In Vivo imaging reveals extracellular vesicle-mediated phenocopying of metastatic behavior

Abstract

Most cancer cells release heterogeneous populations of extracellular vesicles (EVs) containing proteins, lipids, and nucleic acids. In vitro experiments showed that EV uptake can lead to transfer of functional mRNA and altered cellular behavior. However, similar in vivo experiments remain challenging because cells that take up EVs cannot be discriminated from non-EV-receiving cells. Here, we used the Cre-LoxP system to directly identify tumor cells that take up EVs in vivo. We show that EVs released by malignant tumor cells are taken up by less malignant tumor cells located within the same and within distant tumors and that these EVs carry mRNAs involved in migration and metastasis. By intravital imaging, we show that the less malignant tumor cells that take up EVs display enhanced migratory behavior and metastatic capacity. We postulate that tumor cells locally and systemically share molecules carried by EVs in vivo and that this affects cellular behavior.

Graphical abstract

Figure 1

In Vivo Visualization of EV Release and Characterization of the mRNA EV Cargo (A) Images of MDA-MB-231 tumor sections. Boxed areas are shown on the right. Arrows point to different EVs. Images were gamma adjusted to also show dim and small EVs. Scale bars represent 10 μm (left image) and 1 μm (middle and right image). (B) Stills from intravital imaging movies (Movies S1, S2, and S3) of MDA-MB-231 tumors where CFP-marked MDA-MB-231 cells surrounded by DsRed-marked cells release EVs (marked by circles). Images are maximum projections of 5 z planes with a total z volume of 25 μm. Scale bars represent 10 μm. (C) Electron microscopy image of EVs isolated from MDA-MB-231 tumors. Scale bar represents 500 nm. (D) NanoSight particle analysis displaying the size distribution of EVs isolated from MDA-MB-231 tumors. Black line represents mean of five experiments, SD in gray. (E) mRNA profile of EVs and cells isolated from MDA-MB-231 tumors. Smoothed color density representation shows RNA levels of whole cells versus RNA derived from EVs. Red dots represent genes that are significantly and more than 2-fold enriched in EVs; significance is based on replicate experiments (n = 6). (F) These genes in (D) were used as input for WebGestalt analysis (Wang et al., 2013). Bar plot representation of p values from a gene ontology enrichment analysis. Biological processes involved in cell movement are shown in red. See also Movies S1, S2, and S3 and Tables S1 and S2.

Figure 2

The Cre-LoxP System to Measure EV Uptake (A) Cartoon showing the Cre-LoxP system used to report the transfer of Cre recombinase (Cre) activity. A red-to-green color switch is induced in reporter⁺ cells (right) upon the transfer of Cre activity derived from CFP⁺ Cre⁺ cells (left). (B) Electron microscopy analysis of EVs isolated from MDA-MB-231 cell cultures. Scale bar represents 150 nm. (C) NanoSight particle analysis displaying the size distribution of EVs isolated from MDA-MB-231 cell cultures. Black line represents mean of five experiments, SD in gray. (D) Confocal images of reporter⁺ cells that have taken up green-labeled MDA-MB-231-isolated EVs 3 hr after addition of these EVs to the culture. The boxed area is shown on below the image. Arrow points to EVs that are taken up. Scale bars represent 50 μm. (E and F) Characterization of EVs isolated from MDA-MB-231 conditioned media (CM) using high-speed centrifugation. The supernatant (sup) obtained during this centrifugation procedure was used as negative control in (E) western blot analysis of CD63, Hsp70, and Cre protein levels and (F) RT-PCR for Cre mRNA of samples as indicated. See also Movie S4.

