A Role for Endothelin-2 and Its Receptors in Breast Tumor Cell Invasion Free

Endothelins (ETs) are a group (ET-1, ET-2 and ET-3) of small (21-amino acid) vasoactive peptides with diverse paracrine/autocrine actions (1). ET-1 and ET-2 have similar structures, differing by only two amino acids, whereas ET-3 differs in structure at six amino acids (2). ETs bind at least two receptor subtypes, ET-A receptors (ET-RA) and ET-B receptors (ET-RB; Ref. 3), which are seven-transmembrane domain G protein-coupled receptors (4).

ETs are expressed in a broad range of tumors, and there is a growing interest in ETs and cancer; most reports focus on the role of ET-1/ET-RA in tumor growth, progression, and angiogenesis (5). However, we recently proposed a role for ET-2/ET-RB in invasive ductal carcinoma (IDC) of the breast: binding of ET-2 to ET-RB protects tumor cells from hypoxia-associated apoptosis in an autocrine fashion (6). Both ET-2 and ET-RB are rarely detected in the nonneoplastic human breast, but breast carcinomas contain regions of hypoxia that increase production of ET-2 as well as ET-RA and ET-RB by tumor cells (6). Intratumoral injection of BQ-788, an ET-RB selective antagonist (7), causes increased necrosis and decreased tumor volume in a murine model of breast carcinoma (6); BQ-123, a selective ET-RA antagonist (8), also leads to decreased breast tumor volume but without a concomitant increase in necrosis.

High levels of macrophage infiltrate in breast tumors correlate with a poor prognosis (9) and hypoxic tumors have a higher invasive capacity and a poorer prognosis than well-oxygenated tumors (10). Hypoxia-associated ETs expression by tumor cells may also modulate tumor-associated macrophage distribution and activation in breast tumors; ET-1 and ET-2, but not ET-3, are chemoattractant for macrophages and lead to an activated phenotype in vitro (11). In vivo, ET expression in IDC not only correlates with hypoxia (6) but also with the presence of activated “foamy” tumor-associated macrophages (11).

ETs may also play a role in chemotaxis of several additional cell types: ET-1 and ET-3 are chemoattractants for pulmonary artery fibroblasts (12), and ET-2 is a chemoattractant for neutrophils (13).

As well as being chemoattractant for several cell types, ET-1 may also modulate the invasive capacity of some tumor cells including ovarian carcinoma cells (14), Ewing’s sarcoma, and neuroblastoma cells (15). In ovarian carcinoma cells, ET-1 interacts with ET-RA to increase the expression and activation of matrix metalloproteinases (MMPs) whereas decreasing the secretion of tissue inhibitor(s) of MMPs (TIMP; Ref. 14).

In view of the possible role of ETs in invasion of tumor cells, we investigated whether ETs are chemoattractant for breast carcinoma cells and, in conjunction with macrophages, whether they may modulate the invasive capacity of breast tumor cells. By examining human neoplastic breast tissues, we also analyzed whether ETs are associated with progression and invasion of tumor cells in ductal carcinoma in situ (DCIS), IDC, and lymph node metastases. Because of its regulation by hypoxia and increased expression in IDC compared with normal breast tissue (6), we focused on the potential role of ET-2/ET-RB.

ET peptides and receptor antagonists BQ-123 (ET-RA antagonist; Ref. 16) and BQ-788 (ET-RB antagonist; Ref. 7) were obtained from American Peptide Co., Sunnyvale, CA. MAPK kinase (MEK)1/2 inhibitors PD98059 and U0126 were purchased from Cell Signaling Technology Inc., Beverly, MA. Pertussis toxin (PTX; G-protein inhibitor) and FN439 (nonspecific MMP inhibitor) were supplied by Calbiochem-Novabiochem Corporation, San Diego, CA.

HTH-K cells were cultured as described previously (17). Human cell lines (MCF-7, BT20, BT474, SKBR3, and MDAMB468) from the American Type Culture Collection (Manassas, VA) were cultured similarly. peripheral blood mononuclear cells were isolated from human blood and matured into macrophages under endotoxin-free conditions as described previously (11).

RNA was isolated from cell lines, and cDNA was amplified as described previously (6).

