Research ArticleCancer

ABL kinases promote breast cancer osteolytic metastasis by modulating tumor-bone interactions through TAZ and STAT5 signaling

See allHide authors and affiliations

Sci. Signal.  02 Feb 2016:
Vol. 9, Issue 413, pp. ra12
DOI: 10.1126/scisignal.aad3210

Metastatic breast tumors break down bone

Most breast cancer cells activate the breakdown of bone to promote metastases. Wang et al. found that the ABL kinase isoforms ABL1 and ABL2 enhanced the ability of breast cancer cells to invade and break down bone in mice. In breast cancer cells, the ABL kinases activated the transcriptional coactivator TAZ and the transcription factor STAT5, which triggered the transcription of genes encoding factors that activate osteoclasts (cells that break down bone) and those that enhance the survival of breast cancer cells in the bone microenvironment. An ABL-specific inhibitor decreased bone metastasis in mice, thus suggesting that blocking ABL kinase activity could prevent breast cancer cells from promoting bone metastasis.

Abstract

Bone metastases occur in up to 70% of advanced breast cancer. For most patients with breast cancer, bone metastases are predominantly osteolytic. Interactions between tumor cells and stromal cells in the bone microenvironment drive osteolytic bone metastasis, a process that requires the activation of osteoclasts, cells that break down bone. We report that ABL kinases promoted metastasis of breast cancer cells to bone by regulating the crosstalk between tumor cells and the bone microenvironment. ABL kinases protected tumor cells from apoptosis induced by TRAIL (TNF-related apoptosis-inducing ligand), activated the transcription factor STAT5, and promoted osteolysis through the STAT5-dependent expression of genes encoding the osteoclast-activating factors interleukin-6 (IL-6) and matrix metalloproteinase 1 (MMP1). Furthermore, in breast cancer cells, ABL kinases increased the abundance of the Hippo pathway mediator TAZ and the expression of TAZ-dependent target genes that promote bone metastasis. Knockdown of ABL kinases or treatment with ABL-specific allosteric inhibitor impaired osteolytic metastasis of breast cancer cells in mice. These findings revealed a role for ABL kinases in regulating tumor-bone interactions and provide a rationale for using ABL-specific inhibitors to limit breast cancer metastasis to bone.

INTRODUCTION

The ABL family of nonreceptor tyrosine kinases, ABL1 (also known as c-Abl) and ABL2 (also known as Arg), links diverse extracellular stimuli to signaling pathways that control cell growth, survival, adhesion, migration, and invasion (13). ABL tyrosine kinases play an oncogenic role in human leukemias (4, 5) and promote the progression of solid tumors (5, 6). ABL kinases elicit protumorigenic or antitumorigenic effects in breast cancer cells and promote cancer cell invasion (710). However, whether ABL kinases have a role in the regulation of cellular processes critical for metastasis, other than invasion, has not yet been evaluated. Here, we uncovered a critical role for ABL kinases in the regulation of breast cancer metastasis to bone.

Bone metastases occur in up to 70% of patients with advanced breast cancer and are associated with high mortality and morbidity (11, 12). Whereas the mechanisms that drive tumor cell homing, invasion, and colonization to bone are poorly understood, it is increasingly apparent that bone metastasis requires interactions between tumor and stromal cells in the bone microenvironment (13). When breast cancer cells invade the bone microenvironment, they produce molecules that activate osteoclastic bone resorption, leading to the release of growth factors stored in the bone matrix to promote tumor growth. Currently, there are no available therapies to cure breast cancer metastasis. Thus, there is a need to identify molecules that could be targeted simultaneously in tumor and bone to disrupt the tumor cell–stromal cell interactions that drive metastasis.

Here, we report that increased expression of ABL1 and ABL2 correlated with enhanced breast cancer metastasis and decreased metastasis-free survival. Using metastasis models that bypass invasion and intravasation, we uncovered roles for the ABL kinases in the regulation of breast cancer cell survival and colonization in the bone microenvironment. Further, we identified a role for ABL kinases in promoting the expression of multiple pro–bone metastasis genes such as AXL (which encodes a receptor tyrosine kinase), IL6 (which encodes interleukin-6), MMP1 (which encodes matrix metalloproteinase 1), and TNC (which encodes tenascin-C) through TAZ- and signal transducer and activator of transcription 5 (STAT5)–mediated signaling. Moreover, we found that treatment with a selective allosteric inhibitor of the ABL kinases or simultaneous depletion of both ABL kinases in breast cancer cells impaired breast cancer bone metastases and decreased osteoclast activation in vitro and osteolysis in vivo.

RESULTS

Increased expression of ABL kinase–encoding genes correlates with breast cancer metastasis

To evaluate whether altered expression of the ABL genes is associated with breast cancer progression and metastasis, we examined the expression of ABL1 and ABL2 in normal and invasive breast tumor specimens using published TCGA (The Cancer Genome Atlas) data sets (1416). ABL2 DNA and RNA abundance was significantly increased in breast tumor specimens (Fig. 1, A and B). To further evaluate the importance of enhanced ABL abundance in the context of metastasis, we analyzed an integrative database assembled from 22 publicly available data sets containing information on metastasis-related relapse (17). We found that increased ABL2 mRNA abundance correlated with metastasis across all subtypes of breast cancer, primarily the basal type (Fig. 1, C and D), whereas high ABL1 mRNA abundance significantly correlated with metastasis in human epidermal growth factor (EGF) receptor 2 (HER2)–enriched breast cancer but not in other breast cancer subtypes (Fig. 1E). Furthermore, high ABL1 mRNA was associated with bone metastasis in a microarray data set reporting organ-specific metastasis (Fig. 1F) (18). Collectively, these findings support a link between increased expression of the ABL genes and increased breast cancer metastasis.

Fig. 1 Increased expression of ABL genes in invasive breast cancer is associated with metastasis.

(A) ABL2 copy number in 813 normal samples compared with 789 invasive breast tumor samples in the TCGA database. (B) ABL2 mRNA abundance in 61 normal samples compared with 532 invasive breast tumor samples in the TCGA database. Results shown in (A) and (B) are based on the data generated by the TCGA Research Network (http://cancergenome.nih.gov/); whiskers represent 1st and 99th percentile. (C and D) Kaplan-Meier representation of the probability of cumulative overall distant metastasis–free survival (DMFS) in 2830 breast cancer cases (C) or 482 basal breast cancer cases (D) according to ABL2 expression. (E) Kaplan-Meier representation of the probability of cumulative overall distant metastasis-free survival in 279 HER2-enriched breast cancer cases according to ABL1 expression. (F) Kaplan-Meier representation of the probability of cumulative bone metastasis-free survival (BMFS) in 42 breast cancer cases according to ABL1 expression. P values (log-rank test) and hazard ratios (HR) are shown in the graph.

