Payload-binding Fab fragments increase the therapeutic index of MMAE antibody-drug conjugates

Abstract

Monomethyl auristatin E (MMAE) is a potent tubulin inhibitor that is used as the payload for four FDA-approved antibody drug conjugates (ADCs). Deconjugated MMAE readily diffuses into untargeted cells resulting in off-target toxicity. Here we report the development and evaluation of a humanized Fab fragment (ABC3315) that enhances the therapeutic selectivity of MMAE ADCs. ABC3315 increased the IC50 of MMAE against human cancer cell-lines by >500-fold with no impact on the cytotoxicity of MMAE ADCs, including polatuzumab vedotin and trastuzumab-vc-MMAE (TvcMMAE). Co-administration of ABC3315 did not reduce the efficacy of polatuzumab vedotin or TvcMMAE in xenograft tumor models. Co-administration of ABC3315 with 80 mg/kg TvcMMAE significantly (p<0.0001) increased the cumulative amount of MMAE that was excreted in urine 0-4-days after administration from 789.4 ±19.0 nanograms (TvcMMAE alone) to 2625±206.8 nanograms (for mice receiving TvcMMAE with co-administration of ABC3315). Mice receiving 80 mg/kg TvcMMAE and PBS exhibited a significant drop in white blood cell counts (p=0.025) and red blood cell counts (p=0.0083) in comparison to control mice. No significant differences, relative to control mice, were found for white blood cell counts (p=0.15) or for red blood cell counts (p=0.23) for mice treated with 80 mg/kg TvcMMAE and ABC3315. Co-administration of ABC3315 with 120 mg/kg polatuzumab vedotin significantly (p=0.045) decreased the percentage body weight loss at nadir for treated mice from 11.9 ± 7.0% to 4.1 ± 2.1%. Our results demonstrate that ABC3315, an anti-MMAE Fab fragment, decreases off-target toxicity while not decreasing anti-tumor efficacy, increasing the therapeutic window of MMAE ADCs.

Introduction

Increased understanding of tumor biology and immunology has fueled the development of highly targeted therapies for cancer, including antibody-drug conjugates (ADCs) (1–3). ADCs employ monoclonal antibodies (mAb) with specificity for tumor-associated antigens to increase the efficiency and selectivity of the delivery of cytotoxic agents (i.e., payload molecules) to cancer cells (4). Presently, 11 ADCs are approved for use in the US and more than 100 ADCs are in clinical trials (5–7). Although enthusiasm for ADCs remains quite high, ADC therapies have shown a high rate of failure in the clinic that is primarily attributed to their off-target toxicity (6,8–10), which limits tolerable doses below levels needed for tumor eradication for most patients (6,8,11,12).

ADC toxicity is, in many cases, highly correlated with the systemic exposure of released (“free”) payload molecules, which may efficiently diffuse into and kill healthy, non-cancerous cells (10,11,13–16). Free monomethyl auristatin E (MMAE) may be released from MMAE ADCs at off-target sites through extracellular linker catabolism (i.e. proteolysis, hydrolysis) or via cellular uptake of ADC by non-specific mechanisms with subsequent intracellular linker catabolism (Figure 1). Following release from the ADC linker, unconjugated MMAE is predominately eliminated from preclinical species and humans by biliary excretion, which proceeds following distribution of released MMAE into extracellular fluids (e.g., tissue interstitial fluid and plasma) and following uptake into the liver. As such, released MMAE transits through extracellular fluids prior to elimination. Owing to its lipophilic nature, free MMAE in plasma and interstitial fluids may diffuse into untargeted cells, leading to undesirable off-target toxicity. Review of the clinical pharmacology of MMAE ADCs has shown that plasma concentrations of free MMAE are between 1-10 nM for up to 2-weeks after ADC administration (17). Given that MMAE leads to growth inhibition and cytotoxicity at concentrations below 0.1 nM for many cell types, clinical doses of MMAE ADCs yield plasma concentrations of free MMAE that may be expected to lead to substantial off-target toxicity.

Figure 1: Selective antagonism of MMAE toxicity.

Top left: ADC is internalized into a targeted cell by receptor mediated endocytosis and the drug linker is catabolized within the lysosomal space resulting in the release of MMAE. Free MMAE can diffuse out of targeted cells, or cells that non-specifically degrade ADC, and may diffuse into untargeted healthy cells resulting in off-target toxicities. Co-administration of ABC3315, which is specific for free (i.e., unconjugated) MMAE, has the potential to antagonize the off-target toxicities (peripheral neuropathy, neutropenia) associated with free payload with no impact of on-target ADC effects. The pharmacokinetic expectations for ABC3315 are provided within the top right panel. Created with BioRender.com.

Indeed, the role of extracellular, released MMAE in neutropenia, a common dose-limiting toxicity of MMAE ADCs, was demonstrated by Zhao et al. (18). Their work employed an in vitro co-culture model that assessed the effects of free MMAE and MMAE ADCs on the differentiation of hematopoietic stem cells to neutrophils. The observed concentrations leading to 50% inhibition in neutrophil differentiation were ~0.1 nM for free MMAE and 11-30 nM for MMAE ADCs (with cleavable linkers) (18). Further investigation by Zhao et al. demonstrated that serine proteases released by differentiated neutrophils mediated extracellular linker catabolism and MMAE release from ADCs, where free MMAE in extracellular fluid may diffuse into hematopoietic stem cells to inhibit differentiation to neutrophils (18). The Zhao group also demonstrated that nonspecific uptake of intact MMAE ADCs via Fc-gamma receptor-mediated endocytosis or via fluid-phase endocytosis did not contribute to the observed toxicity, further supporting the role of free MMAE in extracellular fluid as a key mediator of MMAE ADC-induced neutropenia. The role of released payload as a key mediator of MMAE ADC toxicity has also been supported through analyses performed by Food and Drug Administration (FDA) scientists (13,15).

We hypothesized that payload-binding antibodies could be applied to reduce the entry of released payload into untargeted cells by inhibiting membrane diffusion without altering ADC-antigen binding, receptor mediated endocytosis, or intracellular processing. To pursue this hypothesis, we have developed a humanized anti-MMAE Fab fragment, ABC3315, that binds with high affinity to free MMAE but with no observable binding to MMAE conjugated to antibody. The results presented here support the hypothesis that payload-binding agents (i.e., payload-binding selectivity enhancers or PBSEs) can widen the therapeutic index of ADCs. A representative graphic that illustrates the payload binding strategy is provided in figure 1.

