MOLECULAR MODELLING OF HUMAN ALDEHYDE OXIDASE AND IDENTIFICATION OF THE KEY INTERACTIONS IN THE ENZYME-SUBSTRATE COMPLEX

Aldehyde oxidase (EC 1.2.3.1), a cytosolic enzyme containing FAD, molybdenum and iron-sulphurcluster, is a member of non-cytochrome P-450 enzymes called molybdenum hydroxylases which isinvolved in the metabolism of a wide range of endogenous compounds and many drug substances. Drugmetabolism is one of the important characteristics which influences many aspects of a therapeutic agentsuch as routes of administration, drug interaction and toxicity and therefore, characterisation of the keyinteractions between enzymes and substrates is very important from drug development point of view. Theaim of this study was to generate a three-dimensional model of human aldehyde oxidase (AO) in order toassist us to identify the mode of interaction between enzyme and a set of phethalazine/quinazolinederivatives. Both sequence-based (BLAST) and inverse protein fold recognition methods (THREADER)were used to identify the crystal structure of bovine xanthine dehydrogenase (pdb code of 1FO4) as thesuitable template for comparative modelling of human AO. Model structure was generated by aligningand then threading the sequence of human AO onto the template structure, incorporating the associatedcofactors, and molecular dynamics simulations and energy minimization using GROMACS program.Different criteria which were measured by the PROCHECK, QPACK, VERIFY-3D were indicative of aproper fold for the predicted structural model of human AO. For example, 97.9 percentages of phi andpsi angles were in the favoured and most favoured regions in the ramachandran plot, and all residues inthe model are assigned environmentally positive compatibility scores. Further evaluation on the modelquality was performed by investigation of AO-mediated oxidation of a set of phthalazine/quinazolinederivatives to develop QSAR model capable of describing the extent of the oxidation. Substrates werealigned by docking onto the active site of the enzyme using GOLD technology and then HASL methodwere used to generate a 3D-QSAR model. Correlation coefficient (r2) between the test set actual andpredicted Km values was 0.65.