ABL family protein kinases are required for bone metastasis

To directly evaluate the relationship between ABL family kinases and metastasis, we analyzed ABL1 and ABL2 protein abundance in MDA-MB-231–derived breast cancer cell lines with different organ metastasis tropisms (19). The MDA-MB-231–derived 1833 cell line, which is characterized by enhanced bone-specific metastasis compared to the parental cell line or cell lines with increased tropism to lung and brain, showed increased abundance of ABL1 and ABL2 (fig. S1A). To examine the functional role of ABL kinases in these cells, we depleted endogenous ABL kinases with previously characterized short hairpin RNAs (shRNAs) specific against ABL1 and ABL2 (20). The lentivirus-encoded shRNAs decreased the abundance of ABL1 and ABL2 by 80% at day 2 after viral transduction, but ABL1 abundance was slightly increased by day 21 after viral transduction (fig. S1B). Depletion of ABL kinases did not affect cell growth in monolayers or colony formation in Matrigel (fig. S1, C to E) but decreased cell invasion in both 1833 triple-negative and HER2-positive SKBR3 breast cancer cells (fig. S2, A to D). Conversely, overexpression of constitutively active forms of ABL1 (ABL1PP) and ABL2 (ABL2PP) enhanced the invasiveness of the parental MDA-MB-231 cells (fig. S2, E to G).

ABL kinases regulate cancer cell invasion (21), but it is unclear whether they play a role in the regulation of subsequent steps of the metastatic cascade. To investigate whether inhibition of ABL kinases interferes with metastatic processes other than invasion, we depleted ABL1 and ABL2 (ABL1/ABL2) in two bone metastatic breast cancer cell lines, 1833 and SCP28 (22), and evaluated the metastatic potential of these cells after intracardiac injection into immunodeficient mice. This mouse model bypasses the initial invasion step and allows the analysis of subsequent steps in the metastatic cascade. The 1833 and SCP28 breast cancer cells were engineered to express reporters with luciferase and green fluorescent protein to monitor metastatic progression by bioluminescence imaging (Fig. 2A). We found that ABL kinase knockdown increased the survival of tumor-bearing mice (Fig. 2B) and markedly inhibited bone metastases by 1833 and SCP28 breast cancer cells as measured by bioluminescence imaging (Fig. 2, C to F) and hematoxylin and eosin (H&E) staining (Fig. 2, G and H). Decreased metastasis by ABL-deficient breast cancer cells was accompanied by a significant reduction in the extent of hindlimb osteolytic lesions, as determined by x-ray and micro–computed tomography (μCT) imaging (Fig. 2, I and J).

Fig. 2 Knockdown of ABL kinases decreases breast cancer bone metastasis.

(A) Experimental design. (B) Survival of mice after intracardiac injection of 1833 (1 × 105) breast cancer cells transduced with control shRNA (Scr) or shRNAs against ABL1 and ABL2 (shAA); n = 10 mice per group. (C to F) Bioluminescent images (C and E) of bone metastasis from representative mice at day 22 after inoculation with 1833 cells (n = 10 mice per group) or day 35 after inoculation with SCP28 cells (n = 8 mice per group). Quantification of bone metastases (D and F). (G and H) Representative H&E staining (G) and quantification of H&E-stained tumor area of bone lesions. Arrows indicate tumor. n = 3 mice per group. Scale bar, 200 μm. Met, metastatic. (I and J) Representative x-ray and μCT reconstruction (I) and quantification of bone volume (BV)/total volume (TV) from μCT analysis of the mouse tibias (J). n = 3 mice per group. (K) Representative immunoblots of 1833 cells transfected with control shRNA, shRNA against ABL1 (shABL1), ABL2 (shABL2), and shRNA#2 against both ABL1 and ABL2 (shAA#2) and ABL1/ABL2 knockdown cells with overexpression of mouse Abl1/Abl2 (shAA + mAbl1/Abl2). n = 3 blots. p, phosphorylated. (L) Bioluminescent images of bone metastases from representative mice at day 18 after inoculation. n = 8 mice per group. (M) Quantification of (L). *P < 0.05, one-way analysis of variance (ANOVA) followed by Tukey’s post hoc test.

To evaluate whether ABL1 and ABL2 are individually responsible for promoting metastasis, we used specific shRNAs to silence either ABL1 or ABL2 in breast cancer cells. We found that ~90% knockdown of ABL1 alone resulted in enhanced ABL2 expression and did not produce a significant decrease in the phosphorylation of CrkL, a reporter for the activation state of the ABL kinases (Fig. 2K), and did not inhibit metastasis (Fig. 2, L and M). Double knockdown of ABL1 and ABL2 was required to decrease the phosphorylation of CrkL by more than 90%, which indicates inactivation of the endogenous ABL kinases (Fig. 2K). Whereas knockdown of ABL2 alone decreased metastasis, knockdown of both ABL1 and ABL2 was required to significantly decrease metastasis (Fig. 2, L and M). To further validate that the decreased metastasis induced by double knockdown of ABL1 and ABL2 was not due to off target effects of the lentivirus-encoded shRNAs, we used a second set of ABL shRNAs (shAA#2) and carried out rescue experiments by expressing mouse Abl1 and Abl2 constructs (mAbl1/Abl2) that are resistant to shRNAs against human ABL1 and ABL2. We found that expression of murine Abl1 and Abl2 kinases in the knockdown cells reversed the decreased metastasis (Fig. 2, K to M). Loss of ABL1 and ABL2 in the lung metastatic 4175 breast cancer cell line did not significantly reduce metastasis (fig. S3). These findings reveal a function for ABL kinases in the regulation of breast cancer bone metastasis and tumor-induced osteolysis in vivo.

Allosteric inhibition of the ABL kinases impairs breast cancer bone metastasis

To date, few studies have directly evaluated the biological consequences of targeting the ABL kinases with selective inhibitors in solid tumors, including breast cancer in vivo. The adenosine triphosphate (ATP)–competitive kinase inhibitors imatinib (STI571; trade name: Gleevec), dasatinib, and nilotinib inhibit multiple tyrosine kinases in addition to ABL1 and ABL2 (5). Moreover, these ATP-competitive inhibitors induce the formation of B-RAF and C-RAF dimers, leading to extracellular signal–regulated kinase (ERK) activation in diverse cancer cell types (23). Therefore, we wished to use a different approach to evaluate whether ABL kinases play a role in breast cancer metastasis using the allosteric inhibitor GNF5, which targets the unique ABL myristate binding site and functions as a non–ATP-site and selective inhibitor of the ABL kinases (24, 25). Notably, GNF5 does not activate the RAF-ERK pathway in breast cancer cells (23) (fig. S4).

Treatment of tumor-bearing mice with GNF5 starting on day 8 after intracardiac injection of breast cancer cells (Fig. 3A) resulted in a significant increase in survival (Fig. 3B) and a decrease in bone metastasis burden as measured by bioluminescence imaging (Fig. 3, C and D). Similarly, histological analyses revealed a decrease in bone tumor burden in mice treated with the allosteric inhibitor of the ABL kinases (Fig. 3, E and F). Notably, we found that bone destruction was decreased and the ratio of bone volume to total volume was increased in tumor-bearing mice treated with GNF5 (Fig. 3, G and H). These results demonstrate that ABL kinase activity is required for osteolytic metastasis in breast cancer and suggest that pharmacological inhibition of the ABL kinases may be an effective treatment for bone metastasis.

Fig. 3 Allosteric inhibition of ABL kinases decreases breast cancer bone metastasis.

(A) Experimental design. (B) Survival of mice after intracardiac injection of 1833 (1 × 105) breast cancer cells and treatment with either dimethyl sulfoxide (DMSO) control or the allosteric ABL inhibitor GNF5. n = 10 mice per group. (C) Bioluminescent images of representative mice at day 22 after inoculation. (D) Quantification of bone metastases. n = 10 mice per group. (E and F) Representative H&E staining (E) and quantification (F) of stained tumor area of bone lesions. Arrows indicate tumor. n = 3 mice per group. Scale bar, 200 μm. (G and H) Representative x-ray and μCT reconstruction (G) of mouse tibias and quantification (H) of bone volume/total volume. n = 3 mice per group.

