Biochemical and biological characterization of exosomes containing prominin-1/CD133

For immunoblotting, microvesicles and prom1-exo resuspended in PBS were checked for consistency by NTA. Aliquots of microvesicles, exosomes and FEMX-I total cell lysates containing 1–10 μg of protein were mixed 1:1 with SDS sample buffer (NuSep, Bogart, GA) containing 2% 2-mercaptoethanol, boiled for 5 min, and loaded onto a 8% Tris/Glycine/SDS gel. Electrophoretic separation of proteins was performed at a constant voltage of 120 V for 2 h, and electrophoretic transfer of the proteins into Hybond ECL membrane (GE Healthcare, Waukesha, WI) was carried out at constant amperage (30 mA) for 15 h. The blots were blocked with 5% dry milk in Tris-buffered saline containing 0.05% Tween 20 (pH 7.5), antibody, and incubated with W6B3C1 anti-prominin-1 (Miltenyi Biotec, Auburn, CA), or alix 3A9 clone (Cell Signaling Technology, Danvers, MA) at 1:1000 dilution in TBS-T for 5 h at room temperature. After washing with TBS-T, blots were incubated with IRDye 800CW secondary antibody (Li-Cor Biosciences, Lincoln, NE) in TBS-T (1:20,000) for 45 min at room temperature. Finally, blots were washed with TBS-T, scanned by Odyssey infrared imaging system and analyzed by Odyssey 2.1 application software (Li-Cor Biosciences). Gel band densitometric quantification was performed employing the ImageJ64 software (http://​rsbweb.​nih.​gov/​ij).

An automated electrospray ionization (ESI)-tandem mass spectrometry approach was used, and data acquisition and analysis were carried out as described previously [27, 28] with modifications. The lipid extracts from the FEMX-I cell pellets were dissolved in 1 ml chloroform. An aliquot of 50 μl of each extract in chloroform was used for analysis. Precise amounts of internal standards, obtained and quantified as previously described [29], were added in the following quantities (with some small variation in amounts in different batches of internal standards): 0.6 nmol di12:0-PC, 0.6 nmol di24:1-PC, 0.6 nmol 13:0-lysoPC, 0.6 nmol 19:0-lysoPC, 0.3 nmol di12:0-PE, 0.3 nmol di23:0-PE, 0.3 nmol 14:0-lysoPE, 0.3 nmol 18:0-lysoPE, 0.3 nmol di14:0-PG, 0.3 nmol di20:0(phytanoyl)-PG, 0.3 nmol di14:0-PA, 0.3 nmol di20:0(phytanoyl)-PA, 0.2 nmol di14:0-PS, 0.2 nmol di20:0(phytanoyl)-PS, and 0.23 nmol 16:0–18:0-PI. The sample and internal standard mixture was combined with solvents, such that the ratio of chloroform/methanol/300 mM ammonium acetate in water was 300/665/35, and the final volume was 1.4 ml. The microvesicle samples were prepared similarly, except that the entire sample was analyzed, 1/3 of the above standard amounts were added, and the final volume was 0.75 ml. The unfractionated lipid samples with internal standards were introduced by continuous infusion into the ESI source on a triple quadrupole MS/MS (API 4000, Applied Biosystems, Foster City, CA). Samples were introduced using an autosampler (LC Mini PAL, CTC Analytics AG, Zwingen, Switzerland) fitted with the required injection loop for the acquisition time and presented to the ESI needle at 30 μl/min. Sequential precursor and neutral loss scans of the extracts produce a series of spectra with each spectrum revealing a set of lipid species containing a common head group fragment. Lipid species were detected with the following scans: PC and lysoPC, [M + H]⁺ ions in positive ion mode with Precursor of 184.1 (Pre 184.1); PE and lysoPE, [M + H]⁺ ions in positive ion mode with Neutral Loss of 141.0 (NL 141.0); PG, [M + NH₄]⁺ in positive ion mode with NL 189.0 for PG; PI, [M + NH₄]⁺ in positive ion mode with NL 277.0; PS, [M + H]⁺ in positive ion mode with NL 185.0; PA, [M + NH₄]⁺ in positive ion mode with NL 115.0. SM was determined from the same mass spectrum as PC (precursors of m/z 184 in positive mode) [27, 30] and by comparison with PC internal standards using a molar response factor for SM (in comparison with PC) determined experimentally to be 0.39. The collision gas pressure was set at 2 (arbitrary units). The collision energies, with nitrogen in the collision cell, were +28 V for PE, +40 V for PC (and SM), +25 V for PA, PI and PS, and +20 V for PG. Declustering potentials were +100 V for all lipids. Entrance potentials were +15 V for PE and +14 V for PC (and SM), PA, PG, PI, and PS. Exit potentials were +11 V for PE and +14 V for PC (and SM), PA, PG, PI, PS. The scan speed was 50 or 100 u per sec. The mass analyzers were adjusted to a resolution of 0.7 u full width at half height. For each spectrum, 9 to 150 continuum scans were averaged in multiple channel analyzer (MCA) mode. The source temperature (heated nebulizer) was 100°C, the interface heater was on, +5.5 kV or -4.5 kV were applied to the electrospray capillary, the curtain gas was set at 20 (arbitrary units), and the two ion source gases were set at 45 (arbitrary units). The background of each spectrum was subtracted, the data were smoothed, and peak areas integrated using a custom script and Applied Biosystems Analyst software, and the data were isotopically deconvoluted. The first and typically every 11^th set of mass spectra were acquired on the internal standard mixture only. Peaks corresponding to the target lipids in these spectra were identified and molar amounts calculated in comparison to the two internal standards on the same lipid class, except for PI, which was quantified in relation to a single internal standard. Ether-linked (alk(en)yl,acyl) lipids were quantified in comparison to the diacyl compounds with the same head groups without correction for response factors for these compounds as compared to their diacyl analogs. To correct for chemical or instrumental noise in the samples, the molar amount of each lipid metabolite detected in the “internal standards only” spectra was subtracted from the molar amount of each metabolite calculated in each set of sample spectra. The data from each “internal standards only” set of spectra was used to correct the data from the following 10 samples. Finally, the data were corrected for the fraction of the sample analyzed and normalized to the mg protein to produce data in the units nmol/mg.

