Comparative modelling and MD-derived energetic fingerprints identify engineering hotspots in Purpureocillium...

Subtilisin-like serine proteases (family S8) secreted by the nematophagous fungus Purpureocillium lilacinum are widely involved in virulence-related processes during nematode infection and are therefore potential targets for biocontrol enhancement. Here, nine extracellular S8A.139 proteases from P. lilacinum PLFJ-1 were analysed using a unified computational workflow that includes AlphaFold2 structure prediction, docking-based pose initiation, explicit-solvent molecular dynamics (MD) refinement, MM/PBSA ranking and per-residue decomposition, hydrogen-bond persistence analysis, residue interaction network mapping, and in silico mutational scanning with targeted MD validation. A short peptide (VAQGGAAGLA) was used as a standardised probe to compare peptide association across paralogs rather than as a confirmed native substrate. Using a consistent MM/PBSA protocol applied to the equilibrated 300-400 ns window, PL-S8P6 exhibited the most favourable peptide association estimate (ΔG_bind = -52.51 kcal/mol), while PL-S8P7 exhibited the least favourable (ΔG_bind = -17.80 kcal/mol). MM/PBSA estimates were derived from 1000 evenly spaced frames (stride 100 ps) and reported with block-averaged uncertainty to support the robustness of the comparative ranking. Residue-level analyses revealed distinct energetic and contact fingerprints across paralogs; notably, the D658Y substitution in PL-S8P7 increased interfacial hydrogen-bond persistence and improved the binding estimate within the protocol (ΔG_bind = -25.19 kcal/mol) without causing large-scale instability in MD. Collectively, these results offer paralog-resolved structural and energetic hypotheses that prioritise candidate residues for biochemical validation and structure-guided optimisation of P. lilacinum S8 proteases in nematode-related contexts.