Computational Simulations Based On Protein-protein Interactions And Its Application To The Discovery Of WD40 Family Proteins And CXCR2 Receptor Active Compounds

Protein-protein interactions(PPIs)are inseparable from most processes of vital activities,such as DNA replication,transcription and translation,cell cycle regulation and signaling transduction,cell-cell adhesion and communication,protein synthesis and degradation.Mutations or abnormal expression of some key targets can lead to dysregulation of PPIs,which in turn can cause the occurrence and development of various diseases such as cancer,infectious diseases and neurodegenerative pathologies.Therefore,drug design targeting PPIs as PPIs can be an important complement to direct intervention targets and can be developed as an effective drug intervention strategy for the treatment of diseases.In the past,PPIs modulation through small molecules is generally considered an impossible task.This is mainly due to the lack of a well-defined ligand-binding pocket on the shallow PPI surface which can be targeted by small molecules.With the discovery of interferable PPI targets and the study of "hot spot" amino acid residues at the PPI interface,the drug design of small molecule antagonists based on PPIs is gradually becoming possible.However,the discovery and optimization of small-molecule antagonists of PPIs are challenging because the PPIs interface is too flat and classical drug discovery methods are often less applicable.With the application of high-throughput screening,fragment-based drug design,structure-based drug design and virtual screening based on computer technology to the design of PPI antagonists,especially the successful development of bcl-2 antagonist Venetoclax(ABT-199),researchers have gradually developed some theoretical and practical experiences for the discovery and optimization of PPI antagonists,which further accelerates the development of PPI antagonists.In the thesis,the computer-aided drug design(CADD)method has been utilized in the PPIs pocket studies of different targets,containing(WD repeat domain 5)、WSB1(WD repeat And SOCS box-containing protein 1)、CXCR(CXC chemokine receptor)1/2.Computational simulations targeting the PPI interface of these proteins were carried out by means of protein structure prediction,molecular docking,and molecular dynamics simulations.For different targets,virtual screening models and protein antagonistic selectivity models were established respectively for the discovery of antagonists,covalent antagonists and selective antagonists.The first part of this article focuses on computational simulations and virtual screening of two proteins of the WD40 family.Virtual screening models were built for the PPI interfaces of WDR5 and WSB1 which were applied to the discovery of WDR5 and WSB1 antagonists.WDR5 protein virtual screening model establishment and its application: a dataset was constructed from WDR5 antagonists with known activity and the generated Decoys molecules to evaluate the screening and discrimination ability of different software and WDR5 crystal structures for WDR5 antagonists.based on the evaluation results,three different virtual screening models-the Le Dock screening model,the multi-method combination screening model,and the covalent screening model-were also developed for screening WDR5 antagonists: 1)Virtual screening of a chemdiv compound library for WDR5 targets was performed by the Le Dock software.2)A Bayesian-based QSAR model and the screening methods of Glide SP and Pose filter were introduced to improve the screening accuracy and screening rationality compared to the first model;3)The screening model for covalent antagonists was based on the exposed261-position cysteine in the PPI interface of WDR5 as the Michael addition site,and the screening was performed by the combination of Structure Filtering,Glide SP,Pose Filter,Covalent Docking and other techniques.Using the established virtual screening model,we conducted a screening of Chemdiv and covalent compound libraries to obtain compounds with reasonable binding conformation and novel backbone for potential WDR5 activity.The results of the bioactivity tests showed that most of the molecules obtained from the screening did not possess WDR5 antagonistic activity,with SZY-27 showing weak covalent antagonistic activity in the tested species.Further molecular dynamics simulations showed that this molecule has potential for further modification and optimization.In addition,we have modeled and conducted virtual screening of WSB1,a novel target against tumor metastasis.In response to the current lack of crystal structure of WSB1,we constructed the three-dimensional structure of WSB1 using homology modelling,machine learning and molecular dynamics simulation.Further work has been carried out on protein-peptide docking,protein-small molecule docking and molecular dynamics simulations to identify "hotspot" interfaces on WSB1 for targeting.Based on these results,we conducted a virtual screening of WSB1 antagonists using Autodock vina and Le Dock software.A series of potential WSB-1 active compounds with novel structures were obtained finally.The second part of this paper is a computational simulation-based work on CXCR2 selective antagonists based on PPIs.First,we used the crystal structure of CXCR2 in the RCSB database to evaluate the ability of different molecular docking software to reduce the antagonist binding conformation in CXCR2 crystals and to determine the computational simulation parameters.The three-dimensional structure of the CXCR1 protein was then predicted with relative accuracy by means of homology modeling and molecular dynamics simulations.Based on the crystal structure of CXCR2 and the predicted CXCR1 protein structure,a set of predictive and analytical models for CXCR2 antagonist selectivity is developed in this paper.The model first could achieve the differentiation of CXCR2 selective antagonists,and its differentiation ability was verified by literature molecules.Molecular dynamics simulations of literature molecules and CXCR1 and CXCR2 were performed,and this paper identified key amino acid residues ASP84,ARG144 and LYS320 on CXCR2 that are associated with antagonist selectivity,of which the side chain length of LYS320 is the main factor affecting antagonist selectivity.In this paper,the validated antagonist selectivity model was applied to the evaluation of active molecules in the laboratory,and a unique selective antagonism mechanism of the antagonist C5 molecule,which showed the best CXCR2 selectivity(910-fold),was identified.To further validate the antitumor efficacy of C5 molecules,we carried out a validation of the antitumor biological activity of C5.The experiments showed that the C5 molecule has strong anti-tumor migration activity and can also effectively antagonize CXCR2 downstream Akt and ERK1 / 2 signaling pathways,which can be used for further development and design.