Exosomes isolated from cancer patients’ sera transfer malignant traits and confer the same phenotype of primary tumors to oncosuppressor-mutated cells

Patients for the current study were recruited form the department of General Surgery at the Royal Victoria Hospital and St-Mary’s Hospital (Montreal, Canada) and underwent a written consent for blood collection in accordance to a protocol approved by the Ethics Committee of our institution (SDR-10-057). Blood samples were collected from both healthy individuals and patients who underwent resection of primary cancer and who were readmitted for metastatic disease treatment (Table 1).

Blood samples (20 ml) were collected from a peripheral vein in vacutainer tubes (Becton Dickinson) containing clot-activation additive and a barrier gel to isolate serum. Blood samples were incubated for 60 min at room temperature to allow clotting and subsequently were centrifuged at 1500 x g for 15 min. Serum was collected and a second centrifugation was performed on the serum at 2000 x g for 10 min, to clear it from any contaminating cells. Serum samples were aliquoted and stored at −80 °C until use.

We used the CRISPR/Cas9 system to establish a stable BRCA1-KO in human fibroblasts as previously described [14]. Cells were maintained as per supplier’s recommendations. When cells reached 30% confluence, they were treated with DMEM-F12 medium (Wisent, Saint-Bruno, Canada) supplemented with antibiotics and 10% cancer patient sera or control sera, which had been filtered through 0.2 μm filters. Otherwise, cells were exposed to an exosome load of 25–40 μg/ml (which corresponded to 2.6–4.1e + 07 particles/ml of culture medium. Cells were maintained in these conditions at 37 °C in humidified atmosphere containing 95% air and 5% CO₂ with medium change every second day for 3 weeks. When cells reached 80–90% confluence, they were passaged 1 in 6 using 0.05% Trypsin-EDTA (Wisent, Saint-Bruno, Canada). To confirm that there was no contamination or carry-over of cells from human serum, aliquots of the culture medium were placed in a culture plate and incubated at 37 °C, 5% CO₂ for 4 weeks.

Exosomes were isolated from serum using the Total Exosome Isolation Kit according to the manufacturer’s protocol (Invitrogen, Burlington, Canada). Exosomes were labeled using the PKH26 dye following the manufacturer recommendations (Sigma, Oakville, Canada). Labeled exosomes were diluted in labeling stop solution (PBS/FBS) and pelleted by ultra-centrifugation for 80 min at 100,000 x g at 4 °C. The pellet was washed in Hank’s Balanced Salt Solution (HBSS) with an ultra-centrifugation using the same parameters. The pelleted exosomes were re-suspended in HBSS and stored at −80 °C. 10 μg of labeled exosomes was added to cells (~5 × 10³) maintained in 8-well chamber slides (VWR, Mont-Royal, Canada). Cells were washed, fixed for 10 min with Paraformaldehyde 4%. Slides were mounted with coverslip in VECTASHIELD Mounting Medium with DAPI (Vector Laboratories, Burlington, Canada). Stained cells were visualized using an LSM780 confocal microscope (Zeiss, Toronto, Canada). Exosomes internalization was quantified using ImageJ software. Where mentioned, a batch of cells was also analyzed by flow cytometry. Cells were acquired using a FACSCalibur flow cytometer (Becton-Dickinson) at a flow rate of ~300 cells/s. Dead cells and cell debris were excluded from acquisition by gating intact cells on a FCS and SSC biparametric plot.

Morphological examination of isolated exosomes was done using transmission electron microscope (JEM-2010, Jeol Ltd., Tokyo, Japan). Briefly, 20 μl of exosomes were loaded on a copper grid and stained with 2% phosphotungstic acid. Samples were dried by incubating them for 10 min under an electric incandescent lamp. Samples were examined under electron microscope and imaged using a Hitachi H-600 TEM operating at 60 kV. In parallel, an aliquot of exosome samples was run on a Nanosight NS500 system (Nanosight Ltd., Amesbury, UK), and size distribution was analyzed using the NTA 1.3 software.

