Differential proteomics argues against a general role for CD9, CD81 or CD63 in the sorting of proteins into extracellular vesicles

Abstract

The tetraspanins CD9, CD81 and CD63 are major components of extracellular vesicles (EVs). Yet, their impact on EV composition remains under-investigated. In the MCF7 breast cancer cell line CD63 was as expected predominantly intracellular. In contrast CD9 and CD81 strongly colocalized at the plasma membrane, albeit with different ratios at different sites, which may explain a higher enrichment of CD81 in EVs. Absence of these tetraspanins had little impact on the EV protein composition as analysed by quantitative mass spectrometry. We also analysed the effect of concomitant knock-out of CD9 and CD81 because these two tetraspanins play similar roles in several cellular processes and associate directly with two Ig domain proteins, CD9P-1/EWI-F/PTGFRN and EWI-2/IGSF8. These were the sole proteins significantly decreased in the EVs of double CD9- and CD81-deficient cells. In the case of EWI-2, this is primarily a consequence of a decreased cell expression level. In conclusion, this study shows that CD9, CD81 and CD63, commonly used as EV protein markers, play a marginal role in determining the protein composition of EVs released by MCF7 cells and highlights a regulation of the expression level and/or trafficking of CD9P-1 and EWI-2 by CD9 and CD81.

Keywords: CD63; CD81; CD9; IgSF8; PTGFRN; exosomes; extracellular vesicles; mass spectrometry; proteomic; tetraspanins.

FIGURE 1

Proteomic analysis of extracellular vesicles secreted by MCF7 cells. (a) Schematic representation of the protocol used to collect EVs from the conditioned medium of MCF7 cells cultured for 16 h in the absence of serum. (b) Distribution of the proteins identified in EVs according to their relative LFQ Value. After selection of the proteins identified and quantified in at least four experiments, the LFQ value of each protein was divided by the mean LFQ value of the corresponding experiment. The graph shows the distribution of the median value of the ratio calculated in the six experiments. (c) Top proteins identified in the EVs released by WT MCF7 cells according to the median of the relative LFQ values in six experiments. Classical EVs markers are in blue. Proteins in bold characters correspond to proteins previously shown to be co‐immunoprecipitated with tetraspanins. The median LFQ ratio of some proteins is shown.

FIGURE 2

Relative enrichment of CD9, CD81 and CD63 in EVs and their subcellular distribution. (a) Western‐blot analysis of CD9, CD81 and CD63 in MCF7 lysate and in EVs they release and calculation of their enrichment in EVs. This enrichment is calculated as the signal in EVs divided by the signal in lysates, with the ratio calculated for CD9 being used as the reference. The graph shows the mean ± SEM and values of replicates. CD63 and CD81 were subsequently probed with specific antibodies and a secondary antibody coupled to Alexa Fluor 680 before incubation with the CD9 antibody and a dylight 800‐coupled secondary antibody. In this particular experiment we also analysed the pool of fractions 11–14 collected during the SEC procedure. (b) Comparison of the cellular distribution of CD9, CD81 and CD63 in MCF7 cells by confocal microscopy. The cells were fixed before labelling of CD9, CD81 and CD63 as described in material in methods. Three different confocal sections are shown. The lower images correspond to the ratio of CD81 to CD9 intensity in each pixel. Bar: 10 μm. (c) Relative levels of CD81 and CD9 at cell‐cell contacts or outside these contacts. Regions of interests were drawn around cell‐cell junctions or on the top/periphery of the cells excluding these contacts. The CD81/CD9 ratio in each region was calculated and divided by the mean of the ratios calculated in the regions drawn around cell‐cell contacts. The results for four cell clusters, similar to that shown in Figure 2b are shown.

FIGURE 3

characterization of MCF7 cells lacking CD9, CD81 or CD63 and the EVs they release. (a) Flow‐cytometry analysis of the surface expression of CD9, CD81 and CD63 in parental MCF7 cells and cells KO for the corresponding tetraspanin. (b) Relative expression of the indicated tetraspanin in WT cells and cells not expressing the two other tetraspanins, as determined by flow‐cytometry. (c) Number and size of particles recovered from parental and KO cells, determined using nano‐particle tracking. The graphs show the mean ± SEM and individual biological replicates. (d) Western‐blot analysis of CD9, CD81 and CD63 in cell lysates and EVs. The lysates and EV samples were loaded on the same gel and are shown using the same settings. (e) Relative enrichment of CD9, CD81 and CD63 in the various cell lines, calculated as the signal in EVs divided by the signal in lysates, with parental cells serving as the reference. The graph shows the mean ± SEM and individual biological replicates. The difference between parental and KO cells was statistically analysed using a ratio paired t‐test. *, p < 0.05.

FIGURE 4

Volcano plots showing the relative expression of the proteins identified in the EVs released by parental and the different MCF7 KO cells. The changes in protein expression (expressed as the Log2 (fold change KO/WT)) are shown according to the statistical significance of the observed variations (expressed as the ‐Log10 (p‐value)). Only proteins identified with at least two peptides in three out of four (CD63) or five out of six replicates were considered. CD63 was never detected in CD63 KO cells. CD9 and CD81 were not detected in one and four replicates, respectively. When detected, the amount of CD9 in CD9 KO EVs was in average 1.4% of the amount in WT EVs.

FIGURE 5

Impact of a deficiency in both CD9 and CD81 on EV production and composition. (a) Flow‐cytometry analysis of the surface expression of CD9 and CD81 in parental MCF7 cells and CD9, CD81 dKO cells. (b) Number and size of particles recovered from parental and CD9,CD81 dKO cells, using nano‐particle tracking. The graphs show the mean ± SEM and individual biological replicates. (c) Volcano plots showing the relative expression of the proteins identified in EVs released by parental MCF7 cells and CD9, CD81 dKO cells (expressed as the Log2 (fold change KO/WT)) according to the statistical significance of the observed variations (expressed as the ‐Log10 (p‐value)). Only proteins quantified with at least two peptides in three out of four replicates in parental cells were analysed.

FIGURE 6

Impact of the deficiency in both CD9 and CD81 on CD9P‐1 and EWI‐2 expression levels. (a and b) Western‐blot analysis of CD9, CD81, CD63 and CD9P‐1 in cell lysates and EVs of parental and KO cells. The specific antibodies were revealed with a secondary antibody labelled with Alexa Fluor 680 except the CD9 antibody which was revealed with a dylight 800‐labeled secondary antibody. The graph shows the mean ± SEM, as well as individual biological replicates, of the relative enrichment of CD9P‐1 in the various cell types. The difference between parental and KO cells was statistically analysed using a ratio paired t‐test. *, p < 0.05. (c) Flow‐cytometry analysis of EWI‐2 and CD9P‐1 levels at the surface of parental and CD9, CD81 dKO cells. The graphs show the mean value of geometric mean fluorescence intensities and individual biological replicates of KO cells relative to parental cells. (d) Relative EWI‐2 and CD9P‐1 expression levels at the surface of parental, CD9 KO and CD81 KO cells, determined by flow‐cytometry. (e and f) After biotin‐labeling of surface proteins, parental and CD9, CD81 dKO cells were lysed and immunoprecipitations were performed as indicated on the top. Immunoprecipitated proteins were visualised using Alexa 680‐labelled streptavidin. The same blot was probed with the anti‐EWI‐2 mAb which was revealed using a secondary antibody coupled to Dylight 800. The graph shows the quantification of the bands corresponding to integrins, CD9P‐1 and EWI‐2 in two independent experiments, relative to the value obtained for WT samples. Int: Integrin; Strept: Streptavidin.