BACKGROUND

Advancing cancer immunotherapy requires engineering synthetic immunomodulators that integrate precise receptor targeting, tunable activity, and compatibility with modular biologic formats. The Inducible T-cell Co-Stimulator (ICOS) is a clinically validated co-stimulatory receptor whose engagement enhances T-cell function. However, the development of ICOS-targeting biologics has been hindered by limited receptor affinity and format-dependent agonist activity. To address this, we applied a protein engineering framework to optimize the ICOS ligand (ICOS-L) as a high-affinity, modular component for precision immune modulation.

RESULTS

Using yeast surface display-based directed evolution, we identified an ICOS-L variant (Y8) containing two synergistic mutations (Q51P and N57H) that improved human ICOS (hICOS) binding affinity by ~ 100-fold relative to wild-type. Structural modeling revealed that Q51P enhances backbone rigidity via a proline-induced conformational constraint, while N57H introduces a salt bridge with Asp86 in hICOS. These mutations reconfigure the receptor-binding interface to support high-affinity engagement. Functionally, Y8 induced potent T-cell proliferation and IFN-γ secretion. When genetically fused to pembrolizumab, Y8 further enhanced T-cell activation and tumor cell lysis, demonstrating synthetic synergy between PD-1 blockade and ICOS agonism. Among fusion formats, light-chain conjugation (pembrolizumab-L-Y8) exhibited superior functional output, highlighting the importance of geometric configuration in optimizing fusion-based agonism.

CONCLUSION

This study establishes Y8 as a high-affinity ICOS-L variant with robust co-stimulatory function, capable of potentiating anti-PD-1 immunotherapy through modular fusion design. The integration of Y8 into therapeutic antibody scaffolds provides a versatile engineering framework for the development of next-generation immunomodulatory biologics, offering opportunities to overcome resistance and enhance clinical efficacy in cancer immunotherapy.