Computer-assisted Virtual Screening Of Hepatocyte Nuclear Factor4Receptor Agonists

Diseases such as fatty liver, liver fibrosis, liver cancer and other chronic liverdiseases, as well as diabetes mellitus, are becoming epidemic in China. It is a seriousthreat to people’s health. Hepatocyte nuclear factor-4receptor is an orphan subfamily ofthe nuclear receptor superfamily of transcription factors that are required for hepatocytedifferentiation and development of the liver; they are expressed at high levels in the liver,and to a lesser degree in kidney, small intestine, colon, and pancreatic island. HNF4αand HNF4γ are the most important members of the HNF4subfamily. They have beenshown to share high amino acid sequence homology and similar tissue distributions.HNF4is a central regulator of the metabolism of carbohydrates and lipids, expression ofgenes critical for hepatocyte proliferation, liver function and xenobiotic metabolism.Abnormal HNF4expression or protein activity will induce diabetes, fatty liver, livercancer etc. On the contrary, enhanced expression by transfecting exogenous HNF4genecan reverse liver fibrosis, increase expression of liver-specific genes, decreaseinvasiveness of hepatocellular carcinoma. Therefore, it has been proposed thatstimulation of HNF4activity can effectively prevent and cure the above diseases. Thisdissertation focuses on virtual screening of agonists targeting mouse and human HNF4αand HNF4γ, which, in addition to helping the identification of their ligands, should alsocontribute to screening of new drugs to treat the relevant diseases.Here, we screened for effective agonists targeting members of HNF4subfamily inhuman and mouse with computational simulation method of homology modeling, virtualscreening and molecular docking approaches. First, amino acid sequences and threedimensional protein structures of human and mouse HNF4α and HNF4γ were searchedin the databases. It turns out that crystal structure about human and murine HNF4α and human HNF4γ is already known, while the structure of murine HNF4γ is not availableyet. The analysis based on sequence similarity showed that human and mouse HNF4γproteins share63.9%similarity, and in the ligand binding region, the homology is up to94.3%. Therefore, three-dimensional structure model of mouse HNF4γ was generated bythe MODELER program. Moreover, using Ramachandran plot and Protein3D programsto evaluate the geometric properties and compatibility of amino acid residues in thisstructural model shows that the predicted structure is credible. Then, to find targetreceptor-binding compounds, AutoDock Vina program was carried out to do virtualscreening of small molecules in the ZDD, ZMD, ZIM sub-databases of the parent ZINCdatabase. The resulting molecules were listed according to their affinity to the targetreceptor, and ZINC04026555and ZINC00577115were the two molecules ranked at thetop, showing the highest affinity to HNF4α and HNF4γ respectively. The strongestbinding agonist of human and murine HNF4α is the same, which could be a reflection ofthat they are highly homologous to each other; similarly, the human and murine HNF4γshare the same strongest binding agonist. Using AutoDock to compare the bindingcapacity of the two molecules identified in our study with the fatty acid positive controlidentified in crystal structure study indicate that they all bind to the same pocket of thereceptor protein. Furthermore, calculation using molecular docking of the binding freeenergy values revealed that the new agonists possess much higher affinity than thepositive long-chain fatty acid controls.