Subconfluent tumor cells were trypsinized and resuspended in DMEM supplemented with 10% FCS (2 × 10⁵ cells/ml) and were then subjected to chemotaxis assays and counted as described previously (11). A 12-μm membrane, that had been treated with 6.5 μg/ml fibronectin solution at 37°C for 2 h and then dried was used. Migration was stopped after 5 h for each experiment.

Protein extracts were prepared from cells that had been serum-starved for 24 h before stimulation with ETs. Western blotting for ET receptors was carried out as described previously (6) and for signaling pathways as described previously (11).

In vitro-invasiveness was measured in a modified Boyden chamber as described in Hagemann et al. ⁶ MCF-7 cells were seeded in gelatinase-free medium on polycarbonate membranes coated with Matrigel (ECM gel; Sigma-Aldrich Co. Ltd., Poole, United Kingdom). Macrophages were added at 2:1 (macrophages:tumor cells) in inserts that allow exchange of liquids but no direct cell-cell contact or invasion of macrophages. ETs (100 ng/ml) were added, and renewed every 24 h. After 96 h, the contents of the lower well were removed, and cells were collected by centrifugation. The supernatant was used for zymography. Cells were spun down onto coverslips and stained with 4′,6′-diamino-2′-phenylindol (DAPI, 200 ng/ml). Intact nuclei were counted in six high power fields using UV-microscopy (Axioskop; Zeiss, Jena, Germany).

Protein content in the supernatant was determined using the bicinchoninic acid assay (Pierce, Rockford, IL). Twenty μg were separated on 10% SDS-polyacrylamide gels containing 1 mg/ml gelatin. After electrophoresis, gels were washed for 1 h in renaturation buffer (2.5% Triton X-100) and then were incubated for 36 h at 37°C in 50 mm Tris, 200 mm NaCl, and 5 mm CaCl₂ (pH 7.5). Gels were stained with 0.05% Coomassie Brilliant Blue, and MMPs were identified by size and comparison with the recombinant proteins (Calbiochem-Novabiochem Corporation).

Intraductal and invasive human breast cancer archival samples and corresponding metastatic lymph node samples were supplied by the Hedley Atkins Breast Pathology Laboratory, Guy’s Hospital, London, United Kingdom. All of the specimens were formalin-fixed and paraffin-embedded. Protein expression of ETs and ET receptors were immunohistochemically examined as previously (11). Sections from representative paraffin blocks of tumors were deparaffinized, rehydrated, and blocked for endogenous peroxidase activity using 3% H₂O₂. Mature ETs were detected using a rabbit antimouse ET polyclonal antibody (Oncogene Research Products, Cambridge, MA) that recognizes all ET isoforms (18). Sections were also immunostained with antibodies raised against the COOH terminus of the ET receptors (19). Sections were stained as described previously (11) using a sheep anti-ET-RA polyclonal antibody (Alexis Biochemicals, San Diego, CA) or a sheep anti-ET-RB polyclonal antibody (Calbiochem-Novabiochem Corporation). The specificity of all antibodies used was proven using blocking peptides by the supplying companies (data not shown). Appropriate biotinylated secondary antibodies were used. The sections were incubated with streptavidin-peroxidase and were developed using diaminobenzidene substrate followed by counterstaining with Harris’s hematoxylin. The sections were independently examined by a consultant histopathologist (A. A.). Immunoreactivity was semiquantitatively evaluated by comparing the level of expression in malignant cells with that of normal breast tissue present in the same section. Tumors were considered to be “weakly positive” if up to 25% of cells expressed the protein and “strongly positive” if more than 25% of cells did. This cutoff point was chosen because it seemed to divide the sections into two distinct groups. No staining was observed with negative control samples (absence of primary antibody, or incubation with an irrelevant antibody or IgG).

Results were tested for statistical significance using one-way ANOVA or Tukey-Kramer multiple comparisons test with InStat Version 3.0 software. Western blots were quantified by densitometry using NIH Image Version 1.63.⁷

The murine breast carcinoma cell line HTH-K expresses ET-2, ET-RA, and ET-RB, and this expression is increased by exposure to hypoxia (6). We examined ET and ET receptor mRNA expression in five human breast tumor cell lines: SKBR3, MCF-7, BT20, BT474, and MDAMB468. Similar to the expression profile seen in IDC of the breast (6), all of the tumor cell lines expressed ET-1 and ET-2, but none expressed ET-3; all of the cell lines expressed ET-RB, whereas all but one, BT20, expressed ET-RA (Fig. 1 A).

