MYTX-011: A pH-Dependent Anti–c-MET Antibody–Drug Conjugate Designed for Enhanced Payload Delivery to c-MET–Expressing Tumor Cells

The field of antibody–drug conjugates (ADC) has experienced a renaissance in recent years. Decades of improvement in ADC technology, primarily focused on the linker–payload and target selection, have culminated in several new approvals (1). ADCs are designed to deliver payload specifically to tumor cells; however, only a small fraction of the administered ADC reaches tumors, and the therapeutic dose is generally limited by off-target toxicity (2). Consequently, for ADCs against many otherwise promising targets, only a small fraction of the patients, such as those expressing the target antigen at higher levels, derive benefit (3). In addition, many ADC targets are expressed on normal tissues, leading to target-mediated drug disposition (TMDD) and release of free payload that further narrows the therapeutic index. Attempts to improve the clinical utility of ADCs against any given target have focused on linker–payload, such as novel payload classes, increased drug-to-antibody ratios (DAR) and altering payload potency. Therefore, although the linker–payload space has been explored extensively, limited efforts have been made to enhance payload delivery to tumors via an innovative approach focused on the antibody component of ADCs.

Engineering pH-dependent binding between antibody complementarity–determining regions (CDRs) and their antigens is a well-established method for modulating antibody/receptor dynamics and pharmacokinetics (PK; ref. 4–7). Indeed, two pH-dependent antibodies (ravulizumab and satralizumab) that exploit these mechanisms have been approved for non-oncology indications. Recently, an anti-HER2 ADC based on a moderately pH-dependent antibody, pertuzumab, was further engineered to improve pH-dependent binding by generating a select set of variants based on the crystal structure (8). However, the success of trastuzumab deruxtecan in treating HER2-low patients and the abundance of other HER2-directed therapies in clinical development (9) led us to take a broader approach to the wider universe of other ADC targets and in other indications in which payload delivery might be limiting for ADC mechanism of action. Moreover, we reasoned that an extensive engineering approach directed toward all residues in the heavy and light chain CDRs could yield better candidates for therapeutic use, eliminating the need for a crystal structure.

c-mesenchymal–epithelial transition (c-MET) factor, a receptor tyrosine kinase encoded by the MET proto-oncogene, is widely expressed in various solid tumors (10). In non–small cell lung cancer (NSCLC), oncogenic MET activation is observed in a small subset of patients (2%–4%), whereas c-MET overexpression is highly prevalent (30%–60%) and has been shown to drive resistance to standard-of-care therapies (11, 12). Several antibodies targeting c-MET, with various mechanisms of actions, have been evaluated in clinical trials but have failed to gain FDA approval (13–15). Multiple ADCs targeting c-MET are in early clinical development, most notably ABBV-400, REGN5093-M114, BYON3521, and AZD9592. In addition, telisotuzumab vedotin (Teliso-V), a valine citrulline monomethyl auristatin E (vcMMAE)–conjugated ADC targeting c-MET, is currently in late-stage development and has recently received breakthrough designation for patients with locally advanced or metastatic nonsquamous NSCLC with EGFR wild-type status and high c-MET expression (16). Although Teliso-V activity in patients with c-MET-high NSCLC is encouraging, there remains a much broader population of patients with NSCLC whose tumors express moderate or low levels of c-MET and for whom there are limited therapies in development with promising clinical data.

We sought to create an anti–c-MET ADC with the potential to benefit a broader population of patients including those whose tumors express moderate to low levels of c-MET. We hypothesized that engineering pH-dependent binding in the antibody component of an anti-c-MET ADC would permit the release of ADC from c-MET in the acidic endolysosomal environment and boost ADC uptake and efficacy in c-MET⁺ cancer cells, thereby widening the therapeutic index. Here, we describe MYTX-011, an ADC that incorporates a novel pH-dependent anti–c-MET antibody conjugated to the clinically precedented vcMMAE linker–payload. We characterized the impact of pH-dependent binding by comparing the improved properties of MYTX-011 relative to its corresponding nonengineered parent ADC and/or a benchmark anti–c-MET ADC. Using this comparative approach, we demonstrated that pH-dependent binding significantly improved the uptake and cytotoxicity of MYTX-011 in vitro as well as the efficacy of MYTX-011 in NSCLC xenograft models across a range of low, moderate, and high c-MET expression levels. In cynomolgus monkeys, MYTX-011 demonstrated a comparatively longer half-life and a toxicity profile expected for an MMAE-based ADC. Taken together, these findings highlight the potential of MYTX-011 as a therapeutic candidate for patients with a broad range of c-MET expression.

