逻辑门控的switch-DARPin T细胞衔接器联合CD2共刺激，可提高对MSLN和EpCAM共表达卵巢癌的安全性和疗效

Introduction The development of T cell engagers (TCEs) for solid tumors is challenged by the scarcity of specific tumor-associated antigens (TAAs), which limits selectivity and increases the risk of on-target/off-tumor toxicity due to antigen expression in normal tissues. In ovarian cancer (OC), immunotherapy has shown limited success, largely due to the lack of “clean” TAAs, dysfunctional T cells, and an immunosuppressive tumor microenvironment. To overcome these challenges, we engineered a multi-specific, logic-gated CD3 TCE targeting the TAAs mesothelin (MSLN) and EpCAM, with CD2-mediated co-stimulation. This conditional Switch-DARPin design employs an AND-gate mechanism, enabling CD3 engagement only upon simultaneous recognition of both TAAs. This strategy aims to deliver potent, sustained antitumor activity while minimizing systemic toxicity and off-tumor effects on single TAA-expressing healthy cells. Methods The MSLNxEpCAM-targeting Switch-DARPin was developed using our DARPin platform. It features a masked CD3 binder that is unmasked upon dual binding to MSLN and EpCAM on tumor cells, a CD2 binder for co-stimulation, and includes a silent Fc domain for extended half-life. Preclinical safety and antitumor efficacy were evaluated in vitro and ex vivo using tumor and reporter cell lines, primary T and mesothelial cells from healthy donors, whole blood cytokine release assays, and in vivo using an OVCAR-3 xenograft model in PBMC-humanized NXG mice. Results Using our computational workflow DARPin Compass, we identified MSLN and EpCAM as a promising TAA pair in OC, with low co-expression in healthy tissues. In vitro , the Switch-DARPin induced selective T cell cytotoxicity against tumor cells co-expressing MSLN and EpCAM, with markedly reduced activity on cells expressing only one antigen. Notably, substantially lower activity against primary human mesothelial cells was observed, compared to other clinically tested MSLN-targeting TCEs. CD2 co-engagement promoted sustained activation and proliferation of T cells, preventing exhaustion. In vivo , the Switch-DARPin exhibited suitable pharmacokinetics and achieved significant tumor regression in MSLN+EpCAM+ xenografts without systemic cytokine release. The absence of cytokine induction in human whole blood assays confirmed the favorable safety profile despite CD2 co-stimulation. Conclusions Co-targeting of MSLN and EpCAM may lower the risk of on-target/off-tumor toxicity compared to single-targeting approaches. Preclinically, our logic-gated CD3xCD2 Switch-DARPin demonstrates selective antitumor activity against MSLNxEpCAM dual-positive tumors such as OC. CD2 co-stimulation further enhances T cell function without systemic activation. Together, our approach has the potential to overcome key limitations that TCE therapy faces in solid tumor indications.