Abstract A011: Characterizing the structure of the immune checkpoint B7-H7 to optimize immune activation with targeted therapeutics

Abstract Introduction: Targeting the PD-1/PD-L1 axis has shown remarkable success in treating renal cell carcinoma (RCC) patients, but additional strategies are needed for most patients. B7-H7 (also known as HHLA2) has emerged as an immune checkpoint and our group has reported that B7-H7 expression is non-overlapping with PD-L1 on RCC tumor cells. Furthermore, B7-H7 has both a stimulatory receptor, TMIGD2, as well an inhibitory receptor, KIR3DL3, expressed on NK and T cells. These receptors have distinct expression patterns in lymphocytes. B7-H7/KIR3DL3 blocking monoclonal antibodies (mAbs) have entered Phase I clinical trials and there is promising preclinical data with B7-H7 T cell engager and antibody drug conjugates. However, little is known how B7-H7 mAbs affect lymphocyte phenotypes or where mAbs bind B7-H7, as the crystal structure of B7-H7 is not known. We have developed and described B7-H7 mAbs with differences in TMIGD2 and KIR3DL3 blocking. Methods: In this study, we report the effects of B7-H7 expression in tumor cell cocultures with NK cells with and without anti-B7-H7 mAbs by single cell RNA-seq (scRNA-seq). We also computationally model the structure of B7-H7 and where blocking mAbs bind B7-H7. To assess the impact of B7-H7 or KIR3DL3 targeting mAbs, K562B7-H7-NK cell co-cultures were treated with anti-B7-H7, anti-KIR3DL3, or isotype control mAbs. Results: Two anti-B7-H7 mAb clones were used, 2C4 and 6F10, which are weak and strong blockers, respectively, of TMIGD2. Both clones showed decreases in inflammatory and cytotoxic phenotypes with XCL1+ XCL2+ NK cells being substantially reduced. Compared to the weak blocker clone 2C4, the strong blocker clone 6F10 resulted in lower inflammatory gene expression. This is consistent with the paradigm that B7-H7 provides a co-stimulatory signal through TMIGD2. Anti-KIR3DL3 had little to no effect relative to isotype control, suggesting that as expected KIR3DL3 has little global relevance in early NK cell activation, due to its rare expression. Using CytoSig, the XCL1+ XCL2+ NK cell subpopulation found across cocultured NK conditions was predicted to be highly responsive to IL-2. We modeled the structures of B7-H7 and its interactors using several structure prediction tools including AlphaFold3 and Boltz-2. Structure predictions of apo-B7-H7 across several different models showed beta-strand sharing between the IgC and IgV2, a uncommon architectural feature among Ig folds. Consistent with other groups’ published data, TMIGD2 and KIR3DL3 have distinct epitopes on B7-H7. TMIGD2 binds the IgV1 domain of B7-H7, whereas KIR3DL3 binds the IgC-IgV2 tandem. 6F10 and 2C4 were predicted to bind the IgV1 domain of B7-H7 consistent with observed TMIGD2 blockade. Conclusions: Our scRNA-seq analyses in this early NK activation assay suggest that B7-H7-targeting strategies may improve anti-tumor responses with IL2-based treatment and highlight the importance of sparing the TMIGD2-B7-H7 interaction to induce highly active NK cells. Citation Format: Michael Gomez, Deepthi Chowbene, Nahuel Perrot, Nikolaos Kalavros, Shuoshuo Wang, Antonella Arruda de amaral, David F. McDermott, Ioannis Vlachos, Gordon J. Freeman, Kathleen M. Mahoney. Characterizing the structure of the immune checkpoint B7-H7 to optimize immune activation with targeted therapeutics [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Innovations in Kidney Cancer Research: From Molecular Insights to Therapeutic Breakthroughs; 2026 Mar 13-16; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(5_Suppl_2):Abstract nr A011.