hIgG1 isotype control was purchased from Bio X Cell (Cat# BE0297). Parent and benchmark anti–c-MET antibodies were generated as a hIgG1 isotype with variable region sequences from the Inxight Drugs database from the National Center for Advancing Translational Sciences (drugs.ncats.io) record numbers MO4K3GDN1I (emibetuzumab) and 976X9VXC3Z (telisotuzumab). Paraformaldehyde (PFA) 4%, BSA, anti–Alexa Fluor 488 quenching antibody (Cat# 710369), ProLong Gold Antifade mountant (Cat# P10144), Zenon Alexa Fluor 488 (Cat# Z25402A), and 4′,6-diamidino-2-phenylindole, dihydrochloride (DAPI; Cat# D1306) were purchased from Thermo Fisher Scientific. For imaging experiments, cells were transfected using Myrus Bio TransIT-2020 according to the manufacturer’s instructions (Cat# MIR-5404).

PBS buffer was purchased from Teknova. Fabfluor-pH antibody (Cat# 4722) was purchased from Sartorius. Recombinant c-MET antigens, i.e., human c-MET (Cat# 10692-H08H), cynomolgus c-MET (Cat# 90304-C08H), and mouse c-MET (Cat# 50622-M08H) were purchased from Sino Biological. mTagRFP-T-Lysosomes-20 (Addgene plasmid # 58022; ref. 17) and Rab4-mRuby (mRuby-Rab4a-7; Addgene plasmid # 55878) were a gift from Michael Davidson. Rab11-RFP (pTag-RFP-C-h-Rab11a-c-Myc) was a gift from James Johnson (Addgene plasmid # 79806; ref. 18).

NCI-H2122, NCI-H1975, and NCI-H1373 cells were purchased from ATCC and cultured in RPMI 1640 medium (Cat# 30-2001, ATCC). Detroit-562 and EBC-1 cells were purchased from ATCC and the Japanese Collection of Research Bioresources, respectively, and cultured in Eagle Minimum Essential Medium (Cat# 30-2003, ATCC). Cell lines were cultured in media supplemented with 10% FBS (Cat# 30-2020, ATCC) and maintained at 37°C and 5% CO₂ in a humidified incubator. All cell line source vials were confirmed to be Mycoplasma negative by IDEXX BioAnalytics.

pH-dependent anti–c-MET antibodies were generated by point mutagenesis of CDR residues to generate more than 700 variants, and extensive biophysical and functional characterization was performed to determine the best combination of residues (4, 19). Briefly, heavy and light chain plasmids were synthesized (Azenta Life Sciences), and recombinant antibodies were expressed in Expi293 cells (Thermo Fisher Scientific). Cell culture supernatants at a 1:40 dilution in 1X PBS buffer with 0.05% Tween 20 (1X PBST buffer) pH 7.4 or purified antibodies at 10 μg/mL in 1X PBST buffer pH 7.4 were immobilized on an anti–human IgG Fc (AHC) sensor (Sartorius) followed by a stable baseline with 1X PBST buffer pH 7.4. Association with human c-MET was performed in 1X PBST pH 7.4 for 120 seconds and dissociation in either 1X PBST pH 7.4 or 1X PBST pH 5.4 for 300 seconds. pH-dependent binding of MYTX-011 to human c-MET was assessed with MYTX-011 immobilized on an AHC sensor and human c-MET at 50 nmol/L, association was performed in 1X PBST pH 7.4 for 120 seconds, and dissociation was performed in either 1X PBST pH 7.4 or 1X PBST pH 5.4 for 300 seconds.

Detroit-562 or NCI-H1975 cells were seeded at 5,000 cells/well a day prior to antibody or ADC treatment. pHrodo reagent (Thermo Fisher Scientific, Cat# Z25611), a secondary anti–hIgG Fab conjugated to a pH-sensitive dye that fluoresces upon internalization, was diluted to 60 nmol/L in media. Twenty nmol/L of the hIgG1 component of MYTX-011, i.e., MYTX-011 mAb, parent and isotype control antibodies, or 21 nmol/L of MYTX-011, parent ADC and nonbinding ADC, was diluted 1:1 with pHrodo reagent. After incubation for 30 minutes, media were removed from preseeded Detroit-562 or NCI-H1975 cells, and the pHrodo-antibody/ADC complex was added to the plate and incubated at 37°C for 24 hours. Cells were trypsinized using TrypLE Express (Thermo Fisher Scientific, Cat# 12605010) and washed twice using ice-cold FACS buffer pH 7.4. Fluorescence was analyzed using an Attune NxT flow cytometer (Thermo Fisher Scientific), and raw data were processed using FlowJo. Fold change was calculated by subtracting the median fluorescence intensity of the isotype control/nonbinding ADC from the median fluorescence intensity of MYTX-011 mAb or MYTX-011 and parent antibody or ADC and taking the ratio.

