Metastasis is the major cause of death in patients with cancer; with no therapeutic cure, treatments remain largely palliative. As such, new targets and therapeutic strategies are urgently required. Here, we show that bone morphogenetic protein-4 (BMP4) blocks metastasis in animal models of breast cancer and predicts improved survival in patients. In preclinical models of spontaneous metastasis, BMP4 acted as an autocrine mediator to modulate a range of known metastasis-regulating genes, including Smad7, via activation of canonical BMP-SMAD signaling. Restored BMP4 expression or therapeutically administered BMP4 protein, blocked metastasis and increased survival by sensitizing cancer cells to anoikis, thereby reducing the number of circulating tumor cells. Gene silencing of Bmp4 or its downstream mediator Smad7, reversed this phenotype. Administration of recombinant BMP4 markedly reduced spontaneous metastasis to lung and bone. Elevated levels of BMP4 and SMAD7 were prognostic for improved recurrence-free survival and overall survival in patients with breast cancer, indicating the importance of canonical BMP4 signaling in the suppression of metastasis and highlighting new avenues for therapy against metastatic disease.

Significance:

Targeting the BMP4–SMAD7 signaling axis presents a novel therapeutic strategy to combat metastatic breast cancer, a disease that has had no reduction in patient mortality over 20 years.

Patients with late-stage breast cancer present with tumors that have metastasized to the lung, liver, brain, or bone, eventually compromising organ function and leading to patient death. With around 1.7 million new cases of invasive breast cancer diagnosed and over 500,000 deaths each year worldwide, the identification and characterization of new biomarkers and therapeutic targets is a high priority (1).

Metastasis is controlled by multiple genetic elements that are often distinct from known oncogenes (2). Members of the TGFβ superfamily—consisting of more than 30 cytokines from the TGFβ, activin, and bone morphogenetic protein (BMP) subfamilies and their associated signaling pathways—are pivotal to metastatic predisposition (3). Indeed, genetic suppression of TGFβ/activin signaling by introduction of a dominant-negative TGFβRII (4), or silencing of downstream canonical SMAD2/3, can inhibit breast cancer metastasis (5, 6). SMAD7, a canonically induced inhibitory SMAD (iSMAD) that antagonizes type-I BMP and TGFβ receptor complexes (7), also demonstrates metastasis-suppressive phenotypes in animal models (8). Impaired BMP signaling through loss of cognate receptor BMP-RII or genetic deletion of canonical BMP-SMAD proteins SMAD1/5, can promote spontaneous metastasis in tumor-prone animal models of breast, prostate, ovarian, and testicular cancer (9–13). These findings indicate that epithelial tumor cells have an inherent capacity to metastasize and may be influenced by specific modulation of the SMAD pathway. Because TGFβ and BMP cytokines utilize overlapping, yet distinct intracellular SMAD signaling pathways, it is of clinical relevance to identify the mediators that can activate relevant pathways that favor metastasis suppression.

As part of our on-going efforts to identify metastasis-regulating elements (14–16), we screened a panel of isogenic mammary tumors of varying metastatic capacity and discovered that bone morphogenetic protein-4 (BMP4) is associated with reduced metastatic capacity. BMP4 is an essential morphogen that regulates a diverse range of developmental processes akin to those observed during metastasis, including cellular differentiation, pluripotency, apoptosis, and migration (17, 18). Recent evidence indicates that BMP4 is expressed in some breast tumors; however, its association with clinical outcome is yet to be established (19). We report here that BMP4 expression is lower in aggressive breast cancer cell lines, orthotopic tumors, and patient samples. Moreover, we demonstrate the functional requirement of an active BMP4–SMAD4–SMAD7 signaling axis to block metastasis in vivo and to predict favorable outcomes in patients with breast cancer.

Cell lines and in vitro assays

The syngeneic Balb/c 4T1 series of mouse mammary tumor lines (67NR, 168FARN, 4T07, 66cl4, and 4T1) were obtained from Dr. Fred Miller (Karmanos Cancer Institute, Detroit, MI) in 1995. 4T1.2 and 4T1.13 are bone-metastatic derivatives of 4T1 that we isolated (14). The normal mouse mammary epithelial line NMuMG was a gift from Dr. Mark Waltham (St Vincent's Institute, Victoria, Australia) in 2002. E0771 cells were derived from a spontaneous mammary tumor in a C57Bl/6 mouse and obtained from Dr. Edward Cohen (University of Illinois, Chicago, IL) in 2005. Promyoblast C2C12 cells, and the HME and MCF10A panels of human breast epithelial lines were obtained from Drs. Naoki Nakayama, Rick Pearson, and Patrick Humbert (Peter MacCallum Cancer Centre, Melbourne, Australia) in 2005. The immortalized human mammary epithelial PMC42-LA line was a gift from Dr. Leigh Ackland (Deakin University, Australia) in 2009. Human breast cancer cell lines MDA-MB-468 (TNBC), MCF7 (ER+), ZR-75-1 (ER+), and MDA-MB-231 (TNBC) were purchased from ATCC in 2012; SUM149 (TNBC), SUM190 (HER2+), and SUM159 (TNBC) were purchased from BioIVT (New York, NY) in 2010. The human MDA-MB-231-LM2 cell line was a gift from Dr. Joan Massagué (Memorial Sloan Kettering Cancer Center, New York, NY) in 2008; IBC3 (HER2+) was a gift from Dr. Wendy Woodward (MD Anderson Cancer Center, Houston, TX) in 2013; KPL4 (HER2+) was a gift from Dr. Junichi Kurebayashi (Kawasaki Medical School, Kurashiki, Japan) in 2011; and FC-IBC-02 (TNBC) was kindly provided by Dr. Massimo Cristofanilli (Fox Chase Cancer Center, Philadelphia, PA) in 2014. Mouse J110 (ER+) cells were a kind gift from Dr. Myles Brown (Dana-Faber Cancer Institute, Boston, MA) in 2015. Cells were maintained in alpha-Modified Eagle Medium (14) supplemented with penicillin/streptomycin and 5% FBS, or according to the supplier's instructions at 37°C in an atmosphere of 5% CO2 for no more than 6 weeks. All cell lines were maintained as frozen stock, with cells brought into culture for no more than 4 weeks prior to experimentation. Mycoplasma testing was completed approximately every 3 months on all cells in use at the time, using in-house testing facilities with commercial Mycoplasma testing kits at Olivia Newton-John Cancer Research Institute (Heidelberg, Australia), Peter McCallum Cancer Centre, and MD Anderson Cancer Center. The mouse cell lines have not been authenticated as we do not have the appropriate benchmarks against which to check these lines. Human cell lines were validated using the short-tandem repeat profiling services of Cell Bank Australia, and the Characterized Cell Line Core Facility (MD Anderson Cancer Center, Houston, TX).

Gene transductions and surrogate in vitro assays of metastasis were completed using established methods (14, 20). Details of the anoikis assay are provided in Supplementary Methods. Primers used for PCR and cloning are shown in Supplementary Table S1.

Animals

All animal procedures were conducted with approval from either the Peter MacCallum Cancer Centre, Austin Health (Melbourne, Australia), or MD Anderson Cancer Center Animal Ethics Committees. Balb/c mice were utilized for in vivo assessment of syngeneic tumor lines 4T1.2, 4T07, 66cl4, 168FARN, 67NR; while NOD-SCID-γ (NSG) mice were utilized for xenograft studies using MDA-MB-231 and SUM159. Metastatic burden was assessed by a previously described multiplexed TaqMan assay (14). For further details, see Supplementary Methods.

Analysis of online gene expression databases

We utilized three independent online gene expression databases to interrogate the expression of gene candidates with respect to clinical correlates and outcome. These are: Oncomine (21), Gene Expression-Based Outcome for Breast Cancer Online (GOBO; refs. 22, 23), and BreastMark (24). Further details are given in Supplementary Methods.

Patient samples for IHC analysis

Breast tumor samples from 535 patients treated at a single institution, Royal Perth Hospital (Perth, Western Australia), between January 1996 and December 2001 were analyzed (25). The use of patient samples and clinical information in this study was approved by the Human Research Ethics Committee of Royal Perth Hospital. For further details, see Supplementary Methods.

For all other methods, please refer to Supplementary Methods.

BMP4 loss is associated with breast tumor aggressiveness, increased metastatic potential, and reduced patient survival

Transcriptional profiling identified Bmp4 as a gene whose expression is low in murine mammary tumors with high metastatic propensity (14, 15), indicating that it may be a functional metastasis suppressor. To explore this, we first confirmed that Bmp4 transcripts in highly metastatic, weakly metastatic, or nonmetastatic primary tumor epithelial cancer cells decreased with increasing metastatic potential (Fig. 1A). In parallel, BMP4 secretion was reduced in the supernatant of lines with high metastatic propensity in vitro (Fig. 1B). Immunostaining revealed that BMP4 was localized to the epithelial cell compartment in normal mammary glands and in noninvasive orthotopic tumors, but absent from aggressive and metastatic 4T1.2 tumors (Fig. 1C; Supplementary Fig. 1A), indicating that BMP4 levels correlate inversely with metastatic propensity.

