Homology-based model building

PREDICTING THE THREE-DIMENSIONAL STRUCTURE OF PROTEINS BY HOMOLOGY-BASED MODEL BUILDING... 

ABSTRACT. A priori prediction of the 3-D stmcture of proteins remains an important unsolved problem. Of the theoretical methods employed today, only homology-based model building produces protein structures of sufficient accuracy to be of use in structure-fimction and related studies. The present work provides a discussion of the features of protein structure which bear on homology-based model building in addition to a detailed description of the homology-based approach, with examples drawn from our own work. 

In contrast, heuristic approaches make use of information on protein structure such as that available in the Brool aven Protein Data Bank (PDB). This information is often supplemented by energy-based procedures. Cuirently, the most reliable and hence most popular heuristic approach is homology-based model building [12,13], which seeks to predict the structure of a protein ("the target") from the known structure of a related protein ("the template"). Implicit in this approach is the notion that protein structure possesses inherent regularities which limit the variety of 3-D structures that can be formed firom related sequences of amino acids. 

To date, more than 14,000 protein sequences have been determined. This number contrasts sharply with the approximately 450 3-D structures that have been determined by x-ray crystallographic and NMR methods. Thus, development of reliable 3-D models by homology-based model building can provide useful structural information that will benefit a variety of structure-function studies. For example, site-specific mutagenesis experiments rely heavily on 3-D structure information, both for design and for the interpretion of results [14]. Information on 3-D structure can also significantly reduce the effort needed to design molecules which specifically influence a protein s function (see e.g. [15]). 

In Section 2, a review of the salient structural features of proteins is presented, followed in Section 3 by a discussion of homology-based model building. An example of the model-building process, as applied to Domain I of CD4 (CD4-I) - the T-cell surface protein [16] which binds to Human Immunodeficiency Virus (HIV), is presented to illustrate the various steps of the process. Section 4 discusses a number of issues which bear on the future of homology-based model building. 

The first step in homology-based model building involves the choice of an appropriate template protein related to the target protein. Two important issues in this choice are the degree of homology between the target and template proteins and the accuracy of the structure of the template protein bodt of these factors will influence the accuracy of the predicted structure. If the target protein is within the same family as one or more... 

Figure 8. Scheme for homology-based model building of protein structures. See text for further details. 

If the sequence of a protein has more than 90% identity to a protein with known experimental 3D-stmcture, then it is an optimal case to build a homologous structural model based on that structural template. The margins of error for the model and for the experimental method are in similar ranges. The different amino acids have to be mutated virtually. The conformations of the new side chains can be derived either from residues of structurally characterized amino acids in a similar spatial environment or from side chain rotamer libraries for each amino acid type which are stored for different structural environments like beta-strands or alpha-helices. 

---