Reassessment and constrained redesign of an AKT1–CTCFL multi-epitope construct for triple-negative breast cancer: a hypothesis-generating immunoinformatics study

Computational multi-epitope constructs require disease-relevant target selection, versioned HLA prediction, junction assessment and conservative structural interpretation. To reassess the original AKT1–CTCFL TNBC proposal and perform a constrained, reproducible sequence redesign. AKT1 and CTCFL were evaluated in a 40-gene receptor-confirmed expression panel comprising 158 TNBC tumors, 766 non-TNBC tumors, 113 TCGA adjacent-normal and 179 GTEx healthy-breast samples. Locked peptide sets were reassessed using NetMHCpan 4.1 BA and NetMHCIIpan 4.1 BA. All 120 CTL orders and 33 spacer combinations were tested, all 26 final junctions were scanned, official class I/class II/combined population-coverage reports were imported, and the exact final sequence was modeled with ColabFold and evaluated using pLDDT, PAE and DSSP. Neither AKT1 nor CTCFL was supported as TNBC-enriched: AKT1 was lower in TNBC than non-TNBC tumors, whereas CTCFL was detected at ≥ 1 TPM in only 1.27% of TNBC tumors and was lower than both normal comparators. The final 264-amino-acid construct retained 18/18 supported class I restrictions, 7/8 supported class II associations and all five CTL and five HTL segments. Matched multi-allelic very-strong junctional peptides decreased from 42 to 7 (83.33%), but residual junctional risk remained. World coverage was 60.15% for class I, 34.44% for the tool-supported class II subset and 73.87% combined. The model had mean pLDDT 47.84, mean PAE 26.98 Å and 71.21% of residues below pLDDT 50; 87.12% of residues were classified as coil/turn. Unsupported TLR4 docking, molecular-dynamics and C-IMMSIM claims were withdrawn. SPACER_COMBO_028 is a sequence-defined hypothesis, not a validated TNBC vaccine. Its target premise requires reconsideration, and experimental presentation, T-cell response, safety and antitumor testing are essential.