Figure 1.

BMP4 expression is reduced in invasive breast cancer and is inversely correlated with poor DMFS. A,Bmp4 transcript levels in mammary cancer cells isolated from orthotopic tumors of varying metastatic potential. B, Secreted BMP4 protein levels from mammary tumor lines representing various stages of progression. C, IHC localization of BMP4 in the normal mammary gland and in orthotopic mammary tumors. Scale bar, 80 μm. D,BMP4 mRNA transcript levels in a panel of 51 breast cancer cell lines. Basal A and basal B are considered more aggressive than luminal cancers (23). E, Secreted BMP4 from human breast cell lines and oncogenic derivatives. F,BMP4 mRNA expression is downregulated in invasive breast cancers compared with normal breast tissue. Descriptions of the datasets accessed through Oncomine and GOBO can be found in Supplementary Table S2. G,BMP4 mRNA expression in breast tumors stratified by tumor grade in a meta-analysis of 11 datasets through GOBO. H, DMFS in a meta-analysis of 1,379 breast cancer patient tumor samples, stratified on the basis of median BMP4 mRNA expression. Data in A, B, and E display the mean ± SEM (n = 3) and the statistical significance calculated by unpaired, two-tailed Student t test. Significance in G was calculated by ANOVA, and by log-rank test in H.

Figure 1.

BMP4 expression is reduced in invasive breast cancer and is inversely correlated with poor DMFS. A,Bmp4 transcript levels in mammary cancer cells isolated from orthotopic tumors of varying metastatic potential. B, Secreted BMP4 protein levels from mammary tumor lines representing various stages of progression. C, IHC localization of BMP4 in the normal mammary gland and in orthotopic mammary tumors. Scale bar, 80 μm. D,BMP4 mRNA transcript levels in a panel of 51 breast cancer cell lines. Basal A and basal B are considered more aggressive than luminal cancers (23). E, Secreted BMP4 from human breast cell lines and oncogenic derivatives. F,BMP4 mRNA expression is downregulated in invasive breast cancers compared with normal breast tissue. Descriptions of the datasets accessed through Oncomine and GOBO can be found in Supplementary Table S2. G,BMP4 mRNA expression in breast tumors stratified by tumor grade in a meta-analysis of 11 datasets through GOBO. H, DMFS in a meta-analysis of 1,379 breast cancer patient tumor samples, stratified on the basis of median BMP4 mRNA expression. Data in A, B, and E display the mean ± SEM (n = 3) and the statistical significance calculated by unpaired, two-tailed Student t test. Significance in G was calculated by ANOVA, and by log-rank test in H.

Close modal

Having established an inverse correlation between BMP4 expression and metastatic potential in mouse mammary tumor lines, we sought to validate this in human lines and tumors. Analysis of 51 breast cancer lines (22, 23) revealed that Bmp4 transcripts were reduced in highly aggressive, basal-like lines compared with the less-aggressive, luminal variants (Fig. 1D). Detection of secreted BMP4 protein confirmed these results, with immortalized human mammary epithelial cells (HME) expressing higher levels of BMP4 than those transformed with oncogenic Ras or large T antigen (Fig. 1E). Similarly, nontumorigenic MCF10A breast epithelial cells displayed elevated BMP4 expression, which was reduced in tumorigenic subclones (MCF10A-CA1h; Fig. 1E). In established breast cancer lines, BMP4 secretion was lower than in nontumorigenic HME cells, with near to undetectable levels in aggressive and metastatic lines (Fig. 1E). Thus, loss of BMP4 expression was associated with increased metastatic potential in preclinical mouse and human breast cancer lines.

We next assessed the relationship between BMP4 mRNA expression and clinical correlates using multiple integrated genomics databases (21, 24). Sixteen analyses, spanning 12 independent datasets that directly compared breast tumor samples with normal tissue were identified in Oncomine (Supplementary Table S2). In line with our preclinical studies, BMP4 was found to be consistently lower in breast tumor samples compared with normal breast tissue (Fig. 1F; Supplementary Table S2). In an independent meta-analysis incorporating eleven profiling studies (22), low BMP4 expression was associated with higher grade breast tumors (Fig. 1G), and those with a “basal-like” or ER subtype (Supplementary Fig. S1B–S1D), consistent with our in vitro observations (Fig. 1D). Furthermore, low BMP4 expression correlated with poor distant-metastasis–free survival (DMFS, Fig. 1H) regardless of ER status, as demonstrated by multivariate analysis (Supplementary Fig. S1E). Similarly, the link between low BMP4 expression and poor recurrence-free survival, DMFS, and overall survival (OS) was noted in a third independent meta-analysis (Supplementary Fig. S1F–S1H; ref. 24). Collectively, these data demonstrate that BMP4 is lost in high-grade and aggressive breast cancers, and is inversely correlated with patient survival, indicating that BMP4 may act to suppress tumor progression and metastasis.

BMP4 is a metastasis suppressor in breast cancer

To evaluate whether BMP4 could functionally regulate metastasis, we expressed BMP4 in highly metastatic 4T1.2 cells (4T1.2-BMP4) and confirmed increased expression by immunoblot analysis (Fig. 2A) and ELISA of the secreted protein (Fig. 2B). Conditioned medium from 4T1.2-BMP4 cells, but not parental 4T1.2 or mock vector-transduced (4T1.2-vector) cells, induced differentiation of mouse C2C12 myoblasts (Supplementary Fig. S2A), demonstrating that these 4T1.2-BMP4 cells secrete mature and functionally active BMP4.

Figure 2.

Restoration of Bmp4 expression in invasive breast cancer cells inhibits metastasis and prolongs survival. A, Immunoblot for BMP4 and tubulin in 4T1.2 cells transduced with Bmp4 (4T1.2-BMP4) or base-vector (4T1.2-vector) constructs. B, Secreted BMP4 detected by ELISA in cells after transduction. C, Orthotopic tumor growth of 4T1.2-vector or 4T1.2-BMP4 cells in Balb/c mice (n = 13/group). D, Plasma calcium levels at experimental endpoint as a surrogate measure of tumor-induced osteolysis, compared with nontumor-bearing mice (n = 10). E, Metastatic burden (measured by genomic qPCR) in various tissues at endpoint. ALN, axillary lymph node. Each point indicates tumor burden in an individual tissue, with the mean ± SEM displayed. F, Hematoxylin and eosin staining of representative lung and spine tissues. m, metastatic lesion. Dashed lines, lesion boundary. G, Kaplan–Meier analysis of survival following resection of size-matched 4T1.2-vector and 4T1.2-BMP4 primary tumors (n = 10 mice/group). H and I, There was no difference in the tumor volume (H) or weight (I) of resected MDA-MB-231 orthotopic xenografts expressing BMP4 or a vector control at the time of excision (n = 10/group). J, Representative fluorescence-based imaging of lungs. K, qPCR-based metastatic burden analysis of lung metastasis in mice bearing MDA-MB-231-vector or MDA-MB-231-BMP4 tumors, 24 days post-resection. Statistical significance was determined by unpaired, two-tailed Student t test (B, D, E, H, I, and K) or log-rank test (G). HR, Cox proportional hazards analysis. Scale bars: lung, 500 μm; spine, 250 μm. n.s., nonsignificant.

Figure 2.

Restoration of Bmp4 expression in invasive breast cancer cells inhibits metastasis and prolongs survival. A, Immunoblot for BMP4 and tubulin in 4T1.2 cells transduced with Bmp4 (4T1.2-BMP4) or base-vector (4T1.2-vector) constructs. B, Secreted BMP4 detected by ELISA in cells after transduction. C, Orthotopic tumor growth of 4T1.2-vector or 4T1.2-BMP4 cells in Balb/c mice (n = 13/group). D, Plasma calcium levels at experimental endpoint as a surrogate measure of tumor-induced osteolysis, compared with nontumor-bearing mice (n = 10). E, Metastatic burden (measured by genomic qPCR) in various tissues at endpoint. ALN, axillary lymph node. Each point indicates tumor burden in an individual tissue, with the mean ± SEM displayed. F, Hematoxylin and eosin staining of representative lung and spine tissues. m, metastatic lesion. Dashed lines, lesion boundary. G, Kaplan–Meier analysis of survival following resection of size-matched 4T1.2-vector and 4T1.2-BMP4 primary tumors (n = 10 mice/group). H and I, There was no difference in the tumor volume (H) or weight (I) of resected MDA-MB-231 orthotopic xenografts expressing BMP4 or a vector control at the time of excision (n = 10/group). J, Representative fluorescence-based imaging of lungs. K, qPCR-based metastatic burden analysis of lung metastasis in mice bearing MDA-MB-231-vector or MDA-MB-231-BMP4 tumors, 24 days post-resection. Statistical significance was determined by unpaired, two-tailed Student t test (B, D, E, H, I, and K) or log-rank test (G). HR, Cox proportional hazards analysis. Scale bars: lung, 500 μm; spine, 250 μm. n.s., nonsignificant.