ABL kinases are required for tumor cell survival and tumor-induced osteolysis in the bone microenvironment

To directly examine whether ABL kinases play a role in regulating the colonization and survival of breast cancer cells in the bone microenvironment, we injected control or ABL1/ABL2 knockdown breast cancer cells directly into the tibia of immunodeficient mice. Depletion of the ABL kinases reduced tumor expansion in the tibia as measured by both bioluminescence imaging (Fig. 4, A and B) and histological staining (Fig. 4, C and D). Moreover, three-dimensional (3D) μCT reconstruction of the tibia revealed that mice injected with control cells had a significantly higher degree of osteolysis with a decreased ratio of bone volume to total volume compared to mice injected with ABL1/ABL2 knockdown breast cancer cells (Fig. 4, E and F). Depletion of the ABL kinases did not affect breast cancer cell proliferation or colony formation in vitro (fig. S1). Therefore, these findings suggest that ABL1/ABL2-dependent expansion of breast cancer cells is mediated by factors present in the bone microenvironment.

Fig. 4 ABL kinases are required for tumor survival and tumor-induced osteolysis in the bone microenvironment.

(A and B) Control or ABL1/ABL2 knockdown 1833 cells (1 × 105) were injected directly into the tibias of the mice. Representative bioluminescent images (A) taken at day 21 after inoculation and quantification (B) of bone lesions are shown. n = 5 mice per group. (C and D) Representative H&E staining (C) and quantification (D) of H&E-stained tumor area of mouse tibias from each group. n = 3 mice per group. Scale bar, 500 μm. (E and F) Representative 3D μCT reconstruction of mouse tibias (E) and quantification (F) of bone volume/total volume from μCT analysis. n = 3 mice. (G and H) Representative images (G) of TUNEL (terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling) staining of cells treated with TRAIL and the indicated shRNA and quantification of the percent of TUNEL-positive cells (H). n = 3 biological replicates. Scale bar, 100 μm. (I) Immunoblotting was performed using the indicated antibodies on whole-cell lysates from cells incubated or not with TRAIL. n = 3 blots.

Bone metastasis requires reciprocal interactions between tumor cells, stromal cells, and bone cells (12, 26). Several soluble factors released by stromal cells within the bone microenvironment promote tumor growth and survival (27). These factors include chemokine (C-X-C motif) ligand 12 (CXCL12), a chemokine produced by bone marrow mesenchymal cells that functions as a chemoattractant and survival factor for cells bearing the chemokine (C-X-C motif) receptor 4 (CXCR4), and insulin-like growth factor 1 (IGF-1), a factor that is stored in the bone matrix and released during osteolysis (28). We and others (6, 21) have previously shown that ABL kinases are activated by the binding of CXCL12 and IGF-1 to their cognate receptors. Thus, we evaluated whether loss of ABL kinases could affect the activation of AKT-mediated survival by these factors. We found that CXCL12- and IGF-1–induced activation of AKT was independent of ABL1 and ABL2 in 1833 breast cancer cells (fig. S5, A and B). In contrast, we observed that ABL kinases protected breast cancer cells from TRAIL [tumor necrosis factor (TNF)–related apoptosis-inducing ligand]–induced cell death (Fig. 4, G to I). TRAIL is a proapoptotic member of the TNF family, which induces apoptosis by binding to the cell death receptors DR4 and DR5 (29). TRAIL and DR5 are present in clinical breast cancer bone metastases specimens, and DR4 and DR5 are present in 80% of patient bone tumor biopsies (28, 30). TRAIL-enhanced apoptosis as measured by cleavage of caspase-3 (Fig. 4I), and knockdown of ABL kinases increased the sensitivity of 1833 breast cancer cells to the proapoptotic effects of TRAIL (Fig. 4, G to I). These data suggest that ABL kinases promote breast cancer metastasis to bone in part by increasing tumor cell survival within the bone microenvironment.

Depletion of ABL kinases impairs tumor-induced osteoclast activation in part by decreased IL-6 secretion

Osteoclast activation plays a central role in the progression of breast cancer bone metastasis. To directly examine whether ABL kinases regulate tumor-induced osteoclast activation, we used an in vitro osteoclastogenesis assay (Fig. 5A). Mouse primary bone marrow cells were treated with conditioned medium from control or ABL1/ABL2 knockdown breast cancer cells and then stained for tartrate-resistant acid phosphatase (TRAP), an osteoclast marker. Bone marrow cells cultured with conditioned medium derived from ABL1/ABL2 knockdown 1833 and SKBR3 breast cancer cells had decreased numbers of TRAP+ cells compared to the control groups (Fig. 5, B and C, and fig. S6, A and B). These data suggest that inactivation of the ABL kinases in breast cancer cells may impair the secretion of soluble factor(s) required for osteoclast activation. Because bone marrow contains a heterogeneous population of cells, factors secreted in an ABL1/ABL2-dependent manner might interact directly with osteoclast progenitors and promote their differentiation or might instead function to regulate osteoclasts indirectly by modulating the activity of osteoblasts (22). To test the first possibility, we carried out the osteoclastogenesis assay using the RAW264.7 murine preosteoclast cell line. Conditioned medium derived from ABL1/ABL2 knockdown 1833 breast cancer cells did not impair tumor-induced RAW264.7 preosteoclast differentiation (Fig. 5D). These findings suggest that ABL kinases regulate osteoclast maturation indirectly, possibly by modulating osteoblast function.

Fig. 5 Depletion of ABL kinases impairs tumor-induced osteoclast activation in part by decreasing IL-6 secretion.

(A) In vitro osteoclastogenesis assay. (B) TRAP staining of bone marrow cells treated with conditioned medium (CM) from 1833 breast cancer cells. Scale bar, 50 μm. (C) Quantification of TRAP+ cells in (B). (D) Quantification of TRAP+ cells derived from RAW264.7 cells; ns, not significant. (E and F) RANKL (E) and OPG (F) expression was detected by reverse transcription polymerase chain reaction (RT-PCR) of the osteoblast cell line 7F2 treated with conditioned medium harvested from the indicated 1833 cells. (G) Identification of differentially expressed cytokines in the conditioned medium of 1833 cells using a cytokine antibody array. n = 2 biological replicates. (H) ELISA quantification of IL-6 in conditioned medium of 1833 cells. (I) Quantification of TRAP+ bone marrow–derived osteoclasts incubated with the indicated doses of IL-6. *Significantly different from 0; P value was calculated using one-way ANOVA followed by Tukey’s post hoc test. (J) TRAP staining of bone marrow treated with conditioned medium from 1833 cells with or without added IL-6. (K) Quantification of TRAP+ osteoclasts in (J). *P < 0.05; **P < 0.01, two-way ANOVA followed by Tukey’s post hoc test. n = 3 biological replicates unless otherwise indicated.