Nonbinding control (anti–HIV gp120 mAb) and MYTX-011 mAb were synthesized as a hIgG1 isotype with variable region sequences from US 5652138 A (example 4) and WO 2022/169975A1 (example 43; ref. 20), respectively. Nonbinding, parent, and MYTX-011 mAbs with site-specific cysteine residues (as described in WO 2006/034488A2, SEQ ID NO: 27) were recombinantly expressed in Chinese hamster ovary cells. Expressed antibody was purified using protein A and size exclusion chromatography with a monomer content >95%. Intact molecular weights were confirmed by mass spectrometry. Twenty mg/mL of the purified antibody was reduced overnight and subjected to mild reoxidation at 10 mg/mL using a dehydroascorbic acid (DHAA)/mAb ratio of 15 in 50 mmol/L Tris, pH 7.5 at 22°C for 3 hours to make the site-specific cysteines available for conjugation. Reoxidized mAb was conjugated to vcMMAE at 10 mg/mL with a 3.0 fold excess of linker–drug in 50 mmol/L Tris, pH 7.5% and 10% dimethylacetamide (DMA) at 22°C for 2 hours. The DAR was confirmed to be ∼ 2.0 with >98% monomer and <0.5% free drug. Benchmark ADC was generated as described with a DAR of 3.1, >99% monomer and ∼2% free drug (21). Nonbinding ADC2 is an anti–CD79b ADC with a DAR of 3.7, >99% monomer and <1% free drug (22). MYTX-011 used for toxicology studies was produced using a similar but proprietary process.

MYTX-011 cytotoxicity was tested in 62 human cancer cell lines with parent and nonbinding control ADCs in assays performed at Crown Bioscience. Cells were seeded in a black polystyrene flat bottom 96-well plate (4,000 cells/well) in their respective medium. The following day, cells were treated with nine different concentrations of the ADCs (from 15 µg/mL to 1.5 ng/mL; 3.16-fold serial dilutions). Positive and negative controls were cisplatin and DMSO, respectively. After 96 hours, cell viability was assessed using CellTiter-Glo assay (Promega), and luminescence intensity was recorded using an EnVision multi label reader (Perkin Elmer). For each cell line, c-MET protein expression was quantified by flow cytometry using Quantibrite beads (Becton Dickinson, Cat# 340495), and mRNA levels were derived from RNA-seq data deposited in the Crown Bioscience’s database OncoExpress (as of August 17, 2023). To calculate the IC₅₀ values, a dose–response curve was fitted using nonlinear regression. The calculations were performed using GraphPad Prism. Cell lines with IC₅₀ values > 15 μg/mL were classified as insensitive.

Formalin-fixed, paraffin-embedded tumor blocks were prepared from tumors isolated from three untreated mice bearing cell line–derived xenografts (CDX) of NCI-H1975, NCI-H1373, EBC-1, and NCI-H2122 cells and patient-derived xenografts (PDX) of CTG-2669, CTG-3414, CTG-1353, CTG-2533, and CTG-2082 tumors with a size of 240 ± 180 mm³. Formalin-fixed, paraffin-embedded sections obtained from the blocks were stained for c-MET expression by IHC using the SP44 in vitro diagnostics (IVD, Cat# 790-4430, RUO only) test (Ventana). c-MET positivity was qualitatively assessed as low, moderate, or high based on intensity of staining (1+, 2+, or 3+, respectively).

NCI-H1975, NCI-H1373, and EBC-1 CDX studies were performed at Charles River Laboratories (CRL). NCI-H2122 xenograft studies were performed at Avastus Preclinical Services. SCID mice (NCI-H1975, NCI-H1373, and EBC-1) and athymic nude mice (NCI-H2122) were used for efficacy studies. For each study, 6- to 8-week-old female mice were subcutaneously injected with 5 to 10 × 10⁶ viable cells into the right flank (0.1 mL suspension of 50% Matrigel). Mice were randomized when the tumors reached an average volume of 100 to 150 mm³ and injected intravenously with vehicle control (PBS) or test articles based on body weight. Tumor volume (TV) and body weights were measured twice weekly. Tumor volume was calculated using the formula TV = (L × W²)/2. Mice were humanely sacrificed if the tumors ulcerated or if they reached a volume of 2,000 mm³. Animals with nontreatment-related deaths were excluded from analysis.

PDX studies were performed at Champions Oncology. Informed consent from patients was obtained through Champion Oncology’s Institutional Review Board protocol for specimen collection and PDX model development. Briefly, 65 mm³ size tumor fragments were implanted in the left upper flank of 6- to 8-week-old female athymic nude mice. Vehicle (PBS) and test articles were injected intravenously when the tumors reached an average size of 150 to 300 mm³. Unless indicated, the animals were sacrificed when the tumors reached an average size of 1,500 mm³. Tumor volumes were calculated using the formula TV = W² × L × 0.52. Animals with nontreatment related deaths were excluded from analysis. All CDX studies conducted at CRL and Avastus Preclinical Services and PDX studies conducted at Champions Oncology were reviewed and approved by their respective Institutional Animal Care and Use Committee (IACUC).

