Deciphering preferential interactions within supramolecular protein complexes: the proteasome case

Abstract

In eukaryotic cells, intracellular protein breakdown is mainly performed by the ubiquitin-proteasome system. Proteasomes are supramolecular protein complexes formed by the association of multiple sub-complexes and interacting proteins. Therefore, they exhibit a very high heterogeneity whose function is still not well understood. Here, using a newly developed method based on the combination of affinity purification and protein correlation profiling associated with high-resolution mass spectrometry, we comprehensively characterized proteasome heterogeneity and identified previously unknown preferential associations within proteasome sub-complexes. In particular, we showed for the first time that the two main proteasome subtypes, standard proteasome and immunoproteasome, interact with a different subset of important regulators. This trend was observed in very diverse human cell types and was confirmed by changing the relative proportions of both 20S proteasome forms using interferon-γ. The new method developed here constitutes an innovative and powerful strategy that could be broadly applied for unraveling the dynamic and heterogeneous nature of other biologically relevant supramolecular protein complexes.

Keywords: affinity purification; correlation profiling; label‐free quantitative proteomics; mass spectrometry.

Figure 1

Protein correlation profiling (PCP) analysis of glycerol density gradient-separated proteasome complexes

1. PCP-MS strategy to identify proteins interacting with specific proteasome subtypes. U937 cells were cross-linked with formaldehyde and lysed, and proteins were concentrated and ultrafiltrated on a 100 kDa cutoff device. Protein complexes were then separated on a 15–40% glycerol gradient. Each fraction of the gradient was analyzed by nano-LC-MS/MS. Protein quantification was performed using the mean XIC of the three most intense validated peptides for each protein, after internal standard calibration using a mix of 8 isotopically labeled peptides. The PCP analysis was performed as described in the Materials and Methods section.

2. PCP analysis of the 19S regulatory complex. Protein abundance profiles of 16 proteins of the 19S RP (Rpt1–6, Rpn1–3, Rpn5, 7–9, 11–13, gray lanes) and of their median abundance (black lane) (left panel). PCP analysis is performed by plotting the χ² values (representing the Euclidian distance between the abundance profile of each protein and the reference profile) of the experimental replicate 2 as a function of the χ² values of the experimental replicate 1 (middle left panel). The median profile of the 19S complex subunits was used as the reference profile for the calculation of the χ² values. Different zooms of the graph are represented (middle right and right panels). Light gray dots represent the proteins quantified in all the fractions of the density gradient and blue dots represent 19S subunits (right panel).

3. PCP analysis of proteasome 20S complex. Protein abundance profiles of 17 proteins of the 20S CP (α1–α7, β1–β7, β1i, β2i, β5i, gray lanes) and of their median abundance (black lane) (left panel). PCP analysis is performed by plotting the χ² values of the experimental replicate 2 as a function of the χ² values of the experimental replicate 1 (middle left panel). The median profile of the 20S complex subunits was used as the reference profile for the calculation of the χ² values. Different zooms of the graph are represented (middle right and right panels). Light gray dots represent the proteins quantified in all the fractions of the density gradient and red dots represent 20S subunits.

Figure 2

Protein correlation profiling (PCP) analysis using the median profile of the PA28αβ regulator as the reference profile

1. Profiles of the PA28α, PA28β, and the β2i proteins (blue, red and green lines, respectively).

2. Plot of the χ² values of the experimental replicate 2 as a function of the χ² values of the experimental replicate 1.

3. A zoom of the graph in (B) is represented and χ² coordinates for PA28α, PA28β, and β2i proteins are highlighted as blue, red, and green dots, respectively. Light gray dots represent the χ² coordinates of the proteins quantified in all the fractions of the gradient. The median profile of the PA28α and PA28β subunits was used as the reference profile for the calculation of the χ² values.

Figure 3

Protein abundance correlation of affinity-purified complexes analyzed by mass spectrometry strategy applied to proteasome complexes

1. A Proteasome complexes were immunopurified from nine formaldehyde-crosslinked human cell lines and analyzed by nano-LC-MS/MS. Protein abundance indexes (PAIs) were used to represent the abundance of proteins in purified proteasome samples. The correlation between two different proteins was quantified using coefficients of determination (R²).

