Peters K P, Fauck J, Frömmel C
Humboldt-University of Berlin, Medical Faculty (Charité), Institute of Biochemistry, Germany.
J Mol Biol. 1996 Feb 16;256(1):201-13. doi: 10.1006/jmbi.1996.0077.
The biological function of a protein typically depends on the structure of specific binding sites. These sites are located at the surface of the protein molecule and are determined by geometrical arrangements and physico-chemical properties of tens of non-hydrogen atoms. In this paper we describe a new algorithm called APROPOS, based purely on geometric criteria for identifying such binding sites using atomic co-ordinates. For the description of the protein shape we use an alpha-shape algorithm which generates a whole family of shapes with different levels of detail. Comparing shapes of different resolution we find cavities on the surface of the protein responsible for ligand binding. The algorithm correctly locates more than 95% of all binding sites for ligands and prosthetic groups of molecular mass between about 100 and 2000 Da in a representative set of proteins. Only in very few proteins does the method find binding sites of single ions outside the active site of enzymes. With one exception, we observe that interfaces between subunits show different geometric features compared to binding sites of ligands. Our results clearly support the view that protein-protein interactions occur between flat areas of protein surface whereas specific interactions of smaller ligands take place in pockets in the surface.
蛋白质的生物学功能通常取决于特定结合位点的结构。这些位点位于蛋白质分子表面,由数十个非氢原子的几何排列和物理化学性质决定。在本文中,我们描述了一种名为APROPOS的新算法,它完全基于几何标准,利用原子坐标来识别此类结合位点。为了描述蛋白质形状,我们使用一种α形状算法,该算法生成一系列具有不同细节程度的形状。通过比较不同分辨率的形状,我们在蛋白质表面发现了负责配体结合的空洞。在一组具有代表性的蛋白质中,该算法能正确定位分子量在100至2000道尔顿之间的配体和辅基的所有结合位点的95%以上。只有在极少数蛋白质中,该方法才会在酶活性位点之外找到单个离子的结合位点。除了一个例外,我们观察到亚基之间的界面与配体结合位点相比呈现出不同的几何特征。我们的结果清楚地支持了这样一种观点,即蛋白质 - 蛋白质相互作用发生在蛋白质表面的平坦区域,而较小配体的特异性相互作用则发生在表面的口袋中。