Tumor–Stroma Interaction: Revealing Fibroblast-Secreted Exosomes as Potent Regulators of Wnt-Planar Cell Polarity Signaling in Cancer Metastasis

Cancer biology is tightly regulated by the tumor-associated stroma, which consists of ECM components and several cell types, including cancer-associated fibroblasts (CAF), immune cells, vascular cells, and bone marrow–derived cells (1). Fibroblasts comprise the most abundant stromal cell population in several carcinomas and are intimately involved in the initiation and progression of cancer (1). Accordingly, the gene expression profile of CAFs changes significantly when compared with that of normal cells (2) and is predictive of clinical outcome (1, 3). Moreover, production of oncogenic growth factors and loss of tumor-suppressing signals by CAFs promote epithelial cell neoplastic transformation and carcinoma progression (1, 4, 5). For example, CAFs secrete several factors, such as hepatocyte growth factor (4) and stromal cell-derived factor 1 (6), which can stimulate tumor growth. In addition, CAFs promote tumor cell invasion by remodeling the ECM, through secretion of matrix metalloproteinases (MMP), including MMP9 (5), and by biomechanical forces (7). CAFs also communicate with other stromal cell types to stimulate tumor-promoting inflammation, angiogenesis, and lymphangiogenesis (6, 8, 9). Importantly, CAFs stimulate cancer cell metastasis (10–12); however, the paracrine signals involved are poorly understood. Because more than 90% of patients with cancer succumb to their disease due to cancer metastasis, it is imperative that we identify CAF paracrine signals that regulate this process.

Cell–cell communication involves more than soluble factors, and recent work has shown that secreted microvesicles, including exosomes, carry biologically active molecules, such as microRNAs, mRNAs, and activated growth factor receptors, which can be horizontally transferred and function in recipient cells (13–15). Exosomes comprise a specific subtype of microvesicles that originate in multivesicular bodies (MVB), are of a typical size of 30 to 100 nm in diameter, and are rich in plasma membrane proteins, including the tetraspanin molecule CD81 (13, 14). Exosomes are secreted by cancer cells and modulate tumor-induced immune suppression, angiogenesis, stromal remodeling, tumor cell invasion, and formation of the premetastatic niche (13–16). Conversely, exosomes are also secreted by tumor-associated stromal cells and play a critical role in tumorigenesis. For instance, exosomes secreted by activated platelets induce the expression of several proangiogenic and invasive factors in lung cancer cells and thus promote angiogenesis and metastasis (17). Importantly, exosomes accumulate in the plasma, ascites, and pleural effusions of patients with cancer, and their use as diagnostic biomarkers for cancer is promising (16, 18). Moreover, exosomes may also be used as therapeutic and drug delivery tools (19).

The planar cell polarity (PCP) pathway is highly conserved from sponges to vertebrates and regulates the organization of subcellular structures, groups of cells, or cell behavior across a tissue in a plane orthogonal to the apical–basal axis (20, 21). For example, in vertebrates, the PCP pathway directs the convergent and extension movements of mesodermal cells that allow for narrowing of the embryonic tissue along the mediolateral axis and extension of the embryo along the anteroposterior axis (20, 21). In particular, during convergent and extension movements, the PCP pathway coordinates polarized cell behavior, including cell polarity, protrusive activity, and directional migration (20). PCP signaling is regulated by a core set of proteins, which include, among others, the seven-pass transmembrane Frizzled (Fzd) receptors, the four-pass transmembrane protein van Gogh-like (Vangl), and the cytoplasmic proteins Dishevelled (Dvl) and Prickle-like (Pk; refs. 20, 21). Several studies have shown that the expression of PCP pathway components is upregulated in cancer and that these components are involved in cancer progression (22). However, it has been unclear whether the PCP pathway per se is coopted in cancer cell migration and metastasis.

Many outstanding questions remain with regards to the extracellular and intracellular molecular mechanisms that regulate Wnt-PCP signaling. In vertebrates, there is evidence that secreted Wnt morphogens, which bind Fzd receptors, regulate PCP signaling (21). However, due to posttranslational modifications, including glycosylation, acylation, and GPI-linkage, Wnt ligands associate tightly with the plasma membrane and the ECM, raising the question of how Wnt ligands might mediate intercellular signaling (23, 24). Several accessory factors, including heparan sulfate proteoglycans, high-density lipoprotein particles, secreted proteins, and exosomes have been shown to solubilize Wnts and promote their signaling (10, 23–26). However, the significance of these mechanisms in different Wnt-PCP pathway-regulated processes is unclear, as is the question of how Wnt-PCP signals are conducted intracellularly. In static tissues such as the sensory epithelium of the inner ear, the PCP pathway is associated with asymmetrical subcellular distribution of core PCP components across the tissue (20). In contrast, in dynamically remodeling tissues such as a gastrulating embryo, such distribution of PCP components has been more difficult to detect. Thus, it remains to be determined whether asymmetrical distribution of core PCP components is a conserved and prominent mark of all PCP signaling.