The miRNA profiling array was carried out using Applied Biological Materials miRNA profiling service (ABM C201). Total RNA from FEMX-I cells and exosomes was prepared byQiazol extraction followed by poly-A tailing reactions and miRNA cDNA synthesis (ABM C204). 250 ng of cell’s total RNA and exosomes’ RNA were used in cDNA synthesis. Both cells’ and exosomes’ cDNA synthesis were carried out simultaneously and equal volume of cDNA synthesis reaction product was used in the subsequent profiling. The Ct values for each miRNA-specific cDNA were compared between FEMX-I cells and exosomes. Real-time qPCR reactions and instrumental analysis was performed using Roche LightCycler480. Lists of miRNAs were generated by pair-wise comparison of our expression data sets (cells vs exosomes). Differentially expressed miRNAs were analyzed by the Ingenuity Pathway Analysis software (Ingenuity Systems, Redwood City, CA) to identify the biological functions that were most significant to the data sets.

Cells were seeded on poly-L-lysine coated chamber slides and grown overnight. Following aspiration of media, cells were fixed in 4% paraformaldehyde (PFA), washed with PBS, permeabilized in 0.5% Tween 20 and blocked with goat serum. After washing with PBS, cells were incubated overnight at 4°C with primary antibodies in 1% BSA-PBS, followed by washes and a 45-minutes incubation at room temperature with fluorochrome-labeled secondary antibody in 1% BSA in PBS. Fluorescent cells were analyzed by a CKX41 fluorescence inverted microscope (Olympus, Center Valley, PA).