For monitoring mAb uptake by live cell imaging, NCI-H1975 cells were seeded on 96-well glass-bottom plates. After spreading, cells were treated with MYTX-011 mAb or parent mAb (10 nmol/L) conjugated with Fabfluor-pH antibody. Real-time mAb uptake was monitored by Incucyte (Sartorius). For confocal microscopy, NCI-H1975 cells transfected with LAMP1-RFP were seeded on glass coverslips. After spreading, cells were treated with MYTX-011 mAb or parent mAb (10 nmol/L) conjugated with Zenon Alexa Fluor 488 according to the manufacturer’s instructions. Cells were washed twice with PBS and fixed with 4% PFA. To quench the fluorescent signal of the conjugated mAbs at the cell surface, cells were incubated with an anti-Alexa Fluor 488 antibody. After 1 hour, coverslips were washed in PBS, incubated with DAPI to stain the nuclei, and mounted on glass slides using ProLong antifade mountant. Cells were imaged using a Leica SP5 spectral confocal microscope controlled by Leica Application Suite (LAS) software. Images were acquired with a 63x Apo lens (NA: 1.45) to achieve lateral sampling of 78 nm and an axial sampling of 250 nm. For each sample, six random fields of view (FOV) and six Z-stacks were acquired. Live cell imaging data were analyzed using Incucyte S3 software. Mander’s colocalization coefficient was calculated using the open-source software Fiji, whereas single-organelle fluorescence intensity was calculated using the open-source software CellProfiler.

For evaluating the endosomal trafficking of internalized mAbs, EBC-1 cells were transfected with Rab11-RFP or Rab4-mRuby and seeded on glass coverslips. After spreading, cells were incubated for 30 minutes at 4°C with MYTX-011 mAb or parent mAb (25 nmol/L) conjugated with Zenon Alexa Fluor 488. Following two washes in complete media, cells were incubated at 37°C for 40 minutes to allow mAb internalization. Cells were fixed with 4% PFA and incubated for 1 hour with an anti–Alexa Fluor 488 to quench the surface fluorescence. Nuclei were stained with DAPI, and coverslips were imaged in PBS using a Nikon AX-confocal system equipped with a Plan APO lD 60x oil lens. Images were acquired using Nikon Elements software. For each condition, up to nine randomly selected FOVs were imaged (lateral sampling: 94 nm; axial sampling: 250 nm). Mander’s colocalization coefficient was calculated using the open-source software Fiji.

Three female cynomolgus monkeys were dosed with a single i.v. dose of 2.7 mg/kg of MYTX-011, parent ADC, and benchmark ADC. Blood samples were collected as described. Samples were analyzed using total mAb and free MMAE LC-MS/MS assays (22). The toxicokinetic profile of MYTX-011 was determined in three male cynomolgus monkeys per group after administration of an i.v. bolus dose of either 6, 12 or 18 mg/kg once every 3 weeks for a total of two doses. Blood samples were collected as described. Samples were analyzed for total mAb, total ADC (MYTX-011), and free MMAE using LC-MS/MS assays (22). All studies with cynomolgus monkeys were reviewed and approved by CRL IACUC.

Data generated in this study are available within the article and supplementary data files.

CDRs from anti–c-MET antibodies were subjected to point mutagenesis to generate antibody variants. Variants were screened in a biolayer interferometry (BLI) assay to identify those that bind at neutral pH but dissociate at pH 5.4 (representing endolysosomal pH). Mutations from pH-dependent antibodies were further combined in a pairwise fashion to generate combinatorial variants, and additional screening using the BLI assay was conducted. Combinatorial variants that exhibited a high degree of pH-dependent binding were assessed for their functional and biophysical characteristics. Briefly, the effect of pH-dependent c-MET binding on net internalization (defined as the overall accumulation of antibodies or ADCs inside cancer cells) was assessed on c-MET⁺ Detroit-562 cancer cells. Enhanced dissociation of pH-dependent antibody variants at pH 5.4 correlated with an increase in net internalization over the non-pH engineered parent antibody confirming that greater pH dependence leads to enhanced accumulation of antibody inside c-MET⁺ cancer cells (Fig. 1A and B). Extensive developability characterization including aggregation, hydrophobicity, and nonspecific binding assessment was conducted as previously described (23). MYTX-011 mAb was shown to have an approximately threefold greater net internalization in c-MET⁺ Detroit-562 cancer cells compared with its parent mAb and showed increased net internalization compared with parent mAb across several NSCLC cell lines with a range of different levels of c-MET expression (Supplementary Fig. S1). Moreover, MYTX-011 mAb exhibited favorable biophysical characteristics in developability assays and was therefore prioritized for ADC generation and further characterization.