We used breast carcinoma cell lines to investigate chemotaxis of breast tumor cells toward ETs in vitro. MCF-7 cells migrated strongly toward 100 ng/ml ET-2 and less robustly toward ET-1 (Fig. 1,B). Tumor cells did not migrate toward ET-3. Murine HTH-K cells had a similar chemotactic profile, and the other human breast tumor cell lines tested (MDAMB468, SKBR3, and BT20) also chemotaxed toward ET-2 (Fig. 1,C). BT20, which does not express ET-RA (Fig. 1,A), had a decreased chemotactic ability toward ET-2, compared with other cell lines (Fig. 1 C).

In a dose-response assay, MCF-7 cells migrated toward ET-2 demonstrating a bell-shaped response curve with maximum migration at 50–100 ng/ml ET-2 (Fig. 1,D). To ensure that ET-2 did not stimulate random migration, the ET-2 gradient was abolished by adding the peptide to both lower and upper wells of the chamber; this revealed that the effects of ET-2 are chemotactic rather than chemokinetic (Fig. 1 E).

To determine which receptor(s) is responsible for tumor cell chemotaxis, we analyzed chemotaxis toward ET-2 in the presence of ET-RA antagonist BQ-123 (16) and/or ET-RB antagonist BQ-788 (7). At a concentration of 100 ng/ml, these modified-peptide antagonists are highly specific for their respective receptor (7, 16) and can distinguish between ET-RA- and ET-RB-induced chemotaxis (11). Incubation of MCF-7 cells with either BQ-123 or BQ-788 before and during the microchemotaxis assay partially decreased, but did not eliminate, migration toward ET-2 (Fig. 2,A). However, combining both antagonists abolished migration toward ET-2 (Fig. 2 A). Neither of the ET receptor antagonists affected viability (assessed by trypan blue exclusion assay; data not shown).

We next analyzed the signaling pathways involved in ET-induced chemotaxis. In macrophages, chemotaxis toward ET-2 is via an ET-RB-mediated MAPK pathway that can be inhibited by hypoxia (11).

Microchemotaxis assays of MCF-7 cells toward ET-2 were carried out in a hypoxic incubator (0.1% oxygen, 10% CO₂, balance nitrogen). Unlike macrophages (11), breast tumor cells were not inhibited by hypoxia from migrating toward ET-2 (Fig. 2 B).

PD98059, which inhibits MEK1 at low concentrations (5 μm) and MEK1/2 at higher concentrations (50 μm; Ref. 20), partially inhibited MCF-7 cell chemotaxis toward ET-2 at 5 μm and completely blocked chemotaxis at 50 μm (Fig. 2,C). The MEK1/2 inhibitor U0126 (10 μm) also inhibited migration of MCF-7 cells toward ET-2 (Fig. 2 C).

ET-RA may increase intracellular calcium by at least two mechanisms: a PTX-insensitive calcium mobilization from intracellular stores and a PTX-sensitive influx of extracellular calcium (21). PTX also inhibits ET-RB-mediated chemotaxis of macrophages toward ET-2 (11). Incubation of MCF-7 cells with PTX (100 ng/ml) led to decreased migration toward ET-2 (Fig. 2,D). Migration of MCF-7 cells toward ET-2 was also inhibited by incubating the cells in calcium-free medium before and during the chemotaxis assay (Fig. 2 D).

Cell viability was not affected by any of the inhibitors used (data not shown).

To demonstrate that activation of the p42/p44 MAPK pathway was responsible for chemotaxis toward ETs, we analyzed the phosphorylation status of key molecules in several distinct signaling pathways. Incubation of serum-starved MCF-7 cells with 100 ng/ml ET-2 for 0–24 h led to a transient activation of p42 and p44 MAPK with maximal phosphorylation between 1 and 5 h (Fig. 2,E). There was no activation of the stress-activated p38 or stress-activated protein kinase (SAPK)/Jun N-terminal kinase (JNK) pathways (Fig. 2 E).