Cells and pelleted exosomes were lysed in RIPA buffer containing protease inhibitors (Sigma, Oakville, Canada). Equal amounts of proteins were resolved on 10% SDS-PAGE and transferred to a nitrocellulose membrane (BioRad, CA, USA). Membranes were blocked in TBS (20 mM Tris, 150 mM NaCl, pH. 7.6) containing 5% non-fat dry milk and exposed overnight at 4 °C to rabbit-anti-GM130 (ab52649) and mouse-anti-TSG101 (ab83) (Abcam, MA, USA), and mouse-anti-Alix (2127, Cell Signaling, MA, USA). Membranes were washed in TBST (TBS-0.05% Tween-20) and incubated with either anti-rabbit or anti-mouse peroxidase-conjugated secondary antibody for 1 h at room temperature. After several washes in TBST, the blots were developed using Immobilon Western HRP Substrate (Millipore, Etobicoke, Canada).

Cells were blocked for 1 h. with different receptor antagonists: Anti-β4 Integrin (10 μg/ml, ASC-8, ab77801, abcam), Cytostatin (1.4 μg/ml, 19,602, Cedarlane), and Heparin (10 μg/ml, H3149, Sigma). In parallel, exosomes were treated for 2 h. with RGD (300 nM, 14,501–1, Cedarlane), and Collagenase I (500 μg/ml, C0130, Sigma). Afterwards, cells were washed, mixed with treated exosomes and incubated for 6 h. Exosomes internalization was analyzed as described in previous section.

Five-week-old female NOD-SCID mice (Jackson Laboratory) were used in compliance with McGill University Health Centre Animal Compliance Office (Protocol 2012–7280). Cells growing in log phase were harvested by trypsinization and washed twice with HBSS. Mice were injected subcutaneously with 2 million cells in 200 μl HBSS/Matrigel. Mice were euthanized one month post-injection. The resulting xenotransplants were photographed and processed as indicated below.

Mice xenotransplants were collected, fixed in 10% buffered formalin, embedded in paraffin, and stained with H&E (hematoxylin and eosin) according to standard protocols or processed for immunohistochemistry. Briefly, 5 μm tissue sections were dewaxed in xylene and rehydrated with distilled water. After antigen unmasking, and blocking of endogenous peroxidase (3% hydrogen peroxide), the slides were incubated with primary antibodies (Additional file 1: Table S1). Labeling was performed using iView DAB Detection Kit (Ventana) on the Ventana automated immunostainer. Sections were counterstained lightly with Hematoxylin before mounting. Histological analyses were performed by a certified pathologist.

For this study, human BRCA1-KO fibroblasts were used as target cells. These cells were exposed for three weeks to different types of cancer patients’ sera (Table 1) and healthy patients’ sera and subsequently cells were injected subcutaneously into NOD/SCID mice. Independently of the cancer serum used, mice developed visible tumors as early as one week following inoculation. None of the mice injected with BRCA1-KO fibroblasts treated with healthy patients’ sera was found to have grown tumors after the mice were euthanized. Histopathological analyses of developing xenotransplants displayed features of adenocarcinomas (H&E staining) with high proliferation index (80–93% Ki67 positivity) (Fig. 1). These data confirm that cancer patients’ sera harbor signaling factors with tumorigenic capabilities that, once delivered to recipient target cells, are capable to complete the cascade of events that eventually lead the cells to acquire malignant traits [10‐14].

We analyzed the expression of specific markers to further characterize these xenotransplants for differentiation patterns based on the primary tumor of the blood donors (Table 1 and Fig. 1). We observed that cancer sera-treated cells had completely changed their fate since all developing tumors stained negative for vimentin, which is normally expressed on fibroblasts (Fig. 1). Notably, xenotransplants obtained with cells exposed to the sera of patients with colorectal cancer, hepatocellular carcinoma, pancreatic cancer, and ovarian cancer displayed phenotypical characteristics of blood donors’ primary tumors (Table 1 and Fig. 1). Tumors, generated with colorectal cancer sera-treated cells, displayed epithelial features typical of colorectal adenocarcinomas, as they all stained negative for CK7, and expressed CEA, CK20, CDX-2 and AE1/AE3, which are universal markers of colorectal cancer (Fig. 1a). Tumors, generated with hepatocellular carcinoma sera-exposed cells, showed convincing hepatocellular carcinoma differentiation, although Hep par-1 staining was negative (Fig. 1b) in keeping with early stages of differentiation [34, 35]. The tumors, generated with cells treated with pancreatic cancer serum, stained strongly positive for CK7 and CK19, which are typical markers of pancreatic adenocarcinoma differentiation. Strikingly enough, different stages of differentiation could be visible in the xenotransplants with areas featuring poorly differentiated pancreatic cancer cells and areas containing cells displaying full differentiation (Fig. 1c). The tumors, produced with ovarian cancer sera-exposed cells, showed convincing differentiation toward ovarian carcinoma (Fig. 1d).

