The goal of this grant is to determine the structural basis for specificity in protein kinase signaling. The hypothesis guiding this research is that specificity in protein kinase signaling is accomplished by the ability of multiple domains or regions of the kinase to simultaneously or sequentially form sequentially form sequence-specific contact with complimentary regions of the target proteins. For example, src-homology 2 and src-homology 3 domains (SH2 and SH3) can mediate interactions between protein-Tyr kinases and their targets. These domains can fold into tertiary structures independent of the whole protein and form pockets that are capable of specific interactions with other proteins or peptides. Specificity in signaling is partially provided by the ability of individual SH2 domains to recognize phosphotyrosine residues on the target protein within a specific sequence context. Likewise, individual SH3 domains are thought to recognize relatively short linear sequences to target proteins that are variations on a common structure. The catalytic cleft of the kinase domain also has selectivity for Tyr residues in a specific sequence context. The combination of individual selectivities of each of these domains insures high specificity in vivo. Here we propose to use a novel oriented peptide library approach to ascertain the optimal peptide sequences for binding to individual kinase catalytic domains as well as SH2 domains and other domains involved in targeting kinases. Since the tertiary structures of several members of each of these families of domains have been solved with high affinity peptides bound, it is possible in many cases to explain the selectivities observed and predict with residues are responsible for binding to side chains of the associated peptides. Mutations at these sites will be made and the oriented peptide libraries will be used to determine whether the selectivity has been altered. By this approach we hope to establish general principles that can be used to predict likely targets of domains on the basis of regions of homology with other domains. Algorithms based on the information from the peptide libraries will be designed to predict the most likely regions of interaction between specific domains and target proteins. These predictions will also be investigated. Ultimately, the models will be tested by the ability of mutations in specific domains of their targets to alter in vivo responses.
Showing the most recent 10 out of 102 publications