Because chemokines may play a role in breast tumor metastasis (22), we analyzed the ability of tumor cells exposed to ET-2 to chemotax toward CXCL12 and CCL21. Primary breast carcinoma cells and cell lines, including MCF-7, express the chemokine receptors CXCR4 and CCR7, which mediate migration toward CXCL12 and CCL21, respectively (22). These chemokines are expressed at sites of metastasis such as the lymph nodes (22). Cells that had been incubated with ET-2 before the chemotaxis assay had a higher migratory capacity toward CXCL12 and CCL21 than did untreated cells (Fig. 3). MCF-7 cells do not express CCR2 (22), and ET-2 did not significantly increase migration toward CCL2.

We analyzed whether the increased migration toward the chemokines CXCL12 and CCL21 was due to up-regulation of the chemokine receptors CXCR4 and CCR7 respectively. Fluorescence-activated cell sorting analysis showed that there was no up-regulation of these receptors at the cell surface (data not shown).

We next investigated whether ETs increase the invasive capacity of breast tumor cells in an in vitro invasion assay. To test whether ETs may induce invasion in a chemotactic manner, ET peptides were added to the lower wells of the invasion chamber during a 96 h assay. ET-1 and ET-2 caused an increase in invasion (Fig. 4 A) whereas ET-3 had no significant effect.

Next, we analyzed whether incubating cells with ETs increases gradient-independent invasion. ET peptide (100 ng/ml) was added to the cells in the upper well of the invasion assay; ET-1 and ET-2 greatly increased the number of tumor cells that had invaded into the lower chamber (Fig. 4 B), whereas ET-3 had no effect.

Coculture of breast tumor cells with macrophages leads to increased invasive capacity of tumor cells compared with tumor cells cultured in isolation (Hagemann et al.).⁶ Furthermore, incubation of macrophages with ET-2 leads to increased activation of these cells (11). Therefore, we evaluated the effect of ETs on cocultured macrophages/tumor cells. The presence of macrophages led to increased MCF-7 tumor cell invasion through Matrigel compared with tumor cells cultured in isolation (Fig. 4,C). Addition of ET-1 or ET-2 to the cocultured cells led to a further increase in invasion by MCF-7 cells, but ET-3 had no effect (Fig. 4 C).

To test which ET receptor(s) was involved in the increased invasion of cocultured MCF-7 cells, we added ET-RA and ET-RB antagonists (singularly and in combination) to the invasion chambers. Unlike the chemotaxis assays, which showed partial inhibition of MCF-7 migration by both ET-RA and ET-RB antagonists and an additive effect when used in combination, both BQ-123 and BQ-788 could entirely abolish the effect of ET-2 on invasion of MCF-7 cells (Fig. 4 D); the effects of BQ-123 and BQ-788 were not additive. Neither BQ-123 nor BQ-788 inhibited macrophage-induced invasion of MCF-7 cells when ETs were not added (data not shown).

The increase in tumor cell invasion induced by ET-1 and ET-2 in coculture suggested that the macrophages were producing a soluble factor that aided tumor cell invasion. We analyzed MMP-2 and -9 production by tumor cells and macrophages using zymography. MCF-7 cells produced little or no MMP-2 or -9 with or without stimulation by ETs (Fig. 5,A). Cultured in isolation, macrophages produce MMP-2 and –9, but production was not altered by the addition of ETs (Fig. 5,B). However, when cocultured tumor cells and macrophages were treated with ETs, all of the ETs caused an increase in MMP-9 and MMP-2 production (Fig. 5,C). The addition of 900 μm FN439, a nonspecific MMP inhibitor, to the coculture invasion assay abolished the increase in tumor cell invasion caused by ET-2 (Fig. 5 D).

We analyzed MMP (MMP-2, -3, -7, and -9) and TIMP (TIMP-1, TIMP-2, and TIMP-3) mRNA expression by macrophages in coculture by real-time PCR. The addition of ETs or receptor antagonists did not alter the macrophage mRNA expression of MMPs or TIMPs (data not shown).

