Background
Breast cancer is the second leading cause of cancer-related death in women [
1]. Compelling evidence points to the tumor microenvironment (TME) as a major factor influencing breast cancer progression and response to therapy [
2]. In the breast cancer TME, cancer and host stromal cells are engaged in complex and dynamic interactions that promote tumor angiogenesis and obstruct the anticancer immune response. A better understanding of these mechanisms and interactive networks is integral for developing more effective therapies.
Tumor-associated fibroblasts (TAFs) are the predominant cell type in the TME of the most aggressive and difficult to treat cancers [
3]. TAFs have been implicated in recruitment of pro-angiogenic myeloid immune cells such as macrophages and myeloid-derived suppressor cells (MDSCs), thereby promoting tumor angiogenesis and metastasis [
4,
5]. Pro-angiogenic activity of myeloid cells depends on metalloproteinase MMP9/gelatinase-B [
5‐
7], which releases VEGFA from matrix-deposited sources increasing recruitment of endothelial cells and pericytes [
6,
8]. In addition, breast carcinoma cells produce MMP9 [
9‐
11] and TAFs enhance expression of MMP9 by tumor cells in the breast TME [
12]. Tumor blood vessels exhibit significant structural and functional abnormalities that provoke tumor hypoxia and metastatic spread [
13]. However, the roles of different cellular components of the TME and their interactions in tumor angiogenesis are not fully understood.
High levels of transforming growth factor-β (TGF-β) and pro-inflammatory cytokines such as tumor necrosis factor (TNF) have been reported for breast cancers [
14‐
17]. These cytokines are upregulated by TAFs and can co-stimulate expression of MMP9 in the breast TME [
12]. TGF-β signaling plays a critical role in breast carcinoma vascularization [
9] as well as in normal vascular and cardiac development [
18]. Mice lacking TGF-β type I receptor (
Tgfbr1/Alk5) exhibit severe defects in the vascular development [
19]. Endothelial cells (ECs) from Tgfbr1-mutant mice show enhanced proliferation, improper migratory behavior, and impaired fibronectin expression. It is still unclear how EC-intrinsic defects lead to vascular abnormalities. Several human syndromes including Marfan and Loeys-Dietz (LDS) are also associated with defects in the TGF-β pathway. However, the cell and molecular mechanisms underlying vascular abnormalities in these patients have not been defined.
Recent studies of breast carcinoma xenografts revealed that TAFs enhance tumor vasculature via a mechanism involving tumor TGF-β signaling [
12]. The current study examined the effects of TAFs on the structure of tumor blood vessels and the role of tumor TGF-β signaling in this response. The study found that TAFs enhanced coverage of endothelium by pericytes, vascular mural cells supporting blood vessels. Tumors with active TGFBR1 receptor increased amounts of TAFs in the TME and enhanced tumor vasculature with improved coverage of endothelium by pericytes. In contrast, inactive TGF-β signaling reduced amounts of fibroblasts and the association of pericytes with endothelium. Accordingly, tumors expressing inactive-TGFBR1 showed signs of hemorrhages. Together, these results indicate that TGF-β signaling is important for TAF-stimulated tumor vascularization.
Discussion
This study revealed that tumor-associated fibroblasts (TAFs) enhance tumor vascularization and tumor TGF-β signaling contributes to this response. We found that TAFs promote tumor growth (Fig.
1) and increase the lumen size of tumor blood vessels (Fig.
2). Inactivation of tumor TGF-β signaling (dnTGFBR1) reduced the microvessel density and lumen sizes (Fig.
2), decreasing tumor growth (Fig.
1). In contrast, tumors with constitutively active TGF-β signaling (caTGFBR1) exhibited greater the microvessel density and lumen sizes (Fig.
2). Inactivation of tumor TGF-β signaling decreased tumor infiltration by TAFs, while tumors with active TGF-β signaling exhibited greater presence of fibroblasts compared to control (Fig.
