Research ArticleCANCER STEM CELLS

VEGF–neuropilin-2 signaling promotes stem-like traits in breast cancer cells by TAZ-mediated repression of the Rac GAP β2-chimaerin

See allHide authors and affiliations

Science Signaling  01 May 2018:
Vol. 11, Issue 528, eaao6897
DOI: 10.1126/scisignal.aao6897

A Rac-TAZ loop drives cancer stem cells

In various cancers, stem-like cells called CSCs may drive tumor growth, metastasis, recurrence, and drug resistance; thus, blocking the survival and proliferation of CSCs may enhance the efficacy of anticancer therapies in patients. Elaimy et al. found that CSC phenotypes in breast cancer cells are supported by VEGF, a growth factor that is associated with poor prognosis in breast cancer patients. Through its receptor NRP2 and subsequent activation of the kinase FAK and the protein Rac1, VEGF activated the Hippo pathway transcription cofactor TAZ, which bound and repressed the promoter of the gene encoding the enzyme β2-chimaerin. β2-chimaerin effectively shuts off Rac1; thus, its loss promoted sustained Rac1 activity in a self-perpetuating loop that promoted CSC-like behavior in cells and correlated with stem-like and metastatic markers in patient tumors.

Abstract

The role of vascular endothelial growth factor (VEGF) signaling in cancer is not only well known in the context of angiogenesis but also important in the functional regulation of tumor cells. Autocrine VEGF signaling mediated by its co-receptors called neuropilins (NRPs) appears to be essential for sustaining the proliferation and survival of cancer stem cells (CSCs), which are implicated in mediating tumor growth, progression, and drug resistance. Therefore, understanding the mechanisms involved in VEGF-mediated support of CSCs is critical to successfully treating cancer patients. The expression of the Hippo effector TAZ is associated with breast CSCs and confers stem cell–like properties. We found that VEGF-NRP2 signaling contributed to the activation of TAZ in various breast cancer cells, which mediated a positive feedback loop that promoted mammosphere formation. VEGF-NRP2 signaling activated the GTPase Rac1, which inhibited the Hippo kinase LATS, thus leading to TAZ activity. In a complex with the transcription factor TEAD, TAZ then bound and repressed the promoter of the gene encoding the Rac GTPase-activating protein (Rac GAP) β2-chimaerin. By activating GTP hydrolysis, Rac GAPs effectively turn off Rac signaling; hence, the TAZ-mediated repression of β2-chimaerin resulted in sustained Rac1 activity in CSCs. Depletion of β2-chimaerin in non-CSCs increased Rac1 activity, TAZ abundance, and mammosphere formation. Analysis of a breast cancer patient database revealed an inverse correlation between β2-chimaerin and TAZ expression in tumors. Our findings highlight an unexpected role for β2-chimaerin in a feed-forward loop of TAZ activation and the acquisition of CSC properties.

INTRODUCTION

Vascular endothelial growth factor (VEGF) was originally characterized as a protein that promotes endothelial growth (1) and increases vascular permeability (2). For these and other reasons, it was presumed that the role of VEGF in cancer was limited to angiogenesis (1, 35). It is evident now, however, that there are angiogenesis-independent functions of VEGF in cancer that are mediated by specific receptors. Tumor cells express VEGF receptor tyrosine kinases (VEGFR1 and VEGFR2) and neuropilins (NRPs), another family of VEGF receptors. NRP1 and NRP2 were identified initially as neuronal receptors for semaphorins, which are axon guidance factors that function primarily in the developing nervous system (6). The finding that NRPs can also function as VEGF receptors and that they are expressed on endothelial and tumor cells launched studies aimed at understanding their contribution to angiogenesis and tumor biology (7). NRPs have the ability to interact with and modulate the function of VEGFR1 and VEGFR2, as well as other receptors (810). There is also evidence that NRPs are valid targets for therapeutic inhibition of angiogenesis and cancer (1114).

A surge of evidence has implicated autocrine VEGF signaling mediated by NRPs in the function of cancer stem cells (CSCs), a subpopulation of cells that function in tumor initiation, the differentiation of multilineage cancer cell hierarchies, therapy resistance, and metastasis (12, 1521). These observations have led to intense investigation into the mechanisms by which VEGF sustains CSCs and how these processes can be exploited therapeutically. Previously, we reported that NRP2 is highly expressed in breast CSCs and that VEGF-NRP2 signaling contributes to breast tumor initiation (22). A key issue that emerges from these findings is the mechanism by which VEGF-NRP2 signaling contributes to the function of CSCs. In pursuit of this issue, we were intrigued by reports that the Hippo pathway transducer TAZ confers stem cell properties and contributes to breast tumorigenesis, especially in high-grade tumors, which are distinguished by high NRP2 expression and VEGF-NRP2 signaling activity (22). Moreover, TAZ expression in breast cancer correlates with tumor grade (23), and high-grade tumors harbor a higher frequency of CSCs than do lower-grade tumors (24). Mechanistic studies have shown that TAZ can induce an epithelial to mesenchymal transition (EMT) in mammary epithelial cells (25), a process that can increase stem cell properties (26). Moreover, TAZ is necessary for the self-renewal of CSCs (23). In contrast, the role of another Hippo pathway effector YAP in breast cancer is less clear, and its expression does not correlate with clinical outcome in breast cancer patients (27).

