Computational Analysis Of Caspase Cleavage Sites In Proteins

By covalently attaching proteins, post-translational modifications （PTMs） play animportant role in regulating a variety of biological processes. Although a large number oftypes of PTMs have been discovered, only a few of them have been well-characterizeddue to the lack of sufficient data for the analysis. Identification of PTM substrates withconventionally experimental approaches, such site-directed mutagenesis, massspectrometry, peptide library, are labor-intensive, time-consuming and expensive.Computational prediction provides an effective approach to narrow down the potentialcandidates and generate useful information for further experiments.In the past two years, my major research interest is computational analysis ofCaspase substrates with the cleavage sites in proteins. The Caspase （Cysteine asparticacid specific protease） can precisely cleave target proteins in specific aspartic acid （D）,and be implicated in numerous biological processes, such as apoptosis, differentiation,proliferation, necrosis and inflammation. At least15Caspases have been identified inmammals. According to the structural and functional analysis, Caspases can be classifiedinto three groups, such as inflammatory Caspase （Caspase-1,-4,-5and-13）, initiatedCaspase （Caspase-2,-8,-9, and10） and effect Caspase （Caspase-3,-6and-7）.Although a number of predictors have been developed, the performance still remains to beimproved, due to the limitation of known training data set. From the scientific literature,we collected580experimentally verified caspase substrates with946cleavage sites. Thealgorithm of position specific scoring matrix（PSSM）was adopted for training, while theleave-one-out and n-fold cross-validations were performed. When the specificity are85%,90%and95%, sensitivity are88.77%,84.76%and71.12%, respectively. By comparisonwith previous studies, our performance is better. The preference for Caspase recognitionwas analyzed with Weblogo, whereas Caspases prefer to cleave a motif of D-E-V-D-（G/S/A）. Moreover, the statistical analysis was performed based on the GO （geneontology） information of known caspase substrates. Beyond apoptosis and inflammation,caspase cleavage proteins can also be invovled in various signaling pathways. We alsodesigned a predictor for Caspase-3, while sequence analysis suggested that Caspase-3also prefer to target the D-E-V-D-（G/S/A） motif, while the informative features of D, E andV at P4, P3and P2positions increase. The functional abundance and diversity ofCaspase-3substrates was also analyzed. Finally, We downloaded the protein sequences ofhuman, rat, Drosophila, C. elegans, yeast and Arabidopsis from the Uniprot database, andconducted a large-scale prediction for the six species. From the results, we observed thatthe Caspase cleavage is ubiquitous in eukaryotes, while the percentage of Caspasecleavage proteins is enhanced during the evolution.Taken together, our prediction of potential Caspase cleavage can generate useful dataset for further experimental consideration. We believe the computational analysis togetherwith followed experimental validations can propel the study of Caspase cleavage into anew phase.