Close modal

The proliferative capacity of 4T1.2 and other breast cancer cell lines was not altered by elevated BMP4 expression or by treatment with recombinant BMP4 (rBMP4; Supplementary Fig. S2B and S2C), while immortalized nontumorigenic mammary epithelial cells, NMuMG, did demonstrate a dose-dependent growth inhibition (Supplementary Fig. S2C). Expectedly, when 4T1.2-vector or 4T1.2-BMP4 cells were implanted into the mammary gland, no difference in the rate of orthotopic tumor growth was noted (Fig. 2C). Tumor-bearing mice were monitored over a 4-week period, and the experiment terminated when mice bearing control tumors displayed signs of respiratory distress and/or hind-limb paralysis. Tumor expression of BMP4 resulted in significantly less morbidity and analysis of peripheral blood revealed reduced hypercalcemia in mice bearing 4T1.2-BMP4 tumors, indicating that BMP4 may diminish the clinical sequelae of osteolytic bone disease (Fig. 2D). We confirmed at autopsy that BMP4 dramatically inhibited spontaneous metastasis to multiple organs including lung, lymph node, and bone (Fig. 2E and F), reducing both incidence and overall metastatic burden (Fig. 2E). In a repeat experiment, mouse survival was monitored following surgical resection of established, size-matched tumors (Supplementary Fig. S2D). As expected, the dramatic reduction in metastatic burden by BMP4 translated into a significant extension in survival, with 75% of the mice alive and disease-free 100 days after primary tumor resection (Fig. 2G). Validating these findings, we demonstrate that constitutive BMP4 expression in human triple-negative breast cancer cells (MDA-MB-231; Supplementary Fig. S2E), did not alter proliferation (Supplementary Fig. S2F), yet drastically inhibited spontaneous metastasis to lung following primary tumor resection (Fig. 2HK).

The reduction in circulating tumor cells (CTC) in peripheral blood from tumor-bearing mice (Fig. 2E) indicates that BMP4 reduces the ability of cancer cells to escape from the primary site, or that it inhibits survival during transit. To explore this further, we assessed whether BMP4 could inhibit the colonization of distant organs following the direct administration of cancer cells into venous circulation. Elevated BMP4 expression resulted in a 2.5-fold decrease in metastatic burden, consistent with the impaired capacity of cancer cells to survive in circulation and/or colonize the lung (Fig. 3A). Moreover, reduced lung metastasis equated to a significant extension in survival (Fig. 3B). Supporting these observations, BMP4 expression blocked experimental lung metastasis in the human triple-negative breast cancer line, SUM159 (Supplementary Fig. S3A), as evidenced by an improved survival rate (Fig. 3C) and reduced formation of experimental lung metastases at necropsy (Fig. 3D).

Figure 3.

BMP4 prevents distant metastatic colonization by sensitizing breast cancer cells to anoikis. A, BMP4 expression blocks experimental lung metastasis of the aggressive 4T1.2 tumor variant (n = 13 mice/group). B, Kaplan–Meier analysis of survival for mice challenged with 4T1.2-vector (n = 18) or 4T1.2-BMP4 (n = 16) experimental lung metastasis. C, Kaplan–Meier analysis of survival for mice challenged with SUM159-vector or SUM159-BMP4 experimental lung metastasis (n = 10/group). D, Hematoxylin and eosin staining of representative SUM159-vector and SUM159-BMP4 lung sections. Scale bar, 500 μm. m, metastatic lesion. E, Fluorescence-based (LI-COR) quantitation of BMP4 protein levels in weakly metastatic 4T07 tumors following stable transduction with a Bmp4-targeting shRNA (shBMP4) and growth in vivo. F, Orthotopic tumor growth of 4T07 tumors engineered to stably express shRNA-targeting BMP4 (4T07-shBMP4) or nontargeting control (4T07-vector; n = 10 mice/group). G, Kaplan–Meier analysis of survival following injection of 4T07-vector or 4T07-shBMP4 cells into the lateral tail vein of Balb/c mice (n = 10/group). H, Representative images of lungs from mice described in G and enumeration of lung metastatic nodules. I, Resistance to anoikis of tumor lines with varying metastatic potential, as measured by clonogenic survival assay. J, Resistance to anoikis following the addition of recombinant BMP4 (rBMP4) or elevated BMP4 expression in metastatic mouse and human breast cancer cells. Data display the mean ± SEM from triplicate independent assays. Statistical significance was calculated by unpaired two-tailed Student t test. Log-rank test was used to assess significance in Kaplan–Meier analyses. HR, Cox proportional hazards analyses.

Figure 3.

BMP4 prevents distant metastatic colonization by sensitizing breast cancer cells to anoikis. A, BMP4 expression blocks experimental lung metastasis of the aggressive 4T1.2 tumor variant (n = 13 mice/group). B, Kaplan–Meier analysis of survival for mice challenged with 4T1.2-vector (n = 18) or 4T1.2-BMP4 (n = 16) experimental lung metastasis. C, Kaplan–Meier analysis of survival for mice challenged with SUM159-vector or SUM159-BMP4 experimental lung metastasis (n = 10/group). D, Hematoxylin and eosin staining of representative SUM159-vector and SUM159-BMP4 lung sections. Scale bar, 500 μm. m, metastatic lesion. E, Fluorescence-based (LI-COR) quantitation of BMP4 protein levels in weakly metastatic 4T07 tumors following stable transduction with a Bmp4-targeting shRNA (shBMP4) and growth in vivo. F, Orthotopic tumor growth of 4T07 tumors engineered to stably express shRNA-targeting BMP4 (4T07-shBMP4) or nontargeting control (4T07-vector; n = 10 mice/group). G, Kaplan–Meier analysis of survival following injection of 4T07-vector or 4T07-shBMP4 cells into the lateral tail vein of Balb/c mice (n = 10/group). H, Representative images of lungs from mice described in G and enumeration of lung metastatic nodules. I, Resistance to anoikis of tumor lines with varying metastatic potential, as measured by clonogenic survival assay. J, Resistance to anoikis following the addition of recombinant BMP4 (rBMP4) or elevated BMP4 expression in metastatic mouse and human breast cancer cells. Data display the mean ± SEM from triplicate independent assays. Statistical significance was calculated by unpaired two-tailed Student t test. Log-rank test was used to assess significance in Kaplan–Meier analyses. HR, Cox proportional hazards analyses.

Close modal

In a reverse-complementary approach, we used shRNA to silence the high level of BMP4 in the weakly metastatic 4T07 line (4T07-shBMP4). Orthotopic tumors developing from this line had 70% reduction in BMP4 protein compared with nonsilenced controls (4T07-vector; Fig. 3E). Consistent with our hypothesis that BMP4 is a metastasis-regulating protein, loss of BMP4 expression did not alter growth of orthotopic tumors (Fig. 3F), but reduced the survival of mice (Fig. 3G) through enhanced metastatic colonization of the lung (Fig. 3H). Finally, to assess whether metastasis suppression is a distinct activity of BMP4, or potentially of other BMP isoforms as well, we compared the effect of elevated Bmp7 expression in 4T1.2 tumors (Supplementary Fig. S3B and S3C). No difference was observed in orthotopic tumor growth rates (Supplementary Fig. S3D), and following tumor resection, we showed that BMP7 can extend survival from metastatic disease, however not to the same extent as BMP4 (Supplementary Fig. S3E). Collectively, these results demonstrate that BMPs can modulate the metastatic potential of breast tumors without impacting primary tumor growth; and as such, we propose that BMP4 is a bona fide breast cancer metastasis suppressor.

As a cytokine with the ability to affect diverse biological processes, we sought to determine how BMP4 suppresses metastasis. With no change in primary tumor growth, we focused on metastatic processes, namely cellular migration, invasion, resistance to anoikis, and distant colonization (1). Expression of BMP4 did not alter the ability of breast cancer cells to migrate or invade through a Matrigel barrier in chemotactic assays (Supplementary Fig. S3F and S3G). Anoikic cell death in circulation is a major barrier that tumor cells need to overcome during metastatic progression (1). Indeed, resistance to anoikis correlated with metastatic capacity in our panel of mammary tumor cells (Fig. 3I) and treatment with rBMP4 or constitutive BMP4 expression (Supplementary Fig. S3A and S3H) enhanced anoikis in aggressive mouse and human breast cancer cells (Fig. 3J). These data are in agreement with the reduced level of CTCs in the blood of mice bearing BMP4-expressing tumors (Fig. 2E) and collectively indicate that BMP4 impairs metastasis by sensitizing metastatic breast cancer cells to anoikic stress induced by cell–substrate detachment and shear flow during systemic dissemination.

