Directed evolution-driven reprogramming of PD-L1 for compact and tunable checkpoint modulation.

BACKGROUND: Immune checkpoint inhibitors (ICIs) targeting the PD-1/PD-L1 axis have achieved major clinical success in cancer therapy. However, IgG-based checkpoint inhibitors face limitations such as large molecular size restricting tumor penetration and Fc effector-mediated depletion of PD-1-expressing T cells. While IgG antibodies can be engineered to modulate these effects, such tunability often requires complex modifications. To overcome these drawbacks, alternative protein scaffolds that retain checkpoint specificity while enabling more compact design and greater engineering flexibility are needed. Wild-type PD-L1, a native ligand of PD-1, exhibits micromolar binding affinity and limited therapeutic utility, highlighting the need for ligand-based engineering approaches that provide improved modularity and predictable receptor engagement. RESULTS: We developed a compact and tunable PD-L1 variant through yeast surface display-based directed evolution. A mutagenized PD-L1 ectodomain library was screened to isolate high-affinity variants, yielding a double mutant (DM) with a 173-fold enhancement in PD-1 binding over the wild-type. In silico immunogenicity analysis predicted that the DM variant maintains low potential immunogenicity, comparable to native PD-L1, despite the engineered mutations. Structural modeling revealed that this affinity gain was driven by a de novo interfacial salt bridge and enhanced hydrophobic complementarity at the PD-1 interface. Functional assays demonstrated that the DM variant restored T cell activity, as evidenced by increased interferon-γ and interleukin-2 secretion in co-culture systems. The DM variant retained favorable biophysical properties and is compatible with modular fusion formats for therapeutic integration. CONCLUSION: This study demonstrates the successful application of directed evolution to generate a compact PD-L1 variant with high-affinity PD-1 binding, low immunogenicity risk, and tunable architecture. The DM variant expands the design space of ligand-based checkpoint inhibitors and offers a versatile platform for next-generation immunotherapeutics engineered through synthetic biology frameworks.