Extracellular Vesicles Secreted by Cancer-Associated Fibroblasts Drive Non-Invasive Cancer Cell Progression to Metastasis via TGF-β Signalling Hyperactivation

Abstract

Metastasis is the leading cause of cancer-related deaths. Cancer-associated fibroblasts (CAFs) are abundant components within the tumour microenvironment, playing critical roles in metastasis. Although increasing evidence supports a role for small extracellular vesicles (sEVs) in this process, their precise contribution and molecular mechanisms remain unclear, compromising the development of antimetastatic therapies. Here, we establish that CAF-sEVs drive metastasis by mediating CAF-cancer cell interaction and hyperactivating TGF-β signalling in tumour cells. Metastasis is abolished by genetically targeting CAF-sEV secretion and consequent reduction of TGF-β signalling in cancer cells. Pharmacological treatment with dimethyl amiloride (DMA) decreases CAFs' sEV secretion, reduces TGF-β signalling levels in tumour cells and abrogates metastasis and tumour self-seeding. This work defines a new mechanism required by CAFs to drive cancer progression, supporting the therapeutic targeting of EV trafficking to disable the driving forces of metastasis.

Keywords: TGF‐β; cancer‐associated fibroblasts; circulating tumour cells; extracellular vesicles; metastasis; therapy; tumour microenvironment.

FIGURE 1

CAF‐sEVs transport TGF‐β signalling components and activate the TGF‐β signalling in breast cancer cells in vitro. (a) 19TT‐sEV morphology was analysed by cryogenic electron microscopy. (b) TGF‐β/SMAD signalling reporter (Ad‐CAGA‐Gluc) activity was quantified in MDA231 cells treated for 24 h ± 19TT‐large EVs (10K pellet), EV‐depleted conditioned medium (100K supernatant), or small EVs (100K pellet). Fractions corresponding to 10K pellet and 100K pellet were resuspended in 100 µL serum‐free medium (SFM) during isolation and before use in this analysis. Five microliters per well 10K pellet, 100K supernatant, and 100K pellet were used in this analysis. Treatment with rhTGF‐β1 was used as positive control. (c) 19TT‐sEV protein extracts were tested for the expression of extracellular vesicle markers and TGF‐β signalling components by western blot. Whole cell lysates (WCL) were used as positive control. Twenty micrograms of protein extracts were loaded to each lane. (d) Nanoparticle tracking analysis (NTA) was used to determine the particle size distribution of sEVs secreted by 19TT CAFs treated ± 5 ng/mL rhTGF‐β1 for 2 h. (e) Luciferase assay was used to quantify the TGF‐β/SMAD signalling reporter (Ad‐CAGA‐Gluc) activity in MDA231 cells as in (b). Cells were treated ± 19TT‐sEVs (1 µg/well total protein) isolated from 19TT cells treated ± 5 ng/mL rhTGF‐β1 for 2 h. Results represent mean ± SD of at least three independent experiments (n = 3). One‐way ANOVA test followed by Dunn's Multiple Comparison test were used to analyse data in (b & e). ns: statistically non‐significant, *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 2

CAF‐sEVs hyperactivate TGF‐β signalling in breast cancer cells in vitro. (a) TGF/SMAD3 signalling reporter (Ad‐CAGA‐Fluc) activity was quantified in MDA231 cells treated with recombinant human (rh)TGF‐β1 or 19TT‐sEVs at increasing concentrations for 24 h. (b) Phosphorylated (p)SMAD2 levels analysed in MDA231 cells treated for 1 h. (c) TGF/SMAD3 signalling reporter (Ad‐CAGA‐Fluc) activity was quantified in MCF7 cells treated with rhTGF‐β1 or 19TT‐sEVs at increasing concentrations for 24 h. (d) pSMAD2 levels analysed in MCF7 cells treated for 1 h. (e–f) TGF/SMAD3 signalling reporter (Ad‐CAGA‐Fluc) activity was quantified in (e) MDA231 and (f) MCF7 cells infected with Ad‐CMV‐GFP (control adenovirus) or Ad‐CMV‐Flag‐SMAD7 treated for 24 h. (g, h) TGF/SMAD3 signalling reporter (Ad‐CAGA‐Fluc) activity was quantified in (g) MDA231 or (h) MCF7 cells challenged with SB431542 and treated with rhTGF‐β1 or 19TT‐sEVs for 24 h. (i, j) pSMAD2 levels in (i) MDA231 and (j) MCF7 challenged with SB431542 and treated with rhTGF‐β1 or 19TT‐sEVs for 1 h. DMSO was used as a vehicle for SB431542. The concentration of 19TT‐sEVs used to treat breast cancer cells in (b) and (d–j) was equivalent to 5 ng/mL TGF‐β activity. Results represent mean ± SD (n ≥ 3). Unpaired Student's t‐test was used to analyse data in (a & c). One‐way ANOVA test followed by Dunn's Multiple Comparison test were used to analyse data in (e–h). ns: statistically non‐significant, *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 3