Methods

Antibodies, cell-lines, and reagents

Polatuzumab vedotin (PV) and trastuzumab were purchased from Millard Fillmore Memorial Hospital (Williamsville, NY). Ramos (CRL-1596, RRID:CVCL_0597) and SKBR3 cells (HTB-30, RRID:CVCL_0033) were purchased from American Type Culture Collection (ATCC, Manassas, VA) and were cultured following ATCC recommendations. NCI-N87 (CRL-5822, RRID:CVCL_WH01) and GFP-MCF7 cells were a generous gift from Dr. Dhaval Shah. SKBR3 cells were purchased from ATCC in September of 2019. Ramos cells were authenticated by ATCC using short tandem repeat profiling and tested negative for mycoplasma using a universal mycoplasma detection kit (ATCC 30-1012K) in July of 2022. NCI-N87 were authenticated and tested negative for mycoplasma in January of 2021. GFP-MCF7 cells were not authenticated or tested for mycoplasma contamination. A breeding pair of HuHER2 mice was kindly provided by Dr. George Sgouros. Mouse studies were approved by the University at Buffalo Institutional Animal Care and Use Committee. MMAE (HY-15162), MMAF (HY15579), d8-MMAE (HY-15162A) and vc-MMAE (HY-15575) were purchased from MedChemExpress (Monmouth Junction, NJ). Trastuzumab was conjugated to vc-MMAE (TvcMMAE) and characterized using hydrophobic interaction chromatography (HIC) following the methods described by Singh et al. (19). The phagemid vector pComb3XSS (RRID:Addgene_63890) was purchased from Addgene (Watertown, MA) and the phagemid vector pADL-10b was purchased from Antibody Design Laboratories (San Diego, CA)

Immunized Phage Library Development

Monomethyl auristatin F (MMAF) was conjugated to keyhole limpet hemocyanin (KLH) or BSA via an EDC linker. Approximately 50 μg of KLH-MMAF immunogen emulsified in Freund’s incomplete adjuvant was used for single animal immunization. Female BALB/c mice (Envigo, Indianapolis, IN, RRID:MGI:5656377) were subcutaneously injected with 200 μL of emulsion and given a booster every three weeks. Anti-MMAF-BSA plasma titers were evaluated with ELISA using an anti-mouse Fc secondary antibody conjugated to alkaline phosphatase (A1418, Sigma-Aldrich, St. Louis, MO). Spleens were isolated from immunized mice, homogenized in TRIzol and RNA precipitated. RNA was resuspended in RNAse-free water and cDNA synthesized using SuperScript^® IV Reverse Transcriptase (Invitrogen, Waltham, MA). Mouse variable heavy chain (VH) and light chain (VL) were amplified via polymerase-chain reaction (PCR) and the amplified VH and VL products (~400 bp) were size selected on a 1% agarose gel and purified. Overlap extension PCR was used to assemble the scFvs with added SfiI cloning sites. Following SfiI restriction digest, scFv DNA was ligated into the pADL-10b phagemid vector overnight at 50 °C for 16 h. Electrocompetent TG-1 cells (Lucigen, Middleton, WI, RRID:CVCL_0P34) were transformed by electroporation with enough DNA to yield a library of at least 10⁷ individual transformants. Transformed bacteria were diluted in recovery media and transferred to a shaking incubator for 1 hour at 37 °C. Subsequently, the bacteria suspension was plated onto LB agar dishes containing 100 μg/mL ampicillin and 2% (wt/v) glucose and grown overnight at 37 °C. The library was recovered by scraping the bacteria into LB medium with sterile glycerol (20%, v/v). Cell library aliquots were stored at −80 °C.

Biopanning and Screening Mouse scFvs

MMAF was conjugated to biotin-PEG2-amine and bound to streptavidin magnetic beads (Invitrogen) preblocked with 2% non-fat dry milk in phosphate buffered saline (MPBS). Phage was rescued from a cell library aliquot as previously described (20). For the first panning input, the stock phage was diluted to 10¹² c.f.u. (colony forming unit)/mL in blocking buffer, and 1 mL of the diluted phage was added to the MMAF-biotin coated streptavidin beads and incubated for 2 hours. For the subsequent panning, the stock phage was diluted 1:1 in blocking buffer. After the incubation, beads were washed with 0.5% tween-20 phosphate buffered saline (PBST) 5 times, 10 times, 15 times, and 15 times for the 1st, 2nd, 3rd, and 4th round of panning. Bound phages were then eluted by incubation with free MMAE at the concentrations of 1 μM, 100 nM, 10 nM, and 1 nM in PBS over 1 hour for the 1st, 2nd, 3rd, and 4th round of panning. Output phages were titrated and TG-1 cells infected for phage production.

Following the fourth round of panning phage-infected TG1 cells were grown overnight, serially diluted in 2xYT media, spread over LB agar plates containing 100 μg/mL ampicillin and 2% wt/v glucose, and incubated overnight at 37 °C. A master plate was generated by inoculating a single colony into wells of a 96-well round-bottom culture plate filled with 100 μL of 2xYT supplemented with 100 μg/mL ampicillin, 2% (wt/vol) glucose and 15% (vol/vol) glycerol, and grown overnight at 37 °C, 300 rpm. Master wells were used to inoculate wells of 96 deep-well plates containing 1 mL of 2xTY medium with 100 μg/mL ampicillin. Plates were incubated at 37 °C and 300 rpm until reaching an optical density at a wavelength of 600 nm of ~0.5, then 1 μL of M13K07 helper phage (PH010L, Antibody Design Laboratories) were added to each well and incubated for 1 hour. Kanamycin was added to a final concentration of 50 μg/mL, and the plates were incubated overnight at 30 °C 250 rpm.

Nunc Maxisorp 96-well ELISA plates were coated with 4 μg/mL NeutrAvidin (Thermo Scientific, Waltham, MA) overnight at 4 °C. Plates were washed five times with PBST, blocked with MPBS for 2 hours at room temperature and 100 μL of 1 μM Biotin-PEG2-MMAF was added for 30 minutes. Plates were then washed five times with PBST and incubated with 4-fold diluted phage supernatant for 2 hours, with or without pre-incubation of free MMAE at concentrations of 10 nM and 100 nM, and with 100 nM of MMAE-ADC. Plates were then washed five times with PBST, and bound scFv-displaying phages were detected using an anti-M13 phage HRP-conjugated antibody (AS003, Antibody Design Labs) diluted 1:1000 in MPBS for 1-hour. Following five washes with PBST, 100 μL of 1-Step Turbo TMB-ELISA (Thermo Scientific) solution was added to each well and incubated for 15 minutes. The reaction was quenched by adding 100 μL of Stop solution (Thermo Scientific) to each well, and absorbance was measured at 450 nm. DNA isolated from positive clones were sent to the Roswell Park sequencing core facility (Buffalo, NY) for sequencing.

