Prediction of New Regulatory Properties For Proteins Sharing Different Functional Motifs

Alex Lyakhovich1, Anatoly Karp2, University of Michigan; 2, University of Wisconsin-Madison

Evolutionary conserved motifs of protein families often mean their structural and/or functional similarities. However, this issue becomes more complex when several motifs are present within a single protein. In attempt to build up a link between functionally different groups of proteins having similar motifs we have developed and confirmed a new algorithm that combines classical sequence similarity searches with scoring and clustering of an obtained set of proteins according to the known functionally active domains. In our test runs we have applied this algorithm to a set of human and E-Coli proteins containing ubiquitin motifs (UIM, UBA, UBX). After scoring and clustering we have found that one of the highest scores is given to a group of proteins that also contain a number of kinase domains. This finding, along with published data, served us as a good hint to test whether proteins that recognize and bind ubiquitin molecules can also be regulated by protein kinases. By performing ubiquitinylation and kinase assays we found that most of the ubiquitin-binding proteins were indeed kinase regulated. The above algorithm therefore allows predicting new regulatory functions for the proteins containing different structural motifs.