The Hippo pathway consists of core kinases and regulatory molecules that facilitate TAZ phosphorylation, cytoplasmic retention, and inactivation (2830). For this reason, identifying upstream receptors that disrupt Hippo signaling and, consequently, enhance TAZ activity is critical for understanding how these effectors contribute to the function of CSCs. Here, we discovered that VEGF-NRP2 signaling contributes to increased TAZ activity by a Rac1-dependent, feed-forward mechanism.

RESULTS

VEGF-NRP2 signaling contributes to TAZ activation

Initially, we assessed the contribution of VEGF-NRP2 signaling to TAZ activation. For this purpose, we used an inducible system to transform MCF10A cells with Src, which generates a population of CD44high/CD24low cells with CSC properties (31). This population is actually composed of distinct epithelial and mesenchymal populations (which we refer to herein as EPTH and MES, respectively) that differ in TAZ activity and tumor-initiating potential. The MES population has enhanced TAZ activity, self-renewal potential, and tumor-initiating capability compared to the EPTH population (32). MES cells express increased amounts of VEGF and NRP2 compared to EPTH cells and are dependent on VEGF-NRP2 signaling for self-renewal (33). Expression of either NRP2 (Fig. 1A) or VEGF (Fig. 1B) was diminished in MES cells using short hairpin RNAs (shRNAs), which resulted in decreased TAZ abundance compared to control cells, as assessed by immunoblotting (Fig. 1, A and B). As reported previously, TAZ abundance is an indicator of its activation status (34, 35), as is its nuclear localization. For this latter reason, we compared TAZ localization in control cells to VEGF- and NRP2-depleted cells by immunofluorescence (Fig. 1C and fig. S1A). TAZ was localized primarily in the nucleus in control MES cells (Fig. 1C and fig. S1A). In contrast, little, if any, TAZ was detected in VEGF- and NRP2-depleted cells, which is consistent with our immunoblotting data (Fig. 1, A and B). The TAZ that was detected in VEGF-depleted cells was localized in the cytoplasm (Fig. 1C and fig. S1A).

Fig. 1 VEGF-NRP2 signaling contributes to TAZ activation.

Expression of NRP2 (A) and VEGF (B) was diminished in MES cells, and the impact on TAZ abundance was quantified by immunoblotting. Representative blots are shown; data are means ± SEM of three biological replicates. shGFP, control shRNA. (C) TAZ localization (cytoplasm, nucleus/cytoplasm, and nucleus) in control and VEGF-depleted MES cells was determined by immunofluorescence confocal microscopy. Data are means of three biological replicates. (D) mRNA expression of the indicated TAZ target genes was quantified by quantitative polymerase chain reaction (qPCR) in NRP2-depleted MES cells. Data are means ± SEM of three biological replicates. (E) MES cells were treated with the indicated concentrations of a function-blocking NRP2 antibody for 6 hours, and the impact on TAZ abundance was quantified by immunoblotting. Representative blots are shown; data are means ± SEM of three biological replicates of 3 μg/ml condition. IgG, immunoglobulin G. (F) Immunoblotting analysis of TAZ abundance in MDA-MB-231 cells in which NRP2 was depleted by one of two shRNAs. Representative blots are shown; data are means ± SEM of three biological replicates. (G) mRNA expression of the indicated TAZ target genes was quantified by qPCR in NRP2-depleted MDA-MB-231 cells. Data are means ± SEM of three biological replicates. (H) NRP2-depleted MDA-MB-231 cells were transfected with an 8xGTIIC-luciferase reporter construct and assayed for TEAD transcriptional activity. Data are means ± SEM of three biological replicates. *P ≤ 0.05 by two-tailed t test.

Depletion of NRP2 in MES cells also reduced the mRNA expression of the TAZ target genes CTGF and CYR61 (Fig. 1D). To substantiate the data obtained with shRNAs, we treated MES cells with a function-blocking NRP2 antibody (11) and observed a concentration-dependent decrease in TAZ abundance (Fig. 1E). Although similar results were observed for YAP (fig. S1, B and C), we focused subsequent experiments on TAZ because convincing data correlating YAP expression and clinical parameters in breast cancer are lacking (27).

We extended this analysis to MDA-MB-231 cells because they exhibit mesenchymal properties and highly express VEGF and NRP2 (33). Similar to MES cells, NRP2 depletion reduced the abundance of TAZ, as well as TAZ target genes (Fig. 1, F and G), suggesting that NRP2 affects TAZ-mediated transcription. TAZ regulates gene expression by associating with the TEAD family of transcription factors (36), which infers that NRP2 should affect TEAD transcriptional activity. The activity of a TEAD luciferase reporter was reduced significantly in MDA-MB-231 cells with diminished NRP2 expression compared to control cells (Fig. 1H).

Rac1 facilitates VEGF-NRP2–mediated activation of TAZ

To investigate the mechanism by which VEGF-NRP2 signaling activates TAZ, we focused on Rac1 for several reasons. This guanosine triphosphatase (GTPase) is a major effector of NRP/plexin signaling in neurons (37, 38), and it has been implicated in VEGF signaling in endothelial cells (39, 40). Moreover, Rac1 has also been implicated in TAZ activation (35, 4143). We observed that depleting VEGF expression or treating MES cells with the NRP2 function-blocking antibody resulted in a substantial decrease in Rac1 activity (Fig. 2, A and B). Similar results were obtained in MDA-MB-231 cells (Fig. 2C). Conversely, stimulating MDA-MB-231 cells with VEGF resulted in an increase in Rac1 activity (Fig. 2D).

Fig. 2 VEGF-NRP2 signaling activates Rac1.