Mutagenesis Library Development, Panning and Screening

Genes for scFv clones 3B8, 1H2 and 1B3 were codon optimized for E. coli and synthesized by GeneArt. The heavy and light chains for each clone were amplified by PCR and subsequently mutated using a PCR based random mutagenesis method. The purified mutagenesis product was combined, and the heavy and light chains ligated using an overlap extension PCR method. The full library was constructed following the methods used for to construct the immunized phage library with the phagemid vector pComb3XSS used in place of the pADL-10b phagemid vector. Phage was panned against MMAF-peg11-biotin-streptavidin coated beads for three rounds, with increasing stringency. Following the third round of panning, phage displaying scFvs were dissociated with 1 μM of MMAE for 3 hours, supernatant removed, and remaining phage eluted with 1 μM MMAE for 24 hours. Two 96-deep well plates were inoculated with single colonies from the 24-hour elution and phage expressed in deep well plates following the protocol provided above. The following day, the phage containing supernatant was diluted 1:10 in MPBS in individual wells of a 96-well plate with or without 1 nM free MMAE for 1-hour. Subsequently, the phage containing solution was transferred to the wells of a Nunc Maxisorb plate containing immobilized MMAF-peg11-biotin-neutravidin and incubated for 2-hours at room temperature on a shaking incubator. Plates were washed and bound phage detected. Colonies with a signal knockdown >70% were sequenced.

Humanization

The amino acid sequence for the variable heavy and variable light chain for clone E2 were input in Abysis (21). Murine framework residues that occur with low frequency in human antibodies were conservatively mutated to an amino acid with high frequency. The murine and humanized sequences were input into Abodybuilder to create a structural model prediction for the human and murine sequences (22). Predicted structures were overlayed in ChimeraX (23) to ensure the human sequence was predicted to have a similar structure as the murine sequence. The humanized sequence was expressed as a Fab fragment (ABC3315) in ExpiCHO-S cells (Gibco^™) and purified using CaptureSelect^™ CH1-XL affinity resin (Thermo Scientific) following manufacturer recommendations

Competitive ELISA

Anti-MMAE Fab was diluted to 1 nM and incubated with MMAE, MMAF, and trastuzumab-vc-MMAE, at concentrations between 15 pM-100 nM. Solutions were added in triplicate to individual wells of an ELISA plate with MMAF-peg11-biotin-streptavidin immobilized. The plate was incubated for two hours at room temperature on a shaking platform set at 300 rpm. Wells were washed 4x with PBST and 250 μL of a 1:1,000 dilution of an anti-Human AP secondary (A8542, Sigma-Aldrich, RRID:AB_258397) added to each well and incubated for 1.5 hours. Wells were washed 2x with PBST and 2x with distilled water. 250 μL of 4 mg/mL p-nitrophenyl phosphate in diethanolamine was added to each well and the change in absorbance overtime at 405 nm was assessed for 10 minutes. The change in absorbance overtime for the MMAE/MMAF and TvcMMAE wells were normalized to wells treated with anti-MMAE Fab alone to determine the bound fraction of anti-MMAE Fab.

Surface Plasmon Resonance

A SR7500DC surface plasmon resonance (SPR; Reichert, Depew, NY) was utilized for binding assessments of anti-MMAE Fab. MMAF-peg11-biotin was flowed over the left channel of a neutravidin immobilized SPR chip (13206065, Reichert, Depew, NY). Subsequently unbound sites on both channels were blocked by injection of free biotin. ABC3315 Fab was serially injected for 3 minutes at concentrations of 1.23, 3.70, 11.11, 33.33 and 100 nM with a 3-hour dissociation step after the final 100 nM injection. To evaluate ABC3315 binding to free MMAE and PV ABC3315 was conjugated to nhs-peg12-biotin (A35389, ThermoFisher) and injected over the left channel of a streptavidin SPR chip (13206071, Reichert, Depew, NY). Subsequently, unbound sites on both channels were blocked by injection of free biotin. MMAE was serially injected for 3 minutes at concentrations of 0.37, 1.11, 3.33, 10, and 30 nM with a 3-hour dissociation step after the final 30 nM injection. A second kinetic titration was completed with PV injected at conjugated MMAE concentrations of 1.23, 3.70, 11.11, 33.33 and 100 nM with a 3-hour dissociation step after the final 100 nM injection. Observed sensorgrams were fit using the kinetic titration module of ClampXP (24) to obtain the association rate constant, dissociation rate constant and equilibrium dissociation rate constant.

Cell-viability assays

100 μL of Ramos cells at a density of 50,000 cells/mL were split into individual wells of a 96-well flat-bottom culture plate. Media containing dilutions to achieve a final concentration of free MMAE between 15 pM-100 nM or PV at concentrations between 45 pM-30 nM with or without 500 nM anti-MMAE Fab was added to individual wells and the cells incubated for 4-days at 37 °C in a humidified incubator with 5% CO2. The 500 nM Fab concentration is equal to a 5-fold molar excess of the highest concentration of MMAE used in the cell viability assay and is within the range of Fab plasma concentrations that are expected following the dosing protocol described in the xenograft efficacy study below. On the fourth day, 25 μL of 4 mg/mL 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) was added to individual wells and the plates incubated for 2-hours. Subsequently, formazan crystals were solubilized overnight following the addition of 100-μL of 10% SDS, 0.1M HCL. The absorbance of individual wells was read at a wavelength of 550 nm and 690 nm. A similar protocol was followed to evaluate the impact of anti-MMAE Fab on the efficacy of TvcMMAE and MMAE on SKBR3 cells. SKBR3 were trypsinsed and diluted to a concentration of 40,000 cells/mL. 100 μL of the cell suspension was added to individual wells of a 96-well u-bottom culture plate and cells allowed to attach overnight. The following day the culture media was aspirated and replaced with fresh media containing MMAE concentrations between 15 pM-100 nM or TvcMMAE at concentrations between 45 pM-30 nM with and without 500 nM anti-MMAE Fab. Cells were incubated for 6-days with fresh media and drug dilutions added on days 3 and 5. Cell-viability was assessed using the same MTT protocol described for the RAMOS cells. Cell viability was determined as the difference in absorbance at 550 nm and 690 nm for the treated wells divided by the difference for untreated wells. The observed cell-viability for each group was fit to a four-parameter inhibitor vs response equation in GraphPad Prism 7 (GraphPad, San Diego, CA, RRID:SCR_002798).

