Experimental validation of in silico methods to predict the recognition properties of active sites: the test-case of a non-heme iron dioxygenase

A non-heme iron dioxygenase, (namely catechol 1,2 dioxygenase -1,2 CTD-) was studied as a test-case to validate computational tools that would predict active site recognition properties. A 3D model of 1,2-CTD was generated by homology modeling and compared to crystallographic coordinates [1]. In addition, 15 active site variants were modeled introducing the alternative approach of rotamer library explorer. GRID and BIOCUBE4mf [2] were applied on the final models to extract some typical features from each mutant. Some of pre-existing experimental data on 1,2-CTD and active site mutants [3] were used to validate the reliability of the predictions. To confirm the applicability of this new strategy to identify novel mutations, two mutants were produced (I101A and I101V) that were predicted to have improved recognition on polychlorinated substrates. Steady-state kinetic studies confirmed that Ile101is crucial for substrate recognition and that alternative residues differently affect KM and kcat. In addition, two novel mutants, L69G and A72C, were also analyzed experimentally, confirming a good agreement with the in silico predicted behaviour. Marked differences were detected by GRID/BIOCUBE4mf in terms of chlorine interactions (CL-) among different mutants and these are in line with the trend of the experimental KM values. Also modulation of hydrogen bonding pattern was correctly predicted by the computational method. The approach is particularly suitable in focusing the effect of amino acid substitution, highlighting subtle differences that might greatly affect enzymological properties, and it is a good predicting tool for protein engineering as well as for selection of improved biocatalyst and study of mutational variants in medicinal chemistry.