

27500

mass 30000

23470

mAb6

26337

25381

%

%

25198

%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Bioconjugate Chemistry

25198

25166 25327 27564

0 20000

22500

25000

27500

mass 30000

25359

25360

27855

0 20000

22500

25000

27500

mass 30000

0 20000

22500

25000

27500

mass 30000

B. ZQG-biotin: +631 Da LC Fd Fc Calc. Observed ∆Mass Calc. Observed Glycan ∆Mass Glycan Calc. Observed ∆Mass mAb1 23216 23213 -3 25072 25069 -3 G0F 25328 25198 -130 G1F 25491 25359 -132 mAb2 23751 23750 -1 25073 25071 -2 G0F 25328 25198 -130 G1F 25491 25360 -131 mAb3 23478 23478 0 26097 26097 0 G0F 25388 25258 -130 G1F 25551 25420 -131 mAb4 23532 23530 -2 27566 27564 G2FS -2 G0F 25296 25166 -130 27857 27855 G2FS2 -2 G1F 25459 25327 -132 mAb5 22655 22653 -2 26340 26337 -3 G0F 25328 25198 -130 G1F 25491 25360 -131 mAb6 23472 23470 -2 25383 25381 -2 G0F 25328 25198 -130 G1F 25491 25359 -132 Figure 1. ESI-MS analysis of antibodies incubated with an acyl donor substrate and MTG was performed by incubation with ZQGbiotin and MTG at 37°C overnight followed by digestion with IdeS to generate (Fab')2 and Fc fragments. (A) The masses of the LC, Fd, and Fc were determined by ESI-MS. (B) The theoretical mass of each fragment was determined by the amino acid sequence subtracted from the observed mass to determine the change in mass (∆mass). A ∆mass of approximately -128 Da is due to cleavage of Lys447. The Fc is glycosylated with one or two oligosaccharides, G0F or G1F.

A single amino acid extension was sufficient for efficient transamidation of Lys447 To benchmark the potential for optimal MTG-mediated transamidation of IgG mAbs, a positive control peptide with two known lysine acyl acceptor sites (GGSTKHKIPGGS)50 was genetically fused to the C terminus of mAb1 HC or LC (HC-KTag or LC-KTag, respectively) and analyzed for transamidation as in Figure 1. The samples were deglycosylated by PNGase F, reduced with DTT, and the HCs and LCs were analyzed by ESI-MS. The LC-KTag antibody was modified with up to two ZQG-biotin molecules, consistent with modification of the two lysines in the KTag (Table 1). The addition of the KTag to the HC surprisingly resulted in the addition of up to three ZQG-biotin molecules per HC. As there are only two lysines in the KTag, a natural lysine within the antibody was the third acyl acceptor site. Given the proximity to the KTag, we inferred that the most likely acyl acceptor site was HC Lys447. Indeed, a similar observation was reported whereby the C-terminal lysine of EGFP was a site of transamidation only in the presence of additional C-terminal residues.51

ACS Paragon Plus Environment

Bioconjugate Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

LC mAb1 HC-KTag LC-KTag

∆Mass 0 0 630 1261

# Biotins 0 0 1.00 2.00

% transamidation 0.0% 0.0% 19.0% 81.0%

HC mAb1 HC-KTag

∆Mass # Biotins % transamidation -127 0 0.0% 627 0.99 18.7% 1262 2.00 42.6% 1894 3.00 38.6% LC-KTag 2 0 0.0% Table 1. Transamidation of an antibody with a C-terminal KTag was analyzed as in Figure 1 except samples were deglycosylation with PNGase F followed by reduction with DTT. A ∆mass of approximately -128 Da is due to cleavage of Lys447 and an approximate ∆mass of +631 Da indicates the addition of one ZQG-biotin. The number of ZQG-biotin molecules conjugated onto the HC or LC was determined by dividing the change in mass by the mass of ZQG-biotin.

Lys447 is typically cleaved by carboxypeptidase B during recombinant IgG1 expression in HEK293 and CHO cells.40-42 However, the addition of the KTag to the HC C terminus blocks removal of Lys447 thereby allowing MTG to utilize Lys447 as an acyl acceptor site. To determine whether transamidation of Lys447 was unique to the KTag, cleavage of Lys447 was blocked by the addition of one or two leucine residues at the C terminus of mAb1 (mAb1-HC-Leu448 or mAb1-HC-LeuLeu449, respectively). Purified antibodies were incubated with ZQG-biotin and MTG and the mass of the deglycosylated HC was analyzed by ESI-MS. The addition of either one or two leucines not only protected Lys447 cleavage but facilitated transamidation by an acyl donor substrate (Table 2). C terminus ∆Mass % transamidation mAb1 ...SPGK* -134 0.0% mAb1-HC-Leu448 ...SPGKL 624 100.0% mAb1-HC-LeuLeu449 ...SPGKLL 624 100.0% Table 2. Transamidation of an HC with a C-terminal leucine(s) was analyzed as in (Table 1). A ∆mass of approximately -128 Da is due to cleavage of Lys447 and an approximate ∆mass of +631 Da indicates the addition of one ZQG-biotin.

To determine if other amino acids could also block cleavage of Lys447 and enable MTG-mediated Lys447 transamidation, the remaining 19 natural amino acids were engineered as single-residue extensions to the C terminus of mAb1. Samples were then analyzed for the ability of MTG to transamidate Lys447 using ZQG-biotin as above. As shown, C-terminal lysine and arginine residues did not protect cleavage of Lys447, as they too are substrates for carboxypeptidase B, while all other amino acids except proline and acidic residues facilitated efficient conjugation to the substrate despite preventing cleavage of Lys447 (Figure 2).

ACS Paragon Plus Environment

Page 6 of 27

Page 7 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Bioconjugate Chemistry

Figure 2. The effect of single C-terminal amino acids on transamidation of HC Lys447 was analyzed as in Figure 1. * indicates the wildtype antibody with a stop codon following the Lys447 codon.

Due to the cleavage of an additional single lysine or arginine C-terminal amino acid, the effect of either amino acid on transamidation on Lys447 could not be assessed. Therefore, a leucine was added to the C terminus of lysine or arginine 448 variants (KL and RL, respectively). The effect of an additional Cterminal leucine was also investigated with the proline (PL), aspartate (DL), and glutamate (EL) 448 variants. As shown in Figure 2, blocking cleavage of the lysine or arginine resulted in 100% transamidation of Lys447. Furthermore, the additional lysine in the KL variant was also found to be 100% transamidated, yielding an antibody with 4 transamidation sites (data not shown). The C-terminal leucine moderately increased transamidation of the proline variant but had no impact on transamidation of Lys447 when preceded by an acidic residue. Optimal transamidation of an LC C-terminal lysine required a spacer between Cys214 and the lysine A lysine was engineered at the C terminus of the LC to determine whether a single C-terminal lysine extension is sufficient to act as an acyl acceptor site. MAb1-LC-K was incubated with ZQG-biotin and MTG and analyzed by ESI-MS as above. In contrast to an HC C-terminal lysine, an LC C-terminal lysine was not cleaved and not an efficient acyl acceptor site (Figure 3). An additional leucine was added to the C terminus of the lysine (KL); however, the lysine was not transamidated likely due to the interchain disulfide bond between LC C-terminal Cys214 residue and the HC where steric hindrance or lack of solvent exposure may have prevented transamidation. To test this hypothesis, a single leucine was added between Cys214 and the lysine to create LK and LKL motifs. No cleavage or transamidation was observed with the LK motif; however, 9.1% of the LCs with the LKL motif were transamidated. To determine if further extension of the lysine from the LC-HC interface could optimize transamidation, a four amino acid Gly-Ser linker was inserted between Cys214 and the lysine (GGSGK). This extension resulted in cleavage of a C-terminal lysine (data not shown). Protecting cleavage of the lysine by the addition of a C-terminal leucine (GGSGKL) resulted in 77.8% of the LCs being transamidated.

ACS Paragon Plus Environment

Bioconjugate Chemistry

100 90 80 70 60 50 40 30 20 10 b1 KL m LC A -L b1 m K A b1 -LC LK m LC A L -G b1 G -L SG C -G K G SG K L A

b1

-L C

m

A m

A

b1 -L C

-K

0

m

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 3. Transamidation of an LC C-terminal lysine was determined as in Figure 1 except samples were reduced with DTT. An approximate ∆mass of +631 Da indicates the addition of one ZQG-biotin.

