Proteins comparative model

A major problem in predicting protein structure is the computational intractability. A short, 100-residue protein will contain at least 100 side-chain-to-side-chain or side-chain-to-solvent interactions. The orientation of each of these interactions will lead to cascading effects throughout the protein. Comparative modeling, threading algorithms, and de novo predictions seek to predict protein structure in reasonable execution times. 

At this time, approximately one-half of all sequences are delectably related to at least one protein of known structure [8-11]. Because the number of known protein sequences is approximately 500,000 [12], comparative modeling could in principle be applied to over 200,000 proteins. This is an order of magnitude more proteins than the number of experimentally determined protein structures (—13,000) [13]. Furthermore, the usefulness of comparative modeling is steadily increasing, because the number of different structural folds that proteins adopt is limited [14,15] and because the number of experimentally determined structures is increasing exponentially [16]. It is predicted that in less than 10 years at least one example of most structural folds will be known, making comparative modeling applicable to most protein sequences [6]. 

All current comparative modeling methods consist of four sequential steps (Fig. 2) [5,6]. The first step is to identify the proteins with known 3D structures that are related to the target sequence. The second step is to align them with the target sequence and pick those known structures that will be used as templates. The third step is to build the model... 

Figure 1 The basis of comparative protein structure modeling. Comparative modeling is possible because evolution resulted in families of proteins, such as the flavodoxin family, modeled here, which share both similar sequences and 3D structures. In this illustration, the 3D structure of the flavodoxin sequence from C. crispus (target) can be modeled using other structures in the same family (templates). The tree shows the sequence similarity (percent sequence identity) and structural similarity (the percentage of the atoms that superpose within 3.8 A of each other and the RMS difference between them) among the members of the family.

This section briefly reviews prediction of the native structure of a protein from its sequence of amino acid residues alone. These methods can be contrasted to the threading methods for fold assignment [Section II.A] [39-47,147], which detect remote relationships between sequences and folds of known structure, and to comparative modeling methods discussed in this review, which build a complete all-atom 3D model based on a related known structure. The methods for ab initio prediction include those that focus on the broad physical principles of the folding process [148-152] and the methods that focus on predicting the actual native structures of specific proteins [44,153,154,240]. The former frequently rely on extremely simplified generic models of proteins, generally do not aim to predict native structures of specific proteins, and are not reviewed here. 

Although comparative modeling is the most accurate modeling approach, it is limited by its absolute need for a related template structure. For more than half of the proteins and two-thirds of domains, a suitable template structure cannot be detected or is not yet known [9,11]. In those cases where no useful template is available, the ab initio methods are the only alternative. These methods are currently limited to small proteins and at best result only in coarse models with an RMSD error for the atoms that is greater than 4 A. However, one of the most impressive recent improvements in the field of protein structure modeling has occurred in ab initio prediction [155-157]. 

To put the errors in comparative models into perspective, we list the differences among strucmres of the same protein that have been detennined experimentally (Fig. 9). The 1 A accuracy of main chain atom positions corresponds to X-ray structures defined at a low resolution of about 2.5 A and with an / -factor of about 25% [192], as well as to medium resolution NMR structures determined from 10 interproton distance restraints per residue [193]. Similarly, differences between the highly refined X-ray and NMR structures of the same protein also tend to be about 1 A [193]. Changes in the environment... 

Figure 9 Relative accuracy of comparative models. Upper left panel, comparison of homologous structures that share 40% sequence identity. Upper right panel, conformations of ileal lipid-binding protein that satisfy the NMR restraints set equally well. Lower left panel, comparison of two independently determined X-ray structures of interleukin 1(3. Lower right panel, comparison of the X-ray and NMR structures of erabutoxin. The figure was prepared using the program MOLSCRIPT [236].

---