Armored IL-18 CAR-T cells: bioinformatics-guided design for cold tumor infiltration

Abstract Chimeric antigen receptor (CAR) T cell therapy has shown remarkable efficacy in hematologic malignancies, yet its translation to solid tumors remains a fundamental challenge. Immunologically “cold” tumors, characterized by poor T cell infiltration, dense stromal barriers, and a profoundly immunosuppressive tumor microenvironment (TME), continue to resist conventional CAR-T cell approaches. Armored CAR-T cells engineered to secrete interleukin-18 (IL-18) represent a promising fourth-generation strategy designed to overcome these limitations. IL-18 operates at the interface of innate and adaptive immunity, driving M1 macrophage polarization, dendritic cell maturation, NK cell activation, and the induction of CXCL9/CXCL10 chemokine gradients that restore lymphocyte trafficking into tumor tissue. To guide the rational design of these constructs, bioinformatics pipelines play an increasingly central role. Multi-omics integration of transcriptomic, proteomic, and single-cell RNA sequencing data enables precise tumor antigen selection, epitope prediction, and structural optimization of CAR constructs using tools such as AlphaFold, RosettaDock, and NetMHCpan. In silico safety screening further evaluates off-target toxicity risks and cross-reactivity against healthy tissues. Controlled IL-18 delivery strategies, including NFAT-inducible promoters, SynNotch logic-gated systems, and Granzyme B-cleavable pro-IL-18 constructs, provide spatial and temporal safeguards against systemic hyperinflammation. Preclinical evidence from xenograft and syngeneic models demonstrates enhanced tumor control, sustained T cell persistence, and induction of endogenous anti-tumor immunity. This review synthesizes current evidence and highlights how bioinformatics-guided engineering of IL-18 armored CAR-T cells offers a translatable framework for converting cold tumors into immunologically responsive, therapeutically tractable targets.