Structural Fingerprints Reveal Which Tumor Mutations Drive Immune Rejection

Neoantigen-based cancer vaccines represent a promising immunotherapy strategy, but identifying which mutated peptides actually trigger immune-mediated tumor destruction remains a major bottleneck. This is especially challenging in high-mutational-burden cancers like cutaneous squamous cell carcinoma (cSCC), where thousands of missense mutations generate an enormous candidate pool. Standard methods prioritize neoantigens based on predicted MHC binding affinity and stability, yet only a small fraction of prioritized candidates generate meaningful T cell responses, and fewer still mediate actual tumor rejection.

To address this gap, researchers first analyzed publicly available human cSCC datasets spanning 149 tumors and found that shared neoantigens between patients were rare—supporting the need for personalized rather than shared-antigen vaccine strategies. They then generated a transplantable mouse cSCC model by exposing BALB/c mice to solar UV light, creating cell lines from resulting tumors. This model closely recapitulated the UV-signature mutational landscape and key driver mutations (including in Trp53, Ras pathway genes, and others) seen in human cSCC, and tumor growth was shown to be constrained by CD8+ T cells.

Using this model, the team systematically prioritized neoantigens using standard MHC-binding prediction tools and then validated candidates through ELISpot assays and prophylactic in vivo vaccination experiments. Two neoantigens were identified that mediated genuine tumor rejection. Mechanistically, one neoantigen exhibited improved MHC class I binding affinity compared with its wild-type counterpart—a classic immunogenicity driver. The second neoantigen, however, showed no significant improvement in MHC binding, yet it still rejected tumors. High-resolution 3D structural modeling of the peptide:MHC complexes revealed that in this second neoantigen, the mutated residue had substantially increased solvent accessibility compared with wild-type—meaning it was more physically exposed and available for T cell receptor (TCR) contact.

Critically, when the researchers extended this structural analysis to a curated set of known neoantigens from other published cancer models—specifically examining those that did not alter MHC binding—increased solvent accessibility of the mutated residue consistently distinguished tumor-rejecting neoantigens from non-immunogenic ones across different MHC alleles and cancer types. This finding suggests that structural modeling captures a dimension of immunogenicity that MHC-binding predictions alone cannot, namely whether the mutated portion of the peptide is physically accessible for TCR recognition.

The study has important implications for neoantigen vaccine design. Current pipelines frequently miss functional neoantigens because they over-rely on MHC binding metrics. Adding a structural modeling step—predicting solvent accessibility of the mutated residue within the peptide:MHC complex—could serve as an additional filter to more accurately identify which candidates will elicit productive T cell responses and tumor rejection. The authors also provide the research community with a well-characterized, clinically relevant mouse cSCC model featuring two defined neoantigens, offering a valuable platform for future mechanistic and therapeutic studies.