3). Examination of the vessel organization showed that TAFs enhanced microvessel coverage by pericytes, vascular mural cells supporting capillaries (Fig.
4). This effect was impaired in tumors with inactive TGF-β signaling, whereas active TGF-β signaling enhanced the pericyte-endothelium association (Fig.
4). Accordingly, tumors with inactive TGF-β signaling exhibited visible hemorrhages, a sign of fragile blood vessels (Fig.
1 and Additional file
2: Figure S2). Biochemical data revealed that TGFβ-SMAD signaling strongly up-regulates expression of fibronectin, which plays a prominent role in the pericyte-endothelium association [
31]. Thus, our findings suggest that the tumor-fibroblast crosstalk enhances tumor vascularization by stimulating the pericyte-endothelium association via a mechanism involving the TGF-β-fibronectin axis (Fig.
6).
The current study expands our understanding of how TAFs may promote tumor vasculature and cancer progression. Previous research has implicated TAFs in recruitment of pro-angiogenic immune cells promoting tumor angiogenesis via a mechanism mediated by matrix metalloproteinase MMP9 and VEGFA [
33,
34]. Recent studies have also implicated the tumor-fibroblast interactions in tumor angiogenesis by increasing expression in tumor cells of MMP9 and pro-angiogenic growth factors such as VEGFA and HB-EGF [
12,
27]. Here we found that TAFs, in addition to the above mechanisms, promote maturation of blood vessels by enhancing the pericyte-endothelium association. Pericytes are embedded within the perivascular matrix and cover the walls of capillaries, providing the structural support to capillaries [
35]. The capillary formation involves the recruitment of pericytes through PDGFB-PDGFRB and SDF1-CXCR4 signaling [
35], while association of pericytes with endothelium is mediated by the perivascular matrix [
35,
36]. Our data indicate that TAFs increased the pericyte-endothelium association but did not change tumor infiltration by pericytes. Thus, TAFs may promote the pericyte-endothelium association by regulating the perivascular matrix rather than recruitment of pericytes.
Loss-of-function and gain-of-function experiments revealed that tumor TGF-β signaling enhances tumor infiltration by fibroblasts and maturation of tumor blood vessels. Our previous work showed that the TGFBR1 activity dramatically affects the tumor cell-intrinsic metastatic potential [
9]. Here, we found that TGF-β signaling augmented both host and admixed fibroblasts. However, the amount of pericytes in tumors with inactive TGF-β signaling (dnTGFBR1) was elevated compared to control and caTGFBR1 tumors. This effect may associate with hypoxic conditions observed in dnTGFBR1-tumors [
9], as hypoxia stimulates recruitment of pericytes [
37]. Importantly, inactivation of TGF-β signaling impaired the pericyte-endothelium association stimulated by TAFs, whereas active TGF-β enhanced the pericyte-endothelium coverage. These findings indicate that TGF-β signaling promotes maturation of blood vessels by enhancing the pericyte-endothelium interaction. Accordingly, dnTGFBR1-tumors exhibited hemorrhages and reduced growth compared to control and caTGFBR1 tumors. This observation is consistent with increased hypoxia and cell death in orthotopic xenografts of dnTGFBR1-tumor [
9].
The current results suggest a possible mechanism by which TGF-β and TAFs may regulate the pericyte-endothelium association. We found that TGF-β signaling and TAFs stimulate expression of fibronectin, an extracellular matrix protein implicated in vascular development [
38]. Several findings support the role of fibronectin in the pericyte-endothelium interaction. Fibronectin-deficient embryos display defects in the formation of vascular lumen and vascular network [
38]. Disruption of fibronectin fibrils impairs the perivascular matrix and decreases the pericyte-endothelium association [
31]. The tumor-fibroblast crosstalk increased expression of fibronectin (Additional file
4: Figure S4 and [
27]). TGF-β signaling, but not BMP, induced deposition of fibronectin fibrils (Fig.