Figure 3

In Vitro Visualization of Transfer of Cre Activity (A) Images of MDA-MB-231 cultures consisting of reporter⁺ cells (red) only or a mixture of Cre⁺ (cyan) and reporter⁺ cells. Note the appearance of eGFP⁺ reporter⁺ cells in the presence of Cre⁺ cells, indicated by asterisks. Scale bars represent 50 μm. (B) Quantification of the percentage of eGFP⁺ MDA-MB-231 cells in several conditions with varying ratios of Cre⁺ and reporter⁺ cells (n = 5 independent experiments). Data are represented as mean ± SEM. (C) For 50 cells, the mean fluorescence intensity (MFI) of eGFP (reporting Cre activity) is plotted against the MFI of CFP (Cre⁺ cells). Intensities of non-fluorescent cells are zero (red dot). In the left plot, cell fusion was experimentally simulated by transfecting eGFP⁺ cells with CFP DNA. In the middle plot, communication was experimentally simulated by transfecting reporter⁺ cells with Cre DNA. The right plot shows the results of the experimental co-culture. The blue dotted line outlines cells that show cell fusion, and the green dotted line shows eGFP expression derived from cell-cell communication other than fusion. (D) Images of MDA-MB-231 reporter⁺ cells (left), T47D reporter⁺ cells (middle), or MCF-7 reporter⁺ cells (right) at the bottom well of a transwell system in the absence (upper images) or presence (lower images) of MDA-MB-231 Cre⁺ cells in the upper well (n = 3 independent experiments). Bar graphs below show quantifications. Data are represented as mean ± SEM. Scale bars represent 50 μm. (E) Quantification of T47D cells (upper graph) and MCF-7 cells (lower graph) that express eGFP in several conditions with varying ratios of Cre⁺ and reporter⁺ cells (n = 5 independent experiments). Data are represented as mean ± SEM.

Figure 4

The Transfer of EVs between Tumor Cells in Living Mice (A) Cells and EVs were isolated from Cre⁺ tumors and Cre⁻ tumors. Shown is the RT-PCR of indicated samples. (B) EVs isolated from Cre⁺ MDA-MB-231 cells and pretreated with RNase and proteinase K (protK) were intratumorally injected in T47D reporter⁺ tumors. Shown are representative images of these tumors and quantification of eGFP⁺ cells (n = 3 mice). Data are represented as mean ± SEM. Scale bars represent 50 μm. (C) Recombinant Cre protein (110 ng) was intratumorally injected in T47D reporter⁺ tumors. Shown are representative images of these tumors and quantification of eGFP⁺ cells (n = 4 mice). Data are represented as mean ± SEM. Scale bars represent 50 μm. (D) Whole-cell lysate of 10% of a Cre⁺ MDA-MB-231 tumor was intratumorally injected in T47D reporter⁺ tumors. Shown are representative images of these tumors and quantification of eGFP⁺ cells (n = 4 mice). Data are represented as mean ± SEM. Scale bars represent 50 μm. (E) Confocal images of tumors containing only reporter⁺ cells in control animals (negative control) and in tumors that contain Cre⁺ cells and reporter⁺ cells. MDA-MB-231 mammary tumor (260 ± 67 eGFP⁺ tumor cells per intravital 620 × 620 × 100 μm z stack, n = 3 mice [for 3D reconstruction see Movie S5]), T47D mammary tumor (513 ± 66 eGFP⁺ tumor cells per 16-μm-thick tumor section, n = 3 mice), PyMT mammary tumor (30 ± 13 confetti⁺ tumor cells per 20 -μm-thick tumor section, n = 3 mice), and B16 melanoma (43 ± 10 eGFP⁺ tumor cells per 100-μm-thick tumor section, n = 3 mice). For the PyMT model, the Cre-confetti-reporter system was used (see Figures S1B and S1C; all colors are shown in green). Scale bars represent 50 μm. See also Movie S5 and Figures S1 and S2.