Lysine scanning mutagenesis of human gamma 1, kappa, and lambda constant regions revealed multiple sites for transamidation Since none of the native putative acyl acceptor sites in recombinant human IgG1 mAbs are transamidated by MTG, lysine scanning mutagenesis was performed on mAb1 to identify regions that could be engineered to introduce one or more efficient acyl acceptor sites. As lysines on exposed flexible loops tend to be preferred sites of transamidation,44-46 mutagenesis was limited to solvent-exposed residues within the loops or turns of the constant regions of human gamma 1, kappa and lambda. Antibodies were initially screened for transamidation using an ELISA-based assay. Following incubation with MTG and ZQG-biotin, antibodies were captured on an anti-Fcγ coated microtiter plate and biotinylated antibodies were detected by HRP-conjugated streptavidin. Wild-type mAb1 was included as a negative control. The fluorescence signal from several samples was approximately three times the mean signal of 2677 RFU (Supplemental Table 1). These included S136K, D221K, T223K, H224K, M252K, N297K and P445K in the HC and S292K in the kappa LC. Identification of the most optimal engineered transamidation sites was determined using ESI-MS. Due to a large number of samples, those most likely to be transamidated were further characterized. To determine the range of fluorescent signal observed in the ELISA assay that correlated to a high percentage of sample transamidation, the masses of all CH1 and upper hinge mutants were analyzed by ESI-MS. Overall, the fluorescence signal from the ELISA screen correlated with the ESI-MS data with signals greater than 7000 RFU corresponding to >70% transamidation of S136K, D221K, T223K, and H224K (Figure 4). The fluorescence signal for T135K and T225K was below 7000 RFU, but those samples were 80.4% and 100% transamidated, respectively. In contrast, G137K had a fluorescence signal of 8616 RFU, but only 38.9% was transamidated. Therefore, while a fluorescence signal >7000 RFU frequently corresponded to a high percentage of transamidation rate, there were a few false positives or negatives, and a lower cutoff of >4000 RFU was used to select CH2 and CH3 samples for analysis. Since only two kappa and no lambda samples were greater than 4000 RFU, LC samples greater than 3000 RFU were also analyzed. The transamidation reaction was performed by incubating ZQG-biotin and MTG with CH2 mutants M252K, E283K, A287K, and N297K, CH3 mutants P343K, G385K, G420K, H433K, L443K, S444K, and P445K, kappa mutants D151K, L201K, S202K, and E213K, and lambda mutants V147K, Q187K, and E213K. Three additional CH1 mutants (S191K, S192K, and L193K) that were not included

ACS Paragon Plus Environment

Page 8 of 27

Page 9 of 27

in the initial ELISA screen were also analyzed by ESI-MS. Of all the mutants screened, the CH1 L193K, CH2 M252K, CH3 P445K, kappa L201K and S202K, and lambda E213K substitutions were found to be transamidated most efficiently (Figure 4). CH2 mutant N297K was inconclusive due to the very low signal.

T1 3 T 1 5K 2 S1 0K 3 G 6K 13 T 1 7K 3 S1 9K 9 S1 1K 9 L1 2K 9 D 3K 22 T2 1K 2 H 3K 22 T2 4K 2 M 5K 25 E2 2K 8 A 3K 28 N 7K 29 P3 7K 4 G 3K 38 G 5K 42 H 0K 43 L4 3K 4 S4 3K 44 P4 K 4 D 5K 15 L2 1K 0 S2 1K 0 E2 2K 1 V1 3K 4 Q 7K 18 E2 7K 13 K Figure 4. Transamidation of single lysine substitutions in gamma 1, kappa, and lambda constant regions was analyzed as in Figure 1. * Not Determined

-T H 135 C -S K H 13 C 6 -D K HC 221 -T K H 223 C -H K H 224 C -T K H 22 C -M 5K H 25 2 C -N K HC 297 -P K L C 445 -L K LC 201 -S K 20 2K

The mutants with the highest transamidation efficiency were reanalyzed in a single experiment using the ZQG-biotin substrate and MTG. The region of transamidation was confirmed by digesting samples with IdeS and reduction prior to ESI-MS analysis to generate Fd (CH1 and hinge), Fc (CH2 and CH3), and LC fragments. Conjugation to N297K was again inconclusive due to low Fc signal despite high Fd and LC signals. Transamidation of samples all correlated to the domain containing the lysine substitution. T135K and P445K were 100% transamidated, and conjugation to M252K was nearly 100% (Figure 5). S136K and S221K were found to be transamidated over 80%. The other three hinge mutants had transamidation efficiencies of less than 50%. Both LC mutants were transamidated greater than 80%.

H C

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Bioconjugate Chemistry

Figure 5. Analysis of transamidation of select single lysine substitutions was conducted by incubating samples with ZQG-biotin and MTG at 37°C overnight. The samples were reduced with DTT and the masses of the HC and LC were determined by LC-MS.

Transamidation of engineered lysines with additional acyl donor substrates

ACS Paragon Plus Environment

Bioconjugate Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 27

One utility of antibody transamidation is for the manufacturing of site-specific ADCs. Conjugation of payloads could be achieved by one of two methods: 1) a two-step method whereby an acyl donor containing a biorthogonal reactive group is conjugated to the antibody, followed by a second conjugation of a payload to the reactive group or 2) a single step method utilizing a payload synthesized with an acyl donor group such as ZQG. The molecular weight of the payloads in both cases would range from small (e.g., ZQG conjugated with an azido group) to large (e.g., ZQG conjugated to a linker-cytotoxin). Therefore, payloads of various molecular weights were tested in transamidation reactions. Two click chemistry substrates were tested as payloads for transamidation. Amino-PEG3-bicyclononyne (BCN) or aminopropylazide was added to the carboxylic acid group of ZQG as detailed in the Methods section. MAb1 with engineered lysines in the HC or LC or with mAb1-HC-Leu448 were incubated with ZQG-azide (ZQG-N3), ZQG-PEG2-BCN, or ZQG-PEG3-BCN and MTG. Reduced samples were analyzed by liquid chromatography-mass spectrometry (LC-MS) to determine addition of the substrate to the antibody. ZQG-N3 was added efficiently (>75% conjugation or DAR 1.5) to most lysine substitutions and mAb1-HC-Leu448 (Table 3). The most permissible sites of transamidation were S135K-HC, L193KHC, D221K-HC, M252K-HC, N297K-HC, P445K-HC, HC-Leu448, and L201K-LC. T136K-HC, T223K-HC, T225K-HC, S202K-LC, and LC-GGSGKL were all transamidated >75% with ZQG-biotin, but not with ZQG-N3. Therefore, not all acyl donors equally transamidate the same acyl acceptor site. Similarly, mAb1-HC-Leu448 was 100% transamidated with ZQG-PEG3-BCN, but transamidation by ZQG-PEG2-BCN onto lysine substitutions was inefficient. The percentage of transamidation by ZQGPEG2-BCN did not vary widely among the lysine substitutions, contrary to ZQG-N3 transamidation. For example, there was a 32% difference in transamidation of T135K-HC and S136K-HC by ZQG-N3, but only a 4% difference with ZQG-PEG2-BCN. HC Modification T135K S136K L193K D221K

ZQG-N3 89.1% 59.6% 81.5% 55.1% 8.5% 48.7% 55.7% 34.1% 89.0% 27.6% 41.3% 23.7% 73.0%

ZQG-PEG2-BCN 27.1% 23.1% 63.7% 31.5%

ZQG-PEG2-AuF 51.4% 16.1% 23.3% 39.7%

P445K

DAR Species 1 1 1 1 2 1 1 1 1 1 2 3 1

0.0% 17.3% 0.0% 15.6% 6.0% 0.0% 0.0% 49.6%

ND ND ND 82.8% 46.5% 30.4% 4.5% 48.7%

HC Modification HC-Leu448

DAR Species 1

ZQG-N3 90.8%

ZQG-PEG3-BCN 100%

ZQG-PEG2-AuF 75.4%

T223K H224K T225K M252K N297K

LC Modification DAR Species ZQG-N3 ZQG-PEG2-BCN ZQG-PEG2-AuF L201K 1 100.0% 52.5% 92.0% S202K 1 58.5% 0.0% 44.5% GGSGKL 1 69.5% 0.0% 44.5% Table 3. Transamidation of single lysine substitutions and Lys447 to various acyl donor payloads was examined as in Figure 5, except the HC-Leu448 + ZQG-PEG2-AuF sample was digested with IdeS followed by reduction with DTT and analysis by LC-MS.

