由数据驱动的肿瘤抗原分析指导的逻辑门控Switch-DARPin免疫细胞衔接器：用于开发癌症免疫疗法的计算工作流程

Precise targeting of tumor-associated antigens (TAAs) and understanding of the tumor microenvironment (TME) are essential for the development of effective and safe immunotherapies. However, the lack of highly specific, clean TAAs remains a challenge for targeted immunotherapy of solid tumors. Switch-DARPins are engineered molecules designed to conditionally activate immune cells based on co-expression of a specific combination of TAAs, enabling localized logic-gated tumor-cell killing while minimizing off-tumor toxicity. To support the rational design of Switch-DARPins, we developed DARPin Compass, a computational workflow that integrates transcriptomic, proteomic, and structural data to identify optimal TAA pairs for dual-targeting strategies. The workflow evaluates TAA prevalence, tissue specificity, and DARPin compatibility, and identifies suitable models for preclinical validation. Additionally, it incorporates machine learning to identify candidate biomarkers for patient stratification and integrates TME profiling (including single-cell [sc] level) to assess antigen expression across TME cell compartments and evaluate therapeutic applicability. Moreover, the workflow identifies existing clinical data of the TAAs. DARPin Compass enables us to identify clinically actionable TAA pairs and extend the Switch-DARPin concept by including co-stimulatory signals. The workflow thereby supports the development of next-generation multi-specific immune cell engagers for solid tumors with complex antigenic and microenvironmental landscapes. Our first logic-gated Switch-DARPin immune cell engager, a CD3 T cell engager (TCE) with CD2 co-stimulation, was designed to be selectively active against solid tumors which co-express mesothelin (MSLN) and EpCAM (AND-gate). We found that MSLN and EpCAM show low co-expression in healthy tissues, while both TAAs are expressed in >50 % of ovarian cancer (OC) cells from in-house curated scRNAseq patient samples. Our DARPin Compass results captured the underlying patient heterogeneity and suggested that combinatorial targeting of both, EpCAM and MSLN, may lower the risk of on-target off-tumor toxicity compared to only targeting one of the TAAs. Our initial preclinical data validated this concept of logic-gated preferential tumor-targeting and supports further investigation. In summary, our DARPin Compass allows to identify potential novel TAAs, profile their expression patterns and assess their safety regarding expression in healthy human cells. Used to develop our first Switch-DARPin TCE for conditional CD3/CD2 activation in OC by targeting MSLN and EpCAM, this workflow is widely applicable to discover and characterize novel TAA pairs for logic-gated immune cell engagers to treat different solid tumors lacking selective single TAAs.