AI-Driven Design of Miniproteins Targeting Mutant p53 for Cancer Immunotherapy

Genetic alterations in the tumor suppressor p53 occur in nearly every type of human cancer, with mutation rates ranging from 10% to nearly 100%, averaging about 50% in cancer patients. Over 80% of these mutations are missense mutations in the p53 DNA-binding domain (DBD), with the p53-R175H mutation being the most common "hot spot," occurring in approximately 5% of p53-mutated cases. Clinically, p53-R175H mutation in cancer patients are significantly associated with decreased overall survival compared to cancer patients with wild-type p53. Biologically, missense mutations in the p53 gene, particularly "hot-spot" mutations, including p53-R175H, result in the high expression of mutant p53 protein in cancer cells. These features suggest that expression of mutant p53-R175H protein could serve as an attractive therapeutic target. However, targeting the intracellular oncoprotein mutant p53-R175H has proven extremely challenging and remains a major unmet need in cancer therapy. Interestingly, the mutant p53-R175H protein can be processed by the proteasome into a 9-mer peptide, presented on the cancer cell surface by human leukocyte antigen (HLA)-A*02:01, a common major histocompatibility complex (MHC) class I allele in the U.S. population (40%), providing a “unique” target for immunotherapy by harnessing immune cells—such as T cells—to eliminate cancer cells carrying mutant p53R-175H. Traditional efforts to develop T cell receptors (TCRs) or monoclonal antibodies (mAbs) against the peptide-MHC-I complex have largely failed due to low affinity, limited specificity, and multi-year development timelines. Recent advances demonstrate that artificial intelligence (AI)-designed miniproteins (less than 150 amino acids) targeting peptide-MHC class I complexes can be engineered into chimeric antigen receptor (CAR)-T cells, enabling robust T-cell activation and potent recognition and elimination of tumor cells. To accelerate the discovery of miniprotein-mediated therapeutics, we have developed and integrated multiple AI-driven platforms for de novo design of p53-R175H peptide-MHC-I-complex binding miniproteins, including diverse backbone generation, protein sequence optimization, and enhanced high-throughput computational screening of peptide-MHC-I-miniprotein complexes. Top candidates have undergone, or are currently undergoing, experimental validation as engineered T/NK cell engagers and CAR-T/NK cells to kill cancer cells. In summary, our integrated AI-driven framework enables rapid and personalized cancer immunotherapy by transforming previously “undruggable” intracellular oncoproteins—including, but not limited to, mutant p53—into actionable immunotherapeutic targets. Citation Format: Fengze Jin, Puja Singh, Hanyong Chen, Christopher Warlick, Yibin Deng, . AI-driven de novo design of miniproteins targeting mutant p53 peptide-MHC-I complex for cancer immunotherapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 5477.