Interestingly, conjugation to D221K-HC and N297K-HC resulted in multiple conjugation sites within the HC. The D221K-HC substitution is adjacent to Lys222, which is not typically an acyl acceptor site (Figure 1); however, the adjacent lysine facilitated a low level of transamidation of Lys222. Of note, only the ZQG-N3 substrate was conjugated at more than one site in D221K-HC. The structure of ZQG-N3 is smaller than the other substrates tested and it may be that steric hindrance of the second conjugation site blocks its transamidation by the other substrates. The N297K-HC mutation removes the glycosylation

ACS Paragon Plus Environment

Page 11 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Bioconjugate Chemistry

site at Asn297, and aglycosylated antibodies have been reported to adopt a different structure than the glycosylated forms.23-28 The N297Q mutation also results in an aglycosylated antibody that perturbs the antibody secondary structure in such a way that Gln295 is then transamidated by a variety of aminocontaining acyl acceptor substrates.15, 16, 18 N297K-HC likely changes the confirmation of the CH2 region resulting in multiple native lysines now becoming acyl acceptors sites for ZQG-N3. ZQG-PEG2-Auristatin F (ZQG-PEG2-AuF) was synthesized to examine the direct transfer of a cytotoxin onto an acyl acceptor site. ZQG-PEG2-AuF was incubated with mAb1 single lysine substitutions or mAb1-HC-Leu448 and MTG and analyzed by LC-MS. Transamidation was most efficient onto acyl acceptor sites M252K-HC, HC-Leu448, and L201K-LC (82.8%, 75.4%, and 92% of the HC or LC conjugated, respectively; Table 3). The efficiency at certain sites did not correspond with the ZQG-N3 data. Conjugation to ZQG-PEG2-AuF was low for T135K-HC, L193K-HC, and D221K-HC but high for ZQG-N3. N297K-HC again demonstrated multiple conjugation sites, but the efficiency was not as high as with ZQG-N3. Disruption of secondary structure by a single lysine insertion facilitated transamidation Residues Ser190 through Thr195 form a beta turn that connects beta strands E and F in the CH1 domain. Lysine scanning mutagenesis across this exposed area found that only one site – position 193 – was an acceptable acyl donor site (Figure 4). This region forms an alpha helix, and it is possible that this secondary structure prevents transamidation in this region. To potentially disrupt the structure, a lysine was inserted between Ser191 and Ser192, Ser192 and Leu193, or Leu193 and Gly194 in mAb1. Samples were incubated with MTG and ZQG-biotin and analyzed by ESI-MS. Insertion of a lysine between Ser191 and Ser192 or Ser192 and Leu193 resulted in 100% transamidation, but an insertion between Leu193 and Gly194 exhibited no transamidation (Table 4). sequence ∆Mass % transamidation S191.K.S192 SSKSLGT 628 100.0% S192.K.L193 SSSKLGT 628 100.0% -3 0.0% L193.K.G194 SSSLKGT Table 4. Transamidation of a lysine insertion in a structured region was determined as in Table 1. An approximate ∆mass of +631 Da indicates the addition of one ZQG-biotin.

Multiple acyl acceptor sites in a single mAb were transamidated to increase the DAR An engineered antibody with one acyl acceptor site yields a theoretical DAR of 2. By combining multiple engineered acyl acceptor sites, the theoretical DAR increases by 2 for each site. Increasing the drug load of an ADC can lead to increased delivery of cytotoxic payloads into target cells per binding event, which may allow for maximal target cell killing and potentially lower patient dosing. To determine whether multiple acyl acceptor sites could yield a DAR greater than 2, mAb1 was engineered to include the LC modifications L201K or S202K in combination with the CH1 T135K or S136K, the CH1-CH2 S136KN297K, the CH1-CH2-CH3 S136K-N297K-P445K, or the CH1-CH3 T135K-L448 modifications to yield antibodies with 4, 6, 8, or 6 acyl acceptor sites, respectively. T135K-based samples were incubated with MTG and ZQG-biotin and their masses were analyzed by LCMS. The L201K-LC was 100% transamidated in all samples (Table 5). S202K-LC in combination with T135K-HC was more than double the efficiency when in combination with T135K-L448-HC (69.8% versus 31.8%). T135K-HC was 100% transamidated when combined with just L201K-LC resulting in a DAR of 4.0. In combination with S202K-LC, transamidation was reduced to 88.7% with an average DAR of 3.17. Combining T135K-HC with the HC-L448 resulted in 100% transamidation of the antibodies. However, the DAR was heterogeneous with 77.1% containing 2 biotins per HC and 22.9% containing only 1 biotin yielding an average DAR of 3.54. When the double-HC modifications were

ACS Paragon Plus Environment

Bioconjugate Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 27

combined with an LC modification, transamidation efficiency of the HC decreased while the amount of DAR 1 species more than doubled and the DAR 2 species dropped 2.3- to 6.5-fold. Despite these two antibodies having a potential DAR of 6, the average DAR was less than 4.

T135K-HC/L201K-LC T135KHC/S202K-LC T135K-L448-HC T135K-L448-HC/L201K-LC T135K-L448-HC/S202K-LC

% HC transamidation DAR 1 DAR 2 100.0% 0.0% 88.7% 0.0% 22.9% 77.1% 56.0% 11.9% 52.8% 33.2%

% LC transamidation DAR 1 100.0% 69.8% 0.0% 100.0% 31.8%

Ave DAR 4.00 3.17 3.54 3.59 3.02

% CH1 transamidation % LC transamidation Ave DAR DAR 1 DAR 1 S136K-HC/S202K-LC 85.5% 72.5% 3.16 S136K-N297K-HC/S202K-LC 88.1% 59.7% 2.95 S136K-N297K-P445K-HC/S202K-LC 84.5% 56.0% 2.80 Table 5. Antibodies with multiple lysine substitutions were analyzed as above. The masses of T135K-HC antibodies were analyzed by LC-MS as in Figure 5. The S136K-HC antibodies were analyzed by ESI-MS as in Figure 1. The signal from the Fc fragments for N297K-HC was too low to analyze.

Samples containing S136K-HC were incubated with MTG and ZQG-biotin and their masses were analyzed by ESI-MS following IdeS digestion and reduction. The CH1 and light chain of all samples were transamidated (Table 5). S136K-HC was 84.5% to 88.1% transamidated for all samples. S202K transamidation was higher in the single S136K-HC versus the double and triple HC substitutions (72.5% versus 56% or 59.7%). The DAR for S136K-HC/S202K-LC was 3.16 out of 4 potential sites. The ESIMS signal for the Fc fragments containing the N297K-HC mutation was very low and the conjugation efficiency could not be determined. Therefore, the DARs for the double- and triple-HC substitutions were at least 2.95 and 2.8, respectively. Transamidation of multiple sites by other acyl donor substrates was also examined. The T135K-HCbased mutants were incubated with ZQG-N3, ZQG-PEG2-BCN, or ZQG-PEG2-AuF and MTG and their masses were analyzed by LC-MS. Transamidation of L201K-LC and S202K-LC by each of the substrates was similar for all multiple-substitution antibodies and the single L201K-LC and S202K-LC modifications shown in Table 3 with the exception of ZQG-PEG2-BCN (Table 6). The two mAbs containing a single T135K-HC modification plus a single LC modification demonstrated that the transamidation efficiency of T135K-HC by acyl donor substrates was similar to the single T135K-HC modification in Table 3. Adding the second HC-L448 modification resulted in mixtures of samples with DAR 1 and DAR 2 on the HC. The samples transamidated most efficiently were T135K-HC/L201KLC+ZQG-N3 (DAR 3.57 out of 4), T135K-L448-HC+ZQG-N3 (DAR 3.1 out of 4), and T135K-L448HC/L201K-LC+ZQG-N3 (DAR 5.04 out of 6).

ACS Paragon Plus Environment

Page 13 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Bioconjugate Chemistry

T135KHC/L201KLC T135KHC/S202KLC T135K/L448HC T135K/L448HC/L201KLC T135K/L448HC/S202KLC

% HC transamidation DAR 1 DAR 2 78.6% 0.0% 74.9% 0.0% 41.6% 56.7% 33.2% 59.3% 39.8% 53.2%

Z-Gln-Gly-N3 % LC transamidation DAR 1 100.0% 70.3% 0.0% 100.0% 75.1%

Ave DAR 3.57 2.90 3.10 5.04 4.43

T135KHC/L201KLC T135KHC/S202KLC T135K/L448HC T135K/L448HC/L201KLC T135K/L448HC/S202KLC

Z-Gln-Gly-PEG2-BCN % HC transamidation % LC transamidation DAR 1 DAR 2 DAR 1 ND ND 0.0% 41.9% 0.0% 0.0% ND ND 0.0% ND ND 0.0% 42.2% 19.9% 0.0%

Ave DAR 0.91* 0.84 ND 0.80* 1.64

Z-Gln-Gly-PEG2-AuF % HC transamidation % LC transamidation Ave DAR 1 DAR 2 DAR 1 DAR T135KHC/L201KLC 58.6% 0.0% 94.5% 3.06 T135KHC/S202KLC 38.4% 0.0% 44.6% 1.66 T135K/L448HC 57.1% 24.7% 0.0% 2.13 T135K/L448HC/L201KLC 61.2% 13.4% 90.8% 3.57 T135K/L448HC/S202KLC 0.5% 0.0% 34.1% 1.76 Table 6. Transamidation of multiple acyl acceptor sites with various acyl donors was analyzed as in Figure 5.

