Interaction and uptake of exosomes by ovarian cancer cells

Human ovarian cancer SKOV3, embryonic kidney HEK293 and neuroglioma H4 cell lines were grown in Dulbecco's Modified Eagle Medium (DMEM) (Sigma) at 37°C, 5% CO₂ supplemented with 10% fetal calf serum (Gibco), 1% penicillin/streptomycin solution (Gibco).

Confluent SKOV3, HEK293 and H4 cells were cultivated for 24 h in serum-free medium. The supernatant was collected and centrifuged, at 500, 10,000 and 100,000 × g 10, 20 and 120 min, respectively, at 4°C. The pellet of the last centrifugation consisted of secreted membrane vesicles. Sucrose-density-fractionation was performed as described previously [21].

Cellular extracts were obtained by solubilization of centrifuged cells in Triton X-100 buffer (50 mM Tris-HCl pH 7.5, 5 mM EDTA, 1% Triton X-100, 0.02% protease inhibitors cocktail, Complete, Roche), for 30 min. Total protein concentration was determined by the bicinchoninic acid (BCA) method.

Glycoproteins from total cellular extracts and secreted membrane vesicles were stained after transfer to polyvinylidene difluoride (PVDF) membrane with lectins. Concanavalin A (Con A) (Sigma), biotinylated Sambucus nigra (SNA) and Maackia amurensis lectin (MAL) (Galab Technologies) were used. Glycoproteins were fixed on the PVDF membrane with 25% (v/v) 2-propanol and 10% (v/v) acetic acid for 5 min. The membranes were blocked for 1 h with TBS, 0.1% Tween-20 (TTBS) for Con A or with TTBS containing 2% BSA for SNA and MAL. For Con A detection the membrane was incubated overnight with 25 μg/ml Con A in TTBS containing 1 mM CaCl₂ and 1 mM MgCl₂ (TTBSS) followed by 1 h incubation with 0.5 μg/ml horseradish peroxidase type I (Sigma) in TTBSS. For SNA and MAL detection, membranes were incubated overnight in TTBS with 0.5 or 5 μg/ml SNA or MAL lectin, respectively. Membranes were further incubated for 1 h with 0.02 μg/ml streptavidin-peroxidase (Sigma). Detection was performed with the Immobilon Western chemiluminescent HRP substrate (ECL) (Millipore).

Immunoblot was performed as previously described [22]. The following antibodies were used: mouse anti-CD9 (1:5000) and mouse anti-L1 (L1-11A) (1:3).

Hydrolysis of α2,3- and α2,6-linked NeuAc from total protein cellular extracts and exosomes was carried out overnight at 37°C by the addition of 15 mU neuraminidase from Vibrio cholerae or from Arthrobacter urefaciens (Roche) in 50 mM sodium acetate pH 5.5 containing 4 mM CaCl₂, and 50 mM sodium acetate pH 5.0, respectively. For specific hydrolysis of α2,3-linked NeuAc, 9 U neuraminidase from Streptococcus pneumoniae (Prozyme, Glyko) in 50 mM sodium phosphate pH 6.0 were used, for 1 hour at 37°C.

SKOV3 exosomes were labeled with carboxyfluoresceine diacetate succinimidyl-ester (CFSE) (Invitrogen) as previously described [12]. Briefly, exosomes (20 μg) collected after a 100,000 × g ultracentrifugation were incubated with 7.5 μM CFSE for 30 min at 37°C in a final volume of 200 μl PBS containing 0.5% BSA. Labeled exosomes (Exos-CFSE) were 65-fold diluted with DMEM supplemented with 10% vesicles-free fetal calf serum and pelleted by ultracentrifugation for 16 h at 10,0000 × g, 12°C. Exos-CFSE were resuspended in DMEM and incubated with SKOV3 cells at 37 or 4°C.

When indicated Exos-CFSE or cells were treated for 30 min with 100 μg/ml proteinase K, or for 2 h with 15 mU neuraminidase from V. cholerae or from A. urefaciens (Roche), before uptake. SKOV3 cells were also incubated, 30 min prior to and during uptake, with the inhibitors 10 μg/ml chlorpromazine, 5 μg/ml cytochalasin D, 50 μM 5-ethyl-N-isopropyl amiloride (EIPA) or 2% methyl-beta-cyclodextrin, or with 150 mM of the monosaccharides D-glucose, D-galactose, α-L-fucose, α-D-mannose, D-N-acetylglucosamine, and the disaccharide β-lactose (Sigma).

Uptake assays were always performed in the presence of the compounds and analyzed after 2 or 4 h by immunofluorescence microscopy or flow cytometry.

