Normal Modes Expose Active Sites in Enzymes

Abstract

Accurate prediction of active sites is an important tool in bioinformatics. Here we present an improved structure based technique to expose active sites that is based on large changes of solvent accessibility accompanying normal mode dynamics. The technique which detects EXPOsure of active SITes through normal modEs is named EXPOSITE. The technique is trained using a small 133 enzyme dataset and tested using a large 845 enzyme dataset, both with known active site residues. EXPOSITE is also tested in a benchmark protein ligand dataset (PLD) comprising 48 proteins with and without bound ligands. EXPOSITE is shown to successfully locate the active site in most instances, and is found to be more accurate than other structure-based techniques. Interestingly, in several instances, the active site does not correspond to the largest pocket. EXPOSITE is advantageous due to its high precision and paves the way for structure based prediction of active site in enzymes.

Fig 1. Accessible surface changes in normal modes.

Upon distortion along a normal mode, different pockets experience different changes of accessible surface. In the shown example, the accessible surface of pockets 1 and 2 does not significantly change. However, the accessible surface of pocket 3 significantly changes to a larger extent than pocket 4. The solvent accessible surface is represented as red squares.

Fig 2. Solvent accessibility changes in normal modes highlight active sites of enzymes.

Shown are nine EXPOSITE predictions for the enzymes (A) 1mbb, (B) 1dj1, (C) 1bvv, (D) 1pgs, (E) 1pmi, (F) 1sca, (G) 1lba, (H) 1a8h, and (I) 132l of the training dataset. The predicted and observed active sites are indicated by green and blue stars, and LIGSITE pockets are displayed as white spheres. In cyan and green are residues experiencing large accessibility changes in normal modes, and in blue, are residues experiencing little or no exposure change. Note that the predicted and observed active site are separated by less than 12Å. The figure was prepared using Pymol.

Fig 3. Line graph of distances between the predicted and observed active sites in the 845 enzyme test dataset.

The distances between the predicted and observed sites are plotted in blue, in green, and in red for EXPOSITE, ENSITE, and LIGSITE respectively. The distribution of distances is shown on a logarithmic scale, and emphasizes the added value of normal modes for prediction of active sites.

Fig 4. Solvent accessibility changes in normal modes highlight the ligand binding site of proteins.

Displayed are EXPOSITE predictions of nine proteins (A) 1inc, (B) 1bid, (C) 1hew, (D) 1hfc, (E) 1imb, (F) 1mrg, (G) 1mtw, (H) 1ulb, and (I) 1rob from the PLD database. The predicted and observed binding sites are indicated by green stars and red ligands respectively, and LIGSITE pockets are displayed as white spheres. In cyan and green, are residues displaying large changes of accessibility in normal modes, and in blue, are residues which display little or no change of exposure. Note that the ligand (in red) is within 4Å of the predicted site (green star). The figure was prepared using Pymol.

Fig 5. Failures to highlight the binding site of proteins.

Displayed are EXPOSITE predictions of four proteins (A) 1igj, (B) 3gch, (C) 3mth, (D) and 2tmn, from the PLD database. The predicted and observed binding sites are indicated by green stars and red ligands respectively. In orange, cyan, and green, are residues displaying large changes of accessibility in normal modes, and in blue, are residues which display little or no change of exposure. Note that EXPOSITE failed to predict the binding site in these cases due to multiple backbone breaks resulting in unusual modes (i.e. 3mth, 3gch), and to odd shaped protein structure (i.e. 1igj). The figure was prepared using Pymol.