THIOMAB antibody technology utilizes cysteine residues engineered onto an antibody to allow for site-specific conjugation. The technology has enabled the exploration of different attachment sites on the antibody in combination with small molecules, peptides, or proteins to yield antibody conjugates with unique properties. As reported previously ( Shen , B. Q. , et al. ( 2012 ) Nat. Biotechnol. 30 , 184 - 189 ; Pillow , T. H. , et al. ( 2017 ) Chem. Sci. 8 , 366 - 370 ), the specific location of the site of conjugation on an antibody can impact the stability of the linkage to the engineered cysteine for both thio-succinimide and disulfide bonds. High stability of the linkage is usually desired to maximize the delivery of the cargo to the intended target. In the current study, cysteines were individually substituted into every position of the anti-HER2 antibody (trastuzumab), and the stabilities of drug conjugations at those sites were evaluated. We screened a total of 648 THIOMAB antibody-drug conjugates, each generated from a trastuzamab prepared by sequentially mutating non-cysteine amino acids in the light and heavy chains to cysteine. Each THIOMAB antibody variant was conjugated to either maleimidocaproyl-valine-citrulline-p-aminobenzyloxycarbonyl-monomethyl auristatin E (MC-vc-PAB-MMAE) or pyridyl disulfide monomethyl auristatin E (PDS-MMAE) using a high-throughput, on-bead conjugation and purification method. Greater than 50% of the THIOMAB antibody variants were successfully conjugated to both MMAE derivatives with a drug to antibody ratio (DAR) of >0.5 and <50% aggregation. The relative in vitro plasma stabilities for approximately 750 conjugates were assessed using enzyme-linked immunosorbent assays, and stable sites were confirmed with affinity-capture LC/MS-based detection methods. Highly stable conjugation sites for the two types of MMAE derivatives were identified on both the heavy and light chains. Although the stabilities of maleimide conjugates were shown to be greater than those of the disulfide conjugates, many sites were identified that were stable for both. Furthermore, in vitro stabilities of selected stable sites translated across different cytotoxic payloads and different target antibodies as well as to in vivo stability.

The reversible attachment of a small-molecule drug to a carrier for targeted delivery can improve pharmacokinetics and the therapeutic index. Previous studies have reported the delivery of molecules that contain primary and secondary amines via an amide or carbamate bond; however, the ability to employ tertiary-amine-containing bioactive molecules has been elusive. Here we describe a bioreversible linkage based on a quaternary ammonium that can be used to connect a broad array of tertiary and heteroaryl amines to a carrier protein. Using a concise, protecting-group-free synthesis we demonstrate the chemoselective modification of 12 complex molecules that contain a range of reactive functional groups. We also show the utility of this connection with both protease-cleavable and reductively cleavable antibody-drug conjugates that were effective and stable in vitro and in vivo. Studies with a tertiary-amine-containing antibiotic show that the resulting antibody-antibiotic conjugate provided appropriate stability and release characteristics and led to an unexpected improvement in activity over the conjugates previously connected via a carbamate.

Previous investigations on antibody-drug conjugate (ADC) stability have focused on drug release by linker-deconjugation due to the relatively stable payloads such as maytansines. Recent development of ADCs has been focused on exploring technologies to produce homogeneous ADCs and new classes of payloads to expand the mechanisms of action of the delivered drugs. Certain new ADC payloads could undergo metabolism in circulation while attached to antibodies and thus affect ADC stability, pharmacokinetics, and efficacy and toxicity profiles. Herein, we investigate payload stability specifically and seek general guidelines to address payload metabolism and therefore increase the overall ADC stability. Investigation was performed on various payloads with different functionalities (e.g., PNU-159682 analog, tubulysin, cryptophycin, and taxoid) using different conjugation sites (HC-A118C, LC-K149C, and HC-A140C) on THIOMAB antibodies. We were able to reduce metabolism and inactivation of a broad range of payloads of THIOMAB antibody-drug conjugates by employing optimal conjugation sites (LC-K149C and HC-A140C). Additionally, further payload stability was achieved by optimizing the linkers. Coupling relatively stable sites with optimized linkers provided optimal stability and reduction of payloads metabolism in circulation in vivo.

Antibody-drug conjugates (ADCs) have a significant impact toward the treatment of cancer, as evidenced by the clinical activity of the recently approved ADCs, brentuximab vedotin for Hodgkin lymphoma and ado-trastuzumab emtansine (trastuzumab-MCC-DM1) for metastatic HER2+ breast cancer. DM1 is an analog of the natural product maytansine, a microtubule inhibitor that by itself has limited clinical activity and high systemic toxicity. However, by conjugation of DM1 to trastuzumab, the safety was improved and clinical activity was demonstrated. Here, we report that through chemical modification of the linker-drug and antibody engineering, the therapeutic activity of trastuzumab maytansinoid ADCs can be further improved. These improvements include eliminating DM1 release in the plasma and increasing the drug load by engineering four cysteine residues into the antibody. The chemical synthesis of highly stable linker-drugs and the modification of cysteine residues of engineered site-specific antibodies resulted in a homogeneous ADC with increased therapeutic activity compared to the clinically approved ADC, trastuzumab-MCC-DM1.