Immunofluorescence microscopy was done as previously described [22]. Primary antibodies were: mouse IgG anti-alpha-tubulin DM1A (1:2000) (Sigma), mouse IgG anti-EEA1 (1:100) (BD Biosciences), mouse IgG anti-LAMP1 H4A3 (1:100) (BD Biosciences) and mouse IgG anti-caveolin-1 (1:50) (Santa Cruz). Secondary antibody was donkey anti-mouse IgG AlexaFluor 594 (1:500) (Molecular Probes).

Images were acquired on a Leica DMRB microscope using a DFC340FX camera coupled to the microscope, and Leica application suite V3.3.0 software. For colocalization, images were acquired on a confocal SP5 microscope. Each picture was acquired with laser intensities and amplifier gains adjusted to avoid pixel saturation. Each fluorophore used was excited independently and sequential detection was performed. Each picture consisted of a z-series of 20 images of 1024-1024 pixel resolution with a pinhole of 1.0 airy unit. Colocalization analysis was performed using the open source Image J version 1.38 http://​rsb.​info.​nih.​gov/​ij/​.

SKOV3 cells incubated with Exos-CFSE for 4 h at 37 or 4°C were washed with PBS, detached using trypsin and resuspended in PBS with 2% fetal calf serum (Gibco). Flow cytometry analysis was performed on a Cyflow ML cytometer (Partec) using Flowmax software 2.56 (Partec). Gate was set on living cells based on forward/side scatter properties and a minimum of 10³ events within the gated live population were collected per sample. Exos-CFSE uptake by SKOV3 cells was measured by the shift in peak fluorescence intensity of CFSE, calculated by the geometric mean of the population. SKOV3 cells with no Exos-CFSE uptake (unlabelled) were used as controls for cell autofluorescence. Results were expressed as means ± S.D. and comparison between two means was performed by Student t test. A P value lower than 0.05 was considered significant.

Since it has been shown that SKOV3 exosomes can be internalized by NK cells [12] here we have investigated if they are internalized by the cell line from where they originated. With that purpose exosomes produced in the absence of fetal bovine serum and collected at 100,000 × g were labeled with CFSE (Exos-CFSE) and incubated with SKOV3 cells. CFSE is a membrane permeable non-fluorescent compound that becomes fluorescent after cleavage of its acetate groups by intracellular esterases [23]. A punctuated green fluorescent pattern corresponding to Exos-CFSE interaction/internalization was observed after 30 min and it increased up to 4 hours of incubation (Figure 1A). To obtain a higher sensitivity exosomes were also biotinylated and detected with streptavidin Alexa-488, and the interaction was observed already 0.5 min after incubation (data not shown). Colocalization studies by immunofluorescence confocal microscopy of Exos-CFSE and alpha-tubulin, a microtubule marker, showed that Exos-CFSE were localized in the same z-stacks as microtubules, thus, confirming their internalization (Figure 1B). Supporting this conclusion was the observation that the exosomes were not removed by acid wash of cells pre-exposed to Exos-CFSE (data not shown).

Several endocytic mechanisms have been described to mediate the entry of material into the cells [24]. Here, we observed that a decrease in temperature from 37 to 4°C caused a reduction of 80 ± 8% (n = 6) in the number of labeled cells as well as a decrease of 77 ± 9% (n = 6) in the uptake for the positively labeled cells, based on the shift in fluorescence intensity corresponding to the geometric mean of the population peaks (Figure 2A). These results indicated that exosomes uptake was mediated by endocytosis in an energy-dependent process. Exos-CFSE were also found to partially colocalize with the endosomal marker EEA1 (Figure 2B), thus showing the participation of clathrin-mediated endocytosis in exosome uptake. This conclusion was further supported by uptake inhibition (19 ± 18%, n = 4) (Figure 2C) with 10 μg/ml chlorpromazine, which blocks clathrin-mediated endocytosis [25]. Colocalization with the lysosomal marker LAMP1 (Figure 2B) indicated that at least a part of the exosomes were targeted to the lysosome.

