Poster: Bioinformatics

Abs # 906: Molecular Modeling of b-Glucosidases in Arabidopsis thaliana

Presenter:	Bevan, David R,, drbevan@vt.edu
Authors	Bevan, David R, (A)   Cunningham, Joy A. (A)   Whitley, Katherine N. (A)   Zmuda, Jory  (A)   Shih, Ming-Che  (C)   Cheng, Chi-Lien  (C)   Poulton, Jonathan E. (C)   Mohamed, Ali  (D)   Winkel, Brenda  (B)   Esen, Asim  (B)
Affiliations:	(A): Dept. of Biochemistry, Virginia Tech (B): Dept. of Biology, Virginia Tech (C): Dept. of Biological Sciences, Univ. of Iowa (D): Dept. of Biology, Virginia State Univ.

Availability of the genome sequence of Arabidopsis thaliana revealed the presence of 46 putative family 1 b-glucosidases. We have undertaken the challenge of determining the structure and function of these enzymes. In this regard, we have applied comparative structure (homology) modeling using the MODELLER program. Template structures consisted of the plant b-glucosidases for which crystal structures are available (maize Glu1, white mustard Myr, and white clover CBG). The white mustard enzyme is a myrosinase, which is important in these modeling studies since 6 of the enzymes in Arabidopsis appear to be myrosinases. Models generated in MODELLER were subjected to energy minimization in AMBER. Examination of the models with PROCHECK revealed that none of the models had >1.5% of residues with f, y angles in disallowed regions of a Ramachandram plot, with most having <1%. The z-scores calculated by ProsaII, another indicator of the quality of the models, were in the range expected. Structural alignment of the models revealed essentially identical 3-dimensional structures, with the active site glutamate residues being nearly superimposable among all of the models. Calculations of the electrostatic properties provided an analysis of differences in the charge distribution among the models. These differences can be assessed qualitatively based on differences in amino acid sequence, but this computational approach provides a more refined and complete analysis. Further analysis of the structures was done with the GRID program, in which probe atoms with different properties were used to determine the physicochemical features of the active site. This analysis provides clues as to the properties of substrates that are likely to bind in the active site of these b-glucosidases.