Rosetta-based Sensor Protein Design And Applications Of Network Pharmacology

Computational protein design and network pharmacology are two fast-developing new research fields in life sciences.Computational protein design could modify protein structure,property and function through energy function constrained rotamer optimization process,to fulfill specific design purpose.Network pharmacology could extract knowledge from vast amount of biological and pharmaceutical data using mathematics and physics methods,to guide target identification and drug discovery.The studies in this thesis focused on these two fields,constructed certain research methods and applied them to specific cases,provided new insights and valuable results for sensor protein design,mechanism study of Chinese herbal formulae and identification of lead compounds.The main content of this thesis was given below.The first chapter epitomized the research backgrounds,related methods and arrangements of the thesis.The second chapter tried to conceive rational design of cAMP and c-di-AMP sensor proteins.Both cAMP and c-di-AMP are second messengers in living organisms,and play important roles in physiological activities.The level of these messenger molecules could reflect live status of organisms,thus it is meaningful to design sensor proteins for detection of these messenger molecules.In this thesis,RosettaLigand and RosettaDesign were combined to optimize binding site amino acids during ligand-protein flexible docking process,thus to discover best mutation sequence.The first design was to enhance binding between Epac2 protein and cAMP.After rapid mutation generation,docking and binding mode analysis,mutations I79R,V85R,C86N,and A106H were found to be optimal sequence.The second design was to enhance the binding between c-di-AMP and c-di-GMP binding protein DGC.After similar computational process,mutations K45R,P46E,N48K,W80R,and E84R were the optimal sequence.Thus these residue sites may affect binding affinity between DGC and c-di-AMP/c-di-GMP.Corresponding experimental tests are currently in progress.A Chinese herbal formula usually consists of several herbs,thus may contain hundreds or even thousands of compounds,and possesses complex pharmacological effects.A challenging problem in scientific study of traditional Chinese medicine is how to analyze the formula's mechanism and identify active key compounds.Thus,in the third chapter,taking herbal formula Tian-Ma-Gou-Teng-Yin for the treatment of Alzheimer's disease(AD)as an example,we tried to investigate its molecular mechanisms through network pharmacology approach.Herbal compounds and marketed drugs along with their target information were first collected from open sources,and network-based models were then constructed.With our previously developed method bSDTNBI,364 new targets were predicted for 494 formula constituents.Twelve compounds were identified to be significantly related to AD through Fisher's exact test.After enriching the targets onto KEGG pathways,it was found that this formula might interact with several pathways,such as neuroactive ligand-receptor interaction pathway and inflammatory mediator regulation of TRP channels,and regulate key targets such as ACHE,HTR2A,NOS2 and TRPA1,to exert neuroprotective and anti-neuroinflammatory effects against the progression of AD.Furthermore,AD and hypertension disease modules were calculated in the context of protein-protein interaction network,overlapping submodules were hence determined,and they might be involved in the polypharmacological effects of herbal formulae.Prostate cancer is becoming one of the most common and lethal cancers among males.Androgen and androgen receptor(AR)play a key role in the progression of prostate cancer.One commonly used treatment for prostate cancer is androgen deprivation therapy.Thus it is meaningful for prostate cancer treatment to discover novel AR antagonists.In the fourth chapter,network pharmacology methods were used to identify potential targets for 93 available novel compounds.A global network model consisting of 12649 compounds and 1799 targets were first constructed,and the above-mentioned bSDTNBI method was used for target prediction of the 93 compounds.These compounds were clustered into 14 groups.Several groups of compounds were predicted to be potential AR ligands.Then known AR agonists and antagonists were collected from the DrugBank database,and similarity assessment was conducted between 93 compounds and those known agonists and antagonists,based on Wilcoxon rank sum test.Compounds in group 12 were identified to be most possible AR antagonists.Then yeast two-hybrid and luciferase methods were used to verify the prediction.M13,M14 and M15 from group 12 were verified as AR antagonists,with IC50 values at 4.969 ?M,3.446 ?M and 0.407 ?M,respectively.Further biological experiments and structural modifications are currently ongoing.The final chapter was a summary of the whole thesis.