Predicting kinase-substrate interactions in the era of proteomics: focus on "Identifying protein kinase target preferences using mass spectrometry".

WHEN THE “OMICS” BEGAN TO EMERGE in the life sciences, some investigators teasingly proposed to institute an “omics police” (2). This does not seem to have prevented a rapid growth of the “omics” (genomics, proteomics, metabolomics). Of these neologisms, proteomics refers to the concurrent identification, quantification, and analysis of a large number of proteins in a functional context (6). Mass spectrometry has been instrumental in the progress of proteomics because of its ability to reliably identify peptides based on their molecular weight (6). Because of the biological importance of phosphorylation, enrichment methods have been developed to specifically identify phosphopeptides by mass spectrometry (“phosphoproteomics”) (5). “Systems biology” has emerged as the term to describe the application of proteomics and other techniques to discover and model biological phenomena such as signaling networks (1). Now that the buzzwords have been introduced and briefly explained, you will appreciate the paper by Douglas, Gunaratne and colleagues in this issue of American Journal of Physiology-Cell Physiology (3). This group reports a mass spectrometry-based method to identify protein kinase target preferences. To do so, they developed a new bioinformatics tool that generates a visual language for kinase sequence preference motifs. In this “sequence logo” the phosphorylated residue is positioned at 0 and the predicted surrounding amino acids are displayed in their alphabet code with font size corresponding to probability (Fig. 1). The anticipated result from this in silico analysis is that some amino acids will be overrepresented on certain positions in the generated motifs so that this will point in the direction of a specific kinase or kinase family. The output comes with a statistical filter that corrects for multiple testing and reduces signal-to-noise ratio. The bioinformatics tool was coined “PhosphoLogo” to emphasize its utility in predicting phosphorylation events. The authors used in vitro experiments to test 16 protein kinases for their target sequence preferences. A tissue lysate containing rat kidney, liver, brain, and small intestine was first dephosphorylated and heat-denatured and subsequently incubated with ATP and a recombinant kinase of interest. The samples were then prepared for analysis by mass spectrometry using trypsin and immobilized ion affinity chromatography to enrich phosphopeptides. Of the 16 kinases, protein kinase A (PKA) was used as the model kinase. Indeed, PhosphoLogo generated a sequence logo that displayed the well-known R-R-X-S motif and was similar to sequence logos from existing sources such as PhosphoSitePlus and NetPhorest. Additional outputs of PhosphoLogo include a charged-grouped logo, size-grouped logo, and anti-logo. The charged-grouped logo shows positional preference for polar, nonpolar, acidic, or basic aminoacids, while the size-grouped logo indicates the preferred size of the residue. Interestingly, the anti-logo shows disfavored residues, which should be especially helpful for investigators to reduce a list of candidate kinases. The authors translated the Address for reprint requests and other correspondence: E. J. Hoorn, Dept. of Internal Medicine Nephrology, Rm. H-436, P. O. Box 2040, 3000 CA Rotterdam, The Netherlands (e-mail: [email protected]).