5) and myofibroblast-like phenotype associated with matrix deposition and remodeling [
39]. In addition, human breast carcinomas show elevated deposits of fibronectin primarily in stromal compartments and this correlates with location of blood microvessels (Additional file
6: Figure S6). Fibronectin may promote the pericyte-endothelium association by regulating the deposition and signaling of pro-angiogenic cytokines. Fibronectin fibrils anchor VEGF-A, FGF2, and TGF-β cytokines regulating availability of active cytokines, while fibronectin-integrin interactions may facilitate cytokine release and signaling [
38]. Fibronectin fibrils can also regulate vascular cell migration, differentiation, proliferation, and survival [
38,
40]. Finally, fibronectin fibrils may physically facilitate the organization of the vascular basement membrane [
40]. Consistent with our findings, TGF-β signaling may play a more complex role by promoting pericyte maturation as well as expression and deposition of VEGF by pericyte precursors interacting with endothelial cells [
41].
Our results shed light on the function of TGF-β signaling in tumor angiogenesis. We found that TGFBR1/ALK5 signaling regulates the pericyte-endothelium association (Fig.
4). Consistent with this idea, tumors with inactive TGFBR1 exhibit hemorrhages and signs of leaky vessels (Fig.
1b and Additional file
2: Figure S2). Tumor blood spots (Additional file
2: Figure S2) are strikingly reminiscent of blood-vessel lesions (red spots or telangiectasia) typically found in patients with hereditary hemorrhagic telangiectasia (HHT) [
42]. HHT is manifested by multiple red spots known as telangiectases around lips, oral mucosa, and fingertips [
43]. Telangiectases consist of abnormally dilated thin-walled vessels that are prone to spontaneous and recurrent bleeding. HHT is the autosomal-dominant trait and about 90% of HHT cases are linked to genetic inactivation of the TGF-β pathway in endothelial cells [
43,
44]. Recent studies have linked the pathogenesis of HHT to excessive angiogenesis and loss of capillary bed between arteries and veins [
45]. In the mouse models of HHT, inactivation of ALK1, an endothelial-specific TGF-β type I receptor, or Endoglin, a TGF-β axillary receptor, results in disruption of the pericyte-endothelium communication leading to insufficient coverage of capillaries by pericytes [
45]. Our data are consistent with these studies implicating TGF-β signaling in the regulation of the pericyte-endothelium association [
45]. Thus, a tumor-fibroblast co-xenograft model may represent a valuable system for examining complex mechanisms underlying angiogenesis and vascular abnormalities in human diseases. In particular, it may help to define the contribution of the SMAD-fibronectin axis to the pericyte-endothelium interaction.
Our findings have broad translational implications for anti-cancer therapy targeting blood vessels. The identified role of TAFs and TGF-β signaling in the maturation of blood vessels raises a question whether targeting TAFs and/or inactivation of TGF-β would improve or worsen cancer treatment. Recent studies indicate that depletion of pericytes decreases tumor growth but markedly increases lung metastasis [
29,
46,
47]. Consistent with our data, these reports show an increased capillary bleeding and a reduced tumor oxygenation. While the TGF-β pathway is a potential target in the metastatic disease, our results seed a doubt for a systemic administration of drugs inhibiting a kinase function of TGF-β receptors. In this regard, TGF-β signaling mediators such as TAK1 or p38, which also contribute to tumor angiogenesis and cancer progression [
12,
20,
27,
48], may provide a better alternative strategy.
In summary, our study uncovered a novel mechanism by which TAFs may regulate tumor angiogenesis (Fig.
6). The identified tumor-fibroblast crosstalk upregulates TGF-β signaling that, in turn, increases production of fibronectin and other matrix proteins. TGF-β-stimulated matrix proteins enhance formation of the perivascular matrix, facilitating the pericyte-endothelium association.