Lysine substitutions in other IgG isotypes were efficiently transamidated The Fc of IgG1, IgG2, IgG3, and IgG4 are 89.2% identical for all human isotypes (Figure 6A). Therefore, it was possible that acyl acceptor sites identified above for IgG1 could also be acyl acceptor sites in IgG2, IgG3, and IgG4. Transamidation of wild-type mAb1-IgG2, IgG3, and IgG4 by MTG and ZQG-biotin was first analyzed by ESI-MS. As with IgG1, no native lysines in IgG2, IgG3, or IgG4 that were efficient acyl acceptor sites for MTG transamidation (Figure 6B).

ACS Paragon Plus Environment

Bioconjugate Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

A. hu hu hu hu

IgG1 IgG2 IgG3 IgG4

X X GGPSVFLFPPKPKDTLMISRTPEVTCVVVDVSHEDPEVKFNWYVDGVEVHNAKTKPREEQYNSTYRVVSVLTVLHQDWLNGKEYKCKVSNKALPAPIEKT AGPSVFLFPPKPKDTLMISRTPEVTCVVVDVSHEDPEVQFNWYVDGVEVHNAKTKPREEQFNSTFRVVSVLTVVHQDWLNGKEYKCKVSNKGLPAPIEKT GGPSVFLFPPKPKDTLMISRTPEVTCVVVDVSHEDPEVQFKWYVDGVEVHNAKTKPREEQYNSTFRVVSVLTVLHQDWLNGKEYKCKVSNKALPAPIEKT GGPSVFLFPPKPKDTLMISRTPEVTCVVVDVSQEDPEVQFNWYVDGVEVHNAKTKPREEQFNSTYRVVSVLTVLHQDWLNGKEYKCKVSNKGLPSSIEKT

hu hu hu hu

IgG1 IgG2 IgG3 IgG4

ISKAKGQPREPQVYTLPPSRDELTKNQVSLTCLVKGFYPSDIAVEWESNGQPENNYKTTPPVLDSDGSFFLYSKLTVDKSRWQQGNVFSCSVMHEALHNH ISKTKGQPREPQVYTLPPSREEMTKNQVSLTCLVKGFYPSDIAVEWESNGQPENNYKTTPPMLDSDGSFFLYSKLTVDKSRWQQGNVFSCSVMHEALHNH ISKTKGQPREPQVYTLPPSREEMTKNQVSLTCLVKGFYPSDIAVEWESSGQPENNYNTTPPMLDSDGSFFLYSKLTVDKSRWQQGNIFSCSVMHEALHNR ISKAKGQPREPQVYTLPPSQEEMTKNQVSLTCLVKGFYPSDIAVEWESNGQPENNYKTTPPVLDSDGSFFLYSRLTVDKSRWQEGNVFSCSVMHEALHNH

hu hu hu hu

IgG1 IgG2 IgG3 IgG4

X YTQKSLSLSPGK YTQKSLSLSPGK FTQKSLSLSPGK YTQKSLSLSLGK

B.

Figure 6. (A) The primary amino acid sequences of human IgG1, 2, 3, and 4 Fc are aligned and differences are highlighted. IgG1 residues Met252, Asn297, and Pro445 are indicated by an X. (B) Transamidation of antibodies was analyzed as in Figure 1.

IgG2, IgG3, and IgG4 versions of mAb1 were made with or without Leu448 or Asp448 and analyzed for transamidation with ZQG-biotin and MTG by ESI-MS. As with IgG1, the addition of a C-terminal Leu448 facilitated efficient transamidation of the antibodies while Asp448 did not (Figure 6B). Lysine substitutions were made at positions analogous to IgG1 (M252, N297, and P445). Except for IgG4 that encodes for a leucine at position 445, there are no differences at these residues between the isotypes (Figure 6A). N297K and M252K were both found to be efficiently transamidated with the N297K mutants yielding more than one conjugation site per HC as with IgG1. IgG2-N297K and IgG4-N297K contained 2 acyl acceptor sites while IgG3 contained 3. P445K was only efficiently transamidated in the IgG2 isotype. P445K transamidation was only 62.6% and 50.6 in IgG3 and IgG4, respectively. Lysine mutagenesis of a Fab' fragment demonstrated acyl acceptor sites not present in analogous fulllength mutant antibodies The hinge of IgG is a flexible linker between CH1 and CH2,52 and it is, therefore, possible that this flexibility could allow transamidation of acyl acceptors in this region. Indeed, upper hinge mutants D221K, T223K, H224K, and T225K were efficiently transamidated; however, there was no transamidation of middle or lower hinge mutants. This lack of transamidation may be due to structural constraints of the interchain disulfide bonds and proximity to CH2, respectively. To determine whether these residues can be transamidated when relieved of any structural constraints, lysine substitutions were

ACS Paragon Plus Environment

Page 14 of 27

Page 15 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Bioconjugate Chemistry

made in the context of a mAb1 Fab' containing the entire hinge region but with Cys226 and Cys229 mutated to alanine (DTHTAPPAPAPELL). The transamination of Fab' fragments was determined by ESI-MS as above using ZQG-biotin as the acyl donor. The masses of the HC portion of the mAb1-Fab' fragments were as expected except for L235K. This modification results in a C-terminal lysine residue that was cleaved, similar to full-length IgG.40-42 While no transamidation of full-length mAb1 IgG was seen in Figure 1, a minor amount of transamidation (4.7%) was seen in the wild-type mAb1 Fab', possibly at Lys222 (Table 7). Scanning lysine mutagenesis of the Fab' revealed several acyl acceptor sites at positions D221, T223, H224, T225, P230, and E233. These Fab' fragments were transamidated with 1 or 2 biotins, one on the engineered lysine and a second on a native lysine. % transamidation +1 biotin +2 biotins Total mAb1-Fab DKTHTAPPAPAPELL 4.7% 4.7% D221A AKTHTAPPAPAPELL 65.9% 65.9% KKTHTAPPAPAPELL 89.6% 89.6% D221K T223K DKKHTAPPAPAPELL 51.1% 34.2% 85.3% T223K-H DKKH 56.6% 25.9% 82.5% 87.7% 2.6% 90.2% H224K DKTKTAPPAPAPELL H224K-T DKTKT 83.6% 5.4% 89.0% T225K DKTHKAPPAPAPELL 69.6% 12.2% 81.8% T225K-A DKTHKA 62.6% 14.2% 76.8% 52.2% 52.2% C226K DKTHTKPPAPAPELL C226K,P228A DKTHTKAPAPAPELL 60.1% 11.2% 71.3% P227K DKTHTAKPAPAPELL 56.8% 56.8% P227K,P228A DKTHTAKAAPAPELL 62.6% 17.9% 80.5% P228K DKTHTAPKAPAPELL 16.5% 16.5% P228K-A DKTHTAPKA 65.8% 28.7% 94.4% C229K DKTHTAPPKPAPELL 32.5% 32.5% C229K,P230A DKTHTAPPKAAPELL 77.3% 13.3% 90.6% P230K DKTHTAPPAKAPELL 63.9% 14.8% 78.7% P230K-A DKTHTAPPAKA 75.7% 17.1% 92.8% A231K DKTHTAPPAPKPELL 22.0% 22.0% A231K,P232A DKTHTAPPAPKAELL 79.3% 13.6% 92.9% P232K DKTHTAPPAPAKELL 40.6% 5.5% 46.2% P232K,E233A DKTHTAPPAPAKALL 85.0% 7.7% 92.6% E233K DKTHTAPPAPAPKLL 51.5% 51.5% E233K-L DKTHTAPPAPAPKL 82.4% 9.0% 91.4% 53.2% 8.5% 61.7% L234K DKTHTAPPAPAPEKL E233A,L234K DKTHTAPPAPAPAKL 75.4% 10.9% 86.3% L235K DKTHTAPPAPAPELK 7.8% 7.8% Table 7. Purified Fab' fragments were screened for transamidation as in Figure 1. Amino acid substitutions are shown in red.