We previously reported that human FEMX-I metastatic melanoma cells released into the extracellular medium prominin-1-expressing microvesicles [9]. To investigate their nature, we cultured FEMX-I cells as spheroids under serum-free conditions for six days and compared a “classical” microvesicle preparation, based on differential centrifugation [9], with a prominin-1⁺ preparation, illustrated in Figure 1A, based on the combination of differential centrifugation, filtration and prominin-1-based immuno-magnetic selection (prom1-exo). The final pH at time of harvest was 6.7, which resembled in vivo tumor growth conditions, where low pH condition is a hallmark of tumor malignancy, particularly for malignant melanoma cells, which, differently from normal cells, can survive in an acidic microenvironment [31]. Low pH conditions reportedly increase exosome release and uptake by cancer cells [32]. We used serum-free medium in the present study because serum supplements (such as fetal calf serum) often contain vesicles as well as aggregates of serum proteins, which may interfere with the isolation and characterization of FEMX-I exosomes. By NTA, we determined both size distribution and relative concentration of microvesicles and prom1-exo in the supernatants of FEMX-I cells. As shown by NTA of PKH67-stained microvesicles, the “classical” microvesicle preparation showed several peaks, ranging from 70 to 550 nm, while prom1-exo yielded a single peak of about 100 nm (Figure 1B); persistent binding of anti-prominin-1 50 nm-immunomagnetic beads resulted in an apparent over-estimation of the exosomal size and broadening of the size distribution peak. The concentration of microvesicles and prom1-exo in FEMX-I supernatant were 3 ± 0.4 ×10⁹/ml and 0.35 ± 0.2 × 10⁹/ml, respectively. We then employed the same methodology (Figure 1A) to investigate whether it was possible to isolate prom1-exo from different prominin-1-expressing cancer cell lines. We found that prominin-1-immunomagnetic selection resulted in isolation of cancer exosomes also from human prominin-1-expressing Caco-2 colon carcinoma cells (Figure 1B). An approximately 10-fold difference in concentration of microvesicles and prom1-exo was found also in the cell supernatants of Caco-2 cells (1.5 ± 0.35 × 10⁹/ml and 0.18 ± 0.05 × 10⁹/ml, respectively). Similarly to what we observed in FEMX-I cells, Caco-2 cells microvesicles had a broad size range, while prom1-exo had a single 100 nm-peak.

Comparison of total cell lysates, microvesicles and prom1-exo from FEMX-I cells by Western blotting (Figure 2) revealed a great enrichment in prominin-1 and in the exosomal protein alix in prom1-exo vs. the FEMX-I cells themselves (53- and 184-fold for prominin-1 and alix, respectively) and vs. microvesicles (78- and 168-fold for prominin-1 and alix, respectively). To investigate whether they had the biochemical characteristics of bona fide exosomes, we analyzed the proteolipidic composition of prom1-exo from FEMX-I cells. Three independent preparations were used to determine their protein composition via MS/MS mass spectrometry (see search methods in Additional file 1: Table S1). A total of 282 proteins were confidently assigned across all three samples (Additional file 2: Table S2 and Additional file 3: Table S3). We further highlighted an ensemble of proteins among which could be verified with two or more stringent peptide spectral matches (PSM) in all three replicates or those observed with three or more stringent PSM in any two replicates (Additional file 4: Table S4). This subset of 154 proteins is highly enriched for physiological processes (Additional file 5: Figure S1), involving membrane bound vesicles [count 40, p-value 6.4E-21] and endocytosis [count 20, p-value 7.3E-11] complexes, and including all of the 14 proteins most expressed in exosomes according to the compilation of peer-reviewed data hosted on the Exocarta site [33] (Table 1). Since the biogenesis of exosomes takes place in late endosomes to end up in multivesicular bodies (MVB), we first checked our list of proteins for those known to be involved with that particular compartment (Table 2). Reassuringly, we identified the bro1 domain-containing proteins alix and brox, known to function in association with the ESCRT (Endosomal Sorting Complex Required for Transport) pathway to help mediate intraluminal vesicle formation at multivesicular bodies and the abscission stage of cytokinesis. Various ESCRT components, central to MVB biogenesis, were identified in prom1-exo, including five ESCRT-I proteins [34], three ESCRT-III proteins and many other ESCRT-associated proteins (Table 2). Other proteins, related to their endosomal origin, were identified, including membrane transport and fusion proteins (GTPases, Annexin A2, A4, A5, A6 and A11); eight tetraspanins (TSPAN 4,6,9,14; CD63; CD81; CD82; CD9), and five Rab proteins (Additional file 4: Table S4). Interestingly, the immunosuppressive Immunoglobulin superfamily member 8 (IgSF8), also named CD81 partner 3, known to interact with CD81, CD9 and CD82 as well as with integrin alpha-3/beta-1 and integrin alpha-4/beta-1, was highly expressed. The absence of endoplasmic reticulum proteins, such as calnexin and Grp78, and of Golgi proteins, such as GM130, indicated no contamination of vesicles of other compartments in prom1-exo preparations.