To detail potential mechanisms between BMP4 and anoikis, we interrogated the genes induced by BMP4 in six breast cancer lines (26), for those that regulate anoikis (Gene Ontology “Anoikis,” GO:0043276). Of 36 genes, we highlight six that are regulated by BMP4, which have diverse effects on the survival of cancer cells in anoikic conditions (Supplementary Table S3). Although not proven functionally, the induction of these genes is consistent with a role of BMP4 in anoikis.

BMP4 alters metastasis-regulating genes via the canonical BMP4–SMAD signaling pathway

Canonical BMP-SMAD signaling is initiated through cytokine-induced pairing of cognate type I and type II receptors (BMP-Rs) that induce transphosphorylation events within the receptor complex (27). Phosphorylation increases the affinity of intracellular receptor R-SMAD protein (SMADs-1/5/8) binding to SMAD4, which enables nuclear translocation and DNA binding of the transcription factor complex (SMAD–4/R-SMADs), ultimately regulating gene expression (28). Serum-starved immortalized or tumorigenic mammary cells display minimal levels of phosphorylated SMAD1/5/8 (pSMAD1/5/8) until exposed to rBMP4 or transduced with BMP4 (Fig. 4A). Conversely, constitutively activated pSMAD1/5/8 in 4T1.2-BMP4 cells is blocked by coincubation with the extracellular BMP inhibitor, Noggin (Fig. 4A; ref. 29). Exposure to rBMP4 triggers localization of pSMAD1/5/8 to the nucleus (Fig. 4B and C) and induction of the immediate BMP response gene, Id1 (Supplementary Fig. S3I), indicating that these breast cancer lines have a functionally intact canonical BMP–SMAD signaling pathway.

Figure 4.

Activated BMP4 signaling triggers a metastasis-suppressive gene signature in breast cancer cells and tumors. A, Western blot of phospho-SMAD1/5/8 (pSMAD1/5/8) in different cell lines in response to a 3-hour treatment with recombinant BMP4 (rBMP4), or to recombinant Noggin (rNoggin) in serum-starved 4T1.2-BMP4 cells. B, Immunofluorescence of pSMAD1/5/8 in serum-starved 4T1.2 cells following treatment with rBMP4 for 3 hours. C, Quantitation of the number of cells positive for nuclear-localized pSMAD1/5/8. D, Hierarchical clustering of a panel of metastasis-regulating genes in 4T1.2 cells in response to the addition of rBMP4 or in response to Bmp4 expression. Genes reported to be metastasis suppressors are marked by black boxes, while those reported to be metastasis promoters are marked by white boxes. E, Chromatin immunoprecipitation of the 5′UTR Smad7 promoter by a SMAD4-specific antibody or IgG control or by beads alone following exposure of 4T1.2 cells to rBMP4. F, Induction of SMAD7 mRNA in a panel of mouse and human breast cancer cells in response to rBMP4 or conditioned medium (CM) recovered from 4T1.2-vector or 4T1.2-BMP4 cells. G, BMP4 and SMAD7 protein levels in formalin-fixed, paraffin-embedded tumors or normal mammary gland as detected by IHC. Scale bar, 100 μm. Bar graphs represent the mean ± SEM of triplicate samples. Statistical significance calculated by unpaired two-tailed Student t test (C, E, and F). *, P < 0.05. n.s., not significant.

Figure 4.

Activated BMP4 signaling triggers a metastasis-suppressive gene signature in breast cancer cells and tumors. A, Western blot of phospho-SMAD1/5/8 (pSMAD1/5/8) in different cell lines in response to a 3-hour treatment with recombinant BMP4 (rBMP4), or to recombinant Noggin (rNoggin) in serum-starved 4T1.2-BMP4 cells. B, Immunofluorescence of pSMAD1/5/8 in serum-starved 4T1.2 cells following treatment with rBMP4 for 3 hours. C, Quantitation of the number of cells positive for nuclear-localized pSMAD1/5/8. D, Hierarchical clustering of a panel of metastasis-regulating genes in 4T1.2 cells in response to the addition of rBMP4 or in response to Bmp4 expression. Genes reported to be metastasis suppressors are marked by black boxes, while those reported to be metastasis promoters are marked by white boxes. E, Chromatin immunoprecipitation of the 5′UTR Smad7 promoter by a SMAD4-specific antibody or IgG control or by beads alone following exposure of 4T1.2 cells to rBMP4. F, Induction of SMAD7 mRNA in a panel of mouse and human breast cancer cells in response to rBMP4 or conditioned medium (CM) recovered from 4T1.2-vector or 4T1.2-BMP4 cells. G, BMP4 and SMAD7 protein levels in formalin-fixed, paraffin-embedded tumors or normal mammary gland as detected by IHC. Scale bar, 100 μm. Bar graphs represent the mean ± SEM of triplicate samples. Statistical significance calculated by unpaired two-tailed Student t test (C, E, and F). *, P < 0.05. n.s., not significant.

Close modal

Approximately thirty metastasis-regulating genes that are distinct from known oncogenes have been shown to regulate discrete steps essential for metastatic progression (1, 30). We hypothesized that BMP4 may induce canonical SMAD signaling to alter the expression of metastasis-regulating genes in breast cancer cells. The effect of BMP4 stimulation on the expression of 18 metastasis-suppressing genes and 13 metastasis-promoting genes was assessed in 4T1.2 cells. Unsupervised hierarchical clustering of the expression data revealed concordance between replicate time points and identified two distinct gene clusters whose expression was either elevated (cluster-A) or reduced (cluster-B) by BMP4 (Fig. 4D). Cluster-A contains a significantly higher proportion of metastasis-suppressing genes (12/18 genes) compared with cluster-B (5/13 genes; P < 0.05, two proportions test) and includes the well-established metastasis suppressors Brms1 (31), Casp8 (32), Nme23 (33), Irf7 (15), Arhgdib (34), and Cav1 (35). Conversely, genes downregulated in cluster-B contain a majority of known metastasis-promoting genes, including Spp1 (36), Npnt (37, 38), and Cxcr4 (39).

Nuclear accumulation of pSMAD1/5/8 in BMP4-stimulated 4T1.2 cells (Fig. 4B and C) indicates a potential interaction between the BMP–SMAD transcriptional complex and regulation of metastasis-associated gene promoters. Of specific interest was Smad7, a metastasis suppressor that contains six BMP-responsive SMAD-binding elements (BR-SBEs; GGAGCC) within a 2.5kb 5′-UTR, and displays elevated expression following exposure to BMP4 (Fig. 4D). To confirm that canonical SMAD signaling was involved in the activation of Smad7 expression, we demonstrated by chromatin immunoprecipitation that stimulation with rBMP4 increased SMAD4 loading onto the 5′-UTR of Smad7, 660bps upstream of the Smad7 coding sequence start site (Fig. 4E), and not at sites distal to the Smad7 promoter. To confirm the robustness of gene activation, we tested a panel of breast cancer lines for their ability to induce Smad7 following BMP4 exposure (Fig. 4F). Treatment with either rBMP4 or conditioned medium from 4T1.2-BMP4 cells significantly elevated Smad7 expression, but not in the SMAD4-null MDA-MB-468 cells or in MCF7 cells with low SMAD4 levels (Fig. 4F; Supplementary Fig. S3J), thus demonstrating the requirement for canonical BMP4-SMAD signaling in the control of Smad7 gene expression.

SMAD7 is required for BMP4-mediated suppression of breast cancer metastasis

To investigate the relationship between BMP4 and SMAD7, protein levels in mammary tumors were assessed by IHC. Similar to our observations with BMP4, high SMAD7 protein was detected in epithelial cells of the mammary gland and in nonmetastatic and low metastatic tumors (Fig. 4G). In highly metastatic tumors, SMAD7 protein was low but could be restored when BMP4 was constitutively expressed. This indicates that BMP4 can activate canonical BMP-SMAD signaling in vivo, leading to the upregulation of the metastasis suppressor, SMAD7.

With a clear link establishing the induction of SMAD7 by BMP4 in aggressive breast cancer cells and tumors, we next evaluated a role for SMAD7 during metastasis. Transient transfection of SMAD7 siRNA into 4T1.2 cells resulted in a reversal of the induction of anoikis by BMP4 (Fig. 5A), indicating a functional requirement of SMAD7 to modulate metastasis suppression by BMP4. As expected, high expression of Flag-SMAD7 in 4T1.2 cells did not alter primary tumor growth but resulted in prolonged survival of mice (Fig. 5BD; Supplementary Fig. S4A). In contrast, stable knockdown of SMAD7 in two 4T1.2-BMP4 cell lines with unique shRNAs (Supplementary Fig. S4B and S4C) resulted in acceleration of the onset of metastatic disease that was confirmed at autopsy and resulted in poorer survival (Fig. 5E) following resection of size-matched tumors (Supplementary Fig. S4D). These results establish a key functional role for SMAD7 in BMP4-mediated suppression of breast cancer metastasis.