(A) Expression of VEGF was diminished in MES cells, and the impact on Rac1 activity was assessed using a glutathione S-transferase (GST) fusion protein containing the Rac/Cdc42-binding domain of p21-activated kinase (PAK) (PBD). Representative blots are shown; data are means ± SEM of three biological replicates. (B) Rac1 activity in MES cells treated with a function-blocking NRP2 antibody (3 μg/ml for 6 hours). Representative blots are shown; data are means ± SEM of three biological replicates. (C) NRP2 expression was diminished in MDA-MB-231 cells, and the impact on Rac1 activity was assessed. Representative blots are shown; data are means ± SEM of three biological replicates. (D) MDA-MB-231 cells were serum-starved for 24 hours, treated with VEGF (50 ng/ml for 30 min), and assayed for Rac1 activity. Representative blots are shown; data are means ± SEM of three biological replicates. (E) MES cells were treated with FAK14 (2 μM) for the indicated time points and assayed for Rac1 activity. Representative blots are shown; data are means ± SEM of three biological replicates of the 6-hour condition. DMSO, dimethyl sulfoxide. (F) NRP2 expression was diminished in MDA-MB-231 cells that were then transfected with constitutively active Rac1 (V12 Rac1-GST). GST abundance was quantified by immunoblotting (blots, left), and mRNA expression of the indicated TAZ target genes was quantified by qPCR (graph, right). Data are means ± SEM of three biological replicates. (G) MES cells were treated with Rac inhibitor (50 μM EHT1864 or NSC23766) and assayed for self-renewal by serial passage mammosphere formation. P1, passage 1; P2, passage 2. Data are means ± SEM of three biological replicates. *P ≤ 0.05 by two-tailed t test.

An important issue is whether VEGFRs contribute to Rac1 activation in breast cancer cells. Treatment of MDA-MB-231 cells with the VEGFR inhibitors pazopanib and sunitinib in the presence of VEGF did not decrease Rac1 activity compared to VEGF alone (fig. S2), suggesting that VEGF-NRP2–mediated activation of Rac1 is VEGFR-independent. This result is not surprising because we reported previously that MDA-MB-231 and other breast cancer cell lines express very low amounts of VEGFRs (44). In contrast, VEGF-NRP2–mediated Rac1 activation appears to be focal adhesion kinase (FAK)–dependent, because the FAK inhibitor FAK14 significantly decreased Rac1 activity in MES cells (Fig. 2E). This result is consistent with previous findings that FAK is a downstream effector of VEGF-NRP2 signaling (22), and it has been implicated in Rac1 activation (4547), as well as Hippo pathway regulation (48, 49).

The results described above prompted us to evaluate whether VEGF-NRP2 signals through Rac1 to promote TAZ activation. On the basis of our finding that depletion of NRP2 in MDA-MB-231 cells decreased the expression of TAZ target genes (Fig. 1G), we found that expression of a constitutively active Rac1 (Rac1-V12) in these cells rescued their expression (Fig. 2F). This result provides evidence that the VEGF-NRP2-Rac1 axis contributes to TAZ activation. Subsequently, we assessed the role of Rac1 inhibition on mammosphere formation in MES cells and observed decreased mammosphere formation upon treatment with the Rac inhibitors EHT1864 and NSC23766 (Fig. 2G).

We next sought to investigate the mechanism by which VEGF-NRP2–mediated regulation of Rac1 contributes to TAZ activation. We focused on the LATS tumor suppressor kinases because they phosphorylate TAZ directly at the Ser89 position and promote its cytoplasmic retention and degradation when phosphorylated and activated on their hydrophobic motifs (Thr1079 in LATS1 and Thr1041 in LATS2) (25, 50, 51). Moreover, LATS can be regulated by Rac1 (35, 41, 42). Treatment of MES cells with the Rac inhibitor EHT1864 resulted in a concentration-dependent increase in the abundance of phosphorylated Ser89 (pSer89) TAZ and a concomitant decrease in TAZ abundance (Fig. 3A). It also increased the abundance of pThr1079 LATS1 and decreased TAZ abundance as early as 15 min after treatment, and this pattern persisted for up to 6 hours (Fig. 3B). Expression of dominant-negative N17 Rac1 also decreased TAZ abundance in MES cells (Fig. 3C). Similar results were obtained with the NRP2 function-blocking antibody (Fig. 3D). NRP2 knockdown in MDA-MB-231 cells also increased the abundance of pSer89 TAZ and pThr1079 LATS1 and decreased the abundance of TAZ (Fig. 3, E and F). NRP2 depletion also increased LATS-mediated YAP phosphorylation (pSer127) (fig. S1B). LATS knockdown rescued TAZ abundance in NRP2-depleted MDA-MB-231 cells, which shows that VEGF-NRP2-Rac1 regulation of TAZ is LATS-dependent (fig. S3A).

Fig. 3 Rac1 facilitates VEGF-NRP2 activation of TAZ through inhibition of LATS.