Osteoblasts modulate osteoclast activity through secretion of the TNF family member RANKL (receptor activator of nuclear factor κB ligand). Binding of RANKL to the RANK receptor on the surface of osteoclasts activates a pathway essential for osteoclast differentiation. Osteoprotegerin (OPG), a soluble decoy receptor for RANKL, is also produced by osteoblasts and antagonizes the activity of RANKL (31). Several tumor-derived bone metastasis factors can increase RANKL production or decrease OPG secretion by osteoblasts, thereby promoting osteoclast differentiation and activation (12, 32). To evaluate whether ABL kinases might regulate the secretion of osteoblast-derived RANKL or OPG leading to osteoclast differentiation, we analyzed RANKL and OPG mRNA abundance in the murine osteoblast cell line 7F2 in response to conditioned medium from control and ABL1/ABL2 knockdown breast cancer cells. Whereas conditioned medium from ABL1/ABL2-depleted breast cancer cells did not affect RANKL abundance in osteoblasts compared with the cells treated with control conditioned medium (Fig. 5E), we found that conditioned medium from breast cancer cells lacking ABL kinases increased OPG abundance in the osteoblast cell line (Fig. 5F). These findings suggest that ABL kinases promote tumor-induced osteoclast activation in part by increasing OPG abundance in osteoblasts.

To identify tumor-secreted cytokines regulated by the ABL kinases that promote breast cancer metastasis to bone, we used a human cytokine antibody array to identify changes in cytokine concentrations in the conditioned medium from control and ABL1/ABL2 knockdown cells. We found that IL-6 concentrations were decreased in the conditioned medium derived from ABL1/ABL2 knockdown cells compared to that from control cells (Fig. 5G), results that were validated by enzyme-linked immunosorbent assay (ELISA) (Fig. 5H). IL-6 is a multifunctional cytokine with pleiotropic functions that include inducing osteoclast activation, enhancing bone resorption, and increasing metastasis (3335). Inhibition of the IL-6 receptor directly blocks osteoclast formation in vitro and in vivo (36). Thus, we evaluated whether addition of IL-6 could in part rescue defective osteoclastogenesis induced by conditioned medium from breast cancer cells depleted of the ABL kinases. We added the optimal dose of IL-6 required to promote maturation of osteoclasts (Fig. 5I) to conditioned medium derived from ABL1/ABL2 knockdown breast cancer cells. Addition of IL-6 to reconstituted conditioned medium from ABL1/ABL2-depleted breast cancer cells partially restored osteoclast activation (Fig. 5, J and K). IL-6 induced RANKL expression but suppressed OPG expression in the 7F2 osteoblast cell line (fig. S7, A and B). Thus, depletion of the ABL kinases in breast cancer cells and the accompanying decrease in IL-6 secretion result in enhanced OPG expression and reduced overall RANKL/OPG ratio, thereby decreasing osteoclast differentiation and bone destruction.

Next-generation sequencing reveals ABL1/ABL2-regulated genes in breast cancer cells

To gain insight into the signaling pathway(s) required for ABL1/ABL2-dependent bone metastasis, we evaluated the consequences of single or double inactivation of ABL1 and ABL2 on the transcriptome of breast cancer cells using next-generation sequencing (RNAseq) analysis (Fig. 6 and fig. S8A). We found that 180 genes showed significantly decreased expression and 40 genes showed significantly increased expression in ABL1/ABL2 knockdown cells (fig. S8B and table S1). Principal components analysis revealed that transcripts altered in breast cancer cells depleted of ABL1, ABL2, or both ABL1 and ABL2 were clustered and were distinct from the control group (fig. S8C). This analysis indicated that the transcriptomes of single- and double-knockdown cells were similar to each other but different from that of the control cells, supporting the quality and validity of the RNAseq analysis. Comparison of the transcripts revealed that breast cancer cells with knockdown of ABL2 alone shared a similar gene expression pattern with that of the ABL1/ABL2 double-knockdown cells (Fig. 6A and fig. S8D). Notably, analysis of cell lysates from ABL2-depleted and ABL1/ABL2 double-knockdown breast cancer cells showed a greater reduction of the phosphorylation of CrkL compared to cells with knockdown of ABL1 alone (Fig. 2K). Thus, the altered gene expression profiles correlate with decreased ABL kinase activity in breast cancer cells.

Fig. 6 ABL kinases regulate the expression of genes in the JAK/STAT and Hippo pathway signatures in metastatic breast cancer cells.

(A) CummeRbund heat map of genes that were differentially expressed in control and single and double ABL1 and ABL2 knockdown cells. (B) GSEA analysis of the indicated gene signatures in ABL1/ABL2 knockdown cells compared with control cells (Scr). NES, normalized enrichment score. (C) Expression of the indicated genes in control, ABL1 or ABL2 single-knockdown, and ABL1/ABL2 double-knockdown cells quantified using Cufflinks CuffDiff. *Significantly different from control cells (P < 0.05 after Benjamini-Hochberg correction for multiple testing). Error bars represent SD. n = 3 biological replicates for (A) and (C).

ABL kinases signal to TAZ and STAT5 to promote breast cancer bone metastasis

To identify the pathways affected by the inactivation of the ABL kinases in metastatic breast cancer cells, we carried out Gene Set Enrichment Analysis (GSEA) using multiple databases (37). In addition to the GSEA oncogenic signature database, we used the KEGG (Kyoto Encyclopedia of Genes and Genomes) database and published breast cancer metastasis data sets. We found that a gene signature consisting of 273 genes important for breast cancer bone metastasis showed decreased expression in ABL1/ABL2 knockdown cells (Fig. 6B) (38). Further, inactivation of the ABL kinases resulted in decreased expression of the genes in the Hippo, Janus kinase (JAK)/STAT, and cytokine/cytokine receptor pathway signatures (Fig. 6B). To identify key molecular mediators of the ABL kinases implicated in the regulation of the ABL1/ABL2-dependent pathways, we analyzed the expression of individual genes for transcripts altered by loss of the ABL kinases. Among transcripts that were decreased in ABL1/ABL2 knockdown cells were TAZ [also known as WWTR1 (WW domain–containing transcription regulator protein 1)], which encodes a transcriptional coactivator in the Hippo pathway, and STAT5A, which encodes a transcription factor (Fig. 6C).

TAZ and the related YAP1 proteins are components of the Hippo pathway and have been implicated in breast cancer progression and metastasis (39, 40). We found that knockdown of the ABL kinases decreased the mRNA expression of TAZ (Fig. 6C) and reduced the protein abundance of TAZ and its downstream target AXL (Fig. 7A and fig. S9, A and B). AXL encodes a receptor tyrosine kinase that promotes breast cancer bone metastasis in mouse models (41). Knockdown of ABL2, but not of ABL1, reduced TAZ abundance to a similar extent as ABL1/ABL2 double knockdown (fig. S9A), suggesting that ABL2 has a predominant role in regulating TAZ abundance. Overexpression of ABL1 and ABL2 in both 1833 and parental MDA-MB-231 breast cancer cells increased TAZ abundance (fig. S9C). Further, immunofluorescence staining analysis indicated that the TAZ protein predominantly (~90%) localized in the nuclei of 1833 breast cancer cells (fig. S10A), an effect decreased by double knockdown of ABL1 and ABL2 (fig. S10B). Similarly, inhibiting ABL kinase activity with the allosteric inhibitor GNF5 decreased TAZ protein abundance (fig. S11). TAZ protein abundance was not decreased by GNF5 treatment in breast cancer cells expressing murine Abl2-E505K, a mutant that is resistant to the GNF5 allosteric inhibitor (fig. S11). Moreover, ABL2 mRNA expression positively correlated with TAZ mRNA expression in a TCGA data set of 971 invasive breast cancer patients (fig. S12). To evaluate whether loss of ABL kinases affected TAZ activity, we performed chromatin immunoprecipitation (ChIP) analysis using primers for TAZ targets identified by ChIP sequencing analysis (42). We found that depletion of the ABL kinases decreased TAZ binding to some of its target genes (fig. S13). Whereas ABL1 has been reported to phosphorylate YAP1 in response to DNA damage (43), we found that ABL1/ABL2 knockdown did not substantially alter YAP1 protein abundance (Fig. 7A), nuclear localization (fig. S14, A and B), phosphorylation of YAP at Tyr357 (fig. S14C), or binding to some of its downstream targets (fig. S13). However, we cannot rule out the possibility that ABL kinases might regulate YAP1-mediated expression of other target genes in breast cancer cells. Regardless, our data support a role for ABL kinases in the regulation of TAZ protein abundance and activity in breast cancer cells. Moreover, we found that expression of a constitutively active TAZ S89A mutant in ABL1/ABL2 knockdown breast cancer cells (1833 and SCP28) restored the abundance of its target AXL (Fig. 7, B and C). Together, these data revealed a functional link between the ABL kinases and TAZ signaling leading to increased AXL abundance in breast cancer cells, and identified a potentially druggable pathway for the treatment of breast cancer bone metastasis.