Xenograft Efficacy Studies

Male and female Nu/J mice (The Jackson Laboratory, RRID:IMSR_JAX:002019 ) were injected above the right hind leg with 5x10⁶ Ramos cells in 100 μL of DPBS. Xenograft growth was monitored using digital vernier calipers and tumor volume calculated as W²xL/2 where L is the longest tumor diameter and W is the tumor diameter perpendicular to L. At a tumor volume of 200-300 mm³ (average ~250 mm³) mice were randomized in groups of 8 to receive (i) PBS+PBS control, (ii) 1 mg/kg PV+PBS, (iii) 1 mg/kg PV+12x ABC3315, (iv) 3mg/kg PV+PBS (v) 3mg/kg PV+12x ABC3315. PV was administered by injection into the retroorbital venous sinus. ABC3315 doses were 12-fold the mole-equivalent doses of MMAE delivered via the ADC, where 13.16 mg/kg ABC3315 was administered with the 1 mg/kg PV dose, and where 39.48 mg/kg ABC3315 was administered with the 3 mg/kg PV dose. ABC3315 was administered by intraperitoneal injection and was divided into 5 injections, with 30% of the dose administered immediately after the PV injection, 25% administered 8 hours after PV, and 15% administered at 24 hours, 32 hours, and 48 hours. The Fab was administered by intraperitoneal injection due to protocol limitations on the number of intravenous injections allowed within the administration time-period. Tumor volumes and body weight were monitored daily, and mice were sacrificed at a tumor volume of 2000 mm³. Kaplan–Meier survival curves were generated in GraphPad Prism 7 and compared using the log-rank test at a significance level of p ≤ 0.05. The impact of ABC3315 on ADC efficacy was also evaluated with use of a previously reported mixed xenograft model of HER2+ NCI-N87 cells and HER2− GFP-MCF7 cells (25). Female Nu/J mice (The Jackson Laboratory, RRID:IMSR_JAX:002019) were injected above the right hind leg with 9x10⁶ NCI-N87 cells and 1x10⁶ GFP-MCF7 cells in incomplete RPMI medium (25). At a tumor volume of ~350 mm³, mice were randomized in groups of 6 to receive (i) PBS+PBS control, (ii) 3 mg/kg TvcMMAE +PBS, or (iii) 3 mg/kg TvcMMAE +12x ABC3315 at the same dosing frequency as used from the RAMOS study described above. One untreated xenograft tumor was resected at a tumor volume of 350 mm³, flash frozen, cryosectioned, and fluorescently imaged using previously reported protocols (26). The MCF7 cells (HER2-regions) were imaged using the GFP cube of an EVOS Fl auto microscope. HER2+ regions were stained for 2 hours at room temperature using 100 nM trastuzumab followed by a 2-hour incubation with a 1:500 dilution of rabbit anti-human (Invitrogen, Carlsbad, CA, SA5-10223, RRID:AB_2665319) in 5% goat serum followed by an incubation of a 1:500 dilution of a goat anti-rabbit 660 conjugated antibody (Invitrogen, Carlsbad, CA, A-21074, RRID:AB_2535735). The two scans were overlayed in ImageJ (RRID:SCR_003070).

Toxicity Studies

The toxicity evaluation of PV following a 120 mg/kg dose with and without co-administration of 3x ABC3315 was contracted to Champions Oncology (Rockville, MD). Swiss-webster mice (n=5/group) were injected with PV by tail-vein injection. ABC3315 was administered via intraperitoneal injection following the same dosing protocol outlined in the xenograft studies methods. Control mice received an equivalent volume of PBS at each time. Mouse body weight was measured daily for fourteen days after the PV injection. Mice that lost greater than 10% of their body weight were provided with diet gel ad libitum. Mice exhibiting >20% net weight loss for a period lasting 3 days or mice that lost greater than 25% of their body weight when compared to Day 0 were considered moribund and euthanized. A second toxicity evaluation was done in a HuHER2 transgenic mouse model. The HuHER2 mouse strain is an immunocompetent mouse model with human HER2 expression observed in various tissues that has been described previously (27,28). A mixture of male and female mice was randomized to receive (i) PBS+PBS control n=5 (ii) 80 mg/kg TvcMMAE+PBS control n=4 (iii) 80 mg/kg TvcMMAE+12x ABC3315 n=4. An 80 mg/kg dose of TvcMMAE was selected as the average DAR for TvcMMAE is greater than PV (4.2 vs 3.5) and others have reported clinical chemistry and hematological toxicities following an 80 mg/kg dose of a vcMMAE ADC (29). Mice were kept in metabolic cages (Tecniplast^®) for collection of urine for 4-days after dosing. Subsequently, mice were sacrificed, blood collected and allocated into tubes containing EDTA or lithium heparin. Blood in EDTA tubes was diluted 10x in 0.9% NaCl and run using an Abaxis VetScan HM5 hematologic counter (Abaxxis). Blood in lithium heparin tubes were analyzed using an Element DC veterinary chemistry analyzer (Heska).

Assay of MMAE by Liquid Chromatography / Mass Spectroscopy (LC-MS/MS)

Urine samples were processed by LC-MS/MS to assess the urinary excretion of MMAE by mice in the 80 mg/kg HuHER2 toxicity study. 100 μl of urine was removed by pipetting, 10 μl of a 150 ng/μl d8-MMAE internal standard added, and 3 μl of 10M NaOH was added to ensure release of MMAE bound to Fab. Samples were mixed well by vortexing. 500 μL of ethyl acetate was added to each urine sample, mixed by vortexing, and centrifuged for 5 minutes at 16,000 xg. Ethyl acetate was removed by pipetting and transferred to glass tubes. A second round of ethyl acetate extraction was performed to increase MMAE recovery. The 1^st and 2^nd extractions were combined, evaporated using nitrogen gas and resuspended in 95% Acetonitrile/5% DH₂O with 0.1% Formic Acid. An MMAE standard curve was prepared in blank urine collected from control mice and was processed identically to the urine samples. MMAE was quantified via LC-MS/MS using the instruments and settings that were previously described by Singh et al. and Chang et al. (30,31).