Altogether, these observations showed that treated cells had completely changed their fate and adopted features mimicking those of primary cancers of the blood donors. These findings suggest that cells in target organs, might have the potential to incorporate oncogenic factors delivered from primary cancers, acquire aberrant phenotypical traits identical or similar to the primary cancer and be therefore responsible for metastatic disease.

It has been shown that exosomes exert different biological effects by transferring their cargo into target cells [14, 36, 37]. Specifically, these microvesicles have been reported to be involved in cancer invasion and metastasis by aiding invasion of cancer cells into the blood stream and conditioning the metastatic niche in target organs [4, 8, 23, 38]. We hypothesized that the malignant transformation we observed on target cells, following cancer patients’ serum exposure, might be due to serum-carried cancer exosomes and therefore we sought to evaluate and confirm the validity of this theory. To test this hypothesis, we isolated exosomes from cancer patients’ sera, and confirmed their identity both physically and phenotypically (Fig. 2). As visualized by electron microscopy, and measured by Nanosight tracking analyses, the isolated entities were rounded structures of 92 to 112 nm in diameter, which is in the range of the expected size for exosomes (102 +/− 6 nm; Fig. 2a, b) [39]. The identity of these exosomes was further characterized by labeling for selective markers that distinguish them from other cellular microvesicles (i.e. Alix, TSG101) and by confirming the absence of other cell organelle markers (i.e. GM130) to rule out contamination with other vesicles or cellular components (Fig. 2c) [39‐41]. In order to exert their effect, exosomes must be internalized by target cells and deliver their cargo. To assess exosomes uptake, exosomes were tagged with the membrane dye PKH-26 and were incubated with BRCA1-KO fibroblasts. We observed that after 6 h incubation, treated cells efficiently internalized exosomes. Internalized exosomes were uniformly dispersed in the cytoplasm and tended to form aggregates in the perinuclear regions (Fig. 2d).

We reported previously that sera of both healthy donors and cancer patients had no effects on the behaviour of wild type fibroblasts [14]. Moreover, herein, we observed that when wild type fibroblasts were exposed to exosomes isolated from both types of sera, none of mice injected with these cells developed visible tumors even after longer latency periods following subcutaneous inoculation. Afterwards, to analyze the effects that exosomes isolated from cancer patients’ sera had on the behaviour of exposed BRCA1-KO cells, we extracted exosomes from the sera of patients with colorectal cancer, hepatocellular carcinoma, pancreatic cancer and ovarian cancer (Table 1). In parallel, exosomes were also isolated from healthy patients’sera. BRCA1-KO fibroblasts were cultured for 3 weeks with cancer exosomes and exosomes isolated from healthy patients. After treatment, the cells were injected subcutaneously into NOD/SCID mice to verify their tumorigenic potential. All mice injected with BRCA1-KO fibroblasts treated with cancer exosomes developed tumors. None of the mice injected with BRCA1-KO fibroblasts treated with healthy patients’ exosomes, developed tumors. Developing tumors had characteristics of highly proliferative adenocarcinomas (H&E staining and 85–90% Ki67 positivity) (Fig. 3). To determine if these xenotransplants displayed features resembling those of the primary tumors of blood donors, we analyzed the cancer masses for the expression of specific immunohistochemistry markers (Table 1 and Fig. 3). All generated tumors stained negative for vimentin, suggesting that treated cells had completely changed their fate (Fig. 3). Moreover, cells treated with exosomes derived from the sera of patients with colorectal cancer gave rise to tumors that displayed epithelial features typical of colorectal adenocarcinomas (negative for CK7, and positive for CEA, CK20, CDX-2 and AE1/AE3) (Fig. 3a). Also, cells treated with exosomes isolated from the sera of patients with hepatocellular carcinoma generated tumors with patterns of hepatocellular carcinoma differentiation (Fig. 3b). Cells treated with pancreatic cancer serum formed tumors strongly positive for cytokeratins CK19 and AE1/AE3, and with CK7 focal positive patches, reflecting early differentiation into pancreatic cancer (Fig. 3c). The tumors produced with ovarian cancer sera-exposed cells, were weakly positive for WT-1 and EMA suggesting a tendency to differentiate toward ovarian carcinoma (Fig. 3d). These data suggest that exosomes are indeed the main carriers of oncogenic factors through the blood and confirm that their cargo is able to transfer malignant traits at distance.