MYTX-011 mAb and parent mAb were conjugated to vcMMAE at an engineered cysteine site to generate ADCs with a DAR of ∼2.0 as confirmed by mass spectrometry. Binding to recombinant human c-MET, performed using BLI showed rapid dissociation of MYTX-011 at pH 5.4 compared with pH 7.4, confirming pH-dependent binding (Fig. 1C). MYTX-011 bound c-MET⁺ NCI-H1975 cancer cells with subnanomolar affinity (IC₅₀ = 0.4 nmol/L) at pH 7.4 and binding was maintained at pH 6.4 (IC₅₀ = 0.5 nmol/L; Supplementary Table S1), which is below the literature estimates for an acidic tumor microenvironment, reviewed in ref. 24. MYTX-011 bound to human and cynomolgus monkey c-MET with similar affinity, whereas no binding to mouse or rat c-MET was detected (Supplementary Fig. S2). MYTX-011 was found to internalize ∼4-fold more than the parent ADC in internalization assays on NCI-H1975 cells (Fig. 1D).

Given that pH-sensitive c-MET binding was shown to increase net internalization, we next sought to determine whether pH engineering resulted in enhanced cytotoxicity in vitro. A total of 62 cell lines were selected across solid tumor types including NSCLC, head and neck, gastric, pancreatic, esophageal, bladder, kidney, and skin cancers and subjected to a nine-point dose response of MYTX-011, parent ADC, or a nonbinding control ADC for 96 hours. The ADC IC₅₀ and maximum killing values are listed in Supplementary Table S2. Although both parent ADC and MYTX-011 showed cytotoxic activity against 15/62 cell lines tested, only MYTX-011 showed activity against an additional 16 cell lines (31/62). No activity was observed with nonbinding control. This result suggests that pH-dependent binding of ADCs confers more potent cytotoxicity in vitro (Fig. 2A and B). Importantly, sensitivity correlated with c-MET expression in the cancer cell lines (Fig. 2C) demonstrating that the increased cytotoxicity of MYTX-011 is on-target. Cell lines that were sensitive to MYTX-011 but not parent ADC exhibited lower levels of c-MET expression, demonstrating that MYTX-011 is active against cancer cells with a broader range of c-MET expression (Fig. 2D; Supplementary Table S2). Notably, cell lines insensitive to MYTX-011 were largely resistant to free MMAE, indicating that payload sensitivity is an independent determinant of MYTX-011 cytotoxicity (Fig. 2E).

As pH-dependent MYTX-011 exhibited enhanced net internalization and cytotoxicity compared with the parent ADC, we sought to resolve the dynamics of internalization over time by treating cells with mAbs conjugated to a pH-sensitive dye (Fig. 3A and B). At early time points (0–4 hours), MYTX-011 mAb and parent mAb showed similar levels of uptake; however after 6 hours, the intracellular levels of MYTX-011 mAb progressively increased reaching the highest intensity at 24 hours corresponding to a threefold increase over the parent mAb. Confocal microscopy revealed that parent and MYTX-011 mAbs were equally trafficked to lysosomes (Fig. 3C and D); however, the latter showed greater accumulation in these organelles (Fig. 3E). Additionally, we evaluated the endosomal trafficking of internalized mAb/c-MET complexes by confocal microscopy which revealed greater colocalization of the parent mAb with the fast-recycling endosomal compartment (Rab4⁺) compared with MYTX-011 mAb (Fig. 3F and G). These results are in line with a previous report showing Rab4⁺ endosomes rather than Rab11⁺ endosomes (slow-recycling) as the preferential recycling route for c-MET and suggest that increased accumulation of MYTX-011 mAb in c-MET⁺ cells can be attributed, at least in part, to reduced recycling (25).

We next evaluated whether the increased net internalization and cytotoxicity enabled by pH-dependent binding of MYTX-011 translated to improved antitumor activity in vivo. MYTX-011 efficacy was investigated in a variety of NSCLC xenograft models that exhibit a range of c-MET expression, characterized using an SP44 antibody based IHC diagnostic assay previously employed for patient selection in clinical trials (26–28). These models also differ in their EGFR, KRAS, and MET status representing the heterogeneity seen in patients with NSCLC (Supplementary Table S3). The NCI-H2122 xenograft was selected as a representative model for tumors with low c-MET expression, NCI-H1975 and NCI-H1373 xenografts for moderate c-MET expression, and EBC-1 xenograft for high c-MET expression (Fig. 4A).