Fig. 7 ABL kinases are required for TAZ and STAT5 signaling in breast cancer cells.

(A to C) Immunoblots with the indicated antibodies were performed on whole-cell lysates of 1833 and SCP28 cells. (D) Immunoblots were performed on whole-cell lysates (pSTAT5, STAT5, and tubulin) or conditioned medium (MMP1, IL-6, and TNC). (E) Immunoblots were performed on whole-cell lysates (pSTAT5, STAT5, and tubulin) or conditioned medium (MMP1, IL-6, and TNC) of parental 1833 and SCP28 cells. (F) Immunoblotting with the indicated antibodies were performed on whole-cell lysates (pSTAT5, STAT5, and tubulin) or conditioned medium (MMP1, IL-6, and TNC) of 1833 cells. For (A) to (F), n = 3 blots. (G) Bioluminescent images of representative mice at day 25 after intracardiac injection of 1833 cells. (H) quantification of bone metastasis. n = 5 mice per group. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant, one-way ANOVA followed by Tukey’s post hoc test. (I and J) Bioluminescent images (I) and quantification (J) of bone metastasis from representative mice at day 25 after intracardiac injection of 1833 cells transfected with control [nonspecific (NS)] or shRNAs against STAT5 and TAZ (shSTAT5/shTAZ). n = 8 mice per group. (K) Immunoblots were performed on whole-cell lysate. n = 3 blots.

We found that inactivation of the ABL kinases in breast cancer cells also decreased STAT5A mRNA and downstream expression of STAT5 target genes, including TNC (Fig. 6D). STAT5 belongs to a family of transcription factors that regulate cytokine-induced gene expression and is constitutively activated in several human cancers including breast cancer, where it promotes the expression of genes encoding cell survival factors (44). STAT5 is also activated by the oncogenic breakpoint cluster region (BCR)–ABL tyrosine kinase and contributes to the transformation of leukemia cells (45). STAT5 promotes metastasis of human prostate cancer cells (46) and has been implicated in the resistance of metastatic breast cancer cells to targeted therapies (47). Moreover, ablation of a STAT5A allele reduces tumor incidence in a mouse model of breast cancer in which mammary epithelial cells express T antigen (48). We found that depletion of ABL kinases in breast cancer cells decreased STAT5A mRNA expression (Fig. 6D) without decreasing total STAT5 protein abundance as measured by Western blotting with antibodies that detect both STAT5A and STAT5B (Fig. 7D and fig. S9, A and B). However, depletion of ABL kinases decreased the phosphorylation of STAT5 (Fig. 7D and fig. S9, A and B). Conversely, overexpression of ABL kinases, predominantly ABL1, in both 1833 and parental MDA-MB-231 breast cancer cells increased STAT5 phosphorylation (fig. S9C). Further, we found that double knockdown of ABL1 and ABL2 decreased the abundance of various secreted proteins, including IL-6, TNC, and MMP1 (Fig. 7D). Both MMP1 and IL-6 have been linked to the regulation of osteoclast activation (32, 33), and depletion of TNC decreases breast cancer metastasis (49). Expression of a constitutively active STAT5A mutant (STAT5A*) reversed the reduction in MMP1, IL-6, and TNC abundance induced by depletion of both ABL kinases in breast cancer cells (Fig. 7, E and F, and fig. S15). These findings support a role for STAT5 in regulating the ABL1/ABL2-dependent secretome.

To evaluate whether the TAZ and STAT5 pathways promote breast cancer bone metastasis downstream of the ABL kinases, we expressed the constitutively active mutants TAZ S89A and STAT5* in ABL1/ABL2 knockdown cells. Expression of either TAZ S89A or STAT5* alone in ABL1/ABL2-depleted breast cancer cells only partially rescued bone metastasis, and expression of both STAT5* and TAZ S89A was required to fully rescue the impaired bone metastasis by ABL1/ABL2-depleted cells (Fig. 7, G and H). To evaluate whether depletion of TAZ and STAT5 in breast cancer could phenocopy the reduced bone metastasis caused by depletion of ABL1 and ABL2, we performed intracardiac injection of control or TAZ/STAT5 double-knockdown 1833 breast cancer cells and found that cells depleted of TAZ and STAT5 exhibited markedly impaired metastasis that phenocopied the inhibitory effects of ABL1/ABL2 knockdown (Fig. 7, I to K). TAZ and STAT5 might regulate each other (fig. S16). TAZ knockdown with two distinct shRNAs decreased the phosphorylation of STAT5 and, to a lesser extent, total STAT5 protein abundance (fig. S16A). Moreover, knockdown of STAT5 with two different shRNAs slightly decreased TAZ protein abundance (fig. S16B). However, these reciprocal decreases were much lower than those induced by knockdown of the ABL kinases (Fig. 7). Future studies will evaluate the pathways that mediate the crosstalk between the TAZ and STAT5 pathways. Together, our findings suggest that ABL kinases activate the TAZ and STAT5 pathways and that coactivation of their downstream targets promote the bone metastasis of breast cancer cells in mouse models. Using a TCGA data set with 971 invasive breast cancer patients, we found that patients with alterations in the expression of ABL2 and eight validated downstream targets (TAZ, AXL, CTGF, STAT5A, STAT5B, TNC, IL6, and MMP1) exhibited decreased disease-free survival (Fig. 8A).

Fig. 8 ABL kinases activate the TAZ and STAT5 pathways to promote breast cancer bone metastasis.

(A) Kaplan-Meier representation of the probability of cumulative overall disease-free survival in TCGA data set with 971 invasive breast cancer patients according to whether the ABL signature (ABL2, TAZ, AXL, CTGF, STAT5A, STAT5B, TNC, IL6, and MMP1) was altered or not. P value was derived by the log-rank test. (B) Model for the role of ABL kinases in the regulation of breast cancer bone metastasis.