As secondary structure plays a large part in determining whether a lysine is transamidated and the secondary structure of the hinge between the Fab' mutants is unlikely to change significantly, it was unexpected that the native Lys222 and mutants C226K, P227K, C229K, A231K, P232K, L234K, and L235K were not transamidated. The primary sequences surrounding the lysine mutations were analyzed. As demonstrated above, the C-terminal (+1) residue to an acyl acceptor effects transamidation; specifically, a +1 acidic or proline residue results in little to no transamidation. The mutants C226K, P227K, C229K, and A231K all have a +1 proline and P232K has a +1 glutamate. In addition, Lys222 and L234K have a -1 acidic residue. Based on these findings it is possible that any acidic residue flanking an acyl acceptor may inhibit transamidation. To test this hypothesis, alanine substitutions were made to modify either the -1 or +1 acidic residue or +1 proline. Indeed, modifying the acidic and proline residues resulted in efficient transamidation of the lysine substitutions (Table 7). The D221A substitution increased transamidation from 5% to 66%. Similarly, modifying the -1 or +1 acid residues in the P232K,E233A and E233A,L234K Fab' fragments resulted in increased transamidation from 46% to 93%

ACS Paragon Plus Environment

Bioconjugate Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

and 62% to 86%, respectively. Modification of a +1 proline had similar results for C226K (52% to 71%), P227K (57% to 81%), C229K (33% to 91%), and A231K (22% to 93%). The effect of the hinge length on transamidation of T223K, H224K, T225K, P228K, P230K, and E233K was analyzed by deleting all but the adjacent +1 residue. Removing the C-terminal hinge residues had no negative effect on transamidation of the engineered acyl acceptor site for T223K, H224K, and T225K, and increased transamidation of P228K, P230K, and E233K (Table 7). DISCUSSION Here we describe an enzymatic-assisted, site-specific conjugation method utilizing a single point mutation or single C-terminal amino acid extension on an otherwise wild-type antibody (e.g., fully glycosylated). Due to MTG’s reported lower contextual specificity for acyl acceptor sites, it was speculated that multiple lysines on the surface of an IgG would be transamidated by an acyl donor substrate.14, 16 However, we observed no transamidation of a wild-type IgG, and by taking advantage of this lower specificity we were able to engineer single efficient acyl acceptor sites in solvent-exposed loops throughout an antibody. This method offers a significant advantage over other enzymatic conjugation technologies described to date that require either insertion of a four to six amino acid peptide or multiple point mutations to generate a single conjugation site. Favorable acyl acceptor sites generated via lysine point mutations were found in each constant region of the antibody (Figure 7). The diversity of engineered acyl acceptor sites allows flexibility when designing ADCs. Not only can these sites be paired to generate DARs greater than 2, but can be integrated into antibody fragments. We demonstrated this with Fab' fragments, but alternative scaffolds derived from CH2 or CH3 fragments could utilize point mutants M252K, N297K, or P447K. Furthermore, the addition of a C-terminal lysine followed by a non-basic, -acidic, -proline residue could also be added to an antibody fragment or alternative scaffold.

Figure 7. The most efficient transamidation sites are highlighted in the 1HZH crystal structure of human IgG1.

While we found some sequence requirements for optimal transamidation (no adjacent acidic or proline residues), no clear consensus sequence emerged to predict the most optimal sites to engineer an acyl acceptor site. The specificity of MTG for antibody lysines was consistently limited by flanking acidic or proline residues. An acidic or proline residue either C-terminal to Lys447 or flanking an engineered lysine in a Fab' fragment hindered the efficiency of transamidation. Similarly, a recent analysis of tripeptide acyl acceptor substrates for MTG also demonstrated a lack of transamidation in the presence of

ACS Paragon Plus Environment

Page 16 of 27

Page 17 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Bioconjugate Chemistry

a flanking acidic residue. However, in the context of a tripeptide, proline residues had no negative impact on transamidation.53 In addition to engineering in an acyl acceptor site, we found that acyl acceptor sites could be created by engineering acidic or proline residues flanking an existing lysine. However, solventexposed lysines in loops not adjacent to proline or acid residues are not the only requirements for efficient transamidation. For example, while we observed 100% transamidation of Lys447 with a serine, glycine, or alanine at the +1 position, identical PGKS, PGKG, and PGKA sequences in solvent exposed FWRH2 or FWRL2 loops of antibodies tested in Figure 1 were not transamidated. A second example is seen in an S-tag peptide whereby a +1 glutamate did not allow transamidation of Lys447 and a lysine in an S-tag peptide was transamidated despite the presence of a +1 glutamate.54 Our results suggest that the structure surrounding a putative MTG substrate residue likely influences its transamidation. We show that by relieving the core and lower hinge region of any structure contributed by CH2 and/or the interchain disulfide bonds through the generation of Fab' fragments, lysines that were not transamidated in the context of an antibody were able to be transamidated in a Fab' fragment. In the context of a full-length antibody with interchain disulfide bonds at cysteines 226 and 229, mutants P230K and E233K were not transamidated. However, in the context of a Fab' fragment and no interchain disulfide bonds, these residues were transamidated. A disulfide bond flanking an acyl acceptor site itself does not necessarily hinder transamidation, as the +1 position in the lambda mutant E213K is an interchain disulfide bond, and the mutant was transamidated. Therefore, either changing the threedimensional structure of the hinge region by removing the interchain disulfide bonds and/or removing possible steric constraints of the nearby CH2 region allowed transamidation of lysines at these sites. Increased efficiency of MTG transamidation was found by pairing different acyl donor substrates with various acyl acceptor sites. While acyl acceptor sites were efficiently conjugated to ZQG-biotin, not all sites were equally conjugated with ZQG-N3, ZQG-PEGn-BCN, or ZQG-PEG2-AuF. In fact, some sites (e.g., M252K) were inefficiently conjugated to ZQG-PEGn-BCN but conjugated well to ZQG-PEG2-AuF while with some sites (e.g., L193K) the preference for acyl donor substrates had an opposite effect. Of all acyl acceptor site mutants identified, HC-Leu448 and LC-L201K were the most efficiently conjugated by multiple acyl donor substrates. In a recent report, Lys447 was deleted to prevent antibody-antibody crosslinking during MTG-mediated conjugation on a C-terminal glutamine-based acyl donor tag.19 The authors concluded that for MTG-mediated antibody conjugation it is desirable to genetically remove Lys447. In direct contrast, our findings demonstrated the effective use of this native lysine as an acyl acceptor for MTG transamidation after protecting its cleavage by the addition of a selected C-terminus residue. MATERIALS AND METHODS Site-directed mutagenesis Mutations were generated using Stratagene's QuikChange XL according to the manufacturer’s protocol. The desired mutations were confirmed by DNA sequencing. Deletion mutagenesis HC fragments for Fab' expression were made by PCR amplifying the HC leader sequence through the hinge that terminated at various 3' codons. The PCR fragments were cloned into a pcDNA3.1-based mammalian expression plasmid using an In-Fusion HD cloning kit according to the manufacturer’s protocol (Clontech). Cell Culture

ACS Paragon Plus Environment

Bioconjugate Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Transfection and stable cell line generation HEK293 cells were transfected using one of two methods. Transient transfections and stable cell lines of 293F cells were performed as previously reported.55 Alternatively, Expi293 cells were transfected using ExpiFectamine according to the manufacturer’s protocol (ThermoFisher). All cells were maintained in a humidified incubator at 37°C in 8% CO2 with shaking at 125 rpm. Antibody and Fab' production Antibody production from stable pools was performed by seeding cells at 0.6 to 1x106 cells/mL in 293 FreeStyle medium. Cells were incubated at 37°C, 8% CO2, shaking at 125 rpm. After the culture reached a density of 4 to 6 x106 cells/mL, cultures were fed as previously reported.55 Cultures were incubated for an additional 7 to 10 days. When the cell viability was less than 50%, cultures were centrifuged for 1 h at 8000 rpm in a Beckman JLA8.1000 rotor. Supernatants were then filtered through a 0.2 µm PES filter and stored at 4°C or -20°C until purification. Antibody and Fab’ purification Antibodies were purified using one of two methods. For antibody and Fab' supernatants less than 10 mL, affinity chromatography was performed using a batch purification method with protein A resin or antikappa resin, respectively. Antibody and Fab' supernatants greater than 25 mL were purified using prepacked protein A or anti-kappa columns, respectively. Batch purification Prosep-vA High Capacity Protein A resin (Millipore) or CaptureSelect™ KappaSelect LC-kappa resin (ThermoFisher) was equilibrated with DPBS, and 100 µL were added to 3 to 6 mL of sample. Following incubation at 4°C for 1 hour to overnight, the resin was washed three times with 1 mL DPBS and centrifuged at 15,000 x g for 30 s. The sample was eluted from the resin by addition of 400 µL 0.1 M Glycine, pH 2.9 followed by centrifugation at 15,000 x g for 30 s. The sample was neutralized with 40 µL of 1 M Tris, pH 8.0. The buffer was exchanged using 0.5 mL Amicon Ultra, 10k cutoff filters (Millipore) by concentrating the sample to ~ 100 µL by centrifugation at 15,000 x g for 3 to 5 minutes. The concentrated sample was diluted in 400 µL DPBS, followed by centrifugation. The process was repeated a total of four times. Column purification A protein A or HiTrap KappaSelect column (GE Healthcare) was equilibrated with 10 column volumes (CV) of 20 mM sodium phosphate, 10 mM EDTA, pH 7.2. The sample was then loaded, followed by washing unbound material with 10 CV of equilibration buffer. The sample was eluted using 5 CV of 0.1 M Glycine, pH 2.9. The fractions containing the antibody were pooled and dialyzed in DPBS using an MWCO 20K Slide-A-Lyzer (ThermoFisher). ZQG substrates ZQG was purchased from Bachem, and ZQG-biotin was purchased from Zedira (Figure 8).

