Molecular Modeling And Dynamics Simulation Of Abacavir-HLA-B*5701-peptide Complex

Recently, more and more genetic association studies have identified strongassociations between idiosyncratic adverse drug reactions (ADRs) and apecific HLAalleles. The most attracting associations is abacavir-induced hypersensitivity syndromein the Caucasian population, which is usually observed in patients carryingHLA-B*5701genetypes. However, limitation of crystallographic technology bringsgreat difficulties in exploring the mechanisms of ADRs. In this paper, the3-D structureof abacavir-B*5701-peptide complex is established by molecular docking andmolecular dynamics simulation methods. Also, the impacts of abacavir onpeptide-binding specificity of HLA-B*5701a are evaluated by predicting binding freeenergy. The methods and techniques adoptted in this paper can provide importanttechnical support for the researches of other drug-induced ADRs, and be important fordrug R&D and safety evaluation. The main contents and results are as follows:①Based on the VHSE (Principal component score vector of hydrophobic, steric,and electronic properties) description method, support vector machine (SVM) togetherwith genetic algorithms (GA) is employed to establish classification models ofpeptide-binding affinities of HLA-B*5701and B*5801. The sensitivity, specificity, areaunder receiver operating characteristics curve (AUC), and matthews coefficient ofcorrelation (MCC) of the optimal linear SVM model are77.29%,93.99%,93.02%,67.65%for HLA-B*5701, and78.08%,89.62%,88.34%,64.73%for HLA-B*5801.The sensitivity, specificity, AUC, and MCC of test set are80.02%,94.53%,94.62%,72.09%(B*5701), and77.43%,90.79%,87.98%,66.20%(B*5801). Based on theweights of variables of the optimal classification models, detailed comparisons ofpeptide-binding affinities of HLA-B*5701with B*5801are also carried out. The resultsobtained by this paper can provide an important reference basis for the research ofabacvair-induced SADR.②By using surflex-dock method,1238drugs are docked into the peptide bindinggrooves of HLA-B*5701, B*5702, B*5703, and B*5801alleles respectively. Bycomparing the optimal conformations and interaction modes, the results show thatabacavir can bind to B*5701by specific H-bond, van der Waals and hydrophobicinteractions with the E and F pockets of B*5701. Also, the binding of abacavir toB*5701don’t interrupt entirely further binding of peptide to the peptide binding groove of B*5701.③The peptide (HV9) is docked into abacavir-HLA complex by surflex-dock. Theobtained abacavir-HLA-HV9complex is further stablized by molecular dynamicssimulation. The lowest energy comformation of abacavir-HLA-HV9complex canreproduce related crystal structure with the RMSD of2.6. Furthermore, the bindingfree energy of abacavir to HLA complex are evaluated by MM-PBSA and PMM-GBSAmethods. The results indicate that the association between abacavir and B*5701wasstructurally dertermined by H-bond, van der waals, and hydrophobic interactions. Theresults of free energy decomposion show that residues Tyr74, Val97, Asp114, Ser116,Ile124are important for the interations between abacavir and B*5701.④The impacts of abacavir on peptide binding specificity of HLA-B*5701. Byusing MM-PBSA and MM-GBSA methods, binding free energies of abacavir toB*5701-HV9and19one-site mutated structures are precdicted. By comparing thebinding free energies of abacavir to20complexes, the results show that, after treatmentwith abacavir, B*5701prefers Asn、Pro、Val、Trp、Ala at C-terminal. The conclusionsobtained in this paper are basically consistent with the experimental results.