PhD Lecture by Martin Lee Miller, CBS

Linear motifs in phosphorylation-dependent signalling

Wednesday 25 June 2008 at 13:00
CBS, DTU, Lyngby, Building 306, Auditorium 31

Assessment Committee:    Professor Ole Lund, DTU (Chairman)
Associate professor Jens S. Andersen, University of Southern Denmark
Professor Bruce J. Mayer, University of Connecticut Health Center
Chair of defense:Associate professor Henrik Nielsen, DTU
Supervisor:Professor Søren Brunak, DTU

Protein phosphorylation is fundamental to intracellular signaling and is involved in the regulation of a variety of physiological processes. The organization of signaling networks is partly mediated by short amino acid patterns or linear motifs that are phosphorylated by kinases and may subsequently be bound by proteins with modular domains such as the Src homology 2 (SH2) domain. Dysregulation of phosphorylation-dependent signaling plays a causal role in human disease such as cancer. Thus, characterizing linear motifs of kinases and phosphobinding domains is essential in the development of therapeutics. This is a challenging task since there are more than 500 kinases and 100 SH2 domains that recognize specific motifs. In addition, mass spectrometry studies have recently mapped thousands of new in vivo phosphorylation sites in different model-organisms. However, the ability to accurately assign the phosphorylation sites to their related kinases or binding modules is currently limited, both experimentally and computationally. Taken together, there is a demand for developing new tools that can characterize linear phosphorylation motifs and their role in cellular signaling. This thesis combines computational and experimental approaches to support such analysis. We have constructed the to date largest collection of computer-based sequence models for classification of linear phosphorylation motifs by integrating both in vivo and in vitro phosphorylation data. The sequence models cover nearly half of all human kinases and phosphotyrosine-binding domains enabling the first systematic analysis of linear phosphorylation motifs and their physiological characteristics. Supporting previous observations that cancer kinases exhibit a high degree of substrate promiscuity, we found that oncogenic tyrosine kinases display weaker substrate specificity than their non-oncogenic counterparts. Also included in this thesis, we developed a new clustering approach and decomposed the uncharacterized phosphotyrosine proteome for interaction motifs. Besides several known SH2 domain motifs that were extracted and experimentally verified, we identified and confirmed a novel hydrophobic motif for the SH2 domain containing inositol phosphatase SHIP2. Finally, we constructed predictors for phosphorylation sites in bacteria and yeast and several predicted bacterial phosphoproteins were experimentally verified. Collectively, this work demonstrates the advantage of integrating experimental proteomics and bioinformatics to advance our understanding of the complexity of phosphorylation-dependent signaling.

Everybody is welcome. Registration is not necessary.