ACS Paragon Plus Environment

Page 18 of 27

Page 19 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Bioconjugate Chemistry

ZQG O

NH2

O O

N H

H N

O OH

O

ZQG-biotin O

NH2

O O

H N

N H

O

O S N H

O

N H HN

NH O

ZQG-N3 O

NH2

O O

H N

N H

O N H

O

N

N+

N-

ZQG-PEGn-BCN O

NH2

O O

H N

N H

O

O N H

O

O n

N H

H O H

N O

ZQG-PEG2-AuF O

NH NH2 N

O O

N H

H N O

O

O N H

H N

O

O

O

H N

O

O

O O

O

N H

N

Figure 8. Structures of ZQG acyl-donor substrates.

ZQG-pentafluorophenyl ester (ZQG-PFP) The synthesis was from Pasternack et al,56 with modifications. ZQG-OH (328.8 mg, 0.975 mmol) and pentafluorophenol (Sigma, 183.3 mg, 0.996 mmol) were dissolved in 10 mL N,N′-dimethylformamide (DMF). EDAC-HCl (Sigma, 201 mg, 1.04 mmol) was then added and the reaction was incubated at room temperature under N2 for 2 hr. 100 mL of cold diethyl ether was added to the reaction and precipitated overnight at -80°C. The crude product was collected by centrifugation and re-crystallized from 20 mL 60°C methanol. The final product was rinsed with cold diethyl ether and dried over a stream of N2. Final yield was 219.04 mg (44.7%). Electrospray ionization-mass spectrometry (ESI-MS) (direct infusion in 50% acetonitrile in 0.1% formic acid) m/z 504.0 ([M+H], 86%), 526.0 ([M+Na], 100%), 542.0 ([M+K], 22%). ZQG-propyl azide (ZQG-N3) ZQG-PFP (21.24 mg, 4.22 x 10-5 mol) and azidopropylamine (Click Chemistry Tools, 42.2 µL of a 0.91 M stock solution in DMF, 3.84 x 10-5 mol) were dissolved in 0.42 mL final volume of DMF. The reaction was stirred under N2 overnight at room temperature. The product was purified by HPLC using a 0.1% formic acid in H2O/0.1% formic acid in acetonitrile mobile phase. The product was dried in vacuo. Final yield was 10.7 mg (60.4%). ESI-MS (gradient purification) m/z 420.2 ([M+H], 100%), 442.1 ([M+Na], 32%). ZQG-PEGn-bicyclononyne (ZQG-PEGn-BCN) ZQG-PFP (18.4 mg, 3.66 x 10-5 mol) and N-[(1R,8S,9S)-Bicyclo[6.1.0]non-4-yn-9ylmethyloxycarbonyl]-1,8-diamino-3,6-dioxaoctane (Sigma Aldrich) or BCN-PEG3-Amine (Conju-

ACS Paragon Plus Environment

Bioconjugate Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 27

Probe) were dissolved in 0.37 mL final volume of DMF. Reaction was stirred under N2 overnight at room temperature. Product was purified by HPLC using a 0.1% formic acid in H2O/0.1% formic acid in acetonitrile mobile phase. Product was dried in vacuo. Final yield was 0.6 mg (2%). ESI-MS (gradient purification) m/z 688.2 ([M+H], 100%), 710.2 ([M+Na], 69%). ZQG-PEG2-Auristatin F (ZQG-PEG2-AuF) ZQG-PFP (22.2 mg, 4.37 x 10-5 mol) was dissolved in 0.85 mL DMF and 1,2-ethylenediamine (2.3 x 10-5 L, 3.5 x 10-4 mol) was added and mixed. The reaction was stirred under N2 overnight at room temperature. The product was purified by HPLC using a 0.1% formic acid in H2O/0.1% formic acid in acetonitrile mobile phase. The product was dried in vacuo. The final yield of ZQG-NH2 was 3.8 mg (23%). ESI-MS (gradient purification) m/z 380.1 ([M+H], 100%). ZQG-NH2 (3.8 mg, 1.01 x 10-5 mol) and NHS-PEG2-AuF (10.3 mg, 1.03 x 10-5 mol, Concortis Biosystems) were dissolved in 0.2 mL DMF. Triethylamine (14 µL, 1 x 10-4 mol) was added and the reaction was incubated under N2 overnight at room temperature. Half of the reaction was purified by HPLC using a 0.1% formic acid in H2O/0.1% formic acid in acetonitrile mobile phase. The product was dried in vacuo. The final yield of CBZQGPEG2-AuF was 3.8 mg (60%). ESI-MS (gradient purification) m/z 634.0 ([M+H]2+,100%), 645.1([M+Na]2+,45%). 1267.0 ([M+H], 16%). MTG reaction Antibodies ranging in concentrations from 100 µg/mL to 2.5 mg/mL were incubated with 785 µM ZQGbiotin (Zedira), ZQG-N3, ZQG-BCN, or ZQG-PEG2-AuF with 1 U/mL MTG (Zedira) in DPBS for at least 16 h at 37°C. High-throughput MTG assay Antibodies ranging in concentrations from 500 ng/mL to 10 µg/mL were incubated with 60 µM ZQGbiotin and 0.1 U/mL MTG (Zedira) in DPBS for at least 16 h at 37°C. A 96-well microtiter plate was coated with 1 µg/mL goat-anti-human IgG Fcγ antibody (Jackson ImmunoResearch) overnight at 4°C. After washing the plate, the overnight MTG reactions were diluted 1:10 in 50 µL DPBS, added to the plate, and incubated for 1 h at 22°C. The plate was then washed, and 0.1 µg/mL of streptavidinhorseradish peroxidase (HRP) (Jackson ImmunoResearch) was added to the wells. The plate was washed again, and streptavidin-HRP-bound biotinylated samples were quantitated in relative fluorescent units (RFUs) using QuantaBlue substrate (ThermoFisher) according to the manufacturer’s protocol. Ultra-performance liquid chromatography (UPLC)/ESI-MS analysis of antibody conjugation Purified antibodies were diluted to 1 mg/mL in DPBS (if below 1.0 mg/mL samples were left at original concentration). Reactions containing dimethylsulfoxide were desalted using a Zeba spin desalting column. The antibodies were then either deglycosylated using PNGase F (New England Biolabs) or digested into Fab'2 and Fc fragments by IdeS (Promega). To deglycosylate the mAbs, G7 buffer (5 or 10 µL) and PNGase F (1 or 2 µL) were added to the antibody (50 or 100 µL). The reaction was incubated in a Discover microwave (CEM) for 2 cycles: 1.) microwave power 10 W, 37°C, 10 min, and then wait for 3-5 min; 2.) microwave power 2 W, 37°C, 10 min. A portion of the deglycosylated sample was reduced by adding DTT to a final concentration of 20 mM, followed by incubation at 60°C for 3 min. To generate Fab'2 and Fc fragments, 50U/µL of IdeS was added to 0.5 mg/mL of antibody and incubated at 37°C for 0.5-1 h. The IdeS samples were not reduced except for mAb1-Cys448 which was reduced as above. Samples were then analyzed using a Waters Acquity UPLC and Q-ToF Premier mass spectrometer. Samples (0.5-2 µg each) were injected onto a MassPrep micro desalting column at 65°C, eluted from the

