Carbamazepine activates private self and viral reactive TCRs through a drug-permissive HLA-B cleft.

Introduction: Carbamazepine (CBZ) can cause severe cutaneous adverse reactions, including Stevens-Johnson syndrome/toxic epidermal necrolysis, and is strongly associated with HLA risk alleles. However, HLA genotype alone incompletely predicts clinical risk. Adverse drug reactions depend on drug-peptide-HLA conformation, immuno-regulatory cell interactions and the presence of drug-specific TCR clonotypes in susceptible patients. An unresolved question is whether patients that develop a specific drug reaction harbor the same drug-specific TCR (public TCR) or generate patient unique TCRs (private TCR). Methods: To address this controversy we expressed four published CBZ-associated TCRs including two public and two private clonotypes, in TCR-null TG40 cells and tested activation using monoallelic 721.221 APCs expressing HLA-B*15:02, HLA-B*57:01, HLA-B*58:01, or HLA-B*58:01(TE→MA), as well as HLA-transgenic/MHC I-deficient murine splenic APCs. T-cell activation was quantified by PD-1, CD69, and 4-1BB, while LC-MS/MS immunopeptidomics/MAPPs was used to assess drug effects on the peptide repertoire. CBZ docking was applied to define drug-peptide-MHC-TCR interactions in a drug reactive peptide specific TCR. Results and discussion: Public TCRs showed no CBZ-dependent activation, whereas private TCRs displayed dose-dependent, HLA class I-dependent responses supported by HLA-B*15:02 and unexpectedly HLA-B*57:01, but not HLA-B*58:01. Amino acid substitutions to HLA-B*58:01 (T45-E46 to M45-A46) mapped drug presentation to restore CBZ responsiveness without altering HLA expression. CBZ did not detectably induce broad abacavir-like peptide motif alterations but amplified HIV Gag KF11 peptide-driven T cell activation in an allele- and mutation-dependent manner. Modeling predicted a candidate CBZ accessible region beneath the peptide near HLA Met67, HLA Tyr9, and a conserved CDR3α tyrosine providing a plausible structural rationale for allele and clonotype-dependent activation. These findings support a model in which CBZ potentiates binding of selected peptide-HLA-TCR interfaces governed by HLA cleft architecture and clonotype-level TCR sensitivity. This mechanism explains why HLA screening alone incompletely predicts CBZ-SJS/TEN risk and highlights peptide-HLA-TCR functional context as a potential determinant for improved risk assessment and mechanistic drug hypersensitivity testing.