Heparin and PNP‐Xyl treatment inhibit TGF‐β signalling activity induced by CAF‐sEVs in breast cancer cells in vitro. (a–c) Effects caused by Heparin on cells treated with 1 ng/mL recombinant human (rh)TGF‐β1 or 19TT‐sEVs. (a) Phosphorylated (p)SMAD2 levels in MDA231 cells treated for 1 h. TGF‐β/SMAD signalling reporter (Ad‐CAGA‐Fluc) activity in (b) MDA231 or (c) MCF7 cells treated for 24 h. (d‐f) Effects caused by PNP‐Xyl on cells treated with 1 ng/mL rhTGF‐β1 or 19TT‐sEVs. (d) pSMAD2 levels in MDA231 cells treated for 1 h. TGF‐β/SMAD signalling reporter (Ad‐CAGA‐Fluc) activity in (e) MDA231 or (f) MCF7 cells treated for 24 h. The concentration of 19TT‐sEVs used to treat breast cancer cells was equivalent to 1 ng/mL TGF‐β activity. Results represent mean ± SD (n = 3). One‐way ANOVA test followed by Dunn's Multiple Comparison test were used to analyse data. ns: statistically non‐significant, *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 4

CAF‐sEVs rely on TGF‐β signalling activation to induce breast cancer cell aggressiveness in vitro. (a) MCF7 cell morphology analysed by bright field microscopy in cells treated with recombinant human (rh)TGF‐β1 or 19TT‐sEvs over 5 days. (20× magnification). Cell cultures were treated thrice (0, 48, and 96 h). Images obtained in cell cultures fixed and permeabilized. (b) E‐cadherin and ZO‐1 localization in MCF7 cells treated as in (a). (c) E‐cadherin and ZO‐1 expression in MCF7 cells treated as in (a). (d) MDA231 cell invasion in transwell inserts. (e–f) Wound healing assay for MDA231 cells challenged with (e) SMAD7 overexpression or (f) Heparin treatment. Ad‐CMV‐GFP: control adenovirus. The concentration of 19TT‐sEVs used to treat breast cancer cells was equivalent to 5 ng/mL TGF‐β activity. Results represent mean ± SD (n = 3). One‐way ANOVA test followed by Dunn's Multiple Comparison test were used to analyse data. ns: statistically non‐significant, *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 5

CAFs require intact sEV secretion to hyperactivate the TGF‐β signalling in poorly metastatic breast cancer cells in vitro. (a–c) TGF‐β/SMAD signalling reporter (Ad‐CAGA‐Fluc) activity in MCF7 cells ± 19TT cells treated as indicated. MCF7 cells were challenged with (b) SMAD7 overexpression or (c) SB431542 treatment. Ad‐CMV‐GFP: control adenovirus. DMSO: vehicle for SB431542. (d) Rab27a expression in parental (WT) and Rab27a knockdown (KD) 19TT cells. (e) Total sEV secretion quantified by BCA assay in 19TT‐sEVs. (f) Particle size distribution evaluated by nanoparticle tracking analysis (NTA) in 19TT‐sEVs. (g) Quantification of 19TT‐sEV secretion by NTA. (h) TGF‐β/SMAD signalling reporter (Ad‐CAGA‐Fluc) activity quantified as in (A) in MCF7 cells co‐cultured with 19TT (Rab27a WT or Rab27a KD) cells. (i) Total sEV secretion quantified by BCA assay in 19TT‐sEVs treated with DMA for 2 h. (j) Particle size distribution evaluated by NTA in 19TT‐sEVs treated as in (i). (k) Quantification of sEV secretion by NTA in 19TT cells treated as in (i). (l‐n) TGF‐β/SMAD signalling reporter (Ad‐CAGA‐Fluc) activity analysed as in (A) in MCF7 cells in cell cultures treated with (l) DMA, (m) Heparin, or (n) PNP‐Xyl for 24 h. Results represent mean ± SD (n ≥ 3). DMSO: vehicle for DMA and PNP‐Xyl. One‐way ANOVA test followed by Dunn's Multiple Comparison test were used to analyse data in (a–c, h & l–n). Unpaired Student's t‐test was used to analyse data in (e, g, i & k). ns: statistically non‐significant, *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 6

CAF‐induced MCF7 cell aggressiveness depends on EV secretion and TGF‐β signalling pathway activation. (a) Schematic illustration for the quantification of cancer cell migration in transwell inserts using Gaussia luciferase‐labelled MCF7 cells co‐cultured ± 19TT CAFs. Alternatively, transwell insert membranes were coated with Matrigel to evaluate cancer cell invasion. (b) Titration of Gaussia luciferase activity by luciferase assay using increasing numbers of Gaussia luciferase‐labelled MCF7 (MCF7.Gluc) cells. (c) Migration and (d) invasion of Gaussia luciferase‐labelled MCF7 (MCF7.Gluc) cells cultured ± 19TT cells and treated ± recombinant human (rh)TGF‐β1. (e) Wound healing assay for GFP‐labelled MCF7 cells cultured with 19TT (Rab27a WT or Rab27a KD) and treated as indicated. (f) Wound healing assay for GFP‐labelled MCF7 cells treated with vehicle (DMSO) or DMA. Results represent mean ± SD (n ≥ 3). One‐way ANOVA test followed by Dunn's Multiple Comparison test were used to analyse the presented data. ns: statistically non‐significant, *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 7