In a preliminary study, we compared protein expression of ET and ET receptors in 9 cases of DCIS, 10 IDCs, and 10 lymph node metastases of primary breast carcinomas. ETs were expressed in the cytoplasm, with an increased level in tumor tissue compared with nonneoplastic breast tissue (Fig. 6; Table 1). The percentage of cases with strong positive staining of both ET and ET-RB increased from DCIS cases to IDC to the metastases (Fig. 6; Table 1). We also observed that carcinoma cells that were infiltrating the mammary fat tissue and those in the lymphovascular space tended to stain more intensely for ETs and their receptors than other parts of the same tumor (Fig. 6). It is also of note that in all neoplastic tissue (DCIS, IDC, or lymph node metastases) that showed ET immunopositivity, ET receptor staining, particularly ET-RB, was also observed (Table 2). All of the lymph node metastases showed immunopositivity for both ET and ET-RB.

In view of the fact that ET and ET receptor staining often colocalized in vivo and in view of the potential autocrine role(s) of ET-2 in breast neoplasia, we analyzed whether ET-2 induced ET receptor production by MCF-7 cells in vitro. Cells stimulated with 100 ng/ml ET-2 showed a rapid and sustained (1–24 h) increase in the production of ET-RB (Fig. 7); ET-RA production by MCF-7 cells exposed to ET-2 was increased by 24 h.

Many tumor-linked factors have been implicated in the development and progression of invasion and there has recently been interest in the role of ET-1/ET-RA, particularly in ovarian and prostate cancer metastasis (14, 23). Normal breast tissue expresses little ET or ET-RB, but intratumoral hypoxia (24, 25) induces expression of ET-2 and both ET receptors by tumor cells (6). We have now shown that ETs induce a chemotactic and invasive phenotype in breast tumor cells in vitro, particularly when tumor cells are cocultured with macrophages. Furthermore, data from the preliminary immunohistochemical study indicates that ET and ET receptor expression increases from normal breast to DCIS to IDC to metastatic deposits. This is the first time a possible role for ETs in breast cancer invasion has been identified and also the first time that ET-2/ET-RB has been implicated in invasion/progression.

It is interesting that tumor cells migrate strongly toward ET-2; however, it is difficult to understand whether this may have a broader role in metastasis. Physiologically, ET-2 expression is limited to the kidney and the intestine and is not released systemically in any great concentration (1). Thus, it is unlikely that chemotaxis toward ET-2 by breast tumor cells could explain organ-selective metastases, although breast cancers may metastasize to the kidney (26). It is also worth noting that ET-1 is expressed at several additional sites of breast cancer metastasis including bone (27).

It is conceivable, however, that migration toward ET-2 or ET-1 has an impact within the tumor, such that cells mass within hypoxic areas and become more aggressive because of exposure to ETs or by selective pressure of hypoxia. It is also possible that tumor cells may use ETs to migrate toward blood vessels, in which ET-1 is likely to be expressed (28), and thus can begin the process of intravasation. The high levels of ET and ET receptor expression by tumor cells within blood vessels and the high levels of gradient-independent invasion induced by ET-1 or ET-2 suggests that ET/ET receptor interactions have a role in intravasation and the preliminary steps of invasion and metastasis.

Systemically, it may be of more significance that treating tumor cells with ET-2 leads to increased chemotaxis toward the chemokines CXCL12 and CCL21, which are expressed by the lymph nodes and other sites of breast cancer metastasis. We were not able to detect an up-regulation of chemokine receptors on MCF-7 cells when stimulated by ET-2, and, thus, it may be that ET-2 impinges on the signaling pathways to amplify the response of tumor cells to chemokines.