ACS Paragon Plus Environment

Page 21 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Bioconjugate Chemistry

column with a 5 min equilibration in 95% of mobile phase A, a 10 min gradient (5-90% B), and a 10 min re-equilibration in 95% of mobile phase A, at 0.05 mL/min. Mobile phase A was 0.1% formic acid in water. Mobile phase B was 0.1% formic acid in acetonitrile. The Q-ToF mass spectrometer was run in positive ion, V-mode with detection in the range of 500-4000 m/z. The source parameters were as follows: capillary voltage, 2.25 kV (intact antibody)-2.50 kV (reduced antibody); sampling cone voltage, 65.0 V (intact antibody) or 50.0 V (reduced antibody); source temperature, 100°C; desolvation temperature, 250°C; desolvation gas flow, 550 L/hr. The protein peak was deconvoluted using the MassLynx MaxEnt 1 function. The percent transamidation was determined by dividing the peak intensity of the transamidated mAb by the sum of the peak intensities of the transamidated and non-transamidated mAb. LC-MS Samples analyzed using reverse phase liquid chromatography (100 µL at 1-2 mg/mL) were reduced with 20 mM DTT at 60°C for 3 minutes. The samples were analyzed using Waters Alliance HPLC with SQD and PDA detectors. The sample was injected onto a Proteomix RP-1000 column (5µ, 4.6 mm diameter or 50-150 mm length, Sepax) at 65°C. Separation of the LC and HC on the 50mm column occurred with a 3.0 min equilibration in 75% of mobile phase A (0.1% TFA in water) and a 19-minute gradient (25-60% mobile phase B [0.1% TFA in water]) at a flow rate of 1 mL/min; separation of the LC and HC on the 150 mm column occurred with a 3.0 min equilibration in 75% mobile phase A, and a 27 minute gradient (2555% mobile phase B) at a flow rate of 1 mL/min. The SQD mass spectrometer was run in positive ion, V-mode with detection in the range of 200-2000 m/z. Source parameters were as follows: capillary voltage, 3.00-3.20 kV; sampling cone voltage, 4045°C; source temperature, 120-150°C; desolvation temperature, 250-350°C; desolvation gas flow, 700800 L/hr. Scan time, 1 second. The protein peak was deconvoluted by the MassLynx MaxEnt 1 function. The PDA detector was at 280 nm. The DAR was calculated based on the relative signal intensity of the unconjugated and conjugated LC and unconjugated and conjugated HC. Total DAR = (DAR LC+DAR HC) x 2.

The authors declare the following competing financial interest(s): All authors are current employees of Morphotek, Inc. SUPPORTING INFORMATION A table of RFU values for ELISA-based transamidation assay ORCID Jared Spidel: 0000-0001-8397-3091 Earl Albone: 0000-0002-7569-1147 J. Bradford Kline: 0000-0001-5429-2318 ABBREVIATIONS ADC antibody-drug conjugate AuF auristatin F BCN bicyclononyne DAR drug-to-antibody ratio ESI-MS electrospray ionization-mass spectrometry

ACS Paragon Plus Environment

Bioconjugate Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

HC heavy chain LC light chain LC-MS liquid chromatography-mass spectrometry MTG microbial transglutaminase mAb monoclonal antibody NNAA non-natural amino acid SPAAC strain-promoted alkyne-azide cycloaddition ZQG carboxybenzyl-glutamine-glycine

REFERENCES 1. Albone, E. F., Spidel, J. L., Cheng, X., Park, Y. C., Jacob, S., Milinichik, A. Z., Vaessen, B., Butler, J., Kline, J. B., and Grasso, L. (2017) Generation of therapeutic immunoconjugates via Residue-Specific Conjugation Technology (RESPECT) utilizing a native cysteine in the light chain framework of Oryctolagus cuniculus. Cancer Biol Ther. 18, 347-357. 2. Spidel, J. L., Albone, E. F., Cheng, X., Vaessen, B., Jacob, S., Milinichik, A. Z., Verdi, A., Kline, J. B., and Grasso, L. (2017) Engineering humanized antibody framework sequences for optimal sitespecific conjugation of cytotoxins. MAbs., 1-9. 3. Junutula, J. R., Raab, H., Clark, S., Bhakta, S., Leipold, D. D., Weir, S., Chen, Y., Simpson, M., Tsai, S. P., Dennis, M. S. et al. (2008) Site-specific conjugation of a cytotoxic drug to an antibody improves the therapeutic index. Nat Biotechnol. 26, 925-932. 4. Chen, X. N., Nguyen, M., Jacobson, F., and Ouyang, J. (2009) Charge-based analysis of antibodies with engineered cysteines: from multiple peaks to a single main peak. MAbs. 1, 563-571. 5. Gomez, N., Ouyang, J., Nguyen, M. D., Vinson, A. R., Lin, A. A., and Yuk, I. H. (2010) Effect of temperature, pH, dissolved oxygen, and hydrolysate on the formation of triple light chain antibodies in cell culture. Biotechnol. Prog. 26, 1438-1445. 6. Hallam, T. J. and Smider, V. V. (2014) Unnatural amino acids in novel antibody conjugates. Future. Med Chem 6, 1309-1324. 7. Tian, F., Lu, Y., Manibusan, A., Sellers, A., Tran, H., Sun, Y., Phuong, T., Barnett, R., Hehli, B., Song, F. et al. (2014) A general approach to site-specific antibody drug conjugates. Proc Natl Acad Sci U S A 111, 1766-1771. 8. Hutchins, B. M., Kazane, S. A., Staflin, K., Forsyth, J. S., Felding-Habermann, B., Schultz, P. G., and Smider, V. V. (2011) Site-specific coupling and sterically controlled formation of multimeric antibody fab fragments with unnatural amino acids. J Mol Biol 406, 595-603. 9. Liu, C. C., Choe, H., Farzan, M., Smider, V. V., and Schultz, P. G. (2009) Mutagenesis and evolution of sulfated antibodies using an expanded genetic code. Biochemistry 48, 8891-8898. 10. Beerli, R. R., Hell, T., Merkel, A. S., and Grawunder, U. (2015) Sortase Enzyme-Mediated Generation of Site-Specifically Conjugated Antibody Drug Conjugates with High In Vitro and In Vivo Potency. PLoS One 10, e0131177.

ACS Paragon Plus Environment

Page 22 of 27

Page 23 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Bioconjugate Chemistry

11. Madej, M. P., Coia, G., Williams, C. C., Caine, J. M., Pearce, L. A., Attwood, R., Bartone, N. A., Dolezal, O., Nisbet, R. M., Nuttall, S. D. et al. (2012) Engineering of an anti-epidermal growth factor receptor antibody to single chain format and labeling by Sortase A-mediated protein ligation. Biotechnol Bioeng. 109, 1461-1470. 12. Swee, L. K., Guimaraes, C. P., Sehrawat, S., Spooner, E., Barrasa, M. I., and Ploegh, H. L. (2013) Sortase-mediated modification of alphaDEC205 affords optimization of antigen presentation and immunization against a set of viral epitopes. Proc Natl Acad Sci U S A 110, 1428-1433. 13. Drake, P. M., Albers, A. E., Baker, J., Banas, S., Barfield, R. M., Bhat, A. S., de Hart, G. W., Garofalo, A. W., Holder, P., Jones, L. C. et al. (2014) 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 25, 1331-1341. 14. Josten, A., Haalck, L., Spener, F., and Meusel, M. (2000) Use of microbial transglutaminase for the enzymatic biotinylation of antibodies. J Immunol. Methods 240, 47-54. 15. Mindt, T. L., Jungi, V., Wyss, S., Friedli, A., Pla, G., Novak-Hofer, I., Grunberg, J., and Schibli, R. (2008) Modification of different IgG1 antibodies via glutamine and lysine using bacterial and human tissue transglutaminase. Bioconjug. Chem 19, 271-278. 16. Jeger, S., Zimmermann, K., Blanc, A., Grunberg, J., Honer, M., Hunziker, P., Struthers, H., and Schibli, R. (2010) Site-specific and stoichiometric modification of antibodies by bacterial transglutaminase. Angew. Chem Int Ed Engl 49, 9995-9997. 17. Strop, P., Liu, S. H., Dorywalska, M., Delaria, K., Dushin, R. G., Tran, T. T., Ho, W. H., Farias, S., Casas, M. G., Abdiche, Y. et al. (2013) Location matters: site of conjugation modulates stability and pharmacokinetics of antibody drug conjugates. Chem Biol 20, 161-167. 18. Dennler, P., Chiotellis, A., Fischer, E., Bregeon, D., Belmant, C., Gauthier, L., Lhospice, F., Romagne, F., and Schibli, R. (2014) Transglutaminase-based chemo-enzymatic conjugation approach yields homogeneous antibody-drug conjugates. Bioconjug. Chem 25, 569-578. 19. Siegmund, V., Schmelz, S., Dickgiesser, S., Beck, J., Ebenig, A., Fittler, H., Frauendorf, H., Piater, B., Betz, U. A., Avrutina, O. et al. (2015) Locked by Design: A Conformationally Constrained Transglutaminase Tag Enables Efficient Site-Specific Conjugation. Angew. Chem Int Ed Engl. 20. Yokoyama, K., Nio, N., and Kikuchi, Y. (2004) Properties and applications of microbial transglutaminase. Appl. Microbiol. Biotechnol. 64, 447-454. 21. Strop, P. (2014) Versatility of microbial transglutaminase. Bioconjug. Chem 25, 855-862. 22. Kieliszek, M. and Misiewicz, A. (2014) Microbial transglutaminase and its application in the food industry. A review. Folia Microbiol. (Praha) 59, 241-250. 23. Mimura, Y., Church, S., Ghirlando, R., Ashton, P. R., Dong, S., Goodall, M., Lund, J., and Jefferis, R. (2000) The influence of glycosylation on the thermal stability and effector function expression of human IgG1-Fc: properties of a series of truncated glycoforms. Mol Immunol. 37, 697-706. 24. Kwon, K. S. and Yu, M. H. (1997) Effect of glycosylation on the stability of alpha1-antitrypsin toward urea denaturation and thermal deactivation. Biochim. Biophys. Acta 1335, 265-272.