Figure 5

Transfer of EVs between Tumor and Healthy Cells in Living Mice (A) B16 melanoma CFP⁺ Cre⁺ cells were injected subcutaneously in Cre reporter B6 mice ubiquitously expressing a floxed-STOP-floxed-tdTomato (R26-LSL-tdTomato mice). (B) Representative confocal images (n = 3 mice) showing tdTomato⁺ cells in indicated tissues (tumors 374 ± 76 cells/mm³, lungs 69 ± 17 cells/mm³, lymph nodes (LNs) 168 ± 69 cells/mm³, spleens 104 ± 22 cells/mm³). CFP is indicated in cyan, tdTomato is indicated in red, and DAPI is indicated in blue. Scale bars represent 50 μm. All numbers represent mean ± SEM. (C) Confocal images of CD45⁺ (top images) and CD45⁻ tdTomato⁺ (bottom images) cells found in the mice described in (A). Boxed areas are shown on the right. CD45 staining is indicated in yellow, tdTomato is indicated in red, CFP is indicated in cyan, and DAPI is indicated in blue. Scale bars represent 25 μm. (D) Confocal images show Gr1⁺ cells (top images) and F4/80⁺ tdTomato⁺ (bottom images) cells found in the mice described in (A). Boxed areas are shown on the right. Gr1(top) and F4/80 (bottom) stainings are indicated in yellow, tdTomato is indicated in red, CFP is indicated in cyan, and DAPI is indicated in blue. Scale bars represent 25 μm. (E) B16 melanoma reporter⁺ cells were injected subcutaneously in B6 ACTB-Cre mice that express Cre ubiquitously. (F) Tumors were isolated from ACTB-Cre mice bearing B16 melanoma reporter⁺ tumors (n = 3 mice). Representative confocal images show eGFP⁺ melanoma cells (2.5 ± 0.6 cells/mm³). Scale bars represent 50 μm. All numbers represent mean ± SEM.

Figure 6

Less Malignant T47D Cells that Take Up MDA-MB-231 EVs Display Enhanced Migration (A) MDA-MB-231 CFP⁺ Cre⁺ cells and T47D reporter⁺ cells were injected as a mixture in one mammary gland (local communication) or separately in contralateral mammary glands (distant communication). (B) Representative intravital images of the tumors in mice as indicated. Scale bars represent 50 μm. (C) Quantification of eGFP⁺ cells in the local and distant communication experiments (n = 26 frozen tumor sections from 3 mice). Data are represented as mean ± SEM. (D) Mammary tumors containing MDA-MB-231 Cre⁺ cells and T47D reporter⁺ cells were intravitally imaged for 3 hr. Shown are migration paths of MDA-MB-231 cells and T47D DsRed⁺ (left) and eGFP⁺ reporter cells (right) within the same imaging field. In the far right, the migration speed of the same cells is plotted. Red lines indicate means (n > 25 cells). (E and F) The in vivo migration speed of T47D eGFP⁺ and DsRed⁺ reporter⁺ cells in tumors that report local (E) or distant (F) transfer of Cre from MDA-MB-231 Cre⁺ cells. Shown are the average values per imaging field. Values for eGFP⁺ and DsRed⁺ cells in the same imaging field are connected with a line (n = 5–6 imaging fields per mouse for 3 independent mice). (G and H) The normalized in vivo migration speed of T47D eGFP⁺ and DsRed⁺ reporter⁺ cells in tumors reporting the local (G) or distant (H) transfer of Cre from MDA-MB-231 Cre⁺ cells. Per imaging field, the values are normalized to the migration speed of neighboring DsRed⁺ cells (n = 17–18 imaging fields in 3 mice). Data are represented as mean ± SEM. See also Figures S3, S4, and S5.

Figure 7

Uptake of MDA-MB-231-Released EVs by T47D Cells Leads to Increased Metastatic Potential (A) MDA-MB-231 CFP⁺ Cre⁺ cells and T47D reporter⁺ cells were injected as a mixture in one mammary gland (local communication) or separately in contralateral mammary glands (distant communication). The lungs and corresponding primary reporter tumors were analyzed, and the ratio of GFP⁺/DsRed⁺ metastases and cells was determined. If EV uptake results in an enhanced metastatic potential, the ratio of eGFP⁺ over DsRed⁺ cells that arrive and grow out in the lung is expected to be higher than the ratio of these populations of cells in the primary tumor. (B and C) Representative images of frozen sections of the primary tumor and lungs of mice of local (B) and distant communication experiments (C). Right dot plots show the fold increase of metastatic potential (increase in eGFP⁺ over DsRed⁺ ratio in lungs compared to the ratio in corresponding primary tumors). Red lines show the median, each symbol represents one cryosection, and different symbols represent different mice (n = 4 mice per condition). Scale bars represent 50 μm. See also Figures S3 and S6.