DISCUSSION

Here, we uncovered a role for the ABL kinases in promoting breast cancer bone metastasis through the regulation of distinct pathways required for tumor colonization and survival in the bone microenvironment. We showed that ABL kinase activity was required for osteolytic metastasis of breast cancer cells and that depletion or pharmacological inhibition of these kinases impaired breast cancer metastasis to bone. We demonstrated that ABL kinases promoted tumor-induced osteolysis in part through the osteoclast-activating cytokine IL-6, increased serum concentrations of which are associated with poor clinical outcome in breast cancer patients (50). IL-6 can induce osteoclast activation indirectly by altering the expression of RANKL and OPG in osteoblasts (51). Depletion of ABL kinases in breast cancer cells decreased IL-6 concentrations and was accompanied by increased OPG expression in osteoblasts. We found that the addition of IL-6 partially enabled medium conditioned by ABL1/ABL2 knockdown breast cancer cells to activate osteoclasts, and suppressed OPG expression in the 7F2 osteoblast cell line. These findings support a role for ABL-mediated IL-6 secretion by breast cancer cells in osteoclast activation through decreased OPG expression in osteoblasts. IL-6 has pleiotropic roles during tumor progression and metastasis that include increased tumor cell survival, expansion of breast cancer stem cells, and resistance to targeted therapies (52). IL-6 may also function downstream of the ABL kinases to protect breast cancer cells from TRAIL-induced cell death.

Depletion of ABL kinases in breast cancer cells also decreased the abundance of MMP1, a protease that cleaves fibrillar collagens and promotes the proteolytic release of bound growth factors (32). High serum concentrations of MMP1 correlate with bone metastasis in breast cancer patients (32). Through its various targets, MMP1 promotes not only tumor invasion but also breast cancer colonization to bone by mechanisms that include the release of membrane-bound EGF-like growth factors from tumor cells, leading to activation of EGF receptor signaling and suppression of OPG expression in osteoblasts, which in turn promotes the differentiation and activation of osteoclasts required for bone destruction and enhanced tumor growth in the bone microenvironment (32). Thus, decreased osteoclast activation and colonization of breast cancer cells lacking ABL kinases may be mediated in part by reduced MMP1 abundance.

Mechanistically, we found that STAT5 was required for the production of the secreted factors MMP1, IL-6, and TNC, downstream of ABL kinases. Similar to IL-6 and MMP1, TNC is abundant in some breast tumors and promotes metastasis in mouse models (49). We showed that inactivation of the ABL kinases in breast cancer cells resulted in decreased expression of genes in the JAK/STAT and cytokine/cytokine receptor pathway signatures, which may be due to decreased STAT5A mRNA expression and reduced STAT5 phosphorylation in ABL1/ABL2-depleted breast cancer cells. Expression of a constitutively active version of STAT5A in ABL1/ABL2 knockdown cells restored the production of secreted factors (IL-6, MMP1, and TNC) and partially rescued the ability of breast cancer cells to promote bone metastasis. These findings support a role for the ABL kinase–STAT5 signaling axis in breast cancer metastasis.

In addition to inhibiting STAT5 signaling, we found that depletion of ABL kinases decreased the expression of the Hippo pathway mediator TAZ and downstream target genes in triple-negative and HER2+ breast cancer cells. Inactivation of ABL kinases inhibited the expression of the TAZ target gene AXL, which shows increased expression in several human cancers and correlates with poor prognosis, increased invasiveness and metastasis, and enhanced drug resistance (53, 54). AXL mediates resistance to TRAIL-induced cell death in esophageal adenocarcinoma (55); it remains to be determined whether ABL-mediated protection from TRAIL-induced apoptosis in breast cancer cells is mediated by AXL or other targets. Coexpression of both activated TAZ and STAT5 pathways was required to fully rescue the bone metastatic phenotype of ABL1/ABL2 knockdown breast cancer cells. These data, together with the finding that knockdown of both TAZ and STAT5 phenocopied the decrease in breast cancer cell metastasis caused by ABL1/ABL2 depletion, supports a model for a requirement of ABL-dependent TAZ and STAT5 signaling networks in promoting breast cancer metastasis (Fig. 8B).

Our data raise the possibility that inhibition of ABL kinases can increase apoptosis of breast cancer cells and block osteoclast activation, which is required for osteolytic metastasis. Several phase 1 and 2 trials have been carried out using nonselective ATP-competitive inhibitors such as imatinib to treat patients with advanced breast cancer. However, patients enrolled in these trials were not evaluated for ABL abundance and/or activation (56, 57). The allosteric inhibitors specific for ABL kinases (which are currently in clinical trials) provide a potentially useful tool for selectively targeting ABL kinases in metastatic breast cancer types with an increase in the ABL pathway signature (58). We found that allosteric inhibition of the ABL kinases effectively impaired breast cancer bone metastasis and blocked tumor-induced osteolysis in mouse models. Future studies will test and compare the efficacy of imatinib and allosteric compounds in mouse models of breast cancer. We will also test whether ABL-directed therapy can shrink established bone metastasis in the mouse models. Together, our data suggest that clinical studies may be warranted to evaluate the therapeutic potential of ABL allosteric inhibitors and to determine whether combination therapies that incorporate these compounds are effective in treating metastatic breast cancer. It is also striking that multiple downstream targets of the ABL kinases in breast cancer cells (TAZ, AXL, STAT5, and IL-6) have been implicated in therapy resistance. It remains to be determined whether activated ABL kinases play a role in the development of intrinsic and/or acquired resistance in breast cancer cells.

MATERIALS AND METHODS

Cell Culture

The human breast carcinoma cell line MDA-MB-231 was purchased from the American Type Culture Collection (ATCC). The 1833 (bone metastasis), 4175 (lung metastasis), and BrM2a (brain metastasis) sublines were derived from the parental cell line MDA-MB-231 (19) and were gifts from J. Massague (Memorial Sloan-Kettering Cancer Center). The SCP28 (bone metastasis) subline was provided by Y. Kang (Princeton University). MDA-MB-231, their derivative sublines, and genetically modified versions were maintained in Dulbecco’s modified Eagle’s medium (DMEM; Life Technologies) supplemented with 10% fetal bovine serum (FBS; Life Technologies), antibiotics, and appropriate selection drugs for transfected plasmids. Human embryonic kidney (HEK) 293T cells, a packaging cell line for lentivirus production, and the preosteoclast cell line RAW264.7 (ATCC) were maintained in DMEM supplemented with 10% FBS and antibiotics. The murine osteoblast cell line 7F2 (ATCC) was cultured in α-MEM with 10% FBS. The human mammary epithelial cell line (HuMEC) was maintained in human basal serum-free medium (Life Technologies) with the HuMEC kit (Life Technologies). The human breast cancer cell line SKBR3 was purchased from the Duke University Cell Culture Facility and was maintained in McCoy’s 5A medium (Life Technologies) supplemented with 10% FBS (Life Technologies) and antibiotics. All cultures were maintained at 37°C in humidified air containing 5% CO2.

Antibodies

Antibodies used for Western blotting included cleaved caspase-3, phosphorylated CrkL (Y207), phosphorylated Akt (Ser473), Akt, TAZ, YAP1, phosphorylated STAT5 (Tyr694), STAT5, and ERBB2 from Cell Signaling; β-tubulin and actin from Sigma-Aldrich; ABL2 (9H5) from Santa Cruz Biotechnology; ABL1 (8E9) from BD Biosciences; IL-6, TNC, and phosphorylated YAP1 (Tyr357) from Abcam; and MMP1 from Calbiochem. Antibodies used for immunofluorescence staining included YAP1 from Cell Signaling and TAZ from BD Biosciences. Antibodies used for the ChIP assays were TAZ (V386) and YAP1 (D8H1X) from Cell Signaling.