Recently, we discovered that conditioned media from a fibroblast cell line, L cells, strongly stimulates the protrusive activity and motility of several breast cancer cell (BCC) lines, including MDA-MB-231 and SUM-159PT carcinoma cells (10). Moreover, L-cell fibroblasts stimulate the metastatic potential of BCCs to the lung upon coimplantation in the mammary fat pad of immunocompromised mice. Identification of the active component secreted by L cells using protein chromatography and mass spectrometry revealed the unexpected finding that exosomes, rather than soluble factors previously shown to regulate cancer cell motility, were the source of the activity (Fig. 1). Subsequent exploration of exosome components important for stimulating migration then revealed that tetraspanin CD81 plays an important role in mediating fibroblast-stimulated BCC motility and metastasis. We also found that human primary CAFs, isolated from breast cancer tissue, secrete CD81-positive exosomes that also promote BCC protrusive activity and motility. Finally, analysis of a publicly available gene expression dataset (27) revealed that CD81 mRNA levels are upregulated in stroma associated with human breast invasive carcinoma. Together, our results indicate that fibroblasts, including CAFs, secrete exosomes that stimulate BCC migration in metastasis in a CD81-dependant manner.

To elucidate the signaling pathway activated in BCCs upon stimulation with fibroblast exosomes, Smurf ubiquitin ligase expression was first knocked down, as Smurfs were previously shown to regulate cell protrusive activity and motility (28). We found that Smurf expression in BCCs is necessary for exosome-stimulated BCC protrusions and motility (10). Importantly, we also previously showed that Smurfs regulate convergent and extension movements and PCP signaling in mouse development (29). Thus, we investigated whether fibroblast exosomes activate the PCP pathway in BCCs (10) and found that core PCP proteins, including Fzd, Vangl, Dvl, and Pk, are necessary for fibroblast-induced protrusive activity and motility in both MDA-MB-231 and SUM-159PT cell lines. Moreover, we showed that Fzd–Dvl and Vangl–Pk complexes display asymmetric subcellular distribution with respect to cell protrusions in BCCs stimulated with fibroblast exosomes (Fig. 1). Furthermore, we showed that Pk expression in BCCs is necessary for fibroblast stimulation of BCC metastasis, but has no effect on primary tumor growth (10). These results suggest that PCP signaling regulates BCC protrusive activity and motility, possibly in a similar manner to its role in convergent and extension movements during vertebrate development.

To elucidate the mechanisms through which fibroblast exosomes stimulate PCP signaling in BCCs, the source of WNT ligand expression was explored. Although fibroblasts can secrete Wnt ligands (30), which could in principle activate the PCP pathway, this source of ligand was ruled out and (10) instead, expression of Wnt11 in the BCCs themselves was found to be important for exosome-stimulated motility (Fig. 1). Moreover, fibroblast exosomes, marked by a fluorescently tagged CD81 protein, were shown to be internalized by BCCs and to colocalize with BCC-produced Wnt11 in what are likely to be endocytic vesicular structures (Fig. 1). Together, these findings revealed that autocrine Wnt ligands, which tether to fibroblast-secreted exosomes within the endocytic system of recipient BCCs, activate Wnt-PCP signaling in cancer cell motility and metastasis (Fig. 1).

Our findings showed that fibroblast-secreted exosomes stimulate BCC motility during metastasis (10). Importantly, it remains to be determined whether fibroblast exosomes can act both locally and systemically, as has been observed for tumor-secreted exosomes (16). It will also be important to establish what role exosomes secreted from CAFs play in human cancers. To do so, first, it will be necessary to identify specific markers for human CAF exosomes to distinguish them from exosomes secreted by normal and tumor cells. Our proteomic analysis of fibroblast exosomes revealed a number of previously reported exosome components as well as several novel ones, which may serve as important functional components of CAF-specific exosomes that drive cancer progression, either as added components of the Wnt-PCP pathway or as parallel pathways that contribute to the metastatic phenotype. As an added benefit, fibroblast-specific exosome components might serve as useful metastatic biomarkers to assess disease progression in patients with cancer based on analysis of bodily fluids. Importantly, the concentration of tumor exosomes in sera from human patients has been correlated with disease progression (16). Moreover, our analysis of published gene expression data of carcinoma-associated stroma (27) indicates that CD81 expression is upregulated in human CAFs, which suggests that CD81 protein levels in CAF-secreted exosomes might correlate with disease stage. Finally, targeting of CAF-secreted exosomes may allow new and more effective therapies against cancer spread.