ACS Paragon Plus Environment

Bioconjugate Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

25. Wang, W., Antonsen, K., Wang, Y. J., and Wang, D. Q. (2008) pH dependent effect of glycosylation on protein stability. Eur J Pharm Sci 33, 120-127. 26. Yamaguchi, Y., Nishimura, M., Nagano, M., Yagi, H., Sasakawa, H., Uchida, K., Shitara, K., and Kato, K. (2006) Glycoform-dependent conformational alteration of the Fc region of human immunoglobulin G1 as revealed by NMR spectroscopy. Biochim. Biophys. Acta 1760, 693-700. 27. Arnold, J. N., Wormald, M. R., Sim, R. B., Rudd, P. M., and Dwek, R. A. (2007) The impact of glycosylation on the biological function and structure of human immunoglobulins. Annu Rev Immunol. 25, 21-50. 28. Zheng, K., Bantog, C., and Bayer, R. (2011) The impact of glycosylation on monoclonal antibody conformation and stability. MAbs. 3, 568-576. 29. Ohtsuka, T., Sawa, A., Kawabata, R., Nio, N., and Motoki, M. (2000) Substrate specificities of microbial transglutaminase for primary amines. J Agric. Food Chem 48, 6230-6233. 30. Ohtsuka, T., Ota, M., Nio, N., and Motoki, M. (2000) Comparison of substrate specificities of transglutaminases using synthetic peptides as acyl donors. Biosci. Biotechnol. Biochem 64, 26082613. 31. Gundersen, M. T., Keillor, J. W., and Pelletier, J. N. (2014) Microbial transglutaminase displays broad acyl-acceptor substrate specificity. Appl. Microbiol. Biotechnol. 98, 219-230. 32. Gautier, V., Boumeester, A. J., Lossl, P., and Heck, A. J. (2015) Lysine conjugation properties in human IgGs studied by integrating high-resolution native mass spectrometry and bottom-up proteomics. Proteomics. 15, 2756-2765. 33. Ellison, J., Buxbaum, J., and Hood, L. (1981) Nucleotide sequence of a human immunoglobulin C gamma 4 gene. DNA 1, 11-18. 34. Ellison, J. and Hood, L. (1982) Linkage and sequence homology of two human immunoglobulin gamma heavy chain constant region genes. Proc Natl Acad Sci U S A 79, 1984-1988. 35. Ellison, J. W., Berson, B. J., and Hood, L. E. (1982) The nucleotide sequence of a human immunoglobulin C gamma1 gene. Nucleic Acids Res 10, 4071-4079. 36. Wang, A. C., Tung, E., and Fudenberg, H. H. (1980) The primary structure of a human IgG2 heavy chain: genetic, evolutionary, and functional implications. J Immunol. 125, 1048-1054. 37. Edelman, G. M., Cunningham, B. A., Gall, W. E., Gottlieb, P. D., Rutishauser, U., and Waxdal, M. J. (1969) The covalent structure of an entire gammaG immunoglobulin molecule. Proc Natl Acad Sci U S A 63, 78-85. 38. Frangione, B., Rosenwasser, E., Prelli, F., and Franklin, E. C. (1980) Primary structure of human gamma 3 immunoglobulin deletion mutant: gamma 3 heavy-chain disease protein Wis. Biochemistry 19, 4304-4308. 39. Pink, J. R., Buttery, S. H., De Vries, G. M., and Milstein, C. (1970) Human immunoglobulin subclasses. Partial amino acid sequence of the constant region of a gamma 4 chain. Biochem J 117, 33-47.

ACS Paragon Plus Environment

Page 24 of 27

Page 25 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Bioconjugate Chemistry

40. Harris, R. J., Wagner, K. L., and Spellman, M. W. (1990) Structural characterization of a recombinant CD4-IgG hybrid molecule. Eur J Biochem 194, 611-620. 41. Harris, R. J. (1995) Processing of C-terminal lysine and arginine residues of proteins isolated from mammalian cell culture. J Chromatogr. A 705, 129-134. 42. Dick, L. W., Jr., Qiu, D., Mahon, D., Adamo, M., and Cheng, K. C. (2008) C-terminal lysine variants in fully human monoclonal antibodies: investigation of test methods and possible causes. Biotechnol. Bioeng. 100, 1132-1143. 43. van den Bremer, E. T., Beurskens, F. J., Voorhorst, M., Engelberts, P. J., de Jong, R. N., van der Boom, B. G., Cook, E. M., Lindorfer, M. A., Taylor, R. P., van Berkel, P. H. et al. (2015) Human IgG is produced in a pro-form that requires clipping of C-terminal lysines for maximal complement activation. MAbs. 7, 672-680. 44. Coussons, P. J., Price, N. C., Kelly, S. M., Smith, B., and Sawyer, L. (1992) Factors that govern the specificity of transglutaminase-catalysed modification of proteins and peptides. Biochem J 282 ( Pt 3), 929-930. 45. Spolaore, B., Raboni, S., Satwekar, A. A., Grigoletto, A., Mero, A., Montagner, I. M., Rosato, A., Pasut, G., and Fontana, A. (2016) Site-Specific Transglutaminase-Mediated Conjugation of Interferon alpha-2b at Glutamine or Lysine Residues. Bioconjug. Chem 27, 2695-2706. 46. Spolaore, B., Raboni, S., Ramos, M. A., Satwekar, A., Damiano, N., and Fontana, A. (2012) Local unfolding is required for the site-specific protein modification by transglutaminase. Biochemistry 51, 8679-8689. 47. Taguchi, S., Nishihama, K. I., Igi, K., Ito, K., Taira, H., Motoki, M., and Momose, H. (2000) Substrate specificity analysis of microbial transglutaminase using proteinaceous protease inhibitors as natural model substrates. J Biochem 128, 415-425. 48. Sugimura, Y., Yokoyama, K., Nio, N., Maki, M., and Hitomi, K. (2008) Identification of preferred substrate sequences of microbial transglutaminase from Streptomyces mobaraensis using a phagedisplayed peptide library. Arch Biochem Biophys. 477, 379-383. 49. Tagami, U., Shimba, N., Nakamura, M., Yokoyama, K., Suzuki, E., and Hirokawa, T. (2009) Substrate specificity of microbial transglutaminase as revealed by three-dimensional docking simulation and mutagenesis. Protein Eng Des Sel 22, 747-752. 50. Takazawa, T., Kamiya, N., Ueda, H., and Nagamune, T. (2004) Enzymatic labeling of a single chain variable fragment of an antibody with alkaline phosphatase by microbial transglutaminase. Biotechnol. Bioeng. 86, 399-404. 51. Tanaka, Y., Tsuruda, Y., Nishi, M., Kamiya, N., and Goto, M. (2007) Exploring enzymatic catalysis at a solid surface: a case study with transglutaminase-mediated protein immobilization. Org Biomol. Chem 5, 1764-1770. 52. Zhang, X., Tang, H., Sun, Y. T., Liu, X., Tan, W. S., and Fan, L. (2015) Elucidating the effects of arginine and lysine on a monoclonal antibody C-terminal lysine variation in CHO cell cultures. Appl. Microbiol. Biotechnol. 99, 6643-6652.

ACS Paragon Plus Environment

Bioconjugate Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

53. Malesevic, M., Migge, A., Hertel, T. C., and Pietzsch, M. (2015) A fluorescence-based array screen for transglutaminase substrates. Chembiochem. 16, 1169-1174. 54. Tanaka, T., Kamiya, N., and Nagamune, T. (2004) Peptidyl linkers for protein heterodimerization catalyzed by microbial transglutaminase. Bioconjug. Chem 15, 491-497. 55. Spidel, J. L., Vaessen, B., Chan, Y. Y., Grasso, L., and Kline, J. B. (2016) Rapid high-throughput cloning and stable expression of antibodies in HEK293 cells. J Immunol. Methods 439, 50-58. 56. Pasternack, R., Laurent, H. P., Ruth, T., Kaiser, A., Schon, N., and Fuchsbauer, H. L. (1997) A fluorescent substrate of transglutaminase for detection and characterization of glutamine acceptor compounds. Anal Biochem 249, 54-60.

ACS Paragon Plus Environment

Page 26 of 27

Page 27 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Bioconjugate Chemistry

TABLE OF CONTENTS GRAPHIC

ACS Paragon Plus Environment