(A) MES cells were treated with the indicated concentrations of the Rac inhibitor EHT1864 for 2 hours, and the impact on pSer89 TAZ and TAZ abundance was quantified by immunoblotting. Representative blots are shown; data are means ± SEM of three biological replicates. (B) MES cells were treated with the Rac inhibitor EHT1864 (100 μM) and lysed at the indicated time points, and the impact on pThr1079 LATS1, LATS1, and TAZ abundance was quantified by immunoblotting. Blots are representative of two biological replicates. (C) MES cells were transfected with dominant-negative N17 Rac1-HA (hemagglutinin), and the impact on TAZ abundance was quantified by immunoblotting. Blots are representative of two biological replicates. (D) MES cells were treated with the indicated concentrations of a function-blocking NRP2 antibody for 6 hours, and the impact on pSer89 TAZ and TAZ abundance was quantified by immunoblotting. Representative blots are shown; data are means ± SEM of three biological replicates of the 3 μg/ml condition. (E) NRP2 expression was diminished in MDA-MB-231 cells, and the impact on pSer89 TAZ and TAZ abundance was quantified by immunoblotting. Representative blots are shown; data are means ± SEM of three biological replicates. (F) NRP2 expression was diminished in MDA-MB-231 cells, and the impact on pThr1079 LATS1 and LATS1 was quantified by immunoblotting. Blots are representative of two biological replicates. (G) NRP2-depleted MDA-MB-231 cells were transfected with empty vector, wild-type (WT) TAZ, or S89A TAZ and assayed for self-renewal by serial passage mammosphere formation. Data are means ± SEM of three biological replicates. (H) VEGF-depleted MES cells were transfected with empty vector, WT TAZ, or S89A TAZ and assayed for self-renewal by serial passage mammosphere formation. Data are means ± SD from three technical replicates, representative of two biological replicates. *P ≤ 0.05 by two-tailed t test.

Given the importance of TAZ in promoting CSC properties (23), we hypothesized that VEGF-NRP2-Rac1–mediated LATS inhibition is a critical upstream regulator of TAZ-mediated mammosphere formation. Depletion of NRP2 in MDA-MB-231 cells significantly reduced mammosphere formation, which was partially rescued by expression of wild-type TAZ (Fig. 3G and fig. S4). Expression of S89A TAZ, which is resistant to LATS-mediated phosphorylation at that site, rescued mammosphere formation significantly more than did expression of wild-type TAZ (Fig. 3G and fig. S4). Similar results were obtained by VEGF depletion in MES cells (Fig. 3H). These results indicate that the wild-type TAZ ectopically expressed is subject to regulation by upstream VEGF-NRP2 signaling, but that the S89A TAZ mutant is not. Together, these data provide functional evidence that VEGF-NRP2-Rac1 promotes a TAZ-dependent stem-like phenotype through inhibition of LATS-mediated phosphorylation of TAZ at Ser89.

To gain insight into the mechanism by which VEGF-NRP2-Rac1 signaling inhibits LATS activity, we postulated that this signaling regulates Merlin, the protein product of the neurofibromatosis type 2 (NF2) gene, because phosphorylation of Merlin on Ser518 by PAK is inhibitory (5254). Moreover, Merlin phosphorylation inhibits LATS phosphorylation (55). These findings are relevant because PAK is a Rac-activated kinase (56). Following these observations, we found that either Rac inhibition or NRP2 depletion in MES cells reduced the abundance of Ser518-phosphorylated Merlin (fig. S3, B and C).

VEGF-NRP2 signaling represses the Rac GAP β2-chimaerin

Rac1 cycles from guanosine triphosphate (GTP)–bound active states to guanosine diphosphate (GDP)–bound inactive states, which, in large part, is regulated by the expression of guanine nucleotide exchange factors (GEFs) and GTPase-activating proteins (GAPs). Therefore, we profiled the expression of known Rac GEFs and GAPs in EPTH and MES cells (Table 1). Notably, we observed that the expression of the Rac GAP β2-chimaerin at the mRNA level (CHN2, hereafter β2-chimaerin) was markedly reduced in MES cells compared to EPTH cells (Table 1), which we verified by immunoblotting (Fig. 4A). β2-chimaerin is a Rac-specific GAP that has been implicated as a tumor suppressor in breast cancer (5761). Given that β2-chimaerin abundance is reduced in MES cells and these cells highly express VEGF and NRP2 (33), we assessed whether VEGF-NRP2 signaling repressed β2-chimaerin expression. We observed that NRP2 depletion increased β2-chimaerin abundance at both the mRNA and protein level in MES cells (Fig. 4, B and C). Treatment of MES cells with the NRP2 function-blocking antibody also increased β2-chimaerin abundance (Fig. 4D). These observations provide evidence that VEGF-NRP2 signaling represses β2-chimaerin expression.

Table 1 mRNA screen of Rac GEFs and GAPs in EPTH and MES cells.

The expression of the indicated Rac GEFs and GAPs was assessed in EPTH and MES cells using qPCR. Values are fold change (±SD; n = 3) in mRNA expression in MES cells upon normalization with each in EPTH cells, which was set as 1. NE, not expressed.

View this table:
Fig. 4 VEGF-NRP2 signaling represses the Rac GAP β2-chimaerin.

(A) Abundance of β2-chimaerin was quantified by immunoblotting in EPTH and MES cells. Representative blots are shown; data are means ± SEM of three biological replicates. (B) Expression of NRP2 was diminished in MES cells, and β2-chimaerin mRNA expression was quantified by qPCR. Data are means ± SEM of three biological replicates. (C) β2-chimaerin abundance was quantified by immunoblotting in NRP2-depleted MES cells. Representative blots are shown; data are means ± SEM of three biological replicates. (D) MES cells were treated with the indicated concentrations of a function-blocking NRP2 antibody for 6 hours, and β2-chimaerin abundance was quantified by immunoblotting. Representative blots are shown; data are means ± SEM of three biological replicates of the 3 μg/ml condition. *P ≤ 0.05 by two-tailed t test.

