TP53 mutations occur in over 50% of human cancers and generate shared neoantigens, including p53R273H, which represents a compelling target for precision immunotherapy. However, HLA allele-specific presentation and the risk of off-target toxicity remain major challenges. Here, we developed a group of bispecific T cell engagers (TCEs), namely, TCE01 and its optimized versions, based on a stimulated patient-derived TCR that specifically recognizes p53R273H presented by HLA-C∗01:02. The immunodominant 9-mer epitope SFEVHVCAC was identified as selectively presented by this allele. Incorporation of the stapled single-chain TCE formats reduced molecular weight by ∼24% compared with full-length constructs, while fully preserving binding affinity and markedly enhancing in vitro T cell activation and cytotoxicity (>90% at 0.625 nM). Machine learning-guided directed evolution using Boltz-2 and EvotProtGrad yielded TCE01Cr3, exhibiting a Kd of 2.51 nM and a subpicomolar EC50. Alanine scanning combined with combinatorial peptide library screening and structural modeling pinpointed peptide residues 2 and 4 as critical for TCR recognition in context of TCE01. TCE01Cr3 exhibited remarkable in vitro functional stability following 128 h of incubation in human serum and demonstrated a prolonged in vivo half-life in mice (t1/2 = 39.40 ± 10.50 h). In 3D model, monotherapy with TCE01Cr3 eliminated >80% of p53R273H-expressing colorectal cancer cells within 48 h while sparing non-malignant stroma, independent of chemotherapy potentiation (oxaliplatin or irinotecan). This off-the-shelf TCE provides a framework for precision immunotherapy in HLA-C∗01:02-positive patients with p53R273H tumors.
Nhan T. Huynh, Thuy Anh Thi Nguyen, Ngoc Thi Bui et al.· Molecular therapy. Oncology· 0 citations
Identifying which peptides bind major histocompatibility complex (MHC) molecules is central to vaccine design, neoantigen prioritization, and precision immunotherapy. Existing deep learning predictors largely encode amino acids as discrete symbols, thereby missing the residue-level chemistry driving molecular recognition. Performance also tends to degrade under class imbalance, for rare alleles, and on peptide– MHC combinations outside the training distribution. We developed CALFP-MHC, a framework that encodes each amino acid as a set of complementary cheminformatics fingerprints capturing functional groups, atomic connectivity, and substructural features, and combines positional encoding with supervised contrastive pre-training to organize the latent space by binding class before fine-tuning a binary classifier. Peptide–MHC interactions are modeled through a hybrid convolutional-transformer backbone. In a large-scale computational benchmark covering ∼18.7 million peptide–MHC pairs across 112 HLA class I and 53 class II alleles, CALFP-MHC achieved AUCs of 0.93-0.97 and PPVs of 0.66–0.94. Critically, performance remained above AUC 0.90 even at a 200:1 negative-to-positive ratio, where competing tools frequently collapsed toward chance. On independent experimental data containing 3,627 class I and 520 class II MS/MS-confirmed ligands and 570 validated neoantigens, the model maintained strong discrimination, correctly prioritizing immunogenic peptides and MHC-presented ligands. Attention and integrated-gradient analyses recovered established anchor positions (P2 and PΩ for class I, P1, P4, P6, and P9 for class II) and highlighted chemically interpretable functional groups consistent with known binding determinants. CALFP-MHC demonstrates that grounding residue representations in molecular chemistry, rather than sequence symbols alone, improves both robustness and interpretability in peptide–MHC binding prediction.
My-Diem Nguyen Pham, T. Ho, H. Nguyen et al.· bioRxiv· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