Figure S1

The Different Mouse Models Used to Study In Vivo Vesicle Transfer, Related to Figure 4 (A) Cartoon of the experimental setup of the human mammary xenograft and melanoma allograft model. MDA-MB-231 Cre⁺ and reporter⁺ cells were injected orthotopically (in the fourth mammary gland of immune-deficient mice for the human mammary tumor cells and subcutaneously in B6 mice for the melanoma model). When tumors had a diameter of 10 mm, the tumors were analyzed. (B) Cartoon of the experimental setup of PyMT tumor model. Mice in which PyMT is driven by a mammary epithelial-specific promoter were crossed with mice expressing either CreER^T2 or the confetti reporter construct to create tumor pieces of Cre⁺ and Confetti-reporter⁺ cells respectively. Small tumor pieces were transplanted into immune-deficient mice, which grew out to large tumors within 2-4 months. Tumors were isolated once they were 10 mm and examined for the presence of fluorescent cells. (C) Cartoon of the Cre⁺ and reporter⁺ confetti cells. Upon Cre activity, confetti-reporter⁺ cells express CFP, GFP, YFP or DsRed, which are all shown in green in the images of Figure 4E.

Figure S2

In Vivo Transfer of Cre Activity Is Not Mediated by Cell-Cell Fusion, Related to Figure 4 For 50 cells, the mean fluorescent intensity (MFI) of eGFP (reporter⁺ cells that received Cre activity) was plotted against the MFI of CFP (Cre⁺ cells). Intensities of non-fluorescent cells were zero (red dot). In the left plot, cell fusion was experimentally simulated by transfecting eGFP⁺ cells with CFP DNA. In the middle plot, communication was experimentally simulated by transfecting reporter cells with Cre DNA. The right plot shows the results of the experimental in vivo local mixture of MDA-MB-231 Cre⁺ and reporter⁺ cells whereby the MFI of CFP and eGFP was measured for individual and randomly chosen CFP⁺ and eGFP⁺ cells (n = 50 cells in 3 mice). The blue dotted line outlines the cells that show cell fusion and the green dotted line shows cell communication.

Figure S3

T47D Cells that Take Up Less Malignant MCF-7 EVs Display a Decrease in Migration, Related to Figure 6 (A) Left: cartoon of the experimental setup: MCF-7 CFP⁺ Cre⁺ cells and T47D reporter⁺ cells were injected separately in contralateral mammary glands. Middle: confocal images (merge of the blue, green and red channel) of the tumors in mice as indicated. Scale bar represents 50 μm. Right: quantification of the % eGFP⁺ cells in the negative control and distant communication experiment (n = 45 frozen tumor sections from 3 mice). Data are represented as mean ± SEM. (B) The in vitro migration speed of T47D and MCF-7 cells. The cell migration was tracked over 15 hr. Experiments include at least 150 cells in five different wells in three independent experiments. Data are represented as mean ± SEM. (C) T47D reporter⁺ tumors in mice described in A were intravitally imaged for three hours and the in vivo migration speed of eGFP⁺ and DsRed⁺ reporter⁺ cells was measured. Left: shown are the average values per imaging field. Values for eGFP⁺ and DsRed⁺ cells in the same imaging field are connected with a line (n = 10 imaging fields in 3 mice). Right: the normalized in vivo migration speed of T47D eGFP⁺ and DsRed⁺ reporter⁺ cells. Per imaging field, the values are normalized to migration speed of DsRed⁺ cells (n = 10 imaging fields in 3 mice). Data are represented as mean ± SEM. (D) Dot plots shows the fold increase in metastatic potential by plotting the increase in the eGFP⁺ over DsRed⁺ ratio in lungs compared to the ratio in corresponding primary tumors. The red line shows the median, each symbol represents one cryosection, and the different symbols represent different mice (n = 7 mice) showing the variation within one mouse and between different mice.