Similar to macrophages (11), the preliminary steps in ET signaling in tumor cells is typical of chemokine signaling (29). Despite the action of ET-2 during cellular stress (6), ET-2 does not signal via the stress-activated p38 or SAPK/JNK pathways, but via the MEK1/2 and p42/p44 MAPK pathway. ET-2 signaling also involves activation of the G-protein-coupled ET receptors and an influx of extracellular calcium. Although ET-2 signaling in tumor cells has parallels with ET-2/ET-RB signaling in macrophages (11), it is clear that there are several differences. Macrophage chemotaxis toward ET-2 involves solely ET-RB, whereas chemotaxis and invasion of breast tumor cells involved both ET-RA and ET-RB. Chemotaxis toward ET-2 of both cell types can be inhibited by PD98059 or PTX, yet only tumor cells can migrate toward ET-2 under hypoxia. Both ET-2 and hypoxia increase ET receptor expression by tumor cells (6), and this may potentiate hypoxic migration. There may also be a role for ET-RA in hypoxic migration, which is expressed by tumor cells but not by macrophages (11). The chemotactic differences could also be due to the time of the chemotaxis assays; macrophage migration under hypoxia is inhibited rapidly by a short-term induction of MKP-1 (30), whereas tumor cells are allowed to migrate over a period of several hours.

Although the complete abolition of ET-induced chemotaxis required the inhibition of both receptors, ET-induced invasion of MCF-7 cells in coculture could be abolished by either receptor antagonist. This may be due to the complex processes and interactions that occur during invasion in macrophage/tumor cell coculture; individual elements of the invasion process may be mediated by one or other ET receptor allowing abolition of invasion by one receptor antagonist alone. The relative contributions of ET-RA and ET-RB signaling when tumor cells are stimulated by ET-2 are yet to be elucidated and may help us to understand the differences between macrophage and tumor cell signaling pathways.

Macrophages, which colocalize with ETs in breast cancer (11), increased MMP production in response to ETs, thus facilitating tumor cell invasion. Hagemann et al ⁶ showed that the increase in MMP production in macrophage/tumor cell cocultures was due to the macrophages and not to the tumor cells. The inhibition of the MMPs was sufficient to prevent MCF-7 cell invasion, which shows the critical role of these enzymes in invasion. However, all of the ETs could increase production of MMP-2 and -9 by macrophages, yet ET-3 did not increase the invasion of tumor cells. This indicates that additional factors produced by cocultured cells and stimulated (or inhibited) by ETs, distinct from MMP-9 and MMP-2, must also play a role in tumor cell invasion. It is probable that there is a complex interaction between ET expression, hypoxia, and tumor-associated macrophages within the breast tumor such that invasion is increased. However, additional interactions may occur such as paracrine ET stimulation of tumor endothelial cells.

As well as the paracrine actions of ETs in breast tumors, those neoplasia that expressed ET in vivo also expressed the ET receptors, which may indicate an autocrine loop. We recently proposed an autocrine survival function for ET-2/ET-RB binding in breast carcinoma cells under hypoxic conditions (6). All of the human cell lines expressed ET-1 and ET-2 in vitro as well as one or more ET receptor; furthermore, production of the ET receptors by tumor cells in vitro was stimulated by ET-2. These data support an autocrine action of ETs to increase the invasiveness of tumor cells.

Examination of the expression of ETs and their receptors in a large number of breast carcinomas and metastatic deposits, and correlation with macrophage infiltrate, is now required to determine whether there is a correlation between ET/ET receptor expression and grade (or currently used conventional prognostic factors) and prognosis. Such studies may also indicate whether treatment of breast cancer and its metastases with ET receptor antagonists may have a therapeutic benefit. Because ETs stimulates not only breast cancer cell invasion but also chemotaxis toward chemokines, it may be that combined treatment with both ET receptor antagonists and chemokine receptor antagonists may potentially prevent breast cancer metastases better than do single treatments alone. Atrasentan, a selective ET-RA antagonist, is already in clinical trials for prostate cancer (31) and may be of therapeutic benefit for the treatment of IDC of the breast.

Up-regulation of ET-2 and its receptors by hypoxia (6), together with this data implying a role for ET-2 in invasion, may be one factor that explains why hypoxic tumors are more aggressive than nonhypoxic ones (10) and why hypoxia increases carcinoma cell invasion (32).

Not only is this the first time that a role for ET-2 or ET-RB in tumor cell invasion and metastasis has been proposed but it is also the first time the role of ET-stimulated macrophages in tumor cell invasion has been studied. The data showing ET-associated induction of MMP release by macrophages in coculture and increased chemotactic/invasive capacity of tumor cells together with increasing expression of ETs and ET receptor protein from DCIS to IDC to tumors in metastatic lymph nodes support a role for ET expression in progression and metastasis.