Selection of proteins predicted to participate in di- or oligomeric complexes based on HCL analysis of SEC fraction abundance patterns.

A: Clusters derived from Hierarchical clustering (HCL) using the Pearson correlation metric in the software MeV featuring groups of proteins with similar SEC fractionation patterns.In each cluster, more than 70% of the proteins were subunits of known homo- or hetero-oligomeric protein complexes with M.W. values in better agreement with a complex than monomer. Examples of such characterized proteins are ferritin and uridine phosphorylase. Clusters are also visualized in Figure S2 (Suppl. Information).B: Gene name, locus tag and protein description are derived from annotations in the E. coli O157:H7 EDL933 genome (UniProt database) or the EcoCyc database.C: Subcellular localization derived from EcoCyc data or annotations in UniProt for the EDL933 genome and multi-protein complexes (MPC). CY: cytoplasm; PP, periplasm; Ho-‘2’, Ho-6’ etc., homo-oligomeric protein with the number of subunits; ppi, proven protein interactions with another protein or a large assembly such as the ribosome (RIB).D: KWS: the Kruskal-Wallis statistic was applied to the entire EHEC proteome dataset (2521 proteins) to select proteins with statistically significant abundance differences comparing. the size exclusion chromatography (SEC) fractions (>280 kDa, 280-80 kDa and 80-10 kDa). After elimination of integral membrane proteins and setting the KWS p-value at E: Average protein abundances (APEXi scores) in each of the three SEC fractions. n = 11; normalization factor: 2.5×106; ‘0’: no peptide spectra were detected for a given protein.F: Functional roles of proteins predicted for the EHEC protein or the respective E. coli ortholog (data from EcoCyc or UniProt annotations).