Rapid generation of cell line collections for isogenic and inducible bait expression using Flp-recombinase-mediated recombination. (A) Schematic overview on the generation of cell line collections. Starting from human Gateway orfeome collections, cDNAs of interest are recombined into an expression construct for tetracycline (tet)-inducible expression of strep-hemagglutinin double-tagged (SH) bait proteins. Isogenic cell lines are generated using Flp-recombinase-mediated recombination through single FRT sites present in the expression construct and the genome of Flp-In HEK293 cells stably expressing the tet repressor. After transfection, cell lines are selected on hygromycin for 2 weeks, tested for tetracycline-inducible expression and used for subsequent affinity purification. (B) Isogenic bait protein expression in HEK293 Flp-In cells. The expression of the indicated recombinant proteins in the absence or presence of 1 μg/ml tetracycline for 24 h was visualized by indirect fluorescence microscopy with an anti-HA antibody. Nuclei were stained with DAPI. (C) Tet-inducible bait expression. Increasing amounts of tetracycline were added to HEK293 cells expressing SH–eGFP for 24 h. Bait expression was monitored by immunoblotting using anti-HA antibodies. (D) Comparison of protein expression levels of SH-tagged bait proteins with endogenous protein levels. HEK293 cell lines expressing SH-tagged bait proteins were analysed by immunoblotting using the indicated antibodies following induction with tetracycline (1 μg/ml) for 24 h. HEK293 cells that do not express the corresponding bait proteins were used as controls. Note that anti-PPP2C and anti-PPP2R1 antibodies do not distinguish between the highly related endogenous proteins PPP2CA and PPP2CB or PPP2R1A and PPP2R1B, respectively.

Analysis of SH-double-affinity purification specificity. (A) Silver stain monitoring of sample complexity. SH-double-affinity purification was performed on 3 × 10⁷ HEK293 cells expressing SH-PPP2R2B, and aliquots of the first (ES) and second (EH) purification step were separated by SDS–PAGE. (B) Quantitative MS analysis of the specificity increase by the second purification step. HEK293 SH-PPP2R2B lysates were split in half. Single and double-affinity purifications were performed as described before. Peptides derived from single (ES) and double-affinity purification (EH) were mixed in a three-step dilution. Following MS analysis, MS spectra were aligned and MS1 signal intensities were used for relative quantification to generate protein abundance profiles across the three samples. All presented protein profiles were generated from aligned MS1 features that correspond to at least five unmodified, fully tryptic peptides. Note that not all observed protein profiles were included here, as they did not pass the indicated filtering criteria (e.g. PPP2CA). Protein abundance profiles were normalized to the bait profile. Yellow lines correspond to specific binding partners that match the bait profile. Orange lines refer to unspecifically co-purifying proteins identified also in SH–eGFP control samples. Profiles in red represent proteins unspecifically co-purified with the streptavidin sepharose beads that are successfully removed after the second purification step. Error bars indicate s.e.m. of MS1 feature ratios of the indicated proteins.

Summary of the identified protein interaction data set. A total of 197 protein–protein interactions identified in this study have been grouped according to available literature information into known interactions, homologous interactions identified in either human or yeast or unknown protein interactions.

Evolutionary conservation of protein interactions within the PP2A network. Interaction data from this study for which the corresponding orthologous protein interactions could be found in yeast are displayed as a network graph and compared with the yeast PP2A network. Human gene symbols and yeast gene names have been used as node identifiers. Dotted red lines indicate the orthology relationships between the shown network components. Note the expansion of orthologous interactions involving regulatory subunits (dark blue nodes) in the human PP2A system.

Monitoring SH-double-affinity purification efficiency. (A) Schematic overview of the purification procedure. HEK293 cells expressing SH-tagged proteins are lysed and first purified from total protein extracts using streptavidin sepharose (Strep-Tactin beads). After several wash steps, purified proteins are released in the presence of 2 mM biotin for subsequent immunoaffinity purification using anti-HA agarose. Finally, protein complexes are eluted with 0.2 M glycine, pH 2.5, and processed for mass spectrometric analysis. (B) Western blot analysis of SH-purification yields. SH-PPP2R2B expressing HEK293 cell line was generated as described in Figure 1 and the purification was performed on 3 × 10⁷ cells. The purification procedure was monitored by immunoblotting using anti-HA antibodies. L: lysate; SNS: supernatant after streptavidin purification; ES: elution from the streptavidin sepharose; SNH: supernatant after anti-HA purification; EH: elution from the anti-HA agarose (final eluate). Information on the percentage of input is given to compare the relative amount of sample loaded on each gel lane. (C) Reproducibility of SH-purification yields. 11 cell lines inducibly expressing SH-tagged bait proteins related to the PP2A phosphatase system were generated as described. Lysate (L) and final eluate (EH) from all 11 bait-specific SH-purifications were immunoblotted using anti-HA antibodies and percentage of loaded sample amount is indicated.

Data processing and reproducibility of the overall workflow for a human PP2A protein interaction network. (A) HEK293 cell lines expressing 11 different bait proteins linked previously to the human PP2A network were generated as described in Figure 1. Inducible expression was tested by western blotting (upper panel). Eleven cell lines were used for two independent SH-double affinity purification experiments directly followed by LC-MS/MS (SH-LC-MS/MS). Obtained mass spectra were analysed with XTandem followed by statistical validation of the search results using PeptideProphet and ProteinProphet. Only proteins with minimum ProteinProphet probabilities of 0.9 were considered. Primary interaction data were filtered against a contaminant database resulting in a final list of 242 (replicate A) and 218 (replicate B) bait–prey interactions with an overlap of 85%. To assess the reproducibility rate, the sum of common interactions found in experiments A and B were divided by the total number of interactions identified. (B) Generation of a database containing co-purifying contaminant proteins used for data filtering. Eight independent control purifications were performed using SH-tagged eGFP as a bait protein. Tryptic peptides were analysed twice by LC-MS/MS and identified proteins from all 16 measurements were consecutively added to a contaminant database. A total of 109 proteins were identified and used to subtract unspecific protein contaminants. (C) Robustness of protein interaction data obtained by the integrated workflow. The number of protein interactions (black line) and the percent overlap (red line) across the two replicate data sets shown as a function of increased filtering stringency (number of unique peptides required for protein identification).

Protein–protein interactions and different classes of protein complexes within the human PP2A interaction network. Protein–protein interaction data identified in two out of two replicate experiments were vizualized using Cytoscape v.2.5.1. Bait proteins are represented as triangles, and prey proteins are represented as rounded squares. Nodes were classified according to the legend. Edge colour indicates known interactions (orange) or newly identified interactions (blue). Thickness of the edges reflects the number of identified unique peptides. Five classes of PP2A protein complexes could be distinguished on the basis of network topology and literature information and are displayed by dotted lines and colour shading (see legend).