Abstract A210: Tumor-associated ADAM10 and ADAM17 produce soluble PD-L1 (sPD-L1, sB7-H1) and affect downstream tumor immunity – A resistance mechanism to PD-1 checkpoint blockade in melanoma

Abstract

Abstract Background: Tumor surface PD-L1 (B7-H1) signals through T-cell surface receptor PD-1 to suppress antitumor immunity. PD-1 inhibitors like pembrolizumab block this pathway to prevent tumor escape from immunosurveillance. These PD-1 checkpoint inhibitors achieve objective response in 20-30% of patients with melanoma (1-3). As the majority of these patients are resistant to this treatment, there is a critical need to combat PD-1 checkpoint inhibitor resistance in melanoma. What is the mechanism of PD-1 checkpoint inhibitor resistance in melanoma? Our group has reported that soluble PD-L1 (sPD-L1, sB7-H1) is elevated in the serum of patients with PD-1 inhibitor-resistant melanoma and non-small cell lung cancer (NSCLC) (4). We have also shown that soluble PD-L1 is produced by cleavage from other tumor types and leads to apoptosis in CD8+ T-cells (5). The aim of this study is to confirm the source of sPD-L1 in human melanoma and its relation to outcomes. Methods: B7H1-expressing transgenic melanoma cells (Mel-B7H1) were treated with an array of protease inhibitors over 24 hours. Cells were then pelleted and supernatants were analyzed by ELISA for sPD-L1 (sB7-H1) production. Tumor serum samples were also analyzed by ELISA for sPD-L1 production.Tumor pathology slides from patients with melanoma (serum donors from the above study) were stained by immunohistochemistry with antibodies against proteases ADAM10 or ADAM17. Slides were visualized by microscopy for the presence of these antigens. Activated human CD8+ T-cells were cultured with recombinant human B7-H1 (PD-L1) protein, sPD-L1-rich, or sPD-L1-depleted supernatants from a tumor cell line in the presence of placebo or PD-1 checkpoint inhibitor antibodies. Cell survival was measured at 48 hours by flow cytometry.Results: ADAM10 inhibitor GI254023X, ADAM17 inhibitor TAPI-0, and broad metalloprotease inhibitor TAPI-2 inhibited Mel-B7H1 production of sB7H1 (p<0.001, 0.095, and 0.001 respectively). Specific inhibitors of other metalloproteases did not significantly alter sB7H1 production. PD-L1 transcription levels were not significantly altered. Tumor pathology slides from patients with high serum sPD-L1 exhibited high ADAM10 and/or ADAM17 staining. Conversely, tumors from patients with low-sPD-L1levels tended to exhibit low ADAM10 and/or ADAM17 staining. Recombinant PD-L1 and PD-L1-rich supernatants induced cell death of CD8+ T-cells, whereas PBS control solution and PD-L1-depleted supernatants did not. Supraphysiologic doses of pembrolizumab decreased cell death in PD-L1-rich supernatant-treated CD8+ T-cells. Conclusions: Matrix metalloprotease inhibitors ADAM10 and ADAM17 mediate sPD-L1 production from melanoma cells. In samples from patients with melanoma, immunohistochemical staining of these proteases in tumor tissue correlates with serum sPD-L1 levels. As we have shown previously, sPD-L1 induces apoptosis of activated CD8+ T-cells, which may explain PD-1 inhibitor resistance in most patients with melanoma.We posit that sPD-L1 outcompetes PD-1 inhibitors to suppress tumor immunosurveillance. Our results suggest that the combination of ADAM10/17 inhibitor and PD-1 checkpoint inhibitor may overcome this mechanism of PD-1 resistance in melanoma and other cancers. Future studies will determine (A) how a combined ADAM10/17 inhibitor and PD-1 inhibitor may function in clinical use, (B) how tumor-derived sPD-L1 signals downstream to cause apoptosis in T-cells, and (C) whether surface ADAM10/17 predict surface PD-L1 staining in melanoma.