| With the development of structural biology, a large number of3D static protein crystal structures are reported and provide valuable insights into structure based drug design. However, only using these protein crystal structures in drug design has some disadvantages. Only in some rare cases, protein motions are limited, and the drug molecules fit into a fairly static binding pocket like a key fits into a lock. In most conditions, molecular recognition and drug binding are very dynamic processes. In this process, the motions are crucial for drug molecule binding. The static protein crystal structures can not provide sufficient information to describe this process, which is not beneficial to improve the accuracy and reasonableness of structure-based drug design. In this condition, it is necessary to carry out MD simulation to study the dynamic process of drug molecule binding to proteins. Therefore, more and more researchers use molecular dynamics simulation method to describe the dynamic changes when drug molecules bind to proteins. So far, molecular dynamics simulation has been successfully applied in protein and drug molecule conformation sampling, free energy calculation, as well as pH-dependent protein-ligand interactions. Besides the above applications, only using protein crystal structures in drug design have other problems, which can be solved by molecular dynamics simulation:(1) It is not beneficial to find the reasons why minor differences in protein structures can affect drug (or substrate) binding by only using protein crystal structures;(2) It is not beneficial to find key residues which only contribute to agonist (or antagonist) recognition by only using protein crystal structures;(3) It is not beneficial to quantitatively determine the contributions of residues to drug molecule binding. The aim of this research is to discuss the above three issues. The main content and results are as follows:In the first chapter, we first point out the disadvantages of using protein crystal structures in drug design and suggest that molecular dynamics simulation can solve these problems to some extent. Then we introduce some molecular dynamics simulation applications in drug design, including:identifying cryptic and allosteric binding sites; sampling protein conformation; sampling drug conformation; calculating the binding free energy between protein and drug molecules; refine protein crystal structures; simulating pseudo-receptor model; researching pH-dependent protein-ligand interactions. After that, we put forward our scientific problems:Besides the above applications, only using protein crystal structures in drug design have other problems, which can be solved by molecular dynamics simulation. Finally, we give a brief introduction of the whole thesis.In the second chapter, some basic theories of molecular dynamics simulation used in this thesis are introduced, which include basic process of molecular dynamics simulation, commonly used molecular mechanics force field, integration methods, temperature coupling methods, pressure coupling methods, periodic boundary conditions and bond constraint algorithms.In the third chapter, the research focuses on human peroxisome proliferator activated receptor-y (PPAR-y). The aim of this work is to determine the residues which interact with rosiglitazone can be affected by Y473point mutations. To achieve this, molecular dynamics simulations of wild type (WT) PPAR-y/rosiglitazone/coactivator complex and its two mutants (Y473A and Y473F) were carried out. The trajectories were then subjected to carry out hydrogen bond probability and interaction energy analyses. Hydrogen bond probability analysis shows that point mutations can strongly disturb hydrogen bond interactions of rosiglitazone with S289, H323, H449and residue473, which leads to reduced hydrogen bond strength between rosiglitazone and these four polar residues. The analysis of the interaction energies between rosiglitazone and some residues in PPAR-y suggests that the weakened interactions of rosiglitazone with F282, Q286and Y327in Y473mutants can lead to decreased binding affinity between PPAR-y and rosiglitazone. The enhanced interactions of rosiglitazone with R288and1341in both of the two mutants, as well as the increased interactions of rosiglitazone with S342in Y473A mutant can make AF-2unstable, which further results in reduced transcriptional activity. These finding indicate that molecular dynamics simulation can be used to find the reasons why minor differences in protein structures can affect drug (or substrate) binding. In the fourth chapter, the research focuses on human protein tyrosine phosphatase1B (PTP1B). The aim of this research is to explore residues which interact with phosphotyrosine substrate (PTR) can be affected by D181point mutations and lead to increased substrate binding. To achieve this goal, molecular dynamics simulations were performed to simulate the interactions between PTPIB and phosphotyrosine substrate in wild type, D181A and D181E