Aurora-A, a mitotic serine/threonine kinase with oncogene characteristics, has recently drawn intense attention because of its association with the development of human cancers and its relationship with mitotic progression. Using the gene expression profiles of Aurora-A as a template to search for and compare transcriptome expression profiles in publicly accessible microarray data sets, we identified HURP (encodes hepatoma upregulated protein) as one of the best Aurora-A-correlated genes. Empirical validation indicates that HURP has several characteristics in common with Aurora-A. These two genes have similar expression patterns in hepatocellular carcinoma, liver regeneration after partial hepatectomy, and cell cycle progression and across a variety of tissues and cell lines. Moreover, Aurora-A phosphorylated HURP in vitro and in vivo. Ectopic expression of either the catalytically inactive form of Aurora-A or the HURP-4P mutant, in which the Aurora-A phosphorylation sites were replaced with Ala, resulted in HURP instability and complex disassembly. In addition, HURP-wild-type stable transfectants were capable of growing in low-serum environments whereas HURP-4P grew poorly under low-serum conditions and failed to proliferate. These studies together support the view that the ability to integrate evidence derived from microarray studies into biochemical analyses may ultimately augment our predictive power when analyzing the potential role of poorly characterized proteins. While this combined approach was simply an initial attempt to answer a range of complex biological questions, our findings do suggest that HURP is a potential oncogenic target of Aurora-A.

One possible path towards understanding the biological function of a target protein is through the discovery of how it interfaces within protein-protein interaction networks. The goal of this study was to create a virtual protein-protein interaction model using the concepts of orthologous conservation (or interologs) to elucidate the interacting networks of a particular target protein. POINT (the prediction of interactome database) is a functional database for the prediction of the human protein-protein interactome based on available orthologous interactome datasets. POINT integrates several publicly accessible databases, with emphasis placed on the extraction of a large quantity of mouse, fruit fly, worm and yeast protein-protein interactions datasets from the Database of Interacting Proteins (DIP), followed by conversion of them into a predicted human interactome. In addition, protein-protein interactions require both temporal synchronicity and precise spatial proximity. POINT therefore also incorporates correlated mRNA expression clusters obtained from cell cycle microarray databases and subcellular localization from Gene Ontology to further pinpoint the likelihood of biological relevance of each predicted interacting sets of protein partners.

No abstract

No abstract