2. B–E Correlations of abundances of α7 and α6 (B), Rpn3 and Rpn1 (C), PA28β and PA28α (D), and Rpn3 and PA28β (E).

Figure 4

Protein abundance correlation of affinity-purified complexes analyzed by mass spectrometry analysis applied to the proteasome complexes and their interacting proteins

1. Heat-map representing the correlations (expressed as the R²) between the abundances of 73 known proteasome-interacting proteins (PIPs) and the abundances of 8 reference proteins or protein complexes, PA28γ, β2i (representing the iP20S), PA28αβ, ncP20S (median of α1–α7, β3, β4, β6, and β7 profiles), 19S (median of Rpt1–6, Rpn1–3, 5–14 profiles), PI31, β5 (representing the sP20S), and PA200. For protein complexes, the median PAI of their subunits in each of the 24 AP-MS experiments was used: α1–α7, β3, β4, β6, and β7 subunits for the ncP20S, Rpt1–6, Rpn1–3, 5–14 for the 19S RP, and PA28α and PA28β subunits for the PA28αβ RP. The R² values were hierarchically clustered. Three distinct clusters of composition detailed hereafter could be obtained. Cluster 1 (from top to bottom): Rpt3, Rpn13, α2, Rpn7, USP14, hHR23B, α1, β6, β3, α4, α7, Rpn6, Rpn3, Rpt4, Rpn10, Rpn5, Rpt5, Rpn1, Rpn11, Rpt1, Rpn9, Rpn2, Rpn8, α3, α6, β4, Rpt6, PDC6, Rpn12, APEH, Ubiquilin-1, α5, β7, CCT7, CCT4, CCT2, CCT3, CCT5, CCT6A, DNAJA1, HSP90AB1, HSP90AA1, PNP, Rpt2. Cluster 2 (from top to bottom): 14-3-3ζ/δ, CAND1, GSR, UBE3C, β2, ATP5A1, ATP5B, β1, PA200, β5, FBXO7, UCHL5, TXNL1, ECM29, PI31. Cluster 3 (from top to bottom): PA28β, PITH1, β2i, PA28α, PA28γ, β5i, β1i.

2. Principal component analysis (PCA) of the abundances of 73 known PIPs. The circles represent the main clusters observed (iP20S, ncP20S/19S, sP20S and the 20S assembly chaperones).

3. Plot of the R² values between the iP20S or the sP20S and 193 protein correlating (R² > 0.8) with the iP20S, the sP20S, or the ncP20S.

Figure 5

Changes in the expression of 20S proteasome catalytic subunits modulate 20S-associated regulators

1. The two HEK EBNA cell lines express only standard proteasome or immunoproteasome subunits. Western blots against the immuno- (β1i, β2i, β5i) and standard (β1, β2, β5) catalytic subunits of the 20S proteasome. Calnexin is used as a loading control. Black lines delineate the boundary between vertically sliced images that juxtapose lanes that were non-adjacent in the gel. Importantly, the bands were assembled from the same blot.

2. Relative normalized abundance indexes of proteasome regulators in HEK EBNA cells containing only immunoproteasome compared to HEK EBNA cells containing only standard proteasome. The normalized abundance indexes for each regulator were set to 1 for standard proteasome conditions (n = 4).

3. Kinetics of IFN-γ treatment on HeLa cells. HeLa cells were stimulated for 0, 24, 48, or 72 h with IFN-γ. Western blots were performed on total cell lysates with antibodies against the β2i, α2, and α5 subunits. IRF-1 was used to control IFN-γ treatment efficiency, and GAPDH was used as a loading control.

4. For each time point of the IFN-γ treatment, proteasome complexes were purified and analyzed by LC-MS/MS. Proteasome complexes dynamics was measured by label-free quantitative proteomics. The normalized abundance index of each protein or protein complex obtained at each time point was compared to the one obtained at the 0 h time point to obtain a regulator relative normalized PAI (n = 3).

Data information: *P < 0.05; **P < 0.01; ***P < 0.001 (Student t-test). Source data are available online for this figure