PTP1B/substrate complexes. The trajectories were then subjected to carry out RMSD, RMSF, dynamic cross-correlation matrix (DCCM), principal component analysis (PCA), MM/PBSA, hydrogen bond probability and interaction energy analyses. RMSD, RMSF, DCCM, PCA analyses suggest that D181A can have influence on the backbone movement of P loop, WPD loop regions and other residues in the active site of PTP1B. D181E only affects the backbone movement of P loop and WPD loop regions. The results of MM/PBSA calculation shows that D181A and D181E point mutations can enhance the binding affinity between PTP1B and substrate, which is accord with the previous experimental results. Comparing the individual components contributing to the binding free energy, it can be concluded that the electrostatic interaction energies (△Eeie) and polar contributions to solvation free energy (△Gpol) dominate the change in the binding strength. Further hydrogen bond probability and interaction energy analyses indicate that D181A can enhance the interactions of substrate with Y46, K120, F182and G218. D181E can strengthen the interactions of substrate with E181and S216. These finding indicate that molecular dynamics simulation can be used to find the reasons why minor differences in protein structures can affect drug (or substrate) binding.In the fifth chapter, the research focuses on human peroxisome proliferator activated receptor-a (PPAR-a) agonist13M and antagonist471. The aim of this work is to find key residues which only contribute to agonist (or antagonist) recognition. To achieve this goal, molecular dynamics simulations were used to simulate the interactions of PPAR-a with agonist13Mand antagonist471. The trajectories were then subjected to carry out hydrogen bond probability and interaction energy analyses. In hydrogen bond probability analysis, residues which form more stable hydrogen bonds with agonist than with antagonist will be considered to only participate in agonist recognition. On the contrary, residues which form more stable hydrogen bonds with antagonist than with agonist will be considered to only participate in antagonist recognition. The results suggest that all of the five residues form more stable hydrogen bonds with agonist13M than antagonist471. In interaction energy analysis, residues which only exhibit strong interactions with13M will be considered to have selectivity for agonist recognition, while residues which only exhibit strong interactions with471will be considered to have selectivity for antagonist recognition. The results indicate that the interactions of agonist13M with C275, Q277, T279, S280, Y314, L321, V332, H440and Y464are much stronger than antagonist471, while the interactions of antagonist471with1272,1317and1354are much stronger than agonist13M. In conclusion, it can be deduced that S280, Y314, H440and Y464only participate in agonist recognition, which is accord with the previous reports. Besides these four polar residues, our research suggests that C275, Q277, T279, L321and V332are only involved in agonist recognition, while1272,1317and1354only contribute to antagonist recognition. These finding indicate that medicinal chemists can use molecular dynamics simulation to find key residues which only contribute to agonist (or antagonist) recognition.In the sixth chapter, the research focuses on human dipeptidyl peptidase-IV (DPP-IV) inhibitors PS4and B1Q. The aim of this work is to determine the residues which make great contributions to inhibitor binding by comparing the interactions between different inhibitors and residues in DPP-IV. Based on the DPP-IV/inhibitor crystal structure alignment analysis, the DPP-IV inhibitor design protocols can fall into two types. We select B1Q (IC50=6.8nM) and PS4(IC50=0.38nM) as representatives. The residues which meet the following criteria can make greater contributions to inhibitor binding than other residues in DPP-IV:(1) Residues which exhibit strong interactions with both two inhibitors;(2) Residues which interact with PS4much stronger than B1Q. Molecular dynamics simulations were used to simulate the interactions of DPP-IV with these two inhibitors. The trajectories were then subjected to carry out hydrogen bond probability and interaction energy analyses. The results suggest that E205, E206, Q553, Y547and Y662make significant contributions to inhibitor binding. This is consistent with the previous experimental studies. F357, S552, K554and Y666also make great contributions to inhibitor binding. Medicinal chemists can make compounds form strong non-bonded interactions (H-bonds, salt bridges, π-π stacking interactions, etc) with them. R125and N710contribute little to inhibitor binding. These finding indicate that molecular dynamics simulations can be used to quantitatively determine the contributions of residues to inhibitor binding.The seventh chapter summarizes the main content of present research, and proposes the innovation. The further research of this scholar field is also expected. |
PDF Full Text Request