Data Availability Statements

The data generated in this study are available within the article. The datasets used during the current study are available from the corresponding author upon reasonable request.

Results

Development of humanized anti-MMAE Fab fragment

To develop antibodies with selective binding for free MMAE, we immunized mice with keyhole limpet hemocyanin conjugated to the C-terminal carboxyl group of MMAF. The carboxyl group of MMAF is the primary structural difference between MMAF and MMAE and is distant from the N-terminal monomethylvaline that is site of conjugation for MMAE in the vedotin linker. The chemical structures for MMAE, MMAF and vc-MMAE are provided in figure 2. A scFv phage library was developed from spleen cells obtained from immunized mice, and the library was screened to identify scFv with selective binding for free MMAE (relative to vc-MMAE conjugates). Shown in Figure 3A are the top 5 hits that demonstrated binding to free MMAE with negligible binding to TvcMMAE. Bars represent the fraction of phage displaying scFv that is bound to MMAF-peg11-biotin-streptavidin with MMAE or TvcMMAE spiked in relative to a control well. A random mutagenesis phage library was developed from clones 1H2, 1B3 and 3B8 and was subsequently panned and screened to identify clones with increased MMAE affinity. Figure 3B shows the fractional binding signal that was observed for 96 individual clones with and without incubation with 1 nM free MMAE. Several clones (wells appearing in purple: B5, C5, E2, G1) were identified with a >70% knockdown in the ELISA binding signal following incubation with 1 nM of free MMAE (Figure 3B). The clones were sequenced and observed to have a high sequence homology. Using the murine sequence of clone E2 as a lead, humanized anti-MMAE Fab was generated by a resurfacing protocol, where the z-score of the variable domains increased following humanization from −0.624 to 0.872 for the heavy chain and from 0.120 to 0.989 for the light chain (ABC3315, sequence provided within supplementary figure 1). ABC3315 was characterized using a competitive ELISA with increasing concentrations of free MMAE/MMAF and TvcMMAE (Figure 3C). ABC3315 binding was decreased by the addition of free MMAE and MMAF with an IC50 of ~1 nM, whereas no knockdown in binding signal was observed with the addition of TvcMMAE. For high affinity interactions, IC50 values observed in competitive ELISA experiments are often greater than the equilibrium dissociation constant. To better characterize the binding affinity of ABC3315, a series of kinetic titration surface plasmon resonance (SPR) analysis was performed. Rate constants were estimated for ABC3315 binding to MMAF-peg11-biotin bound to a neutravidin SPR chip. The observed sensorgram is provided in figure 3D. The estimated equilibrium dissociation constant for ABC3315 binding to MMAF-peg11-biotin-neutravidin was 44.2 pM. A follow up SPR experiment was completed with biotinylated ABC3315 immobilized on a streptavidin chip. Shown in panel 3E is the observed sensorgram for free MMAE injected over immobilized ABC3315. The fit equilibrium dissociation constant for free MMAE binding to ABC3315 is 7.7 pM. The higher affinity for free MMAE is the result of a faster association rate constant for ABC3315 to free MMAE (2.67x10⁵ M⁻¹sec⁻¹) in comparison to immobilized MMAF-peg11-biotin (4.11x10⁴ M⁻¹sec⁻¹). Shown in panel 3F is the observed sensorgram following a kinetic titration of polatuzumab vedotin (PV) over immobilized ABC3315. Consistent with ABC3315 selectively binding free MMAE no binding signal is observed, with minimal change in the response units during the binding analysis (±2 μRU).

Figure 2: Chemical structure diagram.

(A) The structures of MMAE, MMAF, and vc-PAB-MMAE are shown. The N-terminal secondary amine that is used for vc-MMAE synthesis and subsequent thiol conjugation to antibodies is highlighted in the green circle. The carboxylic acid group of MMAF (blue circle) was used for immunogen generation with the expectation that anti-immunogen antibodies would favor binding to MMAE/MMAF with an unconjugated N-terminus. (B) Shown is the reaction scheme of MMAF with EDC to produce KLH-MMAF immunoconjugates.

Figure 3: Development of the anti-MMAE Fab ABC3315.

(A) Provided are the top 5 scFv clones that were identified following biopanning and screening of the immunized murine scFv phage library. The fraction bound represents the ELISA signal for phage bound to MMAF-biotin-streptavidin immobilized on an ELISA plate, for each clone with or without MMAE or TvcMMAE spiked in. Bars represent the mean of duplicate samples with standard deviation error bars. Clones 1B3, 1H2, 3B8, 2E8 and 2C2 had a >50% decrease in the binding fraction with the addition of 30-100 nM free MMAE, with no change in binding with the addition of 100 nM TvcMMAE. (B) A random mutagenesis scFv library was built from clones 1B3, 1H2 and 3B8 to identify clones with increased MMAE binding affinity. Shown is the fractional binding signal for 96 clones with and without 1 nM free MMAE spiked in. Clones B5, C5, E2 and G1 have a greater than 70% knockdown in phage binding with the addition of 1 nM free MMAE. (C) To develop ABC3315, the clone E2 from panel B was humanized, expressed as a Fab fragment in CHO cells and characterized with a competitive ELISA. The binding of ABC3315 to MMAF-biotin-streptavidin was decreased with increasing concentrations of free MMAE and MMAF (IC50: ~1 nM), whereas TvcMMAE did not compete for binding. Points represent the mean of samples in triplicate with standard deviation error bars. Shown are kinetic titration binding SPR sensorgrams for ABC3315 to MMAF-peg11-biotin neutravidin (D), MMAE to ABC3315-peg12-biotin-streptavidin (E) and PV to ABC3315-peg12-biotin-streptavidin (F). Best fit values for the association rate constant (Kon) dissociation rate constant (koff) and equilibrium dissociation rate constant (KD) are provided in the insets. No binding signal was observed for PV to ABC3315-peg12-biotin-streptavidin. The observed intermittent spikes in the μRU signal for all sensorgrams is the result of the pump valve needle refilling with mobile phase.