Figure 5.

BMP4 promotes anoikis and inhibits metastasis, in part, through the induction of Smad7. A, Anoikis assay in 4T1.2 cells following treatment with rBMP4 in combination with Smad7 silencing. B, Western blot detecting Flag-tagged SMAD7 in 4T1.2 cells transduced with a Flag-SMAD7 construct or vector control. C, Growth of orthotopic 4T1.2-vector or 4T1.2-Flag-SMAD7 tumors in Balb/c mice (n = 14/group). D, Kaplan–Meier analysis of survival in mice bearing size-matched, resected tumors from C. E, Kaplan–Meier analysis of survival in mice bearing size-matched, resected tumors (n = 9 mice/group) with stable knockdown of Smad7. F, Growth of orthotopic 4T1.2 tumors following rBMP4 therapy or vehicle control (n = 9/group). G,Smad7 mRNA levels in primary tumors following therapy as determined by RT-PCR. H and I, Metastatic burden in the lung (H) and bone (I) of mice receiving BMP4 therapy (from F). Bar charts and dot plots indicate the mean ± SEM. Statistical significance was calculated by unpaired two-tailed Student t test (A and G–I) or by log-rank test (D and E). *, significance relative to no treatment controls. HR, Cox proportional hazards analyses. n.s., nonsignificant.

Figure 5.

BMP4 promotes anoikis and inhibits metastasis, in part, through the induction of Smad7. A, Anoikis assay in 4T1.2 cells following treatment with rBMP4 in combination with Smad7 silencing. B, Western blot detecting Flag-tagged SMAD7 in 4T1.2 cells transduced with a Flag-SMAD7 construct or vector control. C, Growth of orthotopic 4T1.2-vector or 4T1.2-Flag-SMAD7 tumors in Balb/c mice (n = 14/group). D, Kaplan–Meier analysis of survival in mice bearing size-matched, resected tumors from C. E, Kaplan–Meier analysis of survival in mice bearing size-matched, resected tumors (n = 9 mice/group) with stable knockdown of Smad7. F, Growth of orthotopic 4T1.2 tumors following rBMP4 therapy or vehicle control (n = 9/group). G,Smad7 mRNA levels in primary tumors following therapy as determined by RT-PCR. H and I, Metastatic burden in the lung (H) and bone (I) of mice receiving BMP4 therapy (from F). Bar charts and dot plots indicate the mean ± SEM. Statistical significance was calculated by unpaired two-tailed Student t test (A and G–I) or by log-rank test (D and E). *, significance relative to no treatment controls. HR, Cox proportional hazards analyses. n.s., nonsignificant.

Close modal

Therapeutic administration of rBMP4 blocks metastasis through regulation of SMAD7

Recombinant BMP proteins are bioactive and have been utilized as FDA-approved, therapeutic agents for bone fracture repair (40). More recently, systemic administration of rBMP4 was reported to reverse β-cell dysfunction in animal models of type II diabetes (41). As we have demonstrated that aggressive breast tumors have an intact and actionable canonical BMP–SMAD signaling pathway, we hypothesized that systemic administration of rBMP4 could be efficacious for antimetastasis therapy.

To test this, Balb/c mice bearing orthotopic 4T1.2 tumors received either rBMP4 therapy or vehicle control, and the response of tumor growth and spontaneous metastasis was measured. As anticipated from our previous findings (Fig. 2C; Supplementary Fig. S2B and S2D), rBMP4 therapy did not alter primary tumor growth (Fig. 5F) nor final tumor weight (Supplementary Fig. S4E). Recombinant BMP4 therapy induced Smad7 expression in primary tumors, indicating that canonical BMP-SMAD signaling was active following systemic treatment (Fig. 5G). Evaluation of distant organs revealed that rBMP4 significantly impaired spontaneous metastasis to lung and bone (Fig. 5H and I). Thus, we have demonstrated that systemically administered rBMP4 modulates effector gene targets within primary tumors and is an effective antimetastatic therapy in preclinical models.

BMP4 and SMAD7 expression are inversely correlated with disease-free and OS in patients with breast cancer

Having established the functional requirement for an active canonical BMP4–SMAD7 signaling pathway to block metastasis, we next evaluated the clinical association between BMP4 and SMAD7 protein levels by IHC in a tissue microarray constructed using primary surgical specimens from 535 treatment-naïve patients diagnosed with carcinoma in situ and invasive breast cancer (Supplementary Table S4; ref. 25). Consistent with our mRNA expression analyses (Fig. 1FH), high BMP4 protein levels were less frequent in ductal carcinoma in situ (DCIS; 48.5%, P = 0.00087) and invasive carcinoma (48.0%, P < 0.00001) compared with benign breast epithelium (64.3%; Fig. 6A; Supplementary Table S4). BMP4-negative expression status was more common in ductal malignancies (57.7%) compared with tumors with lobular (35.9%, P = 0.0015) or the less aggressive tubular histology (32.0%, P = 0.0129). No significant associations were seen between BMP4 status and grade, receptor status or lymph node involvement. SMAD7 positivity was also less frequent in DCIS (55.7%, P = 0.0012) and invasive disease (41.3%, P < 0.00001) relative to normal breast epithelium (70.6%). SMAD7-negative status was similarly more common in ductal tumors (66.1%) compared with lobular (53.0%, P = 0.044) or tubular cancer (34.6%, P = 0.0013). Compared with breast tumors that were SMAD7 positive, SMAD7-negative tumors were associated with higher grade (P = 0.0005), ER-negativity (P = 0.0021) and PR-negativity (P = 0.0196).

Figure 6.

BMP4 levels are lower in aggressive breast cancer and in conjunction with SMAD7, correlate with recurrence-free and OS in patients with breast cancer. A, Representative immunostaining of BMP4 and SMAD7 in human breast tissues containing benign epithelium, DCIS, or invasive breast carcinoma. Scale bar, 50 μm. IDBC, invasive ductal breast cancer; ILBC, invasive lobular breast cancer. B, Kaplan–Meier analysis depicting the probability of recurrence-free survival as stratified by BMP4. C and D, Kaplan–Meier analysis depicting the probability of RFS (C) and OS (D) as stratified by BMP4 and SMAD7. E, BMP4 induces a “stop metastasis” gene signature that incorporates the modulation of canonical (blue) and noncanonical (gray) pathways. Blue shaded area, data arising from this study. Red lines, inhibitory regulation. Green, metastasis suppressor genes; red, metastasis enhancer genes. Statistical significance was calculated by log-rank test (B–D). HR, Cox proportional hazards analyses.

Figure 6.

BMP4 levels are lower in aggressive breast cancer and in conjunction with SMAD7, correlate with recurrence-free and OS in patients with breast cancer. A, Representative immunostaining of BMP4 and SMAD7 in human breast tissues containing benign epithelium, DCIS, or invasive breast carcinoma. Scale bar, 50 μm. IDBC, invasive ductal breast cancer; ILBC, invasive lobular breast cancer. B, Kaplan–Meier analysis depicting the probability of recurrence-free survival as stratified by BMP4. C and D, Kaplan–Meier analysis depicting the probability of RFS (C) and OS (D) as stratified by BMP4 and SMAD7. E, BMP4 induces a “stop metastasis” gene signature that incorporates the modulation of canonical (blue) and noncanonical (gray) pathways. Blue shaded area, data arising from this study. Red lines, inhibitory regulation. Green, metastasis suppressor genes; red, metastasis enhancer genes. Statistical significance was calculated by log-rank test (B–D). HR, Cox proportional hazards analyses.

Close modal

Elevated BMP4 expression was associated with significantly improved RFS (HR = 0.70; P = 0.042; Fig. 6B) and approached significance for OS (HR = 0.69; P = 0.051; Supplementary Fig. S4F). Similarly, elevated SMAD7 was associated with significantly improved RFS (HR = 0.65; P = 0.017; Supplementary Fig. S4G) and OS (HR = 0.61; P = 0.012; Supplementary Fig. S4H). In multivariate analysis after adjusting for age, grade, tumor size, and ER status, BMP4 remained prognostic for RFS (HR = 0.68; P = 0.037), whereas SMAD7 became borderline prognostic (HR = 0.69; P = 0.059). The latter change is likely due to the association of SMAD7 with lower grade and ER+ tumors, accepting that the biological role of SMAD7 may contribute to the less aggressive nature of the tumors that express it.

As BMP4 requires SMAD7 to block metastasis in preclinical models, we explored whether expression of both proteins would provide enhanced predictive power in clinical samples. Breast tumors that were positive for both proteins provided a superior RFS (HR = 0.58; P = 0.005; Fig. 6C) and OS (HR = 0.55; P = 0.005; Fig. 6D) compared with tumors with neither protein present. Collectively, the results of our clinical study indicate that activation of canonical BMP4-SMAD7 signaling is associated with the prevention of breast cancer progression, relapse, and the extension of overall patient survival.

