A Cell Surface-Binding Antibody Atlas Nominates a MUC18-Directed Antibody-Drug Conjugate for Targeting Melanoma

Abstract

Recent advances in targeted therapy and immunotherapy have substantially improved the treatment of melanoma. However, therapeutic strategies are still needed for unresponsive or treatment-relapsed patients with melanoma. To discover antibody-drug conjugate (ADC)-tractable cell surface targets for melanoma, we developed an atlas of melanoma cell surface-binding antibodies (pAb) using a proteome-scale antibody array platform. Target identification of pAbs led to development of melanoma cell killing ADCs against LGR6, TRPM1, ASAP1, and MUC18, among others. MUC18 was overexpressed in both tumor cells and tumor-infiltrating blood vessels across major melanoma subtypes, making it a potential dual-compartment and universal melanoma therapeutic target. AMT-253, an MUC18-directed ADC based on topoisomerase I inhibitor exatecan and a self-immolative T moiety, had a higher therapeutic index compared with its microtubule inhibitor-based counterpart and favorable pharmacokinetics and tolerability in monkeys. AMT-253 exhibited MUC18-specific cytotoxicity through DNA damage and apoptosis and a strong bystander killing effect, leading to potent antitumor activities against melanoma cell line and patient-derived xenograft models. Tumor vasculature targeting by a mouse MUC18-specific antibody-T1000-exatecan conjugate inhibited tumor growth in human melanoma xenografts. Combination therapy of AMT-253 with an antiangiogenic agent generated higher efficacy than single agent in a mucosal melanoma model. Beyond melanoma, AMT-253 was also efficacious in a wide range of MUC18-expressing solid tumors. Efficient target/antibody discovery in combination with the T moiety-exatecan linker-payload exemplified here may facilitate discovery of new ADC to improve cancer treatment.

Significance: Discovery of melanoma-targeting antibodies using a proteome-scale array and use of a cutting-edge linker-payload system led to development of a MUC18-targeting antibody-exatecan conjugate with clinical potential for treating major melanoma subtypes.

Figure. 1.

Screening melanoma cell surface mAb-target pairs for target nomination. A, Representative mAb-target pairs with an elevated expression in melanoma cells by flow cytometry. pAb253 and pAb3 are highlighted. B, Relative flow cytometry MFI ratio of representative mAbs on melanoma cells to other cancer cells. Ratio was calculated by mean relative MFI (to isotype control). C, Representative melanoma mAb-target pair immunofluorescence on cells. Protein class is labeled on top. Target protein name is shown after each mAb ID. IF shows endogenous (GAK) and target overexpression (colocalization of red mAb staining and green GFP-tagged target in 293T). Negative staining on the untransfected cells (blue only) is indicated by a red arrow. Scale bars, 20 μm. D, Representative endocytosis–mediated cytotoxicity for melanoma mAb-target pairs. Blue dashed lines separate data for three cell lines. mAb-target pairs show 80% cell killing (indicated by a red dashed line) for each cell line are labeled red. Cytotoxicity was conducted by binding of a vc-MMAE–conjugated anti-mouse IgG antibody to mAbs and incubated with the cells for 72 hours. Data are mean ± SEM from n = 2 independent experiments. E, MUC18 is expressed in both tumor cells and blood vessels (black arrows) in all subtypes of melanoma. Scale bars, 50 μm. F, Quantification (H-score) of IHC staining on TMAs of melanoma subtypes. IHC was conducted by a validated antibody N1238. G, MUC18 is a prognosis factor for relapse-free survival of patients with mucosal melanoma, N = 38. Unpaired two-sided t test; *, P < 0.05; ****, P < 0.0001.

Figure. 2.

MUC18-targeting ADC design and selection for clinical development. A and B,In vivo efficacy of AMT-253 and AMT-253-M in cutaneous (A375; A) and mucosal (GAK; B) melanoma xenograft models. One or two doses of ADCs at indicated doses were intravenously administered when tumor size reached an average of 150 to 200 mm³ and days (red arrows). Each value represents the mean and SEM (N = 5). MUC18 expressions in A375 and GAK are shown as an IHC images on untreated mouse tumor tissue and H-score. Scale bars, 20 μm. C, Representative hematoxylin and eosin staining of toxicity organs of AMT-253. Control was compared with AMT-253 at the highest dose of 60 mg/kg. Scale bar, 20 μm. D, Representative hematoxylin and eosin staining of MUC18-positive organs in the control and AMT-253 toxicity groups (60 mg/kg). Black arrows, blood vessels. Scale bars, 20 μm. E, Pharmacokinetics of AMT-253 in a cynomolgus monkey repeated dose-toxicity study. Total antibody, ADC, and released payload (exatecan) concentration was measured after three dosages of ADC. F, ADC half-life of AMT-253 and AMT-253-M from repeated dose-toxicity studies in cynomolgus monkeys. G, Therapeutic index of AMT-253 and AMT-253-M calculated from in vivo melanoma efficacy and monkey toxicity study.

Figure. 3.