TAZ activates Rac1 by repressing β2-chimaerin through a TEAD-dependent mechanism

On the basis of our observation that VEGF-NRP2 signaling activates TAZ and represses β2-chimaerin, we assessed the possibility that TAZ represses β2-chimaerin. This possibility is supported by studies demonstrating that TAZ can function in transcriptional repression (62, 63). Depletion of TAZ in MES cells increased both mRNA and protein abundance of β2-chimaerin and consequently decreased Rac1 activity (Fig. 5, A to C). TAZ knockdown in MDA-MB-435 cells (Fig. 5D) and MDA-MB-231 cells (Fig. 5E) also increased β2-chimaerin mRNA expression. Conversely, TAZ overexpression repressed β2-chimaerin mRNA expression in MDA-MB-231 cells (Fig. 5F). Given our observations that Rac1 activates TAZ (Fig. 3) and that TAZ represses β2-chimaerin expression, we inhibited Rac1 in MES cells using EHT1864 and observed an increase in the expression of β2-chimaerin mRNA (Fig. 5G). These results provide evidence that VEGF-NRP2-Rac1–mediated TAZ activation maintains increased Rac1 activity by repressing β2-chimaerin in a positive feedback loop.

Fig. 5 TAZ activates Rac1 by repressing β2-chimaerin.

(A) TAZ expression was diminished in MES cells, and the impact on β2-chimaerin mRNA expression was quantified by qPCR. Data are means ± SEM of three biological replicates. (B and C) β2-chimaerin abundance (B) and Rac1 activity (C) were assessed in TAZ knockdown MES cells. Representative blots are shown; data are means ± SEM of three biological replicates. (D) TAZ expression was diminished in MDA-MB-435 cells (blot, left), and the impact on β2-chimaerin mRNA expression was quantified by qPCR. Data are means ± SEM of three biological replicates. (E) TAZ expression was diminished in MDA-MB-231 cells (blot, left), and the impact on β2-chimaerin mRNA expression was quantified by qPCR. Data are means ± SEM of three biological replicates. (F) MDA-MB-231 cells were transfected with TAZ, and the impact on β2-chimaerin mRNA expression was quantified by qPCR. Data are means ± SEM of three biological replicates. (G) MES cells were treated with the indicated concentrations of the Rac inhibitor EHT1864 for 2 hours, and β2-chimaerin mRNA expression was quantified by qPCR. Data are means ± SEM of three biological replicates. *P ≤ 0.05 by two-tailed t test.

TAZ-mediated transcriptional repression is dependent on the TEAD1–4 family of transcription factors (62, 63). TEAD4, in particular, is expressed at relatively high levels in breast cancer, especially triple-negative breast cancer (64, 65). Given this information, we initially searched the encyclopedia of DNA elements (ENCODE) for TEAD4 chromatin immunoprecipitation sequencing (ChIP-seq) data sets and found four cell types [h1-human embryonic stem cells (hESCs), HCT116 colon cancer cells, Ishikawa endometrial adenocarcinoma cells, and SK-N-SH neuroblastoma cells], where TEAD4 bound to the promoter region of the β2-chimaerin gene (Fig. 6A). Specifically, a conserved peak was observed near position 29229000 (chr7) in all of the cell types. These findings are significant because they demonstrate direct binding of TEAD4 to the β2-chimaerin promoter. In addition, h1-hESCs, HCT116, Ishikawa, and SK-N-SH cells have enhanced TAZ/TEAD activity (66, 67). To validate the ENCODE data in our model system, we performed ChIP in MES cells using antibodies specific for TEAD4 and TAZ. The results verify that TEAD4 and TAZ are recruited to the genomic region in the β2-chimaerin promoter identified in ENCODE (Fig. 6B). On the basis of these data, we depleted TEAD1/3/4 expression in MES cells and observed an increase in the mRNA and protein abundance of β2-chimaerin (Fig. 6, C and D) and a decrease in Rac1 activity (Fig. 6E), consistent with our TAZ knockdown results (Fig. 5, A to C). Similarly, expression of dominant-negative TEAD4 in MDA-MB-231 cells increased β2-chimaerin mRNA expression (Fig. 6F).

Fig. 6 TEAD mediates repression of β2-chimaerin by TAZ.

(A) Using ENCODE, TEAD4-binding signals were analyzed from ChIP-seq data sets from h1-hESCs (human embryonic stem cells), HCT116 (colon cancer), Ishikawa (endometrial adenocarcinoma), and SK-N-SH (neuroblastoma) cells in the promoter region of the β2-chimaerin gene. (B) Binding of TEAD4 and TAZ on the β2-chimaerin gene promoter in MES cells was analyzed using ChIP. Data are means ± SD from three technical replicates, representative of two biological replicates. (C) TEAD1, TEAD3, and TEAD4 expression was diminished by shRNA in MES cells, and the impact on β2-chimaerin mRNA expression was quantified by qPCR. Data are means ± SEM of three biological replicates. (D and E) β2-chimaerin abundance (D) and Rac1 activity (E) were assessed in TEAD1/3/4 knockdown MES cells. Representative blots are shown; data are means ± SEM of three biological replicates. (F) β2-chimaerin mRNA expression was quantified by qPCR in MDA-MB-231 cells expressing either a control vector or dominant-negative TEAD4. Data are means ± SEM of three biological replicates. *P ≤ 0.05 by two-tailed t test.