CAF‐sEVs enhance TGF‐β signalling activity in MDA231 cells in vivo and increase CTCs, metastasis, and tumour self‐seeding. (a) Schematic illustration and timeline for NOD‐SCID mice implanted with unlabelled MDA231 and Gaussia luciferase‐labelled MDA231 (MDA.Gluc) cells and treated ±19TT‐sEVs. (b, c) Quantification of TGF‐β/SMAD signalling reporter (Ad‐CAGA‐Fluc) activity in MDA.Gluc tumours by In Vivo Imaging System (n = 3 mice). (d–g) Gaussia luciferase activity in (d) blood, (e) lung, (f) bone, and (g) unlabelled MDA231 tumour samples (n = 5 mice). Animals are color‐coded. Black dashed lines indicate the background activity for the Gaussia luciferase quantified in samples from non‐implanted mice (n = 2 mice). Results represent mean ± SEM. Unpaired Student's t‐test (c), One‐Way ANOVA followed by Tukey's Multiple Comparison Test (d–g), *p < 0.05, **p<0.01, ***p<0.001, ns: statistically non‐significant.

FIGURE 8

TGF‐β signalling hyperactivation in vivo by CAF‐sEVs enables progression of poorly metastatic MCF7 cancer cells. (a) Schematic illustration and timeline for NOD‐SCID mice implanted with unlabelled MCF7 and Gaussia luciferase‐labelled MCF7 (MCF7.Gluc) cells and treated ± 19TT‐sEVs. (b) Detection of TGF‐β/SMAD signalling reporter (Ad‐CAGA‐Fluc) activity in MCF7.Gluc tumours by In Vivo Imaging System (n = 5–6 mice). (c–h) Gaussia luciferase activity in (c) blood, (d) liver, (e) bone, (f) brain, (g) unlabelled MCF7 tumours and (h) lung samples (n = 5–6 mice). Animals are color‐coded. Black dashed lines indicate the background activity for the Gaussia luciferase quantified in samples from non‐implanted mice (n = 2 mice). Results represent mean ± SEM. One‐Way ANOVA followed by Tukey's Multiple Comparison Test. *p < 0.05, **p < 0.01, ***p < 0.001, ns: statistically non‐significant.

FIGURE 9

CAF‐induced TGF‐β signalling hyperactivation and breast cancer progression is impaired by genetically and pharmacologically targeting EV trafficking. (a) Schematic illustration and timeline for NOD‐SCID mice implanted with Gaussia luciferase‐labelled MCF7 (MCF7.Gluc) cells ±19TT (Rab27a wild type/WT or knockdown/KD). (b) Quantification of TGF‐β/SMAD3 signalling reporter (Ad‐CAGA‐Fluc) activity in MCF7.Gluc cells by In Vivo Imaging System (n = 2 tumour/mice; two mice/group). (c) Schematic illustration and timeline for NOD‐SCID mice implanted with unlabelled MCF7 cells and Gaussia luciferase‐labelled MCF7 (MCF7.Gluc) cells ±19TT (Rab27a WT or KD). (d‐h) Gaussia luciferase activity in (d) blood, (e) liver, (f) bone, (g) unlabelled MCF7 tumours, and (h) lung samples (n = 6 mice). Animals are color‐coded. Results represent mean ± SEM. One‐Way ANOVA followed by Tukey's Multiple Comparison Test. *p < 0.05, **p < 0.01, ***p < 0.001, ns: statistically non‐significant.

FIGURE 10

Working model. Human breast CAFs secrete elevated levels of vesicular TGF‐β. Uptake of CAF‐sEVs by breast cancer cells drives TGF‐β signalling hyperactivation to ultimately increase multiorgan metastasis and tumour self‐seeding. Consequently, decreasing Rab27a levels in CAFs reduces sEV secretion and prevents the amplification of TGF‐β signalling levels in breast cancer cells to ameliorate cancer progression. Accordingly, sEV trafficking disruption by treatment with DMA, heparin, or PNP‐Xyl is a novel therapeutic strategy that normalizes the TGF‐β signalling activity in breast cancer cells and efficiently blocks metastatic progression. In this illustration, arrow thickness represents the intensity of a given step. Red and white boxes highlight therapeutic strategies used in this work to normalize TGF‐β signalling levels by decreasing CAF‐sEV secretion and uptake. MVB: multivesicular body (also termed late endosome). ILVs, intraluminal vesicles.