We reported that BRCA1-KO cells are more prone to internalize exosomes when compared to normal cells [14]. Herein, we sought to determine the mechanisms behind this phenomenon. We isolated the plasma membranes from BRCA1-KO and control fibroblasts using a plasma membrane enrichment kit. This kit enriched for plasma membrane proteins. Western blot analyses of the proteins extracts showed that the plasma membrane isolation was very efficient, as there was enrichment for the transmembrane protein Na/K ATPase and the plasma membrane-bound protein β-actin (Fig. 4a). In contrast, there was a low cytoplasmic proteins contamination, and undetectable nuclear proteins, as judged by the mild expression of GAPDH and absence of Histone H3, respectively (Fig. 4a). The efficiency of plasma membrane enrichment was confirmed following MS analyses. Cellular component GO showed that 1092 proteins were membrane-anchored or proteins known to bind to membrane proteins, while few were from other cellular compartments (i.e. ER, Golgi apparatus, nucleus) (Fig. 4b). Overall, MS detected 1598 proteins: 169 were present only in control fibroblasts, 513 were expressed only in BRCA1-KO fibroblasts, and 916 were shared by both cell types (Fig. 4c and Additional file 2: Table S2).

Exosomes uptake involves a group of cell surface proteins that facilitate exosome/cell interactions and subsequent endocytosis [42, 43]. We scanned the MS outputs for proteins involved in exosomes uptake and compared their expression in the plasma membrane purified from BRCA1-KO and control fibroblasts. We found that BRCA1-KO cells, when compared to control fibroblasts, expressed de novo proteins involved in exosomes uptake such as flotillin, galectin, integrins, dynamin and afadin among others that were overexpressed (Fig. 4d, g, and Additional file 3: Table S3). This observation confirms our hypothesis that oncosuppressor gene mutations might induce expression of proteins on the membrane that would allow cancer exosomes to enter the cell, deliver their cargo and damage the genome [14, 44].

In an attempt to determine whether exosomes shed by cancer cells are more prone than their normal counterpart to internalize into target cells, we isolated exosomes from normal fibroblasts exposed to the serum of healthy individuals (i.e. normal cells) and exosomes from BRCA1-KO fibroblasts after treatment with serum from CRC-LM patients (i.e. transformed cells). Cells were first treated for 3 weeks, then conditioned media from both cultures were collected and exosomes were isolated. Isolated exosomes were subjected to MS analyses that showed efficient exosomes isolation. Our MS analyses detected 62% of the top 100 exosomes proteins reported in the Exocarta protein list (http://​www.​exocarta.​org) (Additional file 4: Table S4). Overall, we detected 1265 proteins: 216 were expressed only in control fibroblasts-derived exosomes, 100 were expressed only in BRCA1-KO fibroblasts-derived exosomes, and 949 were shared by both exosome populations (Fig. 4e and Additional file 5: Table S5).

We screened our MS data for exosomes-expressed ligands specific to the cell receptors involved in vesicles uptake (as shown in Additional file 3: Table S3). Notably, out of 41 sorted proteins, 39 proteins (96%) overlapped with the Exocarta protein list (Additional file 6: Table S6). Based on differential expression dataset, we found that transformed BRCA1-KO cells-derived exosomes overexpressed these ligands when compared to control fibroblasts-derived exosomes. These include laminin (integrins ligands), collagen, tenascin, cadherin, heparan-sulfate proteoglycan and filamin (Fig. 4f, h and Additional file 6: Table S6). Taken together these data show that cancer exosomes express proteins, which enable them to interact with receptors, de novo expressed on the membrane of BRCA1 mutated fibroblasts. These proteins are either not expressed or under-expressed in exosomes shed by non-cancerous cells.