In this translational study, we have established a novel BMP4–SMAD7 signaling axis that controls metastatic progression and demonstrated that stimulation of this pathway is an effective strategy against metastatic breast cancer. Using murine and human cell lines, preclinical models, and clinical samples, we show that loss of BMP4 expression is associated with increased tumor aggressiveness and poor recurrence-free survival and OS. Restoration of BMP4 in preclinical models of malignant progression sensitizes breast tumor cells to anoikis and profoundly inhibits spontaneous metastasis in vivo. The antimetastatic capacity of BMP4 requires the activation of canonical BMP4-SMAD signaling and is dependent on the induction of Smad7. Moreover, we establish that administration of recombinant BMP4 limits metastatic progression in tumor-bearing mice, indicating the potential of targeting BMP4-SMAD signaling as a novel therapeutic strategy for metastatic breast cancer.

At a molecular level, we find that BMP4 can alter the expression profile of metastasis-regulating genes to favor metastatic suppression. Metastasis-regulating genes are distinct from oncogenes or tumor suppressors, being defined as genes that affect the propensity of a tumor to metastasize, irrespective of their ability to impact the growth of a primary tumor (42). These genes are critical in the development of late-stage disease, are detectable at altered levels within primary tumors, and can predict patient outcome (43). Metastasis-regulating genes have diverse molecular properties, including involvement with the extracellular matrix, cell adhesion, chemokine receptor activity, signal transduction, and regulation of transcription (1, 30, 42). We show here that an established metastasis suppressor, SMAD7 (8), is activated by canonical BMP4-SMAD signaling and its induction is required to elicit the antimetastatic effect of BMP4, as well as the survival-promoting impact in patients with breast cancer. Consistent with a critical role for canonical BMP signaling in metastasis, MMTV-PyMT mice that were generated on a BMP-RII–impaired background had an increased incidence of lung metastases (11). Similarly, conditional loss of BMP-specific transcription factors, Smad1 and Smad5, in gonadal tissues yielded spontaneously metastatic tumors within the ovary or testis (12). These findings demonstrate that canonical BMP-SMAD signaling is a major pathway controlling metastatic progression and show that activation of this pathway, for example, through BMP4 stimulation, can inhibit metastasis by induction of a “stop metastasis” gene signature in breast cancer cells. A proposed model for the signaling pathway for inhibition of metastasis is depicted in Fig. 6E.

In a previous study, we sought a role for epithelial-to-mesenchymal transition (EMT) in the metastasis-suppressing activity of BMP4 in 4T1.2 cells, but could not find any convincing evidence to support EMT or the reverse mesenchymal-to-epithelial transition (MET; ref. 16). To investigate further, we obtained an EMT signature (44) and used the top five epithelial and top five mesenchymal genes to interrogate the BMP4-regulated genes (26) after exposure of six human breast cancer lines to BMP4. Of the ten genes examined, none displayed a consistent response to BMP4 in a majority of the cell lines. As such, we are confident that the suppression of breast cancer metastasis by BMP4 is not through modulation of EMT/MET.

Genetically engineered mouse models have demonstrated a requirement for canonical BMP signaling in control of tumor growth and metastasis; however, there are conflicting reports on an oncogenic or tumor-suppressive outcome (11, 12, 45). A review of the role of BMP4 in regulating breast cancer metastasis (46) revealed two preclinical studies demonstrating that BMP signaling can suppress metastasis (11). A third study suggested that rBMP4 treatment can inhibit the tropism of MDA-MB-231 cells toward the adrenal glands in an in vivo experimental metastasis model, but the authors also noted a nonsignificant trend toward increased bone metastasis (47). Our results significantly extend on these previous studies, demonstrating that constitutive BMP4 expression or rBMP4 therapy, can inhibit spontaneous metastasis to many organs, in multiple models of metastatic breast cancer, including bone in one model, by preventing cancer cell escape from the primary tumor and reducing survival in circulation.

In terms of overall patient prognosis, high BMP4 was associated with a nonsignificant trend toward increased recurrence in one analysis (19), while a BMP4-induced gene signature was associated with improved outcome in three other datasets (48). Collectively, our translational study provides a strong rationale for further investigation into the utility of BMP4 to suppress metastasis in breast cancer.

TGFβ signaling blocks tumor initiation and growth, yet can enhance the establishment and growth of metastatic disease (3). As BMP4 utilizes an overlapping, yet distinct signaling mechanism, it is plausible that BMP4 signaling may act in direct opposition to TGFβ signaling, by promoting the early events of tumor initiation but suppressing later progression and metastasis. Because SMAD4 is required for canonical signaling, but is mutated or lost in some cancers, especially those of the gastrointestinal tract (49), the discordant results reported so far for the actions of BMP4 in cancer progression may relate to the SMAD4 status of the tumor cells. SMAD4 loss is rare in breast cancer (50), and it is functionally required to transduce pro- and antimetastatic signals, depending on how this signaling pathway is activated (6). Indeed, it is important to note that many studies to date have assessed the requirement of BMP-receptors (BMP-R), or canonical BMP-SMAD signaling elements, in a ligand-independent context. However, with twenty known cytokines and more than ten antagonists of the BMP-R that can affect BMP-SMAD signal diversity and intensity (51), it is critical to evaluate this pathway with respect to specific cytokines and their controlling elements. Indeed, the level of bioactive BMP4 within the local tumor microenvironment can be regulated by antagonistic proteins such as Gremlin, Coco, and Noggin that either sequester BMP4 or prevent its ligation to the BMP-R in the extracellular matrix, thus suppressing BMP4 signaling and enhancing metastasis in preclinical models (13, 52, 53).

In addition to the cancer cell–intrinsic effects demonstrated here, BMP4 also acts in a paracrine manner to regulate tumor progression via actions on stromal cell populations (16). In a previous study, we showed that BMP4 can regulate tumor progression and metastasis through interactions with the local and distant microenvironments. We found that BMP4 inhibits NF-κB–induced GCSF expression, leading to reduced mobilization and recruitment of tumor-associated neutrophils (Fig. 6E, gray shaded area). Also through inhibition of NF-κB, BMP4 can reduce the growth and maturation of adipocytes and blood vessels, leading to inhibition of inflammatory mediators (54). Supporting this, an increase in adipose tissue mass has been reported in transgenic mice that have a mammary gland–specific ablation of Bmp4 (55). As lower BMP4 expression is associated with higher body mass index (BMI) in humans (55), and BMI is associated with increased risk of breast cancer–specific mortality (56), it is plausible that BMP4 may also modulate tumor progression through interactions with local adipose tissue.

The novel finding that BMP4 can impart an antimetastatic phenotype in breast cancer, utilizing a similar molecular machinery to that of TGFβ, opens new avenues for basic and translational research. Our findings indicate opportunities to leverage BMP4 and regulated downstream targets as prognostic factors to stratify patients at risk of progressive disease, and also as novel therapeutic approaches to reduce metastatic burden in patients. A therapy, using either a stabilized form of BMP4 itself or a small-molecule agonist or agonistic antibody targeting the BMP-R, could provide an effective therapy for patients at risk of progression. An alternate approach would be a means of activating the expression or activity of SMAD7, that we have shown to be integral to the action of BMP4 in suppressing metastasis.

In summary, this study highlights the importance of activated canonical BMP4-SMAD7 signaling in the inhibition of breast cancer metastasis. Further understanding of the fundamental features of this pathway will lead to the development of new and more effective therapeutic strategies for metastatic breast cancer.

No potential conflicts of interest were disclosed.

Conception and design: B.L. Eckhardt, Y. Cao, A.D. Redfern, B.S. Parker, R.L. Anderson

Development of methodology: B.L. Eckhardt, Y. Cao, A.D. Redfern, A.D. Burrows, R.L. Anderson

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): B.L. Eckhardt, Y. Cao, A.D. Redfern, L.H. Chi, A.D. Burrows, S. Roslan, E.K. Sloan, B.S. Parker, S. Loi, P.K.H. Lau, B. Latham

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): B.L. Eckhardt, Y. Cao, A.D. Redfern, A.D. Burrows, S. Loi, N.T. Ueno, P.K.H. Lau, B. Latham, R.L. Anderson

Writing, review, and/or revision of the manuscript: B.L. Eckhardt, Y. Cao, A.D. Redfern, E.K. Sloan, S. Loi, N.T. Ueno, B. Latham, R.L. Anderson

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B.L. Eckhardt, Y. Cao, A.D. Redfern, L.H. Chi, S. Roslan

