Unknown proteins, predicting functions

TABLE 16.4. Computational approaches used to predict functions of unknown proteins... 

Bioinformatics uses computers to create and maintain large electronic databases on genomes, protein sequences, and proteomes. With the help of protein prediction software, the computer analysis of genome sequences is producing thousands of new proteins of unknown structure and function. These proteins are called hypothetical proteins because they are predicted from the gene sequence. To know if they really exist would require that they be isolated, purified, and subjected to X-ray crystallography or... 

However, the fast development of mass spectroscopy and other experimental techniques for studying protein interactions has enabled the construction and analysis of protein interaction networks. The interaction maps obtained for one species can be used to predict interaction networks in other species, which is useful for identifying the functions of unknown proteins. The protein interaction databases provide a wealth of information regarding the interacting pairs of protein-protein interactions. 

Recent advances in sequencing technologies have fostered the capacity to sequence large and complex genomes, ESTs, and other libraries, leading to an exponential increase in the volume of DNA and protein sequence data available in sequence databases. Predicting the function of an unknown protein by sequence or motif similarity to a previously characterized protein is an extremely valuable process and is the lynchpin in the data mining approach to... 

It is possible to predict the tertiary structure of a protein by comparing the amino acid sequence of the unknown protein with those whose structure is known. Structures can be assigned based on unique folds or domains that have conserved amino acids and, although the amino acid sequence of a specific fold may differ, amino acids with similar properties are conserved to maintain the structural and functional properties of the protein. It is possible, with relative accuracy, to assign proteins to a particular protein family in this way. 

Fig. 12.5. Schematic summary of the eight T. canis proteins containing predicted SXC (NC6) domains. The consensus is shown in the N-terminal domain of PEB-1 (phosphatidylethanolamine-binding protein-1) as xCxDxxxDC(6x)C(11x) RCxxTCxxC. This consensus is faithfully repeated in MUC-1 (mucin-1), MUC-2, MUC-4 and MUC-5, and in all but the C-terminal domain of MUC-3. This domain (and the C-terminal SXC domain of PEB-1) show consensus spacing but some variation in consensus residues. Two additional proteins with quadrupled SXC domains differ in spacing between cysteines-2, -3 and -4, and show more variation in consensus residues. These are VAH-1 (venom allergen homologue) and HUF-001 (homologue of unknown function-001).

A prerequisite for this model is the existence of both external and internal gates, i.e., protein domains that are capable of occluding access to the substrate binding site from the extracellular and intracellular environment, respectively. Little is known about such domains in this family of transporters. The molecular mechanisms governing the cooperative function of the putative gating domains also remain unknown. It could be predicted, however, that stabilization of the transporter in the outward-facing conformation in the absence of substrate but in the presence of Na+ requires a network of con-... 

Many microbial genes are of unknown function. Even in the well-studied bacterium E. coli, over 38% of the predicted proteins have no experimental data to support an understanding of their functional role in the cell. 

Recently, Hwang et al. have performed conceptually similar predictions using three-dimensional structural information (Hwang et al., 1999). Here, structural similarities between a protein of unknown function and others available in the public databases were used to infer the general function of the URF as a new nucleotide triphosphosphatase. It is expected that as many new structures of URFs are solved over the next few years, the conceptual approaches described here will be useful in interpreting characteristics of such proteins even in the absence of a clear understanding of overall function or identification of specific substrates and products. 

---