Computational Methodology Study On Protein-protein/Peptide Interactions And PROTACs

Protein-protein interactions(PPIs)play a crucial role in a variety of cellular processes,such as DNA replication and transcription,signal transduction,protein translation,etc.It has been estimated that 15-40% of PPIs within the cell are mediated by protein-peptide interactions(Pp Is).Therefore,elucidating the structural details of proteinprotein and protein-peptide complexes is fundamental for the study of molecular mechanisms for important biological processes and the development of therapeutics.However,it is still a great challenge to elucidate protein-protein and protein-peptide complex structures.Proteolysis-targeting chimeras(PROTACs),which selectively induce targeted protein degradation through the ubiquitin-proteasome system,represent an emerging drug discovery technique.Although numerous PROTACs have been reported,it is still quite difficult to design PROTACs with high potency and high selectivity.Therefore,how to accelerate the investigation of protein-protein/peptide recognition and rational PROTACs design by computational methods is an important research topic.In this study,we comprehensively evaluated the performance of docking programs in Pp Is and developed Hawk Dock,an online computing platform to predict and analyze the structures of PPIs by integrating protein-protein docking and MM/GBSA.Moreover,we proposed PROTAC-Model,a computational method based on protein-protein docking and MM/GBSA to predict PROTAC-mediated ternary complex structures.In addition,we constructed PROTAC-DB,a web-based open-access database of PROTACs.The main research contents and results are as follows:(1)In the first part of this thesis,we constructed a large reference dataset Pep Set consisting of 185 protein-peptide complexes with peptide lengths ranging from 5 to 20 residues.Based on this dataset,a total of 14 docking methods were evaluated,including three protein-protein docking methods,three protein-small molecule docking methods and eight protein-peptide docking methods.In addition,a new evaluation parameter,named IL＿RMSD(Root Mean Square Deviation of the Ligand in the Interface),was proposed to measure the docking accuracy of each method.The results show,for global docking,HPEPDOCK achieves the best performance and yields the success rates of 4.3%,24.3% and 55.7% at the top 1,10 and 100 levels,respectively.In local docking,ADCP performs best with the success rates of 11.9%,37.3% and 70.3% at the top 1,10 and 100 levels,respectively.(2)In the second part of this thesis,the online computing platform Hawk Dock was developed to predict and analyze the structures of PPIs by integrating the ATTRACT docking algorithm,the Hawk Rank scoring function and the MM/GBSA free energy decomposition analysis.For the structural modeling of PPIs,Hawk Dock achieves better docking performance than ZDOCK 3.0.2 in the benchmark testing.For the prediction of key interfacial residues,the prediction success rates for the predicted models and crystal structures in the benchmark dataset are ～81.4% and ～95.4%,respectively,when the top10 predicted residues are considered.In the Hawk Dock online server,all prediction and analysis functions are automated and it is accessible at http://cadd.zju.edu.cn/hawkdock/.(3)In the third part of this thesis,considering the importance of structural data of target-PROTAC-E3 ternary complexes for PROTAC design,PROTAC-Model,an integrative computational method by combining FRODOCK-based protocol and Rosetta Dock-based refinement,was developed to predict PROTAC-mediated ternary complex structures and tested on 14 cases.The quality of the models was evaluated using the criteria of CAPRI(Critical Assessment of Predicted Interactions).With the unbound structures as the initial structures,FRODOCK-based protocol can generate the ternary complex structures with medium or high-quality for 8 cases out of 14.With the refinement by Rosetta Dock,the cases with medium or high-quality can be increased to 12.Compared with PRosetta C and the methods developed by Drummond et al.,PROTAC-Model shows better performance in the modeling of target-PROTAC-E3 ternary complexes based on unbound structures.(4)In the fourth part of this thesis,in order to promote the rational design of PROTACs,we developed PROTAC-DB,a web-based open-access database that integrates the structural information and experimental data of PROTACs.After 2 versions of iterative upgrade,PROTAC-DB consists of 3270 PROTACs,365 warheads(small molecules that target the proteins of interest),82 E3 ligands(small molecules capable of recruiting E3 ligases)and 1501 linkers.In addition,given the importance and the limited number of the crystal target-PROTAC-E3 ternary complex structures,we used the PROTAC-Model modeling method to predict the corresponding ternary complex models for PROTACs with good degradation ability.PROTAC-DB is available online at http://cadd.zju.edu.cn/protacdb/.