Book of Abstracts: Albany 2009

category image Albany 2009
Conversation 16
June 16-20 2009
© Adenine Press (2008)

Representing Protein 3D Structures in Spherical Coordinates -- Two Applications: 1. Detection of Invaginations, Protrusions and Potential Ligand Binding Sites; and 2. Separation of Protein Hydrophilic Outer Layer from the Hydrophobic Core

A Fortran 90 program was written to convert a protein 3D structure PDB file in Cartesian coordinates to spherical coordinates (rho, phi, theta), with the centroid (center of mass) of the protein molecule as origin. We investigated the utility of this representation in the (1.) detection of invaginations, protrusions and potential ligand binding sites (LBSs) on the protein surface, and (2.) separation of the hydrophilic outer layer (HOL) from the hydrophobic inner core (HIC) of protein molecules. The dataset of Laskowski et al., (Prot. Sci., 1996), composed of 67 single-chain protein structures, was used as test set in evaluating both applications. Both phi and theta are partitioned into suitable intervals (e.g., 6- and 8-degree intervals, respectively), giving rise to 1,350 phi-theta bins partitioning all of 3D space. The atom with maximum rho in each phi-theta bin is sequestered. In the first application, this step is done in both the liganded and unliganded form of the query protein and the frequency distribution of the maximum rho values from both forms are plotted superimposed on each other. Invaginations on the protein surface give rise to subpeaks or shoulders on the lagging side of the main peak, while protrusions give rise to similar shoulders, but on the leading side of the main peak. We find that most LBSs are associated with such subpeaks and therefore propose that such subpeaks are potential LBSs. In the second application, a suitable cutoff value for rho, e.g., 0.95rho, is adopted for each phi-theta bin: all atoms with rho values less than this cutoff value are considered part of the HIC, and those with equal or greater rho values part of the HOL. Except for a very few special cases, we show that all of the proteins in the Laskowski dataset, after undergoing our HOL-HIC separation procedure, give rise to an outer layer that is significantly more enriched in hydrophilic amino acid residues, an an inner core that is significantly more enriched in hydrophobic amino acid residues. In addition, a quick but effective way of determining active sites in the HIC and protein-protein interaction (PPI) interfaces in the HOL was derived. Once the HIC and the HOL are separated, the HIC may be searched for His, Glu, Asp, or Cys residues as potential active sites, and the HOL searched for clusters of hydrophobic amino acid residues as potential PPI interfaces (data not shown). We conclude that spherical coordinate representation of protein structures is a useful alternative to Cartesian coordinate representation, and may well find other useful applications beyond the ones described here.

Vicente M. Reyes

Dept. of Biological Sciences
Rochester Institute of Technology
Rochester, NY 14623

Ph: 585-475-4115