Impact of ABC3315 on free MMAE and ADC cellular cytotoxicity

As ABC3315 selectively bound free MMAE, we expected that MMAE cytotoxicity would be dramatically decreased by ABC3315, whereas ADC cytotoxicity would be unaffected. This hypothesis is based on the expectation that free MMAE cytotoxicity results from non-specific uptake via diffusion into cells whereas ADC cytotoxicity is governed by antigen binding, endocytosis, and intracellular ADC catabolism. MMAE bound to ABC3315 is unable to diffuse into cells, whereas MMAE that is released in cells following ADC catabolism is free to bind intracellular tubulin. To test this hypothesis, a Burkitt’s lymphoma cell-line (Ramos) was incubated with free MMAE or with PV with and without 500 nM ABC3315. The cell survival fraction was determined following a 4-day incubation and was calculated as the quotient of the MTT signal for treated wells divided by control wells. The addition of ABC3315 increased the observed IC50 of free MMAE in RAMOS cells from 0.12 nM to 95.96 nM. The IC50 for cells treated with PV was 0.12 nM and was 0.13 nM for PV+ABC3315. Comparable results were obtained in the HER2+ SKBR3 cell-line following treatment with free MMAE or TvcMMAE. The IC50 of free MMAE was increased from 0.09 nM to 46.24 nM with ABC3315, and the IC50 of TvcMMAE was 0.04 nM and 0.03 nM with and without co-incubation with ABC3315, respectively. The observed cell-viability curves for free MMAE, PV and TvcMMAE are provided in figure 4.

Figure 4: ABC3315 selectively inhibits MMAE toxicity.

(A) RAMOS cells were incubated with MMAE (10 pM – 100 nM), with or without co-incubation with 500 nM ABC3315. ABC3315 increased the IC50 of free MMAE by 800-fold. (B) PV, a clinically-approved anti-CD79b ADC that incorporates MMAE as the payload molecule, was incubated with CD79b+ Ramos cells at concentrations of 3 pM – 30 nM, with or without coincubation with 500 nM ABC3315. ABC3315 did not alter the on-target cellular cytotoxicity of PV (IC50: 0.12 nM alone vs 0.13 nM with ABC3315). (C) ABC3315 (500 nM) increased the IC50 of free MMAE against SKBR3 cells by greater than 500-fold. (D) HER2+ SKBR3 cells were incubated with TvcMMAE with and without anti-MMAE Fab with negligible changes in cellular cytotoxicity (IC50: 0.03 nM alone vs 0.04 nM ABC3315). Points represent the mean of triplicate wells with standard deviation error bars.

Effects of ABC3315 on ADC efficacy in xenograft bearing mice

The preclinical development of PV included evaluation of PV activity in mice bearing RAMOS xenografts (32). A single 5 mg/kg dose of PV was reported to result in complete tumor regression (32). To evaluate the impact of ABC3315 on PV efficacy in vivo, we used the same tumor model, but with ADC dosing at 1 and 3 mg/kg to increase our sensitivity to detect any possible inhibitory effects of ABC3315 on therapeutic response. Mice were injected with a single dose of PV alone or with co-administration of ABC3315 at a 12-fold molar ratio (relative to conjugated MMAE). Tumor growth curves for each group are shown in figure 5A, and survival curves for each group are provided in figure 5B. Using the Log-Rank test, co-treatment of ABC3315 did not significantly alter survival for mice treated with PV at a dose of 1 mg/kg (p=0.075) or at 3 mg/kg (p=0.89). A mechanism of efficacy for MMAE ADCs relates to the “bystander effect” in which MMAE diffuses out of targeted cells into neighboring untargeted cancer cells, enabling killing of untargeted cells within tumors. Singh et al. reported a xenograft bystander model that was composed of ~50% HER2 positive NCI-N87 cells and ~50% HER2 negative MCF7 cells, and they evaluated the impact of TvcMMAE in isotropic N87 and MCF7 xenografts and in the N87/MCF7 mixed xenograft model (25). We utilized the same mixed xenograft model to assess effects of ABC3315 on the direct and bystander efficacy of TvcMMAE. The TvcMMAE ADC was synthesized using methods reported by Singh et al. (19). The resulting conjugate was determined by HIC to have a DAR of 4.23 (supplementary figure 2). Co-administration of ABC3315 did not significantly impact the tumor growth profile for mice treated with 3 mg/kg TvcMMAE+ABC3315 in comparison to mice treated with 3 mg/kg TvcMMAE+PBS (Figure 5C). Figure 5D displays a tumor section with HER2− negative regions appearing in magenta, and HER2+ positive regions, which were stained using trastuzumab as the primary antibody, in cyan. Consistent with the observations of Singh et al. (25), at a tumor volume of ~350 mm³, the xenografts are composed of relatively equal fractions of both HER2+ and HER2− regions, with the HER2+ regions appearing as dispersed patches distributed throughout the tumor space. The provided image was modified (inversed LUT and window/level altered) for figure aesthetics. Unmodified fluorescence scans are provided in supplementary figure 3.

Figure 5: ABC3315 does not alter the efficacy of PV.

Panel A and B: Nu/J mice bearing Ramos xenografts at a volume of ~250 mm³ were randomized to receive PBS, 1 or 3 mg/kg PV with or without a 12-fold molar excess of ABC3315 (n=8/group). (A) The tumor volumes over time for each group are provided with standard deviation error bars. Mice were sacrificed at a tumor volume of 2000 mm³. (B) The probability of survival over time for each group is provided. Based on the Log-Rank test, the co-treatment of ABC3315 did not significantly alter the group survival for mice treated with PV at a dose of 1 mg/kg (p=0.075) or 3 mg/kg (p=0.89). Panel C and D: Mice were inoculated with HER2+ NCI-N87 cells and HER2− GFP-MCF7 cells to develop a mixed xenograft model for assessment of bystander effects. (C) Provided is the mean tumor volume over time for mice bearing mixed xenografts of HER2+ NCI-N87 cells and HER2− GFP-MCF7 cells. Groups of mice were treated with PBS vehicle control, 3 mg/kg TvcMMAE+PBS, or 3 mg/kg TvcMMAE+12x ABC3315 (n=6/group). Both TvcMMAE groups exhibited decreased tumor growth rates in comparison to the control group, with no clear differences in tumor volume between groups of mice treated with TvcMMAE and PBS or ABC3315. (D) Fluorescent imaging of a tumor section that was harvested from an untreated NCI-N87/GFP-MCF7 xenograft bearing mouse at a tumor volume of 350 mm³. HER2− negative regions are shown in magenta and HER2+ regions are shown in cyan.