Mesothelin (MSLN) is an attractive target for antibody-drug conjugate therapy because it is highly expressed in various epithelial cancers, with normal expression limited to nondividing mesothelia. We generated novel antimesothelin antibodies and conjugated an internalizing one (7D9) to the microtubuledisrupting drugs monomethyl auristatin E (MMAE) and MMAF, finding the most effective to be MMAE with a lysosomal protease-cleavable valine-citrulline linker. The humanized (h7D9.v3) version, aMSLN-MMAE, specifically targeted mesothelin-expressing cells and inhibited their proliferation with an IC 50 of 0.3 nmol/L. Because the antitumor activity of an antimesothelin immunotoxin (SS1P) in transfected mesothelin models did not translate to the clinic, we carefully selected in vivo efficacy models endogenously expressing clinically relevant levels of mesothelin, after scoring mesothelin levels in ovarian, pancreatic, and mesothelioma tumors by immunohistochemistry. We found that endogenous mesothelin in cancer cells is upregulated in vivo and identified two suitable xenograft models for each of these three indications. A single dose of aMSLN-MMAE profoundly inhibited or regressed tumor growth in a dosedependent manner in all six models, including two patient-derived tumor xenografts. The robust and durable efficacy of aMSLN-MMAE in preclinical models of ovarian, mesothelioma, and pancreatic cancers justifies the ongoing phase I clinical trial. Mol Cancer Ther; 13(11); 2630-40. Ó2014 AACR.

The incorporation of cysteines into antibodies by mutagenesis allows for the direct conjugation of small molecules to specific sites on the antibody via disulfide bonds. The stability of the disulfide bond linkage between the small molecule and the antibody is highly dependent on the location of the engineered cysteine in either the heavy chain (HC) or the light chain (LC) of the antibody. Here, we explore the basis for this site-dependent stability. We evaluated the in vivo efficacy and pharmacokinetics of five different cysteine mutants of trastuzumab conjugated to a pyrrolobenzodiazepine (PBD) via disulfide bonds. A significant correlation was observed between disulfide stability and efficacy for the conjugates. We hypothesized that the observed site-dependent stability of the disulfide-linked conjugates could be due to differences in the attachment site cysteine thiol pK. We measured the cysteine thiol pK using isothermal titration calorimetry (ITC) and found that the variants with the highest thiol pK (LC K149C and HC A140C) were found to yield the conjugates with the greatest in vivo stability. Guided by homology modeling, we identified several mutations adjacent to LC K149C that reduced the cysteine thiol pK and, thus, decreased the in vivo stability of the disulfide-linked PBD conjugated to LC K149C. We also present results suggesting that the high thiol pK of LC K149C is responsible for the sustained circulation stability of LC K149C TDCs utilizing a maleimide-based linker. Taken together, our results provide evidence that the site-dependent stability of cys-engineered antibody-drug conjugates may be explained by interactions between the engineered cysteine and the local protein environment that serves to modulate the side-chain thiol pK. The influence of cysteine thiol pK on stability and efficacy offers a new parameter for the optimization of ADCs that utilize cysteine engineering.

Results from the integrated model evaluated with two different TDCs highlight the impact of DAR and site of conjugation on pharmacokinetics and efficacy. The model can be used to guide future drug optimization and in-vivo studies.

Antibody-drug conjugates (ADCs) are target-specific anticancer agents consisting of cytotoxic drugs covalently linked to a monoclonal antibody. The number of ADCs in the clinic is growing, and therefore thorough characterization of the quantitative assays used to measure ADC concentrations in support of pharmacokinetic, efficacy, and safety studies is of increasing importance. Cytotoxic drugs such as the tubulin polymerization inhibiting auristatin, monomethyl auristatin E, have been conjugated to antibodies via cleavable linkers (MC-vc-PAB) through internal cysteines. This results in a heterogeneous mixture of antibody species with drug-to-antibody ratios (DAR) ranging from 0 to 8. In order to characterize the assays used to quantitate total MC-vc-PAB-MMAE ADCs (conjugated and unconjugated antibody), we used purified fractions with defined DARs from 6 therapeutic antibodies to evaluate different assay formats and reagents. Our investigations revealed that for quantitation of total antibody, including all unconjugated and conjugated antibody species, sandwich ELISA formats did not always allow for recovery of all purified DAR fractions (DAR 0-8) to within ±20% of the expected values at the reagent concentrations tested. In evaluating alternative approaches, we found that the recovery of DAR fractions with semihomogeneous assay (SHA) formats, in which sample, capture, and detection reagents are preincubated in solution, were less affected by the antibody's MMAE drug load as compared to traditional stepwise sandwich ELISAs. Thus, choosing the optimal assay format and reagents for total antibody assays is valuable for developing accurate quantitative assays.