Proteomics-Guided Computational Prioritization of Putative RANKL-Binding Peptides from Proteins Identified in Deer Horn Glue

Deer Horn Glue is rich in collagen- and tissue-derived proteins, but the link between its experimentally observed proteome, the theoretical peptide sequence space derived from that proteome, and bone-related molecular targets remains poorly defined. This study established a proteomics-guided multiscale computational workflow to prioritize putative receptor activator of nuclear factor-κB ligand (RANKL)-binding peptide candidates generated from proteins identified in one Deer Horn Glue sample. The Deer Horn Glue proteome was characterized by liquid chromatography–tandem mass spectrometry (LC-MS/MS). Experimentally identified proteins were subjected to in silico tryptic digestion and stepwise activity, safety, and physicochemical screening. All 49 retained candidates underwent global and interface-focused docking. Five prioritized peptides were examined by AlphaFold 3, three independently seeded 200 ns molecular dynamics simulations per complex, entropy-omitted molecular mechanics/generalized Born surface area (MM/GBSA) analysis, residue decomposition, and locally relaxed computational alanine substitution analysis. Proteomic analysis retained 707 target protein groups and generated 24,575 nonredundant theoretical peptide sequences. Stepwise screening retained 49 candidates. Global and interface-focused docking rankings showed modest agreement, and the expanded interface analysis identified additional candidates while retaining GASLQDWDFGK as the top-ranked sequence. Across three independent simulations, GASLQDWDFGK showed consistently low peptide root-mean-square deviation (RMSD), whereas HEFSVDMTCEGCSNAVTR formed the largest mean number of interfacial hydrogen bonds. Their entropy-omitted MM/GBSA estimates were generally the most favorable, although the order varied among simulations. Locally relaxed alanine substitutions highlighted reproducible energy-sensitive positions for experimental testing. By linking an experimentally observed Deer Horn Glue proteome to multiscale structural analysis, this workflow provides a traceable and reproducible strategy for prioritizing testable peptide–RANKL interaction hypotheses. The five candidates remain theoretical products of in silico digestion and require targeted detection, direct binding, and functional validation.