A strategy for detection of structural conserved regions and contacts in protein superfamilies: application to the PLP-dependent enzymes

During the last few years, structural and genomic projects provided us with a wealth of biological information that now opens new prospects of understanding life and evolution at the molecular level. In the next future, one of the main challenges will be the organization and the interpretation of this exponentially growing amount of data. In this work, we show how the rational application of computational techniques can be exploited to derive a deeper knowledge of the mechanisms underlying protein folding and evolution of protein superfamilies. In particular, the application of such techniques to the case of the fold-type I PLP-dependent enzymes is described. Methods and Results A non redundant data set of 23 superposed crystallographic structures belonging to this superfamily was collected from the structural databanks. Only structures displaying a RMSD < 4Å and a sequence identity < 27% with all the other members of the superfamily were included in this set to assure high structural conservation at low sequence identity. For each selected structure, a multiple sequence alignment of orthologous sequences was obtained using the programs BLAST and CLUSTALW. The 23 resulting alignments were merged following the structural superposition to obtain a non redundant, comprehensive multiple alignment of 921 sequences of PLP enzymes of fold-type I. A new tool was developed to extract and analyze from the superposed enzymes the structural conserved regions (SCRs) and the conserved contacts using both sequence and structural information. The results of this study identified a set of hydrophobic contacts significantly conserved in almost all the superfamily structures. Also, this conserved pattern highlights the presence of a nucleus, in this specific fold, in which residues participating in the most conserved native interactions exhibit preferential evolutionary conservation. At present we are extending this comparative analysis to the human genes coding for PLP-dependent enzymes, to investigate the degree to which the relationship between the SCRs and gene structure (intron position, density, phase and length) is conserved between these superfamily members. Finally, we suggest that such a strategy could be extended to other superfamilies for which suitable sequence and structural information is available, to derive rules for the rational design of protein engineering and folding experiments, to identify new superfamily members, and to elucidate how distinct catalytic properties could emerge from a common scaffold.