Figure S4

EV Exchange Leads to Enhanced Migration in Recipient T47D and MCF-7 Cells, Related to Figure 6 (A) Migration speed of T47D DsRed⁺ and eGFP⁺ reporter⁺ cells that are cultured together with MDA-MB-231 Cre⁺ cells in a 1:1 ratio (co-culture; n = 3 independent experiments) or that are cultured with MDA-MB-231 Cre⁺ cells seeded on top of a 400 nm transwell membrane (transwell; n = 3 independent experiments). Data are represented as mean ± SEM. (B) Migration speed of MCF-7 DsRed⁺ and eGFP⁺ reporter⁺ cells that are cultured together with MDA-MB-231 Cre⁺ cells in a 1:1 ratio (co-culture; n = 3 independent experiments) or that are cultured with MDA-MB-231 Cre⁺ cells seeded on top of a 400 nm transwell membrane (transwell; n = 3 independent experiments). Data are represented as mean ± SEM. (C) MDA-MB-231, T47D and MCF-7 reporter⁺ cells were transiently transfected with Cre. The in vitro migration of DsRed⁺ and eGFP⁺ cells was tracked over 15 hr. Experiments include at least 150 cells per condition in five different wells in three independent experiments. Data are represented as mean ± SEM.

Figure S5

Migratory Cells Do Not Have A Priori Higher Capacity to Take Up EVs, Related to Figure 6 (A) The in vitro migration of MDA-MB-231 cells was tracked over 6 hr. After the migration experiment, PKH67-labeled MDA-MB-231-derived EVs were added and the uptake was monitored. The migration speed of the cell was measured and plotted against the vesicle uptake, expressed as the total PKH67 fluorescence per cell (n = 78 cells; trend line shown in gray). The cells that did not take up any PKH67 could not be shown in the graph and are therefore indicated at the red dashed line. Note that the R² is close to zero indicating that there is no correlation between migration speed and EV uptake. (B) The average in vivo migration speed per imaging field and the percentage of T47D eGFP⁺ reporter cells in tumors that report Cre activity from distant MDA-MB-231 Cre⁺ cells transplanted in contralateral glands (Figure 6A) (n = 17 imaging fields in 3 mice; trend line shown in gray).

Figure S6

Biomolecule Exchange between MDA-MB-231 Cells Leads to Enhanced Metastasis Formation, Related to Figure 7 (A) Graph shows the migration speed of MDA-MB-231 DsRed⁺ and eGFP⁺ reporter⁺ cells that are cultured with MDA-MB-231 Cre⁺ cells seeded on top of a 400 nm transwell membrane (n = 3 independent experiments). Data are represented as mean ± SEM. (B) Left: cartoon of the experimental setup, where a mixture of MDA-MB-231 CFP⁺ Cre⁺ cells and reporter cells were injected in the mammary gland and the subsequent tumors were intravitally imaged for three hours. Right: the normalized in vivo migration speed of MDA-MB-231 eGFP⁺ and DsRed⁺ reporter⁺ cells. Per imaging field, the values are normalized to migration speed of DsRed⁺ cells (n = 15 imaging fields in 3 mice). Data are represented as mean ± SEM. (C) Shown are representative images of frozen sections of the primary tumor and lungs of mice injected with a mixture of MDA-MB-231 CFP⁺ Cre⁺ and reporter⁺ cells (n = 3 mice). The images of the lung are maximum projection images of a volume of 150 μm. Scale bars represent 100 μm. (D) Dot plot shows the fold increase of metastatic potential by plotting the increase in the eGFP⁺ over DsRed⁺ ratio in lungs compared to the ratio in corresponding primary tumors. The red line shows the median, each symbol represents one cryosection, and the different symbols represent different mice (n = 3 mice) showing the variation within one mouse and between different mice. The P-values were calculated using Mann-Whitney U-test using the mouse as comparative unit.