Tumor xenografts and analysis

Procedures involving mice were approved and performed following the guidelines of the Institutional Animal Care and Use Committee (IACUC) of Duke University Division of Laboratory Animal Resources (DLAR). Age-matched female athymic NCr nu/nu mice (5 to 6 weeks old) were used for xenograft experiments. For intracardiac injections, cells were harvested from subconfluent culture plates, washed with phosphate-buffered saline (PBS), and resuspended at 106/ml (1833) or 5 × 106/ml (SCP28) in PBS; 0.1 ml of the suspended cells was injected into the left cardiac ventricle using 30-gauge needles. Mice were anesthetized with isoflurane before injection and imaged by bioluminescence imaging. For intratibial injections, mice were anesthetized using a mixture of ketamine (100 mg/kg) and xylazine (10 mg/kg). The injection site was cleaned with a 70% alcohol wipe. Single-cell suspensions (1 × 105 cells) in a final volume of 10 μl were injected into the upper half of the tibia medullary cavity, as felt by a lack of resistance when pushing cells into the cavity. Bioluminescence imaging was used to confirm successful cancer cell inoculation and progression of metastatic bone lesions. The allosteric inhibitor GNF5 was synthesized by the Duke University Small Molecule Synthesis Facility. For drug treatment, the mice were dosed daily with GNF5 in DMSO/peanut oil (1:9) at 50 mg/kg by intraperitoneal injection.

Bone μCT analysis

Hindlimb bones were excised, fixed in 10% neutral-buffered formalin, and imaged using a μCT scanner (SkyScan 1176, Bruker Corp.) at 17-μm resolution and 180° scanning with a rotation step of 0.7° per image, 242-ms exposure time, 55-kV photon energy, and 455-μA current. The images were reconstructed using NReconServer, and bone volume was analyzed by CT analysis software (CTAn, Bruker Corp.).

In vitro osteoclastogenesis assay

Bone marrow cells were flushed out from the femora and tibia of 6-week-old C57BL/6 mice and plated in basal culture medium (α-MEM supplemented with 10% FBS and antibiotics) overnight. Osteoclastogenesis assay and TRAP staining were conducted as described previously (32). Tumor cells were plated at 2 × 105 per well in 12-well plates to obtain conditioned medium for incubation with either bone marrow cells or RAW264.7 cells. RAW264.7 preosteoclasts were plated at 4 × 105 per well in 24-well plates overnight. RAW264.7 medium was replaced by conditioned medium harvested from tumor cells and supplemented with recombinant murine sRANKL (50 ng/ml). The medium was changed every 3 days, and TRAP staining was performed on day 6 following the manufacturer’s instructions (Sigma).

RNAseq analysis

For RNAseq analysis, 3 × 106 breast cancer cells were plated in a 10-cm petri dish in triplicate in complete medium for 24 hours. Cells were harvested, and RNA was isolated using the RNeasy kit (Qiagen); 1 μg of total RNA input was used for each sample. The libraries were sequenced on an Illumina HiSeq 2000 sequencing system using 50–base pair single-ended reads. RNAseq data were mapped to reference genome (HG19) using Bowtie/TopHat. The reads were counted, and the differential expression between distinct experimental groups was quantified using Cuffdiff. Significant genes were extracted using R CummeRbund.

Viral transduction

HEK293T cells were transfected with the following retroviral or lentiviral constructs, with their packaging vectors indicated in the parentheses, using FuGENE 6 reagent (Promega): pMX-puro-STAT5A* (provided by T. Kitamura, University of Tokyo; pCMV-Gag-Pol and pCMV-VSV-G); pLenti-EF-FH-TAZ-S89A (psPAX2 and pVSV-G); PLKO-NS and shTAZ lentiviral construct (provided by C. Linardic, Duke University; pCMV-Rev, pCVM-VSVG; pMDL). Retroviral or lentiviral supernatants were collected and filtered 24 and 48 hours after transfection. 1833 cells were incubated 48 hours with retroviral or lentiviral medium in the presence of Polybrene (8 μg/ml). Cells were cultured at least 3 days in puromycin (3 μg/ml) or blasticidin (5 μg/ml) for selection. Lentiviral shRNA–mediated knockdown of ABL1/ABL2 and expression of mouse Abl1/Abl2 were conducted as described previously (20, 21, 59). Lentiviral shRNA–mediated knockdown of STAT5A and STAT5B was conducted following the manufacturer’s instructions (Dharmacon RHS4531-EG6776, EG6777).

Immunoblotting

Cells were lysed in radioimmunoprecipitation assay buffer with protease and phosphatase inhibitors. Cell debris was removed by microcentrifugation, and protein was quantified. Alternatively, equal cell numbers were plated onto six-well plates, and 24 hours later, conditioned medium was harvested and concentrated using Amicon Ultra Centrifugal Filters (Millipore). Equal amounts of protein or conditioned medium were separated by SDS–polyacrylamide gel electrophoresis and transferred onto nitrocellulose membranes and probed with the indicated antibodies.

Real-time RT-PCR

RNA was isolated from cancer cells using the RNeasy RNA isolation kit (Qiagen), and complementary DNA was synthesized using oligo(dT) primers and Moloney murine leukemia virus reverse transcriptase (Invitrogen). Real-time PCR was performed using iQ SYBR Green Supermix (Bio-Rad). The primers used were as follows: mouse RANKL, 5′-TTGCACACCTCACCATCA (forward) and 5′-TACGCTTCCCGATGTTTC (reverse); mouse OPG, 5′-CACTCGAACCTCACCACA (forward) and 5′-CAAGTGCTTGAGGGCATA (reverse); mouse GAPDH, 5′-CTCATGACCACAGTCCATGC (forward) and 5′-ACACATTGGGGGTAGGAACA (reverse); human ABL1, 5′-GGCTGTGAGTACCTTGCTGC (forward) and 5′-GGCGCTCATCTTCATTCAGGC (reverse); human ABL2, 5′-CCAGCTACTCCCGAGGCTG (forward) and 5′-CTTGATCCCACAGGGTGAAG (reverse); human GAPDH, 5′-GGCTCTCCAGAACATCATCCC (forward) and 5′-GGGTGTCGCTGTTGAAGTCAG (reverse); human MMP1, 5′-GGTCTCTGAGGGTCAAGCAG (forward) and 5′-AGTTCATGAGCAACACG (reverse); human IL6, 5′-AGACAGCCACTCACCTCTTC (forward) and 5′-TTTCACCAGGCAAGTCTCCT (reverse); human TNC, 5′-CCCTACGGGTTCACAGTTTC (forward) and 5′-TTCCGGTTCGGCTTCTGTAAC (reverse); human AXL, 5′-ATCAGACCTTCGTGTCCCAG (forward) and 5′-ATGTCTTGTTCAGCCCTGGA (reverse); human WWTR1, 5′-GGCTGGGAGATGACCTTCAC (forward) and 5′-AGGCACTGGTGTGGAACTGAC (reverse); human YAP1, 5′-ATGAACTCGGCTTCAGCCAT (forward) and 5′-ACCATCCTGCTCCAGTGTTG (reverse). Analysis was performed using a Bio-Rad CFX384 real-time machine and CFX Manager software. PCR assays were performed in triplicate. The expression of each gene was normalized to that of the GAPDH gene.

