Mass Spectrometry-based Absolute Quantification of 20S Proteasome Status for Controlled Ex-vivo Expansion of Human Adipose-derived Mesenchymal Stromal/Stem Cells

20S proteasomes are very heterogeneous protein complexes involved in many cellular processes. In the present study, we combined an MRM-based assay with the production and purification of entire SILAC labelled proteasome to monitor absolute quantities of the different 20S proteasome subtypes in various human cells and tissues. This method applied to adipocyte-derived stem cells (ADSCs) amplified under various conditions highlights an increased expression of immunoproteasome when this type of cell is primed with IFNγ or amplified in a 20% O2 environment.

accuracies, possibly because of differential oxidation of W. Thus, peptides containing M and Woxidable amino acids were skipped in the final SRM assay. The advantage of using whole isotopelabeled proteasome complex as internal standard over the AQUA or even the PrEST approach (9) is that, at least theoretically, all proteotypic peptides of each subunit can be used for the quantification.
The best peptides were however carefully optimized to reach a final SRM method including in total, 206 independent MS transitions (103 "light" transitions and 103 "heavy" transitions corresponding to heavy surrogate peptides) associated with optimized dwell times and voltages applied to the mass spectrometer (Supplementary Data 3).
The ELISA sandwich assay measures fully-assembled 20S proteasome only whereas the SRM results have been obtained independently from individual α and β non-catalytic subunits SRM-signals measured directly in human cell lysates. Thus, the adequation observed between the two independent assays suggests that the fraction of 20S subunits in the free form or in the course of assembly are very low in these cell lines and tissues. This is supported by the experimental observation that SRM signals of α and β non-catalytic subunits are hardly detectable in the low MW fractions of glycerol-gradient separated proteins from U937 cells and correlate with the ELISA quantification in the high MW fractions containing fully-assembled 20S proteasomes (Supplementary Figure 6A). Moreover, α and β non-catalytic subunits, which are sequentially incorporated during the 20S proteasome assembly process (10), are present in stoichiometric amounts both in the panels of eight cell lysates (Supplementary Figure 6B) and of 11 human tissues (Supplementary Figure 6C) which also argues that they are mainly incorporated into canonical fully-assembled proteasomes.
Altogether, these data prove the suitability of the new developed method for absolute quantification of 20S proteasome in human cell lysates and tissues.

2-Validation of the assay for determination of 20S subtypes stoichiometry and dynamics using a model of pro inflammatory cytokines-treated HeLa cells
An important goal of this work was, in addition to 20S proteasome absolute quantification, to determine the stoichiometry of the different proteasome subtypes in biological samples. Together with the detection of peptides corresponding to non-catalytic subunits, we constructed our method to also follow the six standard and immuno catalytic subunits of 20S proteasome. Three peptides and at least three transitions per peptide were optimized for each protein, except for β2 for which only two peptide sequences suited our selection criteria. This illustrates how optimization of peptide sequences is really an issue to meet the expected requirements for IDMS analysis of proteins (1). In this respect, methods relying on the spike-in of the whole heavy-labeled protein as internal standard, like PSAQ (11) or absolute SILAC (12,13), seem superior to the PrEST methodology (9) where only a short sequence of each protein is used.
Internal normalization using equimolar spikes of our two home-made heavy sP20S and iP20S permitted to monitor precisely the dynamics of all 20S proteasome subunits in interferon -treated HeLa cells over a 96h kinetics. The ability of our method to accurately determine 20S proteasome subtypes stoichiometry was then assessed by considering the known stoichiometries of incorporation of catalytic subunits. Indeed, for each catalytic subunit type, the sum of the absolute quantities of the standard subunit and its immuno-counterpart should equal the amount of total proteasome. So, as detailed hereafter, there are four different methods to determine the absolute quantity of total 20S proteasome from our measurements and at each IFN-stimulation time:  Figure 7A). Our optimized SRM method permitted to achieve accuracies above 96% for all three catalytic subunits types. In comparison with SRM, accuracies measured using the TOP3 label-free (LF) MS quantification (metric calculated as the mean of the three highest peptides areas measured for each protein), and obtained from the same biological samples, were much lower (76% for β1/β1i, 43% for β2/β2i, 79% for β5/β5i) (Supplementary Figure 7A). The poor trueness obtained with β2/ β2i label-free quantification may be explained by the fact that one of the TOP3 tryptic peptide analyzed for β2 is detected in at least five different forms arising from missed cleavages. The amount of peptides containing missed cleavages sites has been reported to influence the absolute quantification (4). Global accuracies obtained with the SRM and label-free MS methods are 97 ± 2 % and 78 ± 12 %, respectively (β2/ β2i signals were excluded for the label-free MS method only) (Supplementary Figure 7B). Concerning method variability, both MS quantification methods give mean CVs below 15%, though SRM seems more reliable as no outliers (ie with CVs above 30%) were measured (Supplementary Figure 7C-D). These performances are in accordance with a previous report where we compared the ability of four different label-free MS approaches to determine 26S proteasome stoichiometry in various cell lines (14). So, these data confirm that label-free MS1-based quantification methods, which are straightforward and allow protein identification and quantification with the same data set, can only give a rough estimation of protein complexes stoichiometry (14,15); these approaches are rather dedicated to the high throughput determination of changes in the relative abundances of protein complex subunits (16)(17)(18). Overall, these data indicate that our optimized SRM method is able to accurately and precisely report the absolute amounts of the six different 20S standard and immuno catalytic subunits.

II. Supplementary Figures Legend: Supplementary Figure 1: Linearity, sensitivity, and precision of the LC-SRM method
A-Typical chromatogram obtained for the SRM analysis of ten selected peptides corresponding to four non-catalytic subunits (3, 4, β6, and β7) and six catalytic subunits (β1, β2, β5, β1i, β2i and β5i). The signal corresponds to the sum of three transitions for each peptide. The equivalent of 5 fmoles of trypsin-digested commercial proteasomes purified from human erythrocytes was injected on column.  Five microliters of an equivalent of 2.5 µg total proteins was injected on column. The peptides obtained after trypsin digestion were spiked with 70 fmol of the four AQUA peptides. Light transitions signals were normalized with their respective heavy counterparts. The normalized transitions corresponding to a given peptide sequence were then averaged. The signal from the heavy AQUA peptide was used to determine the quantity of the endogeneous peptide (light signal). The total proteasome concentration measured in the injected samples and given by each peptide sequence is presented. For each biological replicates, results from 3 replicates of injection were averaged. Three biological replicates were analyzed to obtain statistics.
B-and C-Coefficients of variation (B) and accuracies (D) were calculated for each peptide sequence.
Coefficients of variation were calculated as ratios of the standard deviation over the mean of the values and expressed as a percentage. Accuracies were obtained as the ratio of the difference of the experimental value and the reference value over the reference value and expressed as a percentage.
The reference value corresponds to 20S proteasome concentration obtained by the ELISA reference assay.