Single-dose administration of MYTX-011 to mice with moderate c-MET⁺ adenocarcinoma NCI-H1975 (EGFR L858R–mutated and T790-mutated) or NCI-H1373 (KRAS G12C–mutated) xenografts resulted in greater tumor regression and growth delay than the parent ADC, whereas a nonbinding control ADC was inactive (Fig. 4B). These data demonstrate that engineering MYTX-011 for pH-dependent binding conferred significantly greater antitumor activity, in keeping with increased in vitro cytotoxicity and c-MET–dependent net internalization described above. In addition, MYTX-011 was more efficacious than a benchmark ADC given at an equivalent payload (MMAE) dose (Fig. 4B), which suggests that the antibody component of MYTX-011 achieves enhanced payload delivery compared with other anti–c-MET ADCs currently in development. Subsequently, dose titration studies using payload matched doses were conducted in NCI-H1975 and NCI-H1373 xenograft models to quantify the magnitude of enhanced efficacy compared with the benchmark ADC. MYTX-011 achieved at least threefold greater efficacy against NCI-H1975 and NCI-H1373 tumors, in which a single 2 mg/kg (19 µg/kg toxin) dose showed similar or greater efficacy than the highest dose of benchmark ADC tested at 3.8 mg/kg (56 µg/kg toxin; Fig. 4C).

In the squamous NSCLC EBC-1 xenograft model, which has MET gene amplification and high c-MET expression, a single dose of MYTX-011 eradicated tumors in all animals treated, even at the lowest dose tested (0.5 mg/kg and 4.75 µg/kg toxin; Fig. 4D). In contrast, the benchmark ADC administered at a matched toxin dose (0.33 mg/kg, 4.75 µg/kg toxin) caused only modest delay in tumor growth in this study. MYTX-011 also demonstrated antitumor activity in low c-MET–expressing NCI-H2122 xenograft model following a repeat dose of 6 mg/kg, whereas the benchmark ADC was inactive at an equivalent toxin dose (Fig. 4E). Together with the NCI-H1975 and NCI-H1373 dose titration studies described above, these results highlight the substantially greater efficacy of MYTX-011 across both moderate and high c-MET–expressing cell line–derived tumor models.

To evaluate MYTX-011 efficacy in additional biologically relevant NSCLC tumor models, we screened NSCLC PDX tumor tissue microarrays using IHC to assess c-MET expression levels. Five models were prioritized based on c-MET expression and EGFR/MET mutation status (Fig. 5A; Supplementary Table S3). CTG-3414, an EGFR-mutant, MET-amplified adenocarcinoma model, exhibited highly heterogeneous staining with <5% of tumor cells with high c-MET expression, whereas the rest of the tumor section was negative. Adenocarcinoma models CTG-2533 and CTG-1353 exhibited homogeneous moderate c-MET staining. CTG-2669, an adenocarcinoma with MET exon 14 skipping mutation, showed heterogeneous c-MET moderate staining. The squamous cell carcinoma model CTG-2082 was c-MET⁺ in the initial screen by tissue microarray but was determined to be c-MET negative in tumors isolated on study. Interestingly, CTG-3414 is derived from a patient who progressed on erlotinib and osimertinib (an EGFR inhibitor). Single-dose administration of 6 mg/kg MYTX-011 demonstrated antitumor activity in all c-MET⁺ models tested, suggesting the potential to effectively treat a broad range of NSCLC subtypes (Fig. 5B–E). MYTX-011 treatment (6 mg/kg) was not efficacious in the c-MET–negative CTG-2082 model (Fig. 5F). The nonbinding ADC exhibited negligible activity compared with MYTX-011 further demonstrating that the efficacy of MYTX-011 is c-MET–dependent. Taken together, in vivo efficacy studies demonstrated that MYTX-011 is highly active across a range of moderate and high c-MET–expressing NSCLC PDX models.

To assess the impact of pH-dependent binding on TMDD, PK of MYTX-011, parent ADC, and benchmark ADC were evaluated in cynomolgus monkeys, a setting in which these ADCs retain binding to c-MET in normal tissues. All animals were treated with a single dose of 2.7 mg/kg (n = 3 per group), and serial blood samples were collected over 28 days and analyzed by LC-MS/MS to measure total mAb and free MMAE levels. MYTX-011 total mAb exhibited PK consistent with a standard two-compartment model, with an increased half-life of 158 hours compared with the parent or benchmark ADCs, which showed accelerated clearance 10 and 7 days postdosing, respectively (Fig. 6A). Thus, engineering MYTX-011 for pH-dependent binding resulted in reduced TMDD and an extended half-life in monkeys. Total mAb levels were inversely related to the release of free MMAE with benchmark ADC releasing the highest free MMAE, followed by parent ADC and MYTX-011, respectively (Supplementary Fig. S3). The reduced level of free MMAE seen with MYTX-011 is consistent with its extended total mAb half-life, reflecting reduced uptake and processing in normal tissues. There were no MYTX-011-related clinical signs or changes in body weight, clinical chemistry, or hematology parameters.