Effects of ABC3315 on PV and TvcMMAE toxicity in mice

In contrast to human patients, mice tolerate treatment with MMAE-based ADCs at doses that are substantially above those required for tumor response or regression. The limited toxicity of MMAE ADCs in mice may be partially attributed to reduced off-site on-target toxicity, due to the lack of cross-reactivity of ADCs with murine antigens, but reduced toxicity in mice has also been explained by greater resistance of mouse cells (relative to human cells) to the cytotoxic effects of free MMAE (33–35). Nonetheless, a >10% loss in body weight has been reported in mice following single intravenous doses of MMAE ADCs at doses between 100-150 mg/kg (36,37). To evaluate the effect of ABC3315 on PV toxicity, Swiss-Webster mice were intravenously injected with 120 mg/kg PV and co-dosed with either PBS vehicle or ABC3315 at a 3-fold molar ratio relative to conjugated MMAE. Mouse body weight was measured daily for 14-days after administration. The observed time-course of mean body weights for each group is provided in Figure 6A and the nadir body weight for each group is provided in Figure 6B. Administration of ABC3315 with 120 mg/kg PV decreased the nadir mouse body weight loss from 11.9% with a standard deviation of ± 7.0% (for mice treated with the ADC with co-administration of PBS) to 4.1 ± 2.1% (for mice treated with the ADC with co-administration of ABC3315, p=0.045). A second toxicity evaluation was performed in mice that express human HER2 using TvcMMAE and a 12x ratio of ABC3315 given at the same dosing frequency as described in the efficacy study. Four days after TvcMMAE dosing, a terminal blood sample was collected and hematological and clinical chemistry parameters were evaluated. All parameter values are provided in supplementary tables 1 and 2 and observed body weights over time are provided in supplementary figure 4. In comparison to the control group, mice that were treated with TvcMMAE had a significant drop in the total white blood cell count (p=0.028) and a significant drop in the red blood cell count (p=0.0083). Although there was a trend toward lower white blood cells and red blood cells for mice treated with TvcMMAE with ABC3315 (Figure 6D and 6E), the counts were not significantly different from values found in control mice. No other significant differences were identified between control and treated groups with regard to clinical chemistry and hematologic parameters. However, ALP and ALT enzyme values were greater than reported reference ranges for all groups, possibly a result of stress from housing in metabolic cages (38). Urine from TvcMMAE treated mice was collected and MMAE was quantified using LC-MS/MS. Provided in figure 6C is the cumulative MMAE recovered in the collected urine samples. Co-administration of ABC3315 increased the total mass of MMAE collected in the urine from 789.4 ±19.0 nanograms for mice administered TvcMMAE+PBS to 2625±206.8 nanograms for mice administered TvcMMAE+ABC3315.

Figure 6: ABC3315 decreases ADC toxicity in mice.

Panel A and B: Swiss Webster mice were intravenously injected with 120 mg/kg PV with PBS or a 3-fold molar excess of ABC3315 (n=5/group). (A) The body weight loss over time for each group is provided; points represent the mean percent change in body weight and error bars depict standard deviations. (B) The mean percentage weight change at nadir for each group is provided with standard deviation error bars. Mice administered ABC3315 with PV had a significantly decreased group nadir body weight loss from a mean of 11.9 ± 7.0% for mice treated with PV+PBS to 4.1 ± 2.1% for mice treated with PV+ABC3315 (p=0.045). Panel C, D, and E: HuHER2 mice were treated with PBS+PBS (n=5), 80 mg/kg TvcMMAE+PBS (n=4) or 80 mg/kg TvcMMAE+12x ABC3315 (n=4). (C) Co-administration of ABC3315 with TvcMMAE increased the cumulative amount of MMAE that was excreted in the urine four days after administration from 789±19.0 ng for mice treated with TvcMMAE+PBS to 2624.7±206.8 ng for mice treated with TvcMMAE+ABC3315. (D) In comparison to control mice, mice receiving TvcMMAE+PBS had a significant decrease (p=0.025) in WBC from 4.26±0.84x10⁹ cells/L for control mice to 2.7±0.79x10⁹ cells/L for mice receiving TvcMMAE+PBS. The mean WBC count for mice treated with TvcMMAE+ABC3315 was 3.5±0.5 x10⁹ cells/L, which was not significantly different from WBC counts in control mice (p=0.15). (E) In comparison to control mice, mice administered TvcMMAE+PBS had a significant decrease (p=0.0083) in RBC from 9.42±0.55x10¹² cells/L to 8.23±0.39x10¹² cells/L. The mean RBC count for mice administered TvcMMAE+ABC3315 was 8.93±0.57x10¹² cells/L, which was not significantly different from values found for control mice (p=0.23)

Discussion

ADC drugs are effective treatments for several cancers, with 11 FDA-approved agents in current clinical use, and with more than 100 ADCs in current clinical development. However, all ADCs that have undergone clinical testing have led to substantial off-target toxicity, which negatively impacts the quality of life of treated patients. Additionally, tolerable doses of ADCs are often far below levels required for maximal efficacy; thus, ADC efficacy is limited by off-target toxicities.

Much effort has been dedicated to increase the therapeutic index of ADCs through optimization of antibody affinity, antibody structure, payload loading (i.e., the drug-to-antibody ratio, DAR), linker stability, and through optimization of dosing protocols. For example, the valine-citrulline linker employed within MMAE ADCs, including PV, exhibits excellent stability in plasma and efficient intracellular hydrolysis mediated by cathepsin B during endo-lysosomal processing (39). However, regardless of the nature of the antibody (e.g., affinity or selectivity for tumor antigen), and regardless of the DAR or linker stability, all MMAE administered through dosing of MMAE-ADCs is eventually liberated through linker hydrolysis. MMAE is expected to (eventually) diffuse from sites of release into extracellular fluids, such that the cumulative plasma exposure to released MMAE (e.g., as assessed by the area under the MMAE plasma concentration vs. time curve, AUC) is a function of the MMAE dose and the rate of MMAE clearance. As such, regardless of the ADC attributes listed above, payload plasma AUC is expected to be a simple function of ADC dose and DAR. For lipophilic payload molecules such as MMAE, payload toxicities (e.g., peripheral neuropathy, bone marrow suppression) may be expected to be directly related to the AUC of released payload in plasma. Indeed, pharmacokinetic / pharmacodynamic considerations predict, and preclinical and clinical data demonstrate, that MMAE ADC MTD is independent of the cancer treated, antigen targeted, antibody affinity, and linker stability (13,40).

