INTEGRATED STRUCTURAL ANALYSIS OF BACTERIAL OMPA/OMPA-LIKE PROTEINS FOR THE IDENTIFICATION AND PRIORITIZATION OF HIGH-CONFIDENCE LIGAND-BINDING SITES

Bacterial outer membrane proteins of the OmpA/OmpA-like family play critical roles in structural integrity, biofilm organization, and host-pathogen interactions. In periodontal and opportunistic oral infections, their involvement in tissue colonization and immune evasion highlights them as promising targets for novel anti-virulence strategies. Given the rising challenge of antimicrobial resistance, targeting non-essential virulence mechanisms represents a promising alternative. This study evaluated six OmpA/OmpA-like proteins from key oral pathogens - Q9S3R9 (Porphyromonas gingivalis), O51841 (Aggregatibacter actinomycetemcomitans), G8UJN3 (Tannerella forsythia), M2CB51 (Treponema denticola), B9CZU2 (Campylobacter rectus), and A0A250FSS9 (Capnocytophaga gingivalis) - to identify and prioritize potential ligand-binding sites using complementary in silico approaches. Protein sequences were retrieved from UniProt, and 3D models were obtained from the AlphaFold Protein Structure Database. RamPlot showed >92% of residues in favored Ramachandran regions (threshold >90%). ERRAT2 Overall Quality Factor ranged from 84.69% to 93.59% (>80% cutoff). PROCHECK showed 99–100% of residues in allowed regions, with G-factors between −0.03 to 0.08 (acceptable limit ≥−0.5). QMEANDisCo global scores ranged from 0.52 to 0.72 (>0.50 threshold), while MolProbity scores ranged from 0.99 to 1.42, all below the high-quality cutoff of 1.5. WHAT_CHECK indicated satisfactory structural consistency without critical abnormalities. In contrast, none of the models met the classical VERIFY3D criterion of ≥80% of residues with an average score ≥0.1. Together, these validation results supported the structural reliability of the models for ligand-binding site prediction. Ligand-binding sites were predicted using three complementary approaches: hotspot mapping (FTMap), geometry-based pocket prediction (PrankWeb/P2Rank), and cavity detection with druggability assessment (CavityPlus). Spatial and residue-level convergence among all three approaches was observed for Q9S3R9, O51841, M2CB51, and A0A250FSS9, whereas B9CZU2 displayed partial convergence. M2CB51 exhibited the highest drug binding potential, displaying a P2Rank score of 45.96 (probability 0.954), FTMap crossclusters with up to 16 probes from four chemical classes, and a CavityPlus DrugScore of 4722 (druggability cutoff ≥600). In conclusion, this integrated computational workflow provided robust, high-confidence identification of druggable pockets, establishing a prioritized structural framework for rational drug design against oral pathogens.