Histological analysis

Hindlimb bones were excised, fixed in 10% neutral-buffered formalin, decalcified, and embedded in paraffin for H&E staining (28).

TUNEL staining

For TUNEL staining, 2 × 104 cells were seeded onto each single chamber of a four-chamber slide in complete medium. The next day, the medium was replaced with serum-free medium containing TRAIL (2 ng/ml). After 3 days, cells were fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton X-100. TUNEL staining was performed following the manufacturer’s protocol (Roche Applied Science).

ELISA and cytokine array

Conditioned medium was collected, after 24 hours of incubation, from confluent cells and was applied to the Human Cytokine Antibody Array C1000 (RayBiotech) or the IL-6 ELISA kit (R&D) following the manufacturer’s instructions.

Invasion assay

Invasion was evaluated by plating 25,000 cells in the upper chambers of Matrigel chambers (8.0-μm pore size; BD Biosciences) in serum-free medium. Cells were allowed to invade for up to 48 hours in the presence of serum-containing medium in the bottom chamber. Afterward, the remaining cells, medium, and Matrigel were removed from the upper chambers, and cells on the undersurface of the membrane were fixed, stained with Diff-Quik (Dade Behring), and quantified by microscopy.

In vitro cell growth assays

For 2D cell growth, 3000 cells were seeded onto each well of a 96-well plate. Cell growth was measured daily from days 1 to 5 using CellTiter-Glo (Promega) following the manufacturer’s protocol. For 3D cell growth, 75 μl of Matrigel (BD Bioscience) was plated onto each well of a 96-well plate. A single-cell suspension (50 μl) containing 1500 cells was mixed with Matrigel (1:1) and plated on top of the Matrigel base onto wells of a 96-well plate; 50 μl of complete medium was added, and the cells were cultured for 14 days. Colonies were analyzed and counted using a microscope.

ChIP-qPCR analysis

CHIP–quantitative real-time fluorescence PCR (qPCR) was performed using Cell Signaling SimpleChIP Plus Enzymatic Chromatin IP Kit (#9005) according to the manufacturer’s instructions. SimpleChIP human CTGF promoter primers were from Cell Signaling (#14927). Validated TAZ/YAP primers, as well as previously validated primers for AJUBA, AMOTL2, and WTIP were used for the qPCR analysis (42). Antibodies used for ChIP assays were anti-TAZ (V386) and anti-YAP1 (D8H1X) from Cell Signaling. The primers used were as follows: AXL forward, 5′-CAGCCTCCTCCTCACAGACA; AXL reverse, 5′-GAGCCCTGATCATTCCACTG; AJUBA forward, 5′-AAGGAAAGAGTGTGGGGGTAGG; AJUBA reverse, 5′-ACGCTGGGAACAAAGTCACG; AMOTL2 forward, 5′-TGCCAGGAATGTGAGAGTTTC; AMOTL2 reverse, 5′-AGGAGGGAGCGGGAGAAG; WTIP forward, 5′-GCAGCGCCGTCTCCTTCCT; WTIP reverse, 5′-GCGGCGGAGGAATGTAAGCTC.

Mutagenesis

Abl2 E505K mutagenesis was conducted on pBabe-puro-mAbl2 construct using the Q5 Site-Directed Mutagenesis Kit (NEB, E0554S) according to the manufacturer’s instructions. The primers used were as follows: Abl2 forward, 5′-CATCTCTGAAAAGGTAGCTCAG; Abl2 reverse, 5′-CTGGAGTCATGGAACATTG.

Statistical analysis

Statistical analyses were performed using GraphPad Prism 6, JMP Pro, and R 3.2. Comparisons of two groups were performed using Student’s t tests (two-tailed). Comparisons involving multiple groups were evaluated using one-way or two-way ANOVA, followed by Tukey’s post hoc test. For all tests, P < 0.05 was considered statistically significant. For all figures, the P value was calculated using Student’s t test unless otherwise indicated. Data shown represent averages ± SEM unless otherwise indicated.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/9/413/ra12/DC1

Fig. S1. ABL family protein abundance is increased in breast cancer cells with enhanced bone metastatic activity, and ABL1/ABL2 depletion does not affect proliferation in vitro.

Fig. S2. ABL kinases promote breast cancer cell invasion.

Fig. S3. Depletion of ABL kinases does not inhibit metastasis of 4175 breast cancer cells, which show tropism to the lung.

Fig. S4. Treatment of breast cancer cells with imatinib but not GNF5 promotes ERK activation.

Fig. S5. CXCL12- and IGF-1–mediated survival pathways are independent of ABL kinases.

Fig. S6. Depletion of ABL kinases in SKBR3 breast cancer cells decreases tumor-induced osteoclast activation.

Fig. S7. IL-6 affects RANKL and OPG expression in osteoblasts.

Fig. S8. Quality control and global statistics of RNAseq analysis for transcriptome comparison of control versus ABL1/ABL2 knockdown breast cancer cells.

Fig. S9. ABL kinases increase TAZ protein abundance and STAT5 phosphorylation.

Fig. S10. Depletion of ABL kinases reduces the abundance of TAZ in the nucleus.

Fig. S11. Allosteric inhibition of ABL kinase activity decreases TAZ protein abundance.

Fig. S12. ABL2 mRNA expression positively correlates with TAZ mRNA expression in invasive breast cancer patients.

Fig. S13. Depletion of ABL kinases decreases the binding of TAZ to target genes.

Fig. S14. Depletion of ABL kinases does not affect YAP1 protein abundance, localization, or tyrosine phosphorylation in breast cancer cells.

Fig. S15. Expression of a constitutively active STAT5 mutant increases mRNA expression of MMP1, IL6, and TNC.

Fig. S16. Potential interdependence of TAZ and STAT5 in breast cancer cells.

Table S1. Differentially expressed genes in control and ABL1/ABL2 knockdown breast cancer cells.

REFERENCES AND NOTES

Acknowledgments: We thank J. Massague (Memorial Sloan-Kettering Cancer Center) and Y. Kang (Princeton University) for the breast cancer cell lines, and T. Kitamura (University of Tokyo) and C. Linardic (Duke University) for the viral vectors. We are grateful to D. P. McDonnell (Duke University School of Medicine) for support with the analysis of breast cancer data sets. We thank T. Slotkin (Duke University School of Medicine) for support with the statistical analysis. We thank F. Guilack (Duke University School of Medicine), E. Riggs (Duke University School of Medicine), the Duke University Light Microscopy Core Facility, and the Duke Cancer Center Flow Cytometry Shared Resource for technical assistance. We thank the Duke University Department of Pathology for the evaluation of histology specimens. We thank P. Smith-Pearson for supplemental data support. Funding: This work was supported by NIH grants R01 CA155160 (to A.M.P.) and CA174643 (to D.P.M.) and the National Center for Advancing Translational Sciences of the NIH under award number UL1TR001117 (to A.M.P.). Author contributions: J. W. and A.M.P. conceived the study, generated the hypotheses, designed the experiments, and analyzed the data. J.W. performed all the experiments. C.R. provided technical expertise and contributed to the mouse experiments. J.S.J. carried out a meta-analysis of previously published microarray data of breast cancer patients. J.W. and A.M.P. wrote the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: The Gene Expression Omnibus accession number for the RNAseq data reported in this paper is GSE69125.
View Abstract

Navigate This Article