Our strategy employs anti-payload antibody fragments to bind and “neutralize” payload in plasma, and to enable clearance of neutralized payload through renal filtration. There is a long history of use of antibody fragments to prevent or reverse the adverse effects of drugs and toxins (Praxabind, Digibind, DigiFab etc.) (41,42). Additionally, prior work conducted in our laboratory has demonstrated the successful use of systemic administration of anti-drug antibodies and fragments within inverse targeting strategies to increase the pharmacokinetic and pharmacodynamic selectivity of regional chemotherapy (43–45) (e.g., intraperitoneal chemotherapy for treatment of peritoneal tumors). However, the application of binding molecules to increase ADC safety and therapeutic selectivity has not been previously reported. No prior assessments of anti-payload binding agents on ADC toxicity or selectivity, in vitro or in vivo, have been published (to our understanding).

It is notable that several therapies have been FDA-approved to treat toxicities relating to cancer chemotherapy. Successful biological agents with indications for reversing toxicity of anti-cancer drugs include granulocyte colony-stimulating factor (Neopogen, Neulasta, Lonquex, etc.) (46), erythropoietin (Epogen) (47), and granulocyte-macrophage colony stimulating factor (Leukine) (48). While these agents are effective, and highly successful products, they only address one type of toxicity (e.g., G-CSF reverses neutropenia, erythropoietin reverses anemia, etc.). Given that the available evidence suggests that the systemic exposure of released payload is the main driver for most off-target toxicities associated with ADC therapy, a single PBSE agent, for example anti-MMAE PBSE, may be expected to prevent many of the major toxicities associated with MMAE ADCs (i.e., neutropenia, peripheral neuropathy, etc.).

The results observed with ABC3315 support the hypothesis that payload binding antibody fragments can widen the therapeutic window of ADCs by decreasing off-target toxicity while not negatively impacting anti-cancer efficacy. Nonetheless, additional investigations are required to further characterize the effects of ABC3315 on MMAE pharmacokinetics, ADC toxicity, and ADC efficacy. We chose to administer ABC3315 using a 2-day fractionated dosing protocol as Yip et al. reported that up to 60% of MMAE was eliminated in the first two days after PV administration (49). Considering the rapid renal clearance of Fab fragments in mice we expected that excess Fab would be required to efficiently neutralize free MMAE: therefore, a 12x molar ratio of ABC3315 was first evaluated in the xenograft efficacy studies. The 12x molar ratio was empirically selected as the highest dose possible with the mass of Fab purified at the start of the xenograft studies. Previous toxicity studies have demonstrated a steep-dose toxicity relationship for MMAE ADCs and body weight loss in mice. Due to the steep dose-toxicity relationship with body weight we predicted that a 3-fold excess of Fab would be sufficient to demonstrate diminished body weight loss following a 120 mg/kg dose of PV. In contrast, dose-response relationships for other toxicities in mice are poorly defined. Therefore, in the follow up 80 mg/kg TvcMMAE toxicity study in HuHER2 mice, we evaluated the impact of ABC3315 on hematologic and clinical chemistry values using the 12x ratio of Fab. The empirically selected ratios of Fab:MMAE proved sufficient to demonstrate the benefit of payload binding Fab fragments; however, additional development efforts are required to optimize the required ratio of Fab:MMAE. Physiologically-based pharmacokinetic models for Fab fragments (50), MMAE (31), and MMAE ADCs (51) have been developed previously and these models can be adapted to identify optimum Fab doses and dosing schemes.

A critical consideration for our strategy is the fate of MMAE-Fab complexes. The pharmacokinetics of Fab fragments in pre-clinical animal models (52,53) as well as in humans (54,55) have been studied extensively. A majority of Fab is eliminated by the kidney, and prior work demonstrates that administration of anti-toxin Fab fragments increases the urinary elimination of toxins in animal models and humans (i.e., owing to efficient renal filtration of Fab-toxin complexes) (56–58). Following renal filtration, a fraction of Fab is excreted unchanged into the urine and the remaining fraction is catabolized at the surface of proximal tubule cells via brush-border peptidases or within proximal tubule cells following receptor mediated endocytosis. Due to the saturable nature of mechanisms associated with renal catabolism of protein, the fraction of Fab that is eliminated unchanged in urine is dose dependent. For example, in healthy volunteers, the fraction of idarcuizumab that was excreted intact into the urine was 10.7% at a dose of 1 gram and 38.9% at a dose of 5 grams (54). For our payload binding strategy, the preferred route of elimination is urinary excretion of intact Fab-MMAE complexes.

In the event that Fab-MMAE complexes are reabsorbed into proximal tubular cells (e.g., via interaction with the megalin-cublin endocytosis pathway), it may be expected that MMAE released within proximal tubular cells would be efficiently transported into the urine following Fab catabolism. MMAE is a substrate for the drug efflux pump MDR1 (P-gp) (59) which is highly expressed on kidney proximal tubule cells. Cells that express MDR1 have been shown to be resistant to toxicity mediated by MMAE (60). Additionally, in patients administered brentuximab vedotin, 28% of the recovered dose of MMAE was excreted in the urine (59). As nephrotoxicity is not a common adverse event associated with MMAE containing ADCs (14), it is reasonable to hypothesize that increasing the fraction of MMAE eliminated by the kidney would not lead to high risk for kidney damage. Additionally, several MMAE-antibody fragment and peptide conjugates, which may be expected to show dominant elimination by the kidneys, have been developed and evaluated pre-clinically, with no observed kidney toxicity (61–63). In the urine elimination study, we observed that the amount of MMAE that was excreted in urine did not significantly increase with Fab co-administration until days 3 and 4 after TvcMMAE dosing. These results may be consistent with MMAE-Fab complexes being reabsorbed and slowly catabolized by kidney proximal tubular cells, resulting in a delayed increase in MMAE excretion. Further studies are required to elucidate the relationships between Fab administration and MMAE disposition.

The results provided here demonstrate that application of payload binding antibody fragments increases the therapeutic window of MMAE ADCs. Considering that there are four MMAE-based ADCs in current clinical use, and considering that many more are undergoing late-stage clinical evaluation, anti-MMAE Fab may have broad utility in improving outcomes for many cancer patient populations. We hypothesize that the payload binding approach may also be applied to optimize therapy with ADC drugs delivering additional payloads (e.g., SN38, Dxd, DM4, calicheamicin, etc.), potentially enhancing the safety and utility of all marketed agents within this important drug class.