Ultrafiltration combined with size exclusion chromatography efficiently isolates extracellular vesicles from cell culture media for compositional and functional studies

Abstract

Appropriate isolation methods are essential for unravelling the relative contribution of extracellular vesicles (EVs) and the EV-free secretome to homeostasis and disease. We hypothesized that ultrafiltration followed by size exclusion chromatography (UF-SEC) provides well-matched concentrates of EVs and free secreted molecules for proteomic and functional studies. Conditioned media of BEAS-2B bronchial epithelial cells were concentrated on 10 kDa centrifuge filters, followed by separation of EVs and free protein using sepharose CL-4B SEC. Alternatively, EVs were isolated by ultracentrifugation. EV recovery was estimated by bead-coupled flow cytometry and tuneable resistive pulse sensing. The proteomic composition of EV isolates and SEC protein fractions was characterized by nano LC-MS/MS. UF-SEC EVs tended to have a higher yield and EV-to-protein rate of purity than ultracentrifugation EVs. UF-SEC EVs and ultracentrifugation EVs showed similar fold-enrichments for biological pathways that were distinct from those of UF-SEC protein. Treatment of BEAS-2B cells with UF-SEC protein, but not with either type of EV isolate increased the IL-8 concentration in the media whereas EVs, but not protein induced monocyte adhesion to endothelial cells. Thus, UF-SEC is a useful alternative for ultracentrifugation and allows comparing the proteomic composition and functional effects of EVs and free secreted molecules.

Figure 1

Flow chart of the EV isolation protocols. Red arrows indicate where samples were taken for determination of EV recovery based on CD63+ CD81+ bead-coupled flow cytometry. For nano LC/MS-MS, the isolation protocols were scaled up to 120 ml of cell culture medium as starting material. For UF-SEC, the 120 ml were concentrated by 10 kDa filtration and subsequently run over the SEC column in two aliquots of 0.5 ml. The EV containing fractions from both SEC runs (fractions 6 to 10 or 11) were then pooled and concentrated by 10 kDa filtration. For UC, the 120 ml were processed in 3 portions of 40 ml, corresponding to the maximal volume of the UC tubes.

Figure 2

Evaluation of the isolation protocol. (a) A representative standard curve demonstrating that the CD63⁺CD81⁺ bead-coupled flow cytometry signal is proportional to the EV concentration determined by tuneable resistive pulse sensing (TRPS), allowing semi-quantitative EV measurements. (b) EV recovery after 0.22 µm filtration of cell-depleted media. The data represents flow cytometry samples A and B in Fig. 1. EV quantity in unfiltered cell-depleted media was set to 100%. (c) Protein concentration of the 24 SEC fractions determined by Bradford assay (µg/ml) and EV quantity determined by bead-coupled flow cytometry (relative fluorescent units; RFU) for the following combinations of antibodies (capture/detection): anti-CD63/anti-CD81, anti-CD63/anti-CD63, anti-CD81/anti-CD81, anti-CD9/anti-CD9 (sampling step D indicated in Fig. 1). Graph shows medians without error bars, n = 3. (d) EV mode size, mean size and concentration for SEC fractions 6–12 as determined by TRPS. The graph shows the median and range, n = 3. (e) Top: Ponceau S total protein staining of SEC fractions 5 to 18; bottom: Western blot of SEC fractions 5 to 18 for EV marker proteins CD63, CD81, MFGE8 and HSP70. For the lane MFGE8 s.p., a standardized protein concentration of 5 µg was loaded for each fraction. For all other stainings, whole fractions were precipitated and loaded on the gel without standardization of the protein content. (f) EV recovery in the final UF-SEC EV isolate (Fractions 6–11) and the final UF-SEC protein isolate (Fractions 13–19) as determined by TRPS or CD63⁺CD81⁺ flow cytometry. The EV quantity in conditioned culture medium has been set as 100%. (g) Comparison of the EV recovery for UF-SEC, UC or UC-wash, determined by TRPS (overall p-value = 0.0005) and CD63⁺CD81⁺ bead-couple flow cytometry (overall p-value = 0.0009; sampling steps E, G and H, respectively). Data was analysed using the Kruskall Wallis test followed by Dunn’s post-test. ***p < 0.001. (h) The EV-to-protein rate of purity base on TRPS and Bradford measurements of conditioned cell culture media, UF-SEC EVs and UC EVs. The statistical difference between UF-SEC EVs and UC EVs was assessed using the Mann-Whitney U test.

Figure 3

Characterization of EVs in unfiltered conditioned media, UF-SEC EVs and UC EVs. (a) Size distribution histogram of the EVs in (a) unfiltered conditioned media (b) UF-SEC EV isolates and (c) UC EV isolates as determined by TRPS. Graph shows the mean ± SEM, n = 5. (d) The percentage of particles >200 nm in conditioned media, UF-SEC EVs and UC EVs according to TRPS. The bar indicates the median diameter. **p 0.001–0.01 (e) Size distribution of EVs in UF-SEC and UC isolates based on cryo-TEM. The bar indicates the median diameter. (f) Low magnification and (g) detail cryo-TEM recording of UF-SEC isolated EVs from BEAS-2B.

Figure 4

Proteomic characterization of UF-SEC EVs, UC EVs and UF-SEC protein from BEAS-2B cells. (a) Venn diagram of the proteins identified for UF-SEC EVs, UC EVs and UF-SEC protein. (b) Functional enrichment analysis for cellular component GO terms. (c) Functional enrichment analysis for biological process GO terms. (d) Functional enrichment analysis for manually curated biological pathways. (e) Venn diagram of proteins identified in UF-SEC EVs and UC EVs but not UF-SEC protein from BEAS-2B cells (BEAS-2B EVS) compared to the total lung cancer EV associated proteome and the lung cancer specific proteome according to Hurwitz et al.. (f) Functional enrichment analysis for the site of expression for the BEAS-2B EV proteome and the lung cancer-specific proteome proposed by Hurwitz et al..

Figure 5

Differential functional effects of EVs and free secreted protein. Effect of the EVs obtained by UF-SEC or the protein fractions from unexposed (0% CSE) or CSE-stimulated BEAS-2B cells on (a) the IL-8 secretion of naïve BEAS-2B cells and (b) adhesion of THP-1 monocytes to HUVECs (overall p-value < 0.0001). Data was analysed by the Kruskall Wallis test followed by Dunn’s posthoc test. *p = 0.01-0.05, and ***p < 0.001 compared to the unstimulated control.