Our findings also show that tetraspanin CD81 functions in the context of exosomes to regulate BCC motility and metastasis. This finding is in agreement with previous reports that tetraspanins function as components of tumor-secreted exosomes to promote endothelial cell branching and angiogenesis (31). Next, it will be important to elucidate the mechanism through which CD81 functions in the context of fibroblast-secreted exosomes to stimulate BCC motility. Interestingly, tetraspanins have been proposed to mediate exosome fusion with the membrane of target cells (31). Thus, CD81 may be involved in the adhesion of fibroblast exosomes to BCCs and/or internalization by BCCs. In addition, tetraspanins have been proposed to regulate sorting of cargo molecules in exosomes (32), and thus components such as CD81 might act to recruit other exosome components that contribute to BCC motility. Exosome delivered tetraspanins that could also function in BCCs upon assimilation into the recipient BCC plasma membrane, as tetraspanins can associate with multiple proteins that regulate the actin cytoskeleton (33). Finally, CD81 may also regulate sorting of palmitoylated and/or GPI-linked Wnts in fibroblast exosomes within the MVBs of BCCs. All these possibilities will need to be investigated to understand the function of tetraspanins in general and CD81 in particular as components of metastasis-inducing exosomes.

Our study also showed that Wnt-PCP signaling regulates BCC protrusive activity, motility, and metastasis. It will thus be of interest to determine whether the core PCP pathway regulates the motility and metastatic potential of other cancer cell types. Furthermore, we showed that the asymmetrical distribution of core PCP components is conserved at the single-cell level of malignant cells. The PCP pathway has been notoriously difficult to study in in vitro cell culture models as establishing tissue level organization is challenging. Thus, our system may serve as a useful tool to investigate the intracellular molecular mechanisms that underlie PCP signaling in mammals. We further established that autocrine Wnt11 is tethered onto fibroblast exosomes upon their trafficking in BCCs, but how do BCC-produced Wnts associate with fibroblast-secreted exosomes? Currently, it is unclear how exosomes are internalized and what their fate is upon internalization (13). Our work suggests that endocytosis and recycling of exosomes in BCCs are important for mobilizing autocrine Wnt11. Moreover, Wnts are known to localize to MVBs and recycling endosomes within Wnt-producing cells, and this has been suggested to be necessary for Wnt maturation and signaling (23). Thus, we propose that Wnts are loaded onto fibroblast exosomes within the endocytic recycling compartment, with subsequent secretion and activation of Wnt-PCP signaling. It will be of interest to test this hypothesis by interfering with the function of the endocytic system, including the ESCRT and Rab proteins that regulate exosome biogenesis and secretion, respectively (34, 35). As a final note, our work does not exclude the possibility that fibroblast paracrine Wnt signaling may also be important in cancer metastasis. In fact, it was recently shown that CAFs secrete Wnt10b, which promotes endometrial cancer cell motility, likely through the PCP pathway (30). Furthermore, although both noncanonical Wnts (10) and canonical Wnts (25, 26) associate with exosomes, the functional relevance of exosome-mobilized Wnts in different Wnt-dependent processes has to be explored further. Future testing of these hypotheses promises to greatly expand our understanding of the role of exosomes in tumor–stroma communication and Wnt-PCP signaling in cancer metastasis.

No potential conflicts of interest were disclosed.

Conception and design: V. Luga, J.L. Wrana

Writing, review, and/or revision of the manuscript: V. Luga, J.L. Wrana

The authors apologize to those authors whose work could not be cited due to space limitations. The authors thank the members of J.L. Wrana's laboratory for helpful discussions.

This work was supported by funds from Canadian Institutes of Health Research (grant MOP14339), a Terry Fox Research Institute (Vancouver, BC, Canada) group grant, and a Krembil Foundation (Toronto, ON, Canada) grant awarded to J.L. Wrana.