Dynamics Of Dihydrogen Bond In The Electronically Excite State And Rational Drug Design Based On Molecular Modeling

As an emerging technology based on theoretical and computer science, computational modeling has been involved in most of the modern scientific researches. Computational modeling is able to provide deeper insights which are hardly derived from the wet experiments for small chemical molecules or even the bio-macromolecules. In this dissertation, computational chemistry and computational biology were both applied to understand the intermolecular hydrogen (dihydrogen) bonding in both the chemical complex and the bio-macromolecules and the structure-function relationship. Based on the quantum chemistry and bio-macromolecule simulations, a novel and efficient ligand-based virtual screening approach was developed using the HWZ scoring function and an enhanced shape-density model.The dissertation is structured by the following parts:Firstly, the current research status and the future prospects of molecular modeling are introduced as well as its significance for chemistry and biochemistry. The theoretical backgrounds and the popular softwares of molecular modeling were all introduced.Secondly, the dynamics of intermolecular hydrogen bond and dihydrogen bond involving phenol and2-pyridone were theoretically studied using quantum chemistry. Density functional theory and time-dependent density functional theory were applied to investigate their geometric structure, infrared spectrum, electronic transition between low lying excited states and ground state. Upon electronic excitation, all the hydrogen/dihydrogen-bonded complexes are locally excited where only phenol or2-pyridone moiety is electronically excited. Analyzing their changes between ground state and excited state, it can be infered that the intermolecular hydrogen/dihydrogen bond involving phenol are both strengthened upon electronic excitation; whereas both of these intermolecular interactions involving2-pyridone become weaker in the excited state. The carbonyl group of2-pyridone moiety destabilizes the complex in the electronically excited state so that2-pyridone is not an ideal proton donor for dihydrogen bond in the electronically excited state. Moreover, the relationship between the coexistent intermolecular hydrogen and dihydrogen bonds was also investigated. For the first time, we found that the coexistent intermolecular hydrogen and dihydrogen bonds are favorable to stabilize each other in the ground state. However upon the electronic excitation, they become competitive against each other. Thirdly, extensive molecular modeling combining the normal mode analysis and molecular dynamics simulations were carried out to figure out how the N-terminal moiety of Leishmania RNA helicase protein (LmeIF) moiety initiates adjuvant effects. The computational results demonstrated that the LmelF structure may exist in two different forms corresponding to the extended and collapsed states of the entire structure. The structure of LmeIF tends to undergo large fluctuations in a concerted fashion. The large fluctuations can strongly affect the solvent accessible surface of the epitope on the N-terminal structure. The conformational freedom of the C-terminal domain explains why the entire LmeIF protein is not as active as the N-terminal moiety. Thereafter, a comparative genome analysis following by homology modeling and molecular electrostatic potential (MEP) techniques allowed us to predict a novel and plausible RNA helicase from the Listeria source (LI-helicase) with adjuvant property as observed for LmelF protein. The structural folding and MEP maps revealed similar topologies of the epitope of both LmeIF and LI-helicase proteins as well as the striking identity in the local disposition of the charged groups.Finally, we aimed to develop a new ligand-based virtual screening approach SABRE(?) based on an effective scoring function HWZ and an enhanced molecular shape-density model for the ligands. The performance of SABRE(?) has been tested against the40targets in the Database of Useful Decoys. The virtual screening performance was evaluated in terms of the area under the receiver operator characteristic curve, enrichment factor, and hit rate. The virtual screening results using shape-density model (multi queries) demonstrated a favorable improvement and effectiveness compared with the performance which just concerned HWZ score (mono query). Our multi-queries method is suitable for virtual screening and yields a prediction accuracy superior to the study using the data fusion of multi queries.Concerning deeper on the chemical information of the set of active ligands, our method is less dependent on the choice of the query. Therefore, our novel ligand shape-based screening method constitutes a robust and efficient approach to the3D similarity screening of small compounds and opens the door to a new approach to drug design by implementing this method in the structure-based virtual screening.