In addition other inhibitors have been tested. First, cells incubated with 5 μg/ml cytochalasin D, which is known to inhibit actin polymerization and consequently inhibit phagocytosis [26] as well as other endocytic pathways [27] showed an uptake reduction of 32 ± 7% (n = 4) (Figure 2C). EIPA at 50 μM, which is known to block macropinocytosis [28], caused an uptake reduction of 36 ± 13% (n = 5) (Figure 2C), thus suggesting that exosomes were internalized via macropinocytosis. Methyl-beta-cyclodextrin, that is used to deplete cholesterol from cellular membranes [29], decreased Exos-CFSE uptake (44 ± 8%, n = 5) (Figure 2C). However, there was no colocalization with caveolin-1 (Figure 2B), which is a marker of caveolae that are enriched in cholesterol rich domains. These results indicated that exosome uptake could occur via a cholesterol associated pathway independent of caveolae or that methyl-beta-cyclodextrin affected exosomes membrane integrity, thus decreasing uptake efficiency.

The recognition between exosomes and the target cell has been reported to involve proteins present at the cell surface of both exosomes and target cells [11, 14]. Here, Exos-CFSE or SKOV3 cells were digested with proteinase K, a broad specificity protease, and uptake efficiency was analyzed by flow cytometry analysis. Uptake levels of digested Exos-CFSE were found to be lower (45 ± 12%, n = 6) than Exos-CFSE without treatment (Figure 3A). As control, fluorescence of Exos-CFSE was not affected by proteinase K treatment (data not shown). Furthermore, a decrease of 32 ± 8% (n = 6) in uptake was observed in SKOV3 cells treated with proteinase K (Figure 3B). Therefore, proteins present in exosomes and also in SKOV3 cells are required, at least in part, for internalization by target cells.

Glycan-lectin interactions have been suggested to play a role in the uptake of exosomes by target cells [19, 20], therefore, we have investigated if glycans would play a role in exosome uptake by SKOV3 cells.

First, glycoproteins of cellular extracts and secreted vesicles from SKOV3 cells were detected with the lectins concanavalin A (Con A; binds α-mannosyl containing-branched glycans predominantly of the high-mannose followed by hybrid- and biantennary complex type structures to a lower extent) [30], Sambucus nigra lectin (SNA; recognizes NeuAcα2,6Gal/GalNAc) and Maackia amurensis lectin (MAL; binds NeuAcα2,3Galβ1,4GlcNAc/Glc) [31]. The profile from exosomes was distinct from that of cellular extracts with the three lectins (Figure 4A). Detection with SNA and MAL was almost totally abolished after digestion with V. Cholerae and A. urefaciens neuraminidases (cleave terminal α2,3- and α2,6-linked NeuAc), whereas only MAL binding was abolished after S. pneumoniae (cleaves only α2,3-linked NeuAc) digestion (Figure 4B), thus confirming that SNA and MAL binding to secreted vesicles glycoproteins was specific. The sialic acid was not present in sialylated Lewis epitopes, since there was no detection with anti-sialyl-Lewis^a or anti- sialyl-Lewis^x antibodies (data not shown).

SKOV3 secreted vesicles are constituted by exosomes and apoptotic vesicles, which can be fractionated by using a sucrose gradient, as previously described [21, 22]. CD9 is a tetraspanin protein that has been used as exosome marker [32, 33]. After sucrose gradient separation of the secreted vesicles all the SNA and MAL binding was found in the exosome-containing fractions (fractions 2-5) and not in the apoptotic blebs (fractions 8-11) (Figure 4C).

Specific patterns of protein glycosylation were also found for exosomes from two other human cell lines, embryonic kidney HEK293 and neuroglioma H4 cells (Figure 5).

To investigate a possible role for glycosylation in exosome uptake Exos-CFSE were desialylated with V. cholerae and A. urefaciaens sialidases and exosomes uptake was monitored by immunofluorescence microscopy and flow cytometry analysis of cell fluorescence intensity. Uptake efficiency of exosomes after neuraminidase treatment was slightly increased (16 ± 14%, n = 6) when compared with exosomes incubated with neuraminidase buffer (Figure 6A), however, the observed increase was not statistically significant using the Student t test (P = 0.0764). In addition, increases in the uptake of exosomes incubated with neuraminidase (38 ± 13%, n = 6) or neuraminidase buffer (23 ± 14%, n = 6) were detected when compared with Exos-CFSE without treatment (Figure 6A).

The uptake assay was also performed in the presence of 150 mM of the monosaccharides D-glucose (control), D-galactose, α-L-fucose, α-D-mannose and D-N-acetylglucosamine, and the disaccharide β-lactose. Decreases of 23 ± 7%, 24 ± 14%, 25 ± 16%, 27 ± 8%, 19 ± 15% and 20 ± 8% (n = 6), respectively, in the uptake of Exos-CFSE in comparison with control without sugar were observed. The incubation with α-D-mannose led to a higher decrease in uptake relatively to the control sugar D-glucose, however the difference was not statistically significant (Figure 6B).