Study supervision: B.L. Eckhardt, B.S. Parker, R.L. Anderson

We are grateful to Drs. F. Miller, M. Waltham, N. Nakayama, R. Pearson, and P. Humbert for providing cell lines; the Animal Facility and Histology Core (Dr. Sarah Ellis) at Peter MacCallum Cancer Centre (Melbourne, Australia) for technical assistance. We thank Dr. B Haibe-Kains for assistance with R codes, Dr. T. Kogawa for instruction with Rcmdr, Zoe Lai for her aid in animal experiments, and Dr. S. Madden for his analysis of the BreastMark database. We also acknowledge the members of the Anderson and Ueno laboratories for insightful discussion and for critical reading of the manuscript. This project was supported by a National Health and Medical Research Council of Australia (NHMRC) Project Grants (APP400037 and APP1121199 to R.L. Anderson), a Peter MacCallum Cancer Centre Foundation Grant (R.L. Anderson), a US Army Department of Defense Concept Award (BC045396 to R.L. Anderson), Cancer Council Victoria Grant-in-Aid (APP1006425 to R.L. Anderson), a Rare and Aggressive Cancer Research Appropriations State of Texas Grant (N.T. Ueno), an MD Anderson Cancer Centre Seed Funding Grant provided through the Morgan Welch Inflammatory Breast Cancer Research Program (to B.L. Eckhardt), a Susan G. Komen for the Cure Fellowship PDF82506 (to B.L. Eckhardt), and a GlaxoSmithKline post doctorate support grant (to B.L. Eckhardt). E.K. Sloan was supported by an Early Career Fellowship from the National Breast Cancer Foundation of Australia (NBCF), NHMRC 1008865, & NCI CA160890. P.K.H. Lau was supported by the Health Department of Western Australia Cancer Research Fellowship. R.L. Anderson was supported by a senior fellowship from NBCF. Olivia Newton-John Cancer Research Institute (Heidelberg, Australia) acknowledges the support of the Operational Infrastructure Program of Victorian Government.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Eckhardt
BL
,
Francis
PA
,
Parker
BS
,
Anderson
RL
. 
Strategies for the discovery and development of therapies for metastatic breast cancer
.
Nat Rev Drug Discov
2012
;
11
:
479
97
.
2.
Nguyen
DX
,
Massague
J
. 
Genetic determinants of cancer metastasis
.
Nat Rev Genet
2007
;
8
:
341
52
.
3.
Padua
D
,
Massague
J
. 
Roles of TGFbeta in metastasis
.
Cell Res
2009
;
19
:
89
102
.
4.
Katsuno
Y
,
Hanyu
A
,
Kanda
H
,
Ishikawa
Y
,
Akiyama
F
,
Iwase
T
, et al
Bone morphogenetic protein signaling enhances invasion and bone metastasis of breast cancer cells through Smad pathway
.
Oncogene
2008
;
27
:
6322
33
.
5.
Petersen
M
,
Pardali
E
,
van der Horst
G
,
Cheung
H
,
van den Hoogen
C
,
van der Pluijm
G
, et al
Smad2 and Smad3 have opposing roles in breast cancer bone metastasis by differentially affecting tumor angiogenesis
.
Oncogene
2010
;
29
:
1351
61
.
6.
Kang
Y
,
He
W
,
Tulley
S
,
Gupta
GP
,
Serganova
I
,
Chen
CR
, et al
Breast cancer bone metastasis mediated by the Smad tumor suppressor pathway
.
Proc Natl Acad Sci U S A
2005
;
102
:
13909
14
.
7.
Souchelnytskyi
S
,
Nakayama
T
,
Nakao
A
,
Moren
A
,
Heldin
CH
,
Christian
JL
, et al
Physical and functional interaction of murine and Xenopus Smad7 with bone morphogenetic protein receptors and transforming growth factor-beta receptors
.
J Biol Chem
1998
;
273
:
25364
70
.
8.
Azuma
H
,
Ehata
S
,
Miyazaki
H
,
Watabe
T
,
Maruyama
O
,
Imamura
T
, et al
Effect of Smad7 expression on metastasis of mouse mammary carcinoma JygMC(A) cells
.
J Natl Cancer Inst
2005
;
97
:
1734
46
.
9.
Shepherd
TG
,
Mujoomdar
ML
,
Nachtigal
MW
. 
Constitutive activation of BMP signalling abrogates experimental metastasis of OVCA429 cells via reduced cell adhesion
.
J Ovarian Res
2010
;
3
:
5
.
10.
Buijs
JT
,
Rentsch
CA
,
van der Horst
G
,
van Overveld
PG
,
Wetterwald
A
,
Schwaninger
R
, et al
BMP7, a putative regulator of epithelial homeostasis in the human prostate, is a potent inhibitor of prostate cancer bone metastasis in vivo
.
Am J Pathol
2007
;
171
:
1047
57
.
11.
Owens
P
,
Pickup
MW
,
Novitskiy
SV
,
Chytil
A
,
Gorska
AE
,
Aakre
ME
, et al
Disruption of bone morphogenetic protein receptor 2 (BMPR2) in mammary tumors promotes metastases through cell autonomous and paracrine mediators
.
Proc Natl Acad Sci U S A
2012
;
109
:
2814
9
.
12.
Pangas
SA
,
Li
X
,
Umans
L
,
Zwijsen
A
,
Huylebroeck
D
,
Gutierrez
C
, et al
Conditional deletion of Smad1 and Smad5 in somatic cells of male and female gonads leads to metastatic tumor development in mice
.
Mol Cell Biol
2008
;
28
:
248
57
.
13.
Gao
H
,
Chakraborty
G
,
Lee-Lim
AP
,
Mo
Q
,
Decker
M
,
Vonica
A
, et al
The BMP inhibitor Coco reactivates breast cancer cells at lung metastatic sites
.
Cell
2012
;
150
:
764
79
.
14.
Eckhardt
BL
,
Parker
BS
,
van Laar
RK
,
Restall
CM
,
Natoli
AL
,
Tavaria
MD
, et al
Genomic analysis of a spontaneous model of breast cancer metastasis to bone reveals a role for the extracellular matrix
.
Mol Cancer Res
2005
;
3
:
1
13
.
15.
Bidwell
BN
,
Slaney
CY
,
Withana
NP
,
Forster
S
,
Cao
Y
,
Loi
S
, et al
Silencing of Irf7 pathways in breast cancer cells promotes bone metastasis through immune escape
.
Nat Med
2012
;
18
:
1224
31
.
16.
Cao
Y
,
Slaney
CY
,
Bidwell
BN
,
Parker
BS
,
Johnstone
CN
,
Rautela
J
, et al
BMP4 inhibits breast cancer metastasis by blocking myeloid-derived suppressor cell activity
.
Cancer Res
2014
;
74
:
5091
102
.
17.
Kallioniemi
A
. 
Bone morphogenetic protein 4-a fascinating regulator of cancer cell behavior
.
Cancer Genet
2012
;
205
:
267
77
.
18.
Mundy
GR
. 
Metastasis to bone: causes, consequences and therapeutic opportunities
.
Nat Rev Cancer
2002
;
2
:
584
93
.
19.
Alarmo
EL
,
Huhtala
H
,
Korhonen
T
,
Pylkkanen
L
,
Holli
K
,
Kuukasjarvi
T
, et al
Bone morphogenetic protein 4 expression in multiple normal and tumor tissues reveals its importance beyond development
.
Mod Pathol
2013
;
26
:
10
21
.
20.
Torres-Adorno
AM
,
Vitrac
H
,
Qi
Y
,
Tan
L
,
Levental
KR
,
Fan
YY
, et al
Eicosapentaenoic acid in combination with EPHA2 inhibition shows efficacy in preclinical models of triple-negative breast cancer by disrupting cellular cholesterol efflux
.
Oncogene
2019
;
38
:
2135
50
.
21.
Rhodes
DR
,
Kalyana-Sundaram
S
,
Mahavisno
V
,
Varambally
R
,
Yu
J
,
Briggs
BB
, et al
Oncomine 3.0: genes, pathways, and networks in a collection of 18,000 cancer gene expression profiles
.
Neoplasia
2007
;
9
:
166
80
.
22.
Ringner
M
,
Fredlund
E
,
Hakkinen
J
,
Borg
A
,
Staaf
J
. 
GOBO: gene expression-based outcome for breast cancer online
.
PLoS One
2011
;
6
:
e17911
.
23.
Neve
RM
,
Chin
K
,
Fridlyand
J
,
Yeh
J
,
Baehner
FL
,
Fevr
T
, et al
A collection of breast cancer cell lines for the study of functionally distinct cancer subtypes
.
