The Application Of Flex Pep Dock In Peptide-HLA Docking Researches

The interactions between flexible peptide and human leukocyte antigen(HLA) play a key role in mediating human immune responses. However, there are a small amount of pHLA crystal strctures available, and the structural determinations of pHLA complexes remain challenging due to the conformational complexity and diversity. Consequently, peptide docking and computational modelings of pHLA complexes are essential for studying the T cell immune mechanism.Due to the structural complexities and conformation flexibilities of peptide, peptide docking is one of the key issues and hot spots in the domain of molecular modeling. As one of peptide docking methods, Flex PepDock has been applied to an extensive benchmarking database and obtained high-quality models in a series of peptide-protein docking assays. In order to investigate the performance of Flex PepDock on pHLA modeling, two docking protocols of FlexPepDock, i.e., refinement and ab-initio, were applied to peptide-HLA and peptide- murine MHC docking studies. The docking results in this paper can provide important technical support for the researchs of peptide-HLA interaction.The main contents and results are as follows:①FlexPepDock refinement was applied to the docking studies of 30 nonapeptide-HLA-I complexes. Approximate starting conformations of target peptide can be templated by native peptides cocrystallized with the same and different HLA serotypes. The docking results showed that the near-native pHLA-I models with peptide bb-RMSD less than 2 ? were ranked in the top 1 lowest-energy model for 100%(70/70) docking cases. A total of 46 docking cases(46/70, 65.7%) sampled sub-angstrom bbRMSD models in the top 5 lowest-energy models and 10 docking cases sampled sub-angstrom all-atom models in the top 5 models. The all-atom RMSDs of the P2 and P9 dominant anchors reached, in most cases, sub-angstrom level. The Flex PepDock refinement has the ability to generate high-quality pHLA models when peptide was templated by native peptides cocrystallized with the same HLA serotype.②FlexPepDock ab-initio combined with Score12, Score13 and Score_ad were applied to the 15 peptide-HLA-I complexes docking simulations. The modeling results showed that 7 docking cases(46.6%) rank near-native models into the top 10 lowest-energy models by Score12, 6 docking cases(40%) by Score13, 9 docking cases(60%) by Score_ad. The Score_ad improved distinctly the models quality in both backbone and all-atom levels.③Flex PepDock refinement was applied to the docking simulation of 10 peptide-HLA-II complexes. The peptide was docked starting from an entire extended backbone conformation. Score12, Score13 and Score_ad were applied to the peptide docking cases. The docking results showed that 40% of cases, 50% of cases and 60% of cases obtained near-native models in the top 10 lowest-energy models, respectively. Score_ad, to some extent, improved the quality of pHLA-II models. In addition, the pre-optimization option combined with Score_ad obtained more accurate models, where 80% of docking cases sampled near-native models in the top 10 lowest-energy models.④Flex PepDock refinement was applied to the docking studies of 10 peptidemurine MHC complexes. The starting conformations of peptides were entire extended peptide chain. Score12, Score13 and Score_ad were applied to the peptide docking cases. The docking results showed that the 10% of cases, 40% of cases and 40% of cases ranked near-native models into the top 10 lowest-energy models, respectively. Besides, the results also showed that Score_ad can improve the quality of peptide-MHC models, but pre-optimization option did not result in better models, only 20% of docking cases obtained near-native models in the top 10 lowest-energy models.