To determine if the de novo expressed cell receptors after oncosuppressor mutation (Additional file 3: Table S3) and the newly identified cancer exosome ligands (Additional file 6: Table S6) played a role in the increased cancer exosomes uptake, displayed by BRCA1-KO fibroblasts, we used a panel of pharmacological antagonists. For this purpose, BRCA1-KO fibroblasts were treated with the anti-β4 integrin-neutralizing antibody (ASC-8), with Cytostatin (an inhibitor of cell adhesion to extracellular matrix; i.e. laminin and collagen) [45], and with heparin (a mimetic of the heparan sulfate in the heparan sulfate proteoglycan) [46]. In parallel, exosomes were exposed to RGD (an integrins tripeptide binding site found within fibronectin), and Collagenase I, before culturing them with the BRCA1-KO fibroblasts for 6 h. Non-treated BRCA1-KO fibroblasts exposed to non-treated exosomes were used as control. Cells were analyzed by flow cytometry (Fig. 5a). We noted that the percentage of cells that internalized exosomes (i.e. PKH-26 positive cells) dropped by 25% following treatments with all antagonists without collagenase I. Addition of collagenase I to the antagonists cocktail decreased this percentage to 93% (Fig. 5a). Also, when compared to control cells, we observed that the mean fluorescence intensity (MFI) decreased by 1.5 to 2.6 times following treatments with the antagonists (Fig. 5a). This finding suggests that the blocking treatment had decreased both the percentage of cells internalizing the exosomal cargo and the number of exosomes internalized per cell.

To rule-out the possibility that the observed inhibitory effect was due to decreased cell viability or secondary to exosomes damage following antagonist treatments, we performed cell counting following trypan blue labeling and NanoSight analyses of exosomes (Fig. 5b, c). We found that the treatment with antagonists had a feeble effect on cell viability (Fig. 5b), and no effect on exosomes integrity (Fig. 5c). Altogether theses findings suggest that the newly expressed membrane proteins on both BRCA1-KO fibroblasts and cancer exosomes might have a role in exosomes uptake in target cells.

To determine if inhibition of the receptors involved in the exosomes uptake may inhibit cancer exosomes-induced cell transformation, we isolated exosomes from the serum of CRC-LM patients. Exosomes uptake was blocked using the same protocol shown in Fig. 5a. Antagonists-treated and non-treated cells were transplanted into NOD/SCID mice to analyze their tumorigenic behavior. Mice were followed up for tumor growth, and they were euthanatized 4 weeks after cell inoculation. Mice injected with non-treated cells displayed visible tumors as early as 10 days following inoculation, which continue growing until euthanasia. In contrast, mice injected with treated cells developed minimal palpable masses (0.73 +/− 0.09 cm³ vs. 0.26 +/− 0.07 cm³, P = 0.031; Fig. 5d, e). Histopathological analyses of tumors generated with non-treated cells showed a uniform sheet of poorly differentiated adenocarcinoma. In contrast, tumors obtained following injection of treated cells displayed zones of poorly differentiated adenocarcinomas with discrete areas of necrotic and non-proliferating cells (Fig. 5f).

We recently reported that BRCA1-KO cells were able to “sense” neoplastic factors in the serum of patients with cancers even at early stages and turn malignant regardless of the presence of positive tumor markers (i.e. pancreatic cancer in situ and early colon cancer with negative CEA) [47]. Furthermore, the evidence reported of patients presenting with metastasis years after the resection of dysplastic lesions doesn’t seem to fit in the conventional metastatic model, where cells are supposed to invade the basal membrane in order to metastasize [48]. Herein, we report data obtained with sera from three patients that presented with dysplastic lesions only. The dysplastic lesions were located in the gallbladder, in the colon and in the common bile duct (Fig. 6a). Strikingly enough, the three sera transformed BRCA1-KO cells, as confirmed by tumor formation, following transplantation in NOD/SCID mice (Fig. 6b). Histopathological analyses of these tumors indicated that they had histological appearances compatible with gallbladder cancer, colon cancer and bile duct cancer and all masses showed high mitotic index (over 90%, Fig. 6c). Taken together, these data show that dysplastic lesions, which are premalignant lesions, also shed transforming factors that turn BRCA1-KO cells into cancer. This finding suggests that the metastatic process might not be secondary to migrating cells since dysplastic lesions, by definition, have not yet invaded the basal membrane. Moreover, this outstanding evidence suggests that horizontal transfer of malignant traits occurs even before cancer cells invasion.