Cancer Cell
2006
;
10
:
515
27
.
24.
Madden
SF
,
Clarke
C
,
Gaule
P
,
Aherne
ST
,
O'Donovan
N
,
Clynes
M
, et al
BreastMark: an integrated approach to mining publicly available transcriptomic datasets relating to breast cancer outcome
.
Breast Cancer Res
2013
;
15
:
R52
.
25.
Nguyen Huu
NS
,
Ryder
WD
,
Zeps
N
,
Flasza
M
,
Chiu
M
,
Hanby
AM
, et al
Tumour-promoting activity of altered WWP1 expression in breast cancer and its utility as a prognostic indicator
.
J Pathol
2008
;
216
:
93
102
.
26.
Rodriguez-Martinez
A
,
Alarmo
EL
,
Saarinen
L
,
Ketolainen
J
,
Nousiainen
K
,
Hautaniemi
S
, et al
Analysis of BMP4 and BMP7 signaling in breast cancer cells unveils time-dependent transcription patterns and highlights a common synexpression group of genes
.
BMC Med Genomics
2011
;
4
:
80
.
27.
Chen
D
,
Zhao
M
,
Mundy
GR
. 
Bone morphogenetic proteins
.
Growth Factors
2004
;
22
:
233
41
.
28.
Miyazono
K
,
Kusanagi
K
,
Inoue
H
. 
Divergence and convergence of TGF-beta/BMP signaling
.
J Cell Physiol
2001
;
187
:
265
76
.
29.
Hannallah
D
,
Peng
H
,
Young
B
,
Usas
A
,
Gearhart
B
,
Huard
J
. 
Retroviral delivery of Noggin inhibits the formation of heterotopic ossification induced by BMP-4, demineralized bone matrix, and trauma in an animal model
.
J Bone Joint Surg Am
2004
;
86
:
80
91
.
30.
Nguyen
DX
,
Bos
PD
,
Massague
J
. 
Metastasis: from dissemination to organ-specific colonization
.
Nat Rev Cancer
2009
;
9
:
274
84
.
31.
Seraj
MJ
,
Samant
RS
,
Verderame
MF
,
Welch
DR
. 
Functional evidence for a novel human breast carcinoma metastasis suppressor, BRMS1, encoded at chromosome 11q13
.
Cancer Res
2000
;
60
:
2764
9
.
32.
Stupack
DG
,
Teitz
T
,
Potter
MD
,
Mikolon
D
,
Houghton
PJ
,
Kidd
VJ
, et al
Potentiation of neuroblastoma metastasis by loss of caspase-8
.
Nature
2006
;
439
:
95
9
.
33.
Leone
A
,
Flatow
U
,
King
CR
,
Sandeen
MA
,
Margulies
IM
,
Liotta
LA
, et al
Reduced tumor incidence, metastatic potential, and cytokine responsiveness of nm23-transfected melanoma cells
.
Cell
1991
;
65
:
25
35
.
34.
Gildea
JJ
,
Seraj
MJ
,
Oxford
G
,
Harding
MA
,
Hampton
GM
,
Moskaluk
CA
, et al
RhoGDI2 is an invasion and metastasis suppressor gene in human cancer
.
Cancer Res
2002
;
62
:
6418
23
.
35.
Sloan
EK
,
Stanley
KL
,
Anderson
RL
. 
Caveolin-1 inhibits breast cancer growth and metastasis
.
Oncogene
2004
;
23
:
7893
7
.
36.
Kang
Y
,
Siegel
PM
,
Shu
W
,
Drobnjak
M
,
Kakonen
SM
,
Cordon-Cardo
C
, et al
A multigenic program mediating breast cancer metastasis to bone
.
Cancer Cell
2003
;
3
:
537
49
.
37.
Steigedal
TS
,
Toraskar
J
,
Redvers
RP
,
Valla
M
,
Magnussen
SN
,
Bofin
AM
, et al
Nephronectin is correlated with poor prognosis in breast cancer and promotes metastasis via its integrin-binding motifs
.
Neoplasia
2018
;
20
:
387
400
.
38.
Eckhardt
BL
,
Pouliot
N
,
Anderson
RL
. 
Influence of the bone microenvironment on breast cancer metastasis to bone
.
In:
Meadows
GG
,
editor
.
Integration/interaction of oncologic growth cancer growth and progression
.
The Netherlands
:
Kluwer Press
; 
2005
.
39.
Muller
A
,
Homey
B
,
Soto
H
,
Ge
N
,
Catron
D
,
Buchanan
ME
, et al
Involvement of chemokine receptors in breast cancer metastasis
.
Nature
2001
;
410
:
50
6
.
40.
Friedlaender
GE
,
Perry
CR
,
Cole
JD
,
Cook
SD
,
Cierny
G
,
Muschler
GF
, et al
Osteogenic protein-1 (bone morphogenetic protein-7) in the treatment of tibial nonunions
.
J Bone Joint Surg Am.
2001
;
83-A
:
S151
8
.
41.
Goulley
J
,
Dahl
U
,
Baeza
N
,
Mishina
Y
,
Edlund
H
. 
BMP4-BMPR1A signaling in beta cells is required for and augments glucose-stimulated insulin secretion
.
Cell Metab
2007
;
5
:
207
19
.
42.
Smith
SC
,
Theodorescu
D
. 
Learning therapeutic lessons from metastasis suppressor proteins
.
Nat Rev Cancer
2009
;
9
:
253
64
.
43.
Hennessy
C
,
Henry
JA
,
May
FE
,
Westley
BR
,
Angus
B
,
Lennard
TW
. 
Expression of the antimetastatic gene nm23 in human breast cancer: an association with good prognosis
.
J Natl Cancer Inst
1991
;
83
:
281
5
.
44.
Tan
TZ
,
Miow
QH
,
Miki
Y
,
Noda
T
,
Mori
S
,
Huang
RY
, et al
Epithelial-mesenchymal transition spectrum quantification and its efficacy in deciphering survival and drug responses of cancer patients
.
EMBO Mol Med
2014
;
6
:
1279
93
.
45.
Owens
P
,
Pickup
MW
,
Novitskiy
SV
,
Giltnane
JM
,
Gorska
AE
,
Hopkins
CR
, et al
Inhibition of BMP signaling suppresses metastasis in mammary cancer
.
Oncogene
2015
;
34
:
2437
49
.
46.
Chi
LH
,
Burrows
AD
,
Anderson
RL
. 
Bone morphogenetic protein signaling in breast cancer progression
.
Growth Factors
2019
;
37
:
12
28
.
47.
Ampuja
M
,
Alarmo
EL
,
Owens
P
,
Havunen
R
,
Gorska
AE
,
Moses
HL
, et al
The impact of bone morphogenetic protein 4 (BMP4) on breast cancer metastasis in a mouse xenograft model
.
Cancer Lett
2016
;
375
:
238
44
.
48.
Shee
K
,
Jiang
A
,
Varn
FS
,
Liu
S
,
Traphagen
NA
,
Owens
P
, et al
Cytokine sensitivity screening highlights BMP4 pathway signaling as a therapeutic opportunity in ER(+) breast cancer
.
FASEB J
2019
;
33
:
1644
57
.
49.
Korkut
A
,
Zaidi
S
,
Kanchi
RS
,
Rao
S
,
Gough
NR
,
Schultz
A
, et al
A pan-cancer analysis reveals high-frequency genetic alterations in mediators of signaling by the TGF-beta superfamily
.
Cell Syst
2018
;
7
:
422
37
.
50.
Xie
W
,
Mertens
JC
,
Reiss
DJ
,
Rimm
DL
,
Camp
RL
,
Haffty
BG
, et al
Alterations of Smad signaling in human breast carcinoma are associated with poor outcome: a tissue microarray study
.
Cancer Res
2002
;
62
:
497
505
.
51.
Feng
XH
,
Derynck
R
. 
Specificity and versatility in tgf-beta signaling through Smads
.
Annu Rev Cell Dev Biol
2005
;
21
:
659
93
.
52.
Sun
J
,
Zhuang
FF
,
Mullersman
JE
,
Chen
H
,
Robertson
EJ
,
Warburton
D
, et al
BMP4 activation and secretion are negatively regulated by an intracellular gremlin-BMP4 interaction
.
J Biol Chem
2006
;
281
:
29349
56
.
53.
Tarragona
M
,
Pavlovic
M
,
Arnal-Estape
A
,
Urosevic
J
,
Morales
M
,
Guiu
M
, et al
Identification of NOG as a specific breast cancer bone metastasis-supporting gene
.
J Biol Chem
2012
;
287
:
21346
55
.
54.
Baraban
E
,
Chavakis
T
,
Hamilton
BS
,
Sales
S
,
Wabitsch
M
,
Bornstein
SR
, et al
Anti-inflammatory properties of bone morphogenetic protein 4 in human adipocytes
.
Int J Obes
2016
;
40
:
319
27
.
55.
Qian
SW
,
Tang
Y
,
Li
X
,
Liu
Y
,
Zhang
YY
,
Huang
HY
, et al
BMP4-mediated brown fat-like changes in white adipose tissue alter glucose and energy homeostasis
.
Proc Natl Acad Sci U S A
2013
;
110
:
E798
807
.
56.
Chan
DS
,
Vieira
AR
,
Aune
D
,
Bandera
EV
,
Greenwood
DC
,
McTiernan
A
, et al
Body mass index and survival in women with breast cancer-systematic literature review and meta-analysis of 82 follow-up studies
.
Ann Oncol
2014
;
25
:
1901
14
.