Cellular dynamics, bystander killing effect, and mechanism of action of AMT-253. A, Correlation of exatecan release and target expression. MUC18 expression on each cell line was determined by flow cytometry and exatecan concentration in the culture media at 24 hours after treatment with 100 nmol/L AMT-253 that was determined by LC-MS/MS (N = 3). B, Bystander killing effect of AMT-253 in coculture conditions in vitro. SKOV3 and A2058 cells were cocultured and treated with 10 nmol/L ADCs for 5 days. After collecting adherent cells, cell number and ratio of MUC18‐positive and MUC18‐negative cells were determined by a cell counter and a flow cytometer, respectively. Each bar represents the mean and SD (n = 3). C,In vitro DNA damage induction and apoptosis. A375 cells treated with 10 nmol/L AMT-253 or controls for up to 72 hours were analyzed by Western blot for DNA damage (γH2A.X, pKAP1) and apoptosis (cleaved caspase-3) markers. D–G, γH2A.X and cleaved caspase-3 and -8 foci induction in the MUC18⁺ A375 tumor model by IHC analysis. Tumors were collected at the indicated time points and FFPE for IHC analysis. D, Representative images are shown for indicated days. Positive foci of γH2A.X, cleaved caspase-3 and -8 (black spots) are indicated by red arrows. Scale bar, 20 μm. E–G, Time course of γH2A.X, cleaved caspase-3 and -8 H-score (n = 3).

Figure. 4.

In vivo anti-melanoma efficacy of AMT-253. A, Scheme of in vivo efficacy studies in mouse CDX/PDX models. Dosages are indicated in each individual figure. B–D, Efficacy of AMT-253 as single agent in PDX models. B, A cutaneous melanoma (recurrent and resistant). C, An uveal melanoma. D, An acral melanoma. E, A summary of best tumor volume change (ΔTV) or tumor growth inhibition (TGI) of AMT-253 in melanoma models. CDX models are labeled with a #. The rest is PDX. MUC18 expression is shown as a heat map of H-score. The models (from left to right) were acral melanoma PDX_0234, mucosal melanoma CDX_GAK, uveal melanoma PDX_0208, cutaneous melanoma PDX_0222, and cutaneous melanoma CDX_A375. And the time points used for measurement were 32, 16, 46, 70, and 26 days after the administration of AMT-253. In vivo efficacy of AMT-253 and control ADCs in melanoma models. Tumor type is labeled and MUC18 expression in each model is shown as an IHC image on untreated mouse tumor tissue and H-score. Scale bars, 20 μm. Tumor-bearing mice were intravenously administered with indicated ADCs (10 mg/kg unless otherwise labeled) and on day 0 (tumor size reached an average of 150–200 mm³) and subsequent dates (red arrows). Each value represents the mean and SEM (N = 5). Unpaired two-sided t test. *, P < 0.05; ***, P < 0.001.

Figure. 5.

Targeting tumor vasculature with MUC18-targeting ADC alone or in combination with antiangiogenesis agent. A, pAb253 and Ab15A binds to human and mouse MUC18, respectively (no cross-reactivity). B–D, Distribution of AMT-253 (B) and Ab15A–T1000-exatecan (C) in GAK CDX. GAK xenograft mice (tumor volume at 200 mm³) were given 10 mg/kg of AMT-253 (B), Ab15A–T1000-exatecan (C), or Isotype IgG–T1000-exatecan (D). After 24 hours, green FITC-dextran (50 mg/kg) was injected into the mice 15 minutes before tumor harvest. Fresh-frozen sections were stained with red-colored Alexa594 conjugated anti-human IgG antibodies to visualize ADCs. AMT-253 was localized to tumor cells (red arrow in B) but not blood vessel (white arrow in B). Ab15A–T1000-exatecan was localized to blood vessel (white arrow in C) but not on tumor cells. No staining of IgG–T1000-exatecan on either tumor cells or blood vessels was observed (D). Scale bars, 20 μm. E, Antitumor efficacy of Ab15A–T1000-exatecan versus AMT-253 in GAK model. Left, tumor growth inhibition curve. Right, tumor volume at day 46 for each group. F, Combination treatment of AMT-253 and bevacizumab in GAK xenograft model. Left, tumor growth curve. Right, tumor volume at day 33 for each group. For in vivo experiment, mice were intravenously administered with indicated ADCs (10 mg/kg unless otherwise labeled) and on day 0 (tumor size reached an average of 150–200 mm³) and subsequent dates (red arrows). Each value represents the mean and SEM (N = 5 or 8). Unpaired two-sided t test. *, P < 0.05; **, P < 0.01; ***, P < 0.001. For combination treatment, the drug dose and schedule is labeled for each drug.

Figure. 6.

In vivo antitumor efficacy of AMT-253 in MUC18-expressing solid tumors. A–E,In vivo efficacy of AMT-253 in PDX models. A–C, Squamous cell carcinoma of cervix (cervical SCC; A), head and neck (H&N SCC; B), and esophagus (ESCC; C). D, Small-cell lung cancer (SCLC). E, Ovarian cancer (OVC). F, A summary of best tumor volume change (ΔTV) or tumor growth inhibition (TGI) of AMT-253 in solid tumors. CDX models are labeled with a #. The rest is PDX. MUC18 expression is shown as a heat map of H-score. Data include all models in this figure and from Supplementary Fig. S12. The models (from left to right) are PDX_0615, CDX_SCC-9, PDX_2755, PDX_2797, PDX_0617, PDX_0615 (squamous cell carcinoma of lung), CDX_Huh-7, PDX_0073, and PDX_2343. And the time points used for measurement were 39, 20, 21, 18, 28, 17, 22, 35, and 35 days after the administration of AMT-253. For all in vivo models, PDX tumor type is labeled. MUC18 in each model is shown as an IHC image on untreated mouse tumor tissue and H-score. Scale bars, 20 μm. PDX mice were intravenously administered with indicated ADCs (10 mg/kg unless otherwise labeled) and on day 0 (tumor size reached an average of 150–200 mm³) and subsequent dates (red arrows). Each value represents the mean and SEM (N = 4 or 5). Unpaired two-sided t test. *, P < 0.05; **, P < 0.01.