Toxicity of MYTX-011 was assessed in a non-Good Laboratory Practice (GLP) cynomolgus monkey study (n = 3 per group). Animals were administered two doses of MYTX-011 on every 3-week schedule at 6, 12 or 18 mg/kg. Toxicokinetic analysis demonstrated that MYTX-011 was quantifiable up to 504 hours postdose on day 1, up to 72 hours postdose (the last timepoint) at 6 and 12 mg/kg on day 22, and up to 1,008 hours postdose at 18 mg/kg on day 22 (recovery; Fig. 6B). Total mAb and total ADC concentrations were similar at all doses tested, and data for 6 mg/kg dose are shown in Supplementary Fig. S4. Exposure of total mAb and total ADC was proportional to the dose. MYTX-011-related changes in hematology parameters included mild to marked decreases in neutrophils (Fig. 6C). Neutrophil counts following dosing at 12 or 18 mg/kg were decreased compared with baseline with nadirs on day 8 or 15 and generally increasing but still lower than baseline on day 22 for most animals, returning to near or above baseline counts on day 25. Neutrophil counts at 6 mg/kg/dose fluctuated but were generally within the range considered normal variation or near the individual baseline (Fig. 6C). No MYTX-011–related clinical observations, changes in body weight, qualitative food consumption, ophthalmology, coagulation, clinical chemistry, urinalysis, and macroscopic or organ weight changes were observed. Histopathologic assessment performed in this study yielded unremarkable results. The effect of MYTX-011 on neutrophils and the overall toxicity profile of MYTX-011 in monkeys are consistent with those of previously described site-specific DAR2 vcMMAE ADCs, with significantly increased tolerability compared with hinge-conjugated MMAE ADCs with higher DAR (29). Although c-MET is expressed at low levels in normal cells, we did not observe evidence for on-target toxicity to c-MET–expressing normal tissues in this study or in a subsequent multidose GLP toxicology study.

ADCs are an exciting modality today owing to major advancements in linker and payload technologies. Surprisingly, engineering the antibody component of the ADC has often been overlooked, and few have taken the approach of engineering antibodies to enhance ADC-specific mechanisms, namely, uptake and antibody/receptor trafficking. A c-MET–targeting ADC in late-stage clinical development, Teliso-V, has shown impressive efficacy in c-MET–high patients with NSCLC with nonsquamous histotype and EGFR (wild-type) genotype; however, response rates in tumors with moderate to low expression or histotype/genotype leave substantial room for improvement. Moreover, Teliso-V showed limited efficacy in patients with nonsquamous NSCLC using an every-3-week dosing schedule compared with an every-2-week dosing schedule (30). This is expected because exposure of c-MET–targeting antibodies (31) and therefore c-MET–targeting ADCs are negatively impacted by TMDD, necessitating more frequent dosing. We hypothesized that a pH-dependent anti–c-MET ADC would be released from c-MET inside the tumor cells directing more ADC toward the endolysosomal compartment while also mitigating TMDD and improving ADC exposure. Our goal was to broaden the efficacy of a c-MET–targeting ADC beyond a subset of c-MET–high patients and enable the treatment of additional patients with moderate and low c-MET expression.

We generated MYTX-011 by engineering pH dependence into a non–pH-dependent anti–c-MET parent antibody and conjugating it to vcMMAE at site-specific cysteines. MYTX-011 binds human and cyno c-MET with single-digit nM affinity at pH 7.4 and does not lose binding appreciably at pH 6.4 (Supplementary Table S1, corresponding to an acidic tumor microenvironment) but rapidly dissociates at pH 5.4 (corresponding to endolysosomal pH). Our in vitro assays showed up to a fourfold higher net internalization of MYTX-011 compared with the parent antibody ADC. Among the engineered antibody variants, we found that the degree of pH-dependent binding correlated strongly with the degree of increased antibody net internalization, with pH-dependent loss in binding at pH 5.4, which translated to increased cell uptake. c-MET has been shown to be a recycling receptor, with activated receptors able to traffic back to the plasma membrane after endocytosis (25, 32). By dissociating in the acidic endosomal environment, MYTX-011 may enable recycled c-MET to bind and internalize additional ADC molecules. After internalization, the pH-engineered MYTX-011 mAb showed less colocalization with the fast-recycling endosomal compartment compared with parent mAb, suggesting reduced recycling contributes to its greater accumulation in c-MET–expressing cancer cells. Precisely, how pH-dependent binding impacts MYTX-011/c-MET receptor dynamics and intracellular trafficking to increase ADC uptake in cells is the subject of ongoing and future studies.

