Network generation enhances interpretation of proteomic data from induced apoptosis

Thirteen proteins (identified with 2-D gels and MALDI-TOF MS) are significantly altered during staurosporine-induced apoptosis in SH-SY5Y cells. To gain further insight into the integrated cellular response to apoptosis, we have investigated whether a network can be generated of direct and indirect interactions between these 13 proteins. A network that contains 12 out of the 13 proteins was generated using Ingenuity Pathway Analysis (IPA) and this network is dominated (89%) by direct protein-protein interactions. This network scored 34 with IPA. Bootstrapping 1000 random lists of 13 proteins suggested that the frequency of this score occurring by chance was 1 in 500. We examined whether subsets of proteins such as HSPs, which were elevated after staurosporine, had a disproportionate impact on the network generated. There was no evidence that any subset of 8 from the 13 proteins contributed disproportionately to the network. Network generation, using IPA, identified common features (such as endoplasmic reticular stress protein interactions) in apoptotic studies from different laboratories. The generation of protein interaction networks clearly enhances the interpretation of proteomic data, but only when interpreted cautiously, particularly in respect of statistical analyses.