Other cancer-related proteins and/or proteins implicated in cancer progression were identified, including CD44 [35], Hsp70 [36], annexin A2 [37‐40], as well as components involved in Wnt (SFRP1 = secreted frizzled-related protein 1) and Ras signaling, including the GTP-binding proteins Rap1b and Rap2b, reportedly involved in the activation of ERKs [41], the 14-3-3 protein, a family of exosomal proteins that have a matrix metalloproteinase-1 stimulating effect for dermal fibroblasts [42], and disintegrin and metalloproteinase domain-containing protein 10 (ADAM 10) (Additional file 4: Table S4). A perinuclear pool of prominin-1, associated with integrin-beta 1 (CD29), expressed in FEMX-I exosomes, was detected by fluorescence microscopy (Figure 3). Interestingly, a striking correspondence between prominin-1 and CD29 in FEMX-I cells was observed, suggesting their co-localization in endosomal compartments.

A lipid composition analysis of prom1-exo and parental FEMX-I cells was performed through ESI MS/MS (Figures 4 and 5). A typical lipid raft composition of prom1-exo was observed, with 400% increase in sphingomyelin, 240% increase in phosphatidylserine, 290% in phosphatidylglycerol, 2150% in lyso-phosphatidylethanolamine) and 1190% in lyso-phosphatidylchoxline. A great number of membrane lipids were significantly different between prom1-exo and the membrane compartment of parental FEMX-I cells (Table 3). To offset the elevated sphingolipid levels, phosphatidylcholine levels were decreased by 26%, resulting in similar choline-containing lipid levels between prom1-exo and the FEMX-I plasma membrane. A 45% decrease in phosphatidylinositol content of prom1-exo also partially accounted for the observed increase in raft-associated lipid species of the exosomes.

The miRNA “cargo” of prom1-exo was significantly different from the parental cell content. Of the 1,058 miRNA species investigated, only 49 were over-expressed in prom1-exo (Table 3), including miRNAs known to mediate immune tolerance, and 13 cancer/metastasis-associated miRNAs. In particular, miR-216b, a well-known tumor and metastasis suppressor mi-RNA, that targets Ras [43, 44], was highly expressed in prom1-exo and undetectable in FEMX-I cells, indicating a detoxification role for prom1-exo; let-7i, associated with metastatic progression [45‐47], was found to be expressed at levels 53-fold higher in prom1-exo than in FEMX-I cells. Also, miR-10a, reportedly involved in the metastatic process and immune-escaping [48, 49] was 3.2-fold higher in prom1-exo than in parental cells.

Exposure of FEMX-I cells to PKH-67-labeled prom1-exo for 3 h resulted in massive green perinuclear fluorescence (Figure 6A), co-localized with the red fluorescence of the intracellular pool of prominin-1 upon incubation of the cells with phycoerythrin-conjugated monoclonals (Figures 6A and 3A-B). Interestingly, also exposure of human MSC to PKH-67-labeled prom1-exo for 3 h resulted in intra-cellular localization of fluorescent prom1-exo and of prom1-exo-associated prominin-1 (Figure 6A); for MSC, the punctate pattern differed from the perinuclear accumulation for FEMX-I cells, presumably for the lack of an endogenous pool of prominin-1 in MSC. Since a shorter (1 h)-exposure of FEMX-I to PKH-67-labeled prom1-exo resulted in exclusive, although minor, perinuclear accumulation of green fluorescence (data not shown), the complete absence of puncta in FEMX-I cells may be due to the rapid kinetics of intracellular exosome trafficking or turnover. To confirm the intracellular delivery of prominin-1 by prom1-exo, MSC were incubated with prom1-exo prepared from FEMX-I cells transiently transfected with a prominin-1-GFP fusion plasmid. After 3 h, extensive fluorescence from prominin-1-GFP was detected in the intracellular compartment of MSC (Figure 6B), confirming that prominin-1 was effectively delivered to MSC. To investigate whether direct transfer of prom1-exo occurred from FEMX-I to MSC in mixed cultures, we co-cultured the cells at 5:1 ratio (FEMX-I:MSC) for 24 h, and analyzed the expression of prominin-1 by immunofluorescence. Figure 6 clearly shows transfer of prominin-1 from FEMX-I to the intracellular compartment of MSC. The apparent contrast between the massive uptake of exosomes in Figure 6 and the relatively low transfer of exosomes from FEMX-I to MSC in Figure 7 may be explained by the technical differences of the two experiments (sudden addition of exosomes in Figure 6 and gradual release of exosomes in Figure 7).

We then investigated whether exposure of MSC to prom1-exo resulted in biological effects, such as changes in their invasiveness, measured by the capacity of MSC to pass through a Matrigel layer. Interestingly, a 90% increase in Matrigel invasion was observed after a standard 24 h-assay, compared with mock-treated MSC (Figure 8).