Supplementary Figure 4: Biochemical fractionation coupled to mass spectrometry (BF/MS) strategy to identify E. coli cePPIs.

(a) Schematic showing the identification of PPIs from BF/MS and selection criteria used to filter the dataset to verify the reliability of the predicted cePPIs from AP/MS. (b) Detergent solubilized CEPs extracted from E. coli using 0.020% Triton and, separately, 0.05% DDM (below the critical micelle concentration for each detergent) were fractionated into 84 biochemical fractions using size-exclusion chromatography (SEC-HPLC), proteolytically digested and the resulting peptides analyzed, in duplicate, by precision tandem MS. Hierarchical clustering of co-eluting proteins from SEC-HPLC identified by LC-MS/MS. Shading indicates spectral counts (SPC) recorded by LC-MS/MS. (c) Performance (true positive rate, TPR, vs false positive rate, FPR) of cePPI scoring algorithms (co-apex; weighted cross-correlation, WCC; Pearson correlation coefficient, PCC) that calculates correlation measures between all possible pairs of proteins to capture their tendency to co-elute was used to select a threshold cut-off for each score based on capture of curated cePPIs derived from EcoCyc (area-under-the curve; AUC). (d) Pearson correlation (r = 0.8) profiles between Triton (red) and DDM (blue) detergent solubilized proteins (extracted from cultured E. coli membranes) identified by tandem MS is plotted using normalized spectral counts. (e) Independent testing of cePPI pairs by (i) yeast (Y2H) or bacterial (B2H) two-hybrid assays, and (ii) B2H-based β-galactosidase activity (miller units represented as mean ± SD from at least three biological replicates); dashed line indicates background signal, p-value according to Student's t-test.