β2-chimaerin repression contributes to enhanced TAZ activity

An important question that arises from the data thus far is whether β2-chimaerin repression has a causal role in TAZ activation. Expression of β2-chimaerin in MDA-MB-231 cells decreased Rac1 activity, as well as the abundance of TAZ itself (Fig. 7A) and TAZ target genes (Fig. 7B). Conversely, β2-chimaerin knockdown in EPTH cells increased Rac1 activity, TAZ abundance (Fig. 7C), and TAZ target genes (Fig. 7D). Notably, β2-chimaerin–depleted EPTH cells exhibited increased mammosphere formation compared to control EPTH cells (Fig. 7E). Expression of β2-chimaerin also reduced TAZ-mediated, but not S89A TAZ-mediated, mammosphere formation in MDA-MB-231 cells (Fig. 7F), providing further evidence that Rac1 inhibition of LATS contributes to TAZ activation and CSC properties.

Fig. 7 β2-chimaerin repression contributes to enhanced TAZ activity.

(A) MDA-MB-231 cells were transfected with Myc-tagged β2-chimaerin, and Rac1 activity and TAZ abundance were assessed. Representative blots are shown; data are means ± SEM of three biological replicates. (B) mRNA expression of the indicated TAZ target genes was quantified by qPCR in MDA-MB-231 cells transfected with Myc-tagged β2-chimaerin. Data are means ± SEM of three biological replicates. (C) Expression of β2-chimaerin was diminished in EPTH cells, and the impact on Rac1 activity and TAZ abundance was assessed. Representative blots are shown; data are means ± SEM of three biological replicates. (D) mRNA expression of the indicated TAZ target genes was quantified by qPCR in β2-chimaerin–depleted EPTH cells. Data are means ± SEM of three biological replicates. (E) β2-chimaerin–depleted EPTH cells were assayed for self-renewal by serial passage mammosphere formation. Data are means ± SEM of three biological replicates. (F) MDA-MB-231 cells were transfected with empty vector, WT TAZ, or S89A TAZ with and without Myc-tagged β2-chimaerin and assayed for self-renewal by serial passage mammosphere formation. Data are means ± SD from three technical replicates, representative of two biological replicates. (G) The cBioPortal for Cancer Genomics resource was used to compare TAZ expression with β2-chimaerin, NRP2, and VEGFA expression in the invasive breast carcinoma database from the TCGA. Bottom: TAZ and β2-chimaerin expression were also compared within the glioblastoma and colorectal adenocarcinoma databases in the TCGA and the cancer cell line encyclopedia (CCLE). Log odds ratios and P values were calculated using Fisher’s exact t test with the mutual exclusivity tool on cBioPortal. (H) The cBioPortal for Cancer Genomics was used to analyze the expression of β2-chimaerin in ER+ versus ER breast cancer patients from the invasive breast carcinoma database in the TCGA. *P ≤ 0.05 by two-tailed t test; ***P < 0.0001 by Welch t test.

These in vitro data indicating an inverse causal relationship between β2-chimaerin and TAZ and a positive causal relationship between VEGF-NRP2 and TAZ were substantiated by analysis of their expression in invasive breast carcinomas in The Cancer Genome Atlas (TCGA) database obtained from cBioPortal (Fig. 7G) (68, 69). The expression of TAZ and β2-chimaerin was inversely correlated. In contrast, the expression of TAZ correlated with that of VEGF and NRP2. An inverse correlation between TAZ and β2-chimaerin was also detected in glioblastoma and colorectal adenocarcinoma samples in the TCGA. These findings are significant because both glioblastoma and colon cancer exhibit enhanced TAZ activity (66, 70, 71). Last, TAZ and β2-chimaerin exhibited an inverse correlation in the CCLE obtained from cBioPortal (967 cell lines) (Fig. 7G), which provides further evidence of a repressive role (68, 69).

Our data indicate that TAZ-mediated repression of β2-chimaerin is associated with a mesenchymal phenotype. To substantiate this conclusion, we analyzed a microarray (GSE48204) that used transforming growth factor–β (TGF-β)–treated NMuMG mammary epithelial cells to induce an EMT (72). In support of our conclusion, we found that the EMT reduced β2-chimaerin expression and increased expression of VEGF and NRP2, as well as TAZ target genes (fig. S5). We also used cBioPortal to stratify breast cancer patients in the TCGA database based on their expression of the estrogen receptor (ER), which is associated with an epithelial phenotype. Comparison of β2-chimaerin expression in the ER-positive (ER+) and ER subgroups revealed that ER patients have lower expression of β2-chimaerin compared to ER+ patients (Fig. 7H).

DISCUSSION

The results of this study establish a causal role for VEGF-NRP2 signaling in sustaining the activation of TAZ, a critical effector molecule of the Hippo pathway that contributes to breast tumorigenesis and is associated with aggressive, high-grade tumors. An essential component of this mechanism is the repression of β2-chimaerin, a Rac GAP, by TAZ and the consequent activation of Rac1, resulting in a positive feedback loop driven by VEGF-NRP2 signaling that sustains TAZ activation (Fig. 8). These findings increase our understanding of autocrine VEGF signaling in tumor cells, and they substantiate the importance of Rac in the biology of CSCs and TAZ regulation.

Fig. 8 Model depicting the major findings of this study.

VEGF-NRP2 signaling promotes FAK-mediated Rac1 activation, which inhibits LATS. Consequently, TAZ is located in the nucleus where it associates with TEAD and represses the Rac GAP β2-chimaerin to maintain increased Rac1 activity in a positive feedback loop.

