Structure-Guided Engineering of Novel Peptide Inhibitors Targeting PD-L1 Through Molecular Docking, MM-PBSA, and Molecular Dynamics Simulation

Abstract

Immune checkpoint blockade targeting the PD-1/PD-L1 axis has revolutionized cancer therapy, yet monoclonal antibodies suffer from high cost, limited tissue penetration, and immune-related toxicities. Peptide-based inhibitors offer a promising alternative due to their tunable specificity, improved tumor penetration, and ability to disrupt broad protein–protein interfaces. This study aimed to rationally engineer and computationally validate novel peptide inhibitors of PD-L1 with enhanced binding affinity, conformational stability, and translational feasibility.

A focused peptide library (P0–P9) was generated through targeted hydrophobic, aromatic, and electrostatic substitutions. Three-dimensional peptide structures were predicted using PEP-FOLD3 and refined under MMFF94 force fields. Docking was performed with AutoDock Vina (grid centered at PD-L1 interface residues Tyr56, Met115, Asp122; grid size 30 × 30 × 30 Å; exhaustiveness = 32) and HADDOCK for cross-validation. Top-ranked complexes were subjected to 100 ns molecular dynamics simulations in GROMACS 2023 using the CHARMM36 force field, TIP3P water model, and 0.15 M NaCl. Binding free energies were estimated via MM-PBSA with residue-level decomposition. Engineered peptides enriched with aromatic and hydrophobic residues demonstrated superior interface occupancy, persistent hydrogen-bond networks, and lower binding free energies compared with the parent scaffold. RMSD convergence (<2.5 Å), stable radius of gyration, and hydrogen-bond occupancy (>70%) confirmed dynamic stability. Residue decomposition highlighted Tyr56, Asp122, and Met115 as critical hotspots mediating peptide–PD-L1 recognition. MM-PBSA analysis revealed ΔG binding values ranging from −45 to −62 kcal/mol, with P7 and P9 outperforming parent peptide P0. Preliminary ADMET profiling indicated acceptable solubility, protease resistance, and low predicted immunogenicity.

Enhanced binding was driven by π–π stacking of aromatic residues, hydrophobic groove packing, and electrostatic complementarity with conserved PD-L1 residues. These features collectively stabilized peptide–protein complexes and improved inhibitory persistence.

This integrated computational framework establishes rational peptide engineering as a viable strategy for PD-L1 inhibition. The identified candidates exhibit strong binding affinity, structural stability, and favorable drug-likeness, supporting their potential as next-generation peptide-based immune checkpoint inhibitors. Future work should focus on experimental validation, peptide stapling, and advanced delivery strategies to accelerate clinical translation.