MYTX-011 was broadly active in vitro across cell lines with moderate and lower levels of c-MET expression, consistent with the increased net internalization imparted by pH-dependent binding. Compared with the parent ADC, more cancer cell lines, especially those with moderate or low c-MET levels, were sensitive to MYTX-011 and had a greater depth of cell killing. Importantly, the increased cytotoxic potency of MYTX-011 was on-target and correlated with both c-MET RNA and protein expression across cell lines. The broad increase in MYTX-011 cytotoxicity indicates that the observed impact of pH-dependent binding on potency applies across diverse cancer cell lines and is not restricted to biology unique to a particular cell line or cancer type.

Engineering pH-dependent binding markedly improved the antitumor activity of MYTX-011 in vivo, in which its efficacy was substantially higher than that of the parent ADC. This was not driven by greater ADC exposure, as both MYTX-011 and parent ADCs did not cross-react with mouse c-MET (Supplementary Fig. S2). Importantly, MYTX-011 outperformed the c-MET–targeting benchmark ADC by at least threefold in xenograft models of NSCLC, further highlighting that pH engineering is a promising approach for best-in-class next-generation ADCs. MYTX-011 showed strong activity against CDX and PDX models selected to represent a range of NSCLC subtypes, including adenocarcinoma and squamous histology, EGFR and KRAS oncogenic mutations, and different levels of c-MET expression. Encouragingly, MYTX-011 was active against NSCLC tumor models with limited treatment options, such as squamous lung carcinoma, prior progression on EGFR kinase inhibitors, and moderate or low c-MET expression.

pH-dependent binding has been extensively explored for “naked” antibodies in which pH-engineered antibodies exhibit a longer half-life and therefore greater exposure over their nonengineered parent antibodies (6, 7, 33). Similar to the mechanism described for other pH-dependent antibodies with improved PK, we hypothesize that dissociation of MYTX-011 from c-MET in acidic endosomes of normal tissues expressing c-MET (such as endothelial cells) enables greater recycling by FcRn back into circulation. Accordingly, in cynomolgus monkeys, at a single dose of 2.7 mg/kg, MYTX-011 showed PK consistent with a standard two-compartment model, and an increased half-life of 158 hours compared with the parent and benchmark ADCs, suggestive of reduced TMDD. In a non-GLP toxicology study, MYTX-011 showed a dose proportional increase in exposure at higher doses. The longer half-life of MYTX-011 may allow for less frequent dosing, perhaps once every 3 weeks instead of a weekly or biweekly schedule currently implemented with many clinical-stage MMAE-conjugated ADCs, potentially enabling a differentiated safety profile, increased patient convenience, and combinations with checkpoint inhibitors.

We believe that engineering pH-dependent binding could be a general strategy for ADCs, particularly for targets that are expressed at medium to low levels in patients. This would allow greater delivery of toxins to tumor cells and unlock the use of lower potency payloads to limit off-target toxicity. This technology provides an opportunity to gain on-target ADC potency without changing parameters that have been proven to adversely affect ADC tolerability (e.g., increasing DAR and increasing intrinsic payload potency). A highly precedented ADC design, higher efficacy in CDX and PDX models, longer half-life, and lower free MMAE release in cynomolgus monkeys suggests that MYTX-011 could serve a wide variety of patients who otherwise have limited treatment options. MYTX-011 is currently being evaluated in a clinical trial (NCT05652868).

N. Gera reports a patent for WO2021/022039A1 pending and a patent for WO2022/169975A1 pending. B.P. Fiske reports a patent for WO2022/169975A1 pending and a patent for WO2021/022039A1 pending. No disclosures were reported by the other authors.

N. Gera: Conceptualization, resources, data curation, supervision, methodology, writing–original draft, writing–review and editing. K.M. Fitzgerald: Data curation, supervision, methodology, writing–review and editing. V. Ramesh: Data curation, writing–review and editing. P. Patel: Data curation, writing–review and editing. D. Kanojia: Data curation, writing–review and editing. F. Colombo: Data curation, writing–review and editing. L. Kien: Data curation, writing–review and editing. S. Aoyama: Data curation, writing–review and editing. L. Xu: Data curation, supervision, writing–review and editing. J. Jean: Data curation, writing–review and editing. A.M. Deshpande: Data curation, supervision, writing–review and editing. W.C. Comb: Data curation, supervision, writing–original draft, writing–review and editing. T. Chittenden: Resources, supervision, writing–original draft, writing–review and editing. B.P. Fiske: Conceptualization, resources, supervision, writing–original draft, writing–review and editing.