Our conclusion that repression of β2-chimaerin contributes to TAZ activation and self-renewal indicates that this Rac GAP is an important gatekeeper that impedes the acquisition of stem cell properties. On the basis of the hypothesis that breast CSCs are dedifferentiated and exhibit features of an EMT (26), this conclusion infers that repression of β2-chimaerin is a consequence of the EMT and that its expression is associated with an epithelial phenotype. We uncovered that β2-chimaerin is repressed by the TGF-β–induced EMT of mammary epithelial cells. We also demonstrated that ER+ patients have higher β2-chimaerin expression compared to ER patients. These observations contrast with the report that β2-chimaerin reduces E-cadherin levels in an in vitro overexpression system (60). However, this report also demonstrated that low expression of β2-chimaerin is associated with reduced relapse-free survival of breast cancer patients, which supports our findings.

Our results need to be discussed in the context of the report that NRP2 binds β2-chimaerin directly, and that semaphorin 3F–NRP2 signaling reduces this association to activate β2-chimaerin and regulate axonal pruning in the hippocampus (38). Although β2-chimaerin and NRP2 exhibit an inverse expression pattern in breast cancer, we tested the hypothesis that residual β2-chimaerin may be sequestered and inactivated by NRP2 as a mechanism of Rac1 regulation. However, we were unable to co-immunopurify NRP2 with β2-chimaerin. This is not definitive, but it does suggest that NRP2 regulation of β2-chimaerin differs in breast cancer cells and neurons. This difference may reflect the fact that different ligands (semaphorin 3F and VEGF) engage NRP2 in these cell types. It is also worth mentioning that we previously demonstrated that semaphorin 3A and VEGF compete for NRP binding in breast cancer cells and that these two ligands have opposite effects on the behavior of these cells (73). Nonetheless, the existing data highlight an important causal effect of NRP2 on β2-chimaerin that is executed by distinct mechanisms.

A major conclusion of this study is that VEGF-NRP2 signaling contributes to TAZ activation by a Rac1-dependent mechanism. This role for Rac1 differs from its more established role in regulating cell invasion and migration in cancer, but it is consistent with other reports implicating Rac1 in the function of CSCs (7476), as well as in YAP/TAZ activation (35, 4143). Our results on the ability of VEGF-NRP2 signaling to inhibit LATS by a Rac1-dependent mechanism support these observations, and they identify a novel ligand-receptor interaction that can mediate this regulation. Furthermore, our data suggest that the ability of Rac1 to inhibit LATS is mediated by its regulation of Merlin, which is an important organizer of the membrane-cytoskeleton interface (5255, 7779). Although we are not ruling out the possibility that VEGF-NRP2 signaling may activate RhoA, which has also been implicated in YAP/TAZ activation (35, 41, 7981), we focused our attention on Rac1 because β2-chimaerin is Rac-specific and does not have GAP activity against Rho (57, 58).

Although many studies have implicated autocrine VEGF signaling in the function of CSCs (20), its ability to contribute to Hippo-TAZ regulation provides a new dimension to our understanding of VEGF biology. While this manuscript was in review, however, it was reported that VEGF-VEGFR2 signaling contributes to YAP/TAZ activation during developmental angiogenesis (82). Our data support the role of VEGF in promoting YAP/TAZ activation, but the mechanism used by breast cancer cells is distinct because it appears to be dependent on VEGF-NRP2 activation of FAK-Rac1 but independent of VEGFR. Moreover, our findings reveal a pivotal role for β2-chimaerin as a repressive intermediary between VEGF-NRP2 and TAZ activation. They also reinforce the hypothesis that targeting VEGF-NRP signaling is a viable therapeutic strategy for tumor cells that are dependent on TAZ activation.

MATERIALS AND METHODS

Reagents and antibodies

EHT1864 was purchased from Tocris, NSC23766 was purchased from Selleckchem, FAK14 was purchased from Sigma, sunitinib and pazopanib were purchased from LC Laboratories, human VEGFA165 was purchased from R&D Systems, and the function-blocking NRP2 antibody was provided by Genentech (11). Immunoblotting antibodies were acquired as follows: actin (A2066, Sigma), TAZ (560235, BD Biosciences), YAP/TAZ (8418S, Cell Signaling Technology), pSer89 TAZ (sc-17610, Santa Cruz Biotechnology), pSer127 YAP (4911A, Cell Signaling Technology), NRP2 (sc-7242, Santa Cruz Biotechnology), β2-chimaerin (CHN2) (HPA018989, Sigma), pThr1079 LATS1 (8654S, Cell Signaling Technology), LATS1 (9153S, Cell Signaling Technology), VEGF (sc-152, Santa Cruz Biotechnology), Rac1 (610650, BD Biosciences), Pan-TEAD (13295S, Cell Signaling Technology), pSer518 Merlin (9163S, Cell Signaling Technology), Merlin (6995S, Cell Signaling Technology), HA-Tag (3724S, Cell Signaling Technology), GST (sc-138, Santa Cruz Biotechnology), and Myc-Tag (2278S, Cell Signaling Technology).

Constructs

The following lentiviral shRNA vectors were used: VEGF (TRCN0000003343, TRCN0000003344, and TRCN0000003345), NRP2 (TRCN0000063309, TRCN0000063312, and TRCN0000063310), β2-chimaerin [provided by A. Kolodkin, Johns Hopkins Medical Institute (38)], and TEAD 1/3/4 [provided by J. Mao, University of Massachusetts Medical School (66)]. Retroviral shTAZ vectors were used as previously described (32). Stable shRNA expression was accomplished by selecting cells in puromycin (2 μg/ml) for 2 to 4 days. Myc-tagged (6×) β2-chimaerin was provided by A. Kolodkin, Johns Hopkins Medical Institute (38); Myc-tagged dominant-negative TEAD4 was provided by J. Mao, University of Massachusetts Medical School; and dominant-negative Rac1 (N17Rac1) and constitutively active Rac1 (V12Rac1) were described previously (21). Human TAZ was cloned into pcDNA3.1 vector, and site-directed mutagenesis was performed to generate TAZ S89A.

Cell culture

ER-SRC–transformed MCF10A cells were provided by K. Struhl (Harvard Medical School). To generate puromycin-sensitive MCF10A ER-SRC cells, v-SRC was cloned from the complementary DNA (cDNA) pool of the original MCF10A ER-SRC cell line. Subsequently, pWZL Blast Twist ER plasmid (Addgene plasmid #18799) was digested by Bam HI to remove the Snai1 cDNA, and replaced with v-SRC cDNA, resulting in the expression of the fusion protein, v-SRC-ER, by the new recombinant plasmid. This plasmid was subsequently used to produce retrovirus for infecting MCF10A cells. Stable clones were selected by blasticidin. Isolation of the EPTH and MES populations of CD44+CD24−/low MCF10A ER-SRC–transformed cells using flow cytometry has been previously described (33). EPTH and MES cells were cultured as subclones for two to three passages and used for experiments. MDA-MB-231 and MDA-MB-435 cells were obtained from the American Type Culture Collection. All experiments were performed at a cell density of 25 to 35%.

Transfection and siRNA knockdown

For overexpression, plasmids were transfected using Lipofectamine 2000 (Thermo Fisher Scientific). Cells were processed for immunoblotting, qPCR, or mammosphere formation about 24 hours after transfection. For LATS1/2 small interfering RNA (siRNA) knockdown, MDA-MB-231 cells were transfected using DharmaFect 4 (Dharmacon). Cells were processed for immunoblotting 48 hours after transfection. LATS1/2 siRNA has been previously described (32).

Mammosphere assay

Cells were plated in UltraLow attachment six-well plates in Dulbecco’s modified Eagle’s medium/F12 medium supplemented with B27, epidermal growth factor, and fibroblast growth factor as previously described (33). For serial passaging, mammospheres were pelleted and dissociated with 0.05% trypsin for 15 min at 37°C to obtain single cells. These cells were washed in 1× phosphate-buffered saline (PBS), counted, and replated in UltraLow attachment six-well plates.

Immunoblotting

Cells were washed in 1× PBS and scraped on ice in radioimmunoprecipitation assay buffer with EDTA and EGTA (BP-115DG, Boston BioProducts) supplemented with protease and phosphatase inhibitors (Pierce, #88669). Laemmli 6× SDS sample buffer (BP-111R, Boston BioProducts) was added to each sample, and the protein lysate was boiled for 10 min and separated using SDS–polyacrylamide gel electrophoresis. Rac activity was assessed using a GST fusion protein containing the Rac/Cdc42-binding domain of PAK (PBD) as previously described (21, 83).

Luciferase reporter assay

TEAD transcriptional activity was assessed using a luciferase reporter construct (8xGTIIC, Addgene #34615) with the Dual-Luciferase Reporter Assay System (#E2940, Promega). About 24 hours after transfection, luciferase activity was measured as the average ratio of firefly to Renilla luciferase.

Real-time qPCR

RNA extraction was accomplished using an RNA isolation kit (BS88133, Bio Basic Inc.), and cDNAs were produced using a qScript cDNA synthesis kit (#95047, Quantabio). SYBR Green (Applied Biosystems) was used as the qPCR master mix. Experiments were performed in triplicate and normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH). qPCR primer sequences were obtained from the Massachusetts General Hospital/Harvard Medical School PrimerBank (http://pga.mgh.harvard.edu/primerbank/).

Chromatin immunoprecipitation

ChIP was performed using a ChIP-IT Express Chromatin Immunoprecipitation kit (Active Motif). Antibodies used were TEAD4 (N-G2, Santa Cruz Biotechnology) and TAZ (70148S, Cell Signaling Technology). The following qPCR primer sequence was used to amplify the region of the TEAD4 peak in the β2-chimaerin promoter identified in ENCODE: primers 5′-GCTTACAGCTGGCTTCACTT-3′ (forward) 5′-GGCCGAGAGAGAGAGAGTTT-3′ (reverse).

Immunofluorescence confocal microscopy

TAZ localization was performed by fixing cells with paraformaldehyde (4%) and permeabilizing them with Triton X-100 (0.1%) as described (23, 32). Cells were blocked with 1% bovine serum albumin and horse serum (2.5%) and incubated with TAZ antibody (1:100 dilution; sc-48805, Santa Cruz Biotechnology) overnight at 4°C. Subsequently, cells were washed with 1× PBS and incubated with fluorochrome-conjugated secondary antibodies. Images were captured at ×20 magnification using a confocal microscope (Zeiss).

ENCODE data analysis

ENCODE TEAD4-binding signals were downloaded from www.encodeproject.org in bigwig format. The coverages of duplicate samples were pooled and then plotted along the promoter region of the β2-chimaerin (CHN2) gene.

cBioPortal analysis

cBioPortal (www.cbