Bioinformatic Studies Of Plant Phosphoproteomes

Protein phosphorylation catalyzed by protein kinases (PKs) is a process that transfers a phosphate group from an ATP to a specific amino acid residue. It is a reversible reaction which can also be dephosphorylated by protein phosphatases(PPs) through removing the phosphate group from the substrate. As one of the most important and well-studied post-translational modifications (PTMs), the reversible phosphorylation plays an essential role in almost all biological progresses in plants such as metabolism, signal transduction and environmental response. Recently, with the remarkable progresses of high-throughput proteomic techniques, the phosphorylation data of plants has shown an explosive growth, yet it presents new challenges such as how to extract useful information from massive amounts of data of the plant phosphorylation. My major progresses are focused on computational studies of protein phosphorylation in plants, including the database construction, software development and network analysis.The substrates with phosphorylatable sites are regulated by protein phosphorylation, so the identification of phosphorylation sites (p-sites) is fundamental to dissect the molecular mechanisms of protein phosphorylation. Through a manual curation of the scientific literature from PubMed, the experimentally identified p-sites in proteins from multiple plants were collected, and the datasets from other public databases were also integrated to develop a comprehensive database of dbPPT. In total, dbPPT contained82,175p-sites in31,012phosphoproteins from20plant organisms. Compared with other databases, dbPPT is superior in both the number of p-sites and the coverage of plant species. Besides, all the phosphoproteins and p-sites in the dbPPT database have been comprehensively and critically annotated.As key regulators responsible for the phosphorylation reaction, we described the computational methods and protocols for the identification of PKs by using Oryza sativa subsp. japonica(japonica) as an example. Frist, based on the collected PKs from the literature and public databases, we constructed the Hidden Markov Model (HMM) profiles for PK families. Then, the potential PKs were computationally identified by pre-constructed HMM profiles. Furthermore, we conducted ortholog searches to identify additional PKs for the families without HMM profiles. We totally characterized1,296PKs with9groups and46families in japonica.Based on the understanding of the PKs and substrates with p-sites, it is critical to identify the PK-specific substrates for the research of protein phosphorylation in plants. Since computational predictions can provide useful information for further investigations, then we developed a web server of GPS3.0to predict potential PK-specific p-sites. Frist, we manually collected the experimentally identified PK-specific p-sites from the scientific literature. Then, a previously developed GPS algorithm was employed to conduct the computational prediction. Through critical evaluations and comparisons, GPS3.0achieved a better performance than other existing tools. In general, GPS3.0was able to predict potential p-sites for464PKs in human, with the predicted results being visualized in multiple options.After the development of a plant phosphorylation database and a prediction software for PK-specific p-sites, we further proposed a computational method for the further study of potentially in vivo site-specific kinase-substrate relations (ssKSRs) in plants. This methodology was comprised of GPS algorithm which can predict the PK-specific p-sites, phosphoproteomic datasets and physical interations between PKs and substrates in plants. The prediction performance was evaluated by the experimentally indentified kinase-specific p-sites of plants, and results demonstrated that it is feasible and reliable for the prediction of plant ssKSRs in vivo. Then, in combination with the identified PKs by HMM profiles and ortholog searches, the methodology was adopted for the prediction of in vivo ssKSRs with the experimentally identified p-sites in japonica, Oryza sativa subsp. Indica(indica) and Oryza rufipogon (O. rufipogon), respectively. Furthermore, we constructed the rice kinase-substrate phosphorylation networks (KSPNs) and found out that the networks had scale-free topologies. Moreover, the statistical analyses demonstrated that the proteins in the KSPN of japonica were highly associated with disease resistance but not enriched in stress response. In addition, the evolutionary analyses indicated that the evolution rate of phosphorylation regulation in the domesticated rice species was more rapid than in non-domesticated species. Taken together, througth the integration of phosphoproteomic datasets, PKs identification, software development for predicting the ssKSRs and phosphorylation network analysis, we systematically analyzed plant protein phosphorylation by various bioinformatic approaches. Our study would provide new leads for further investigations of both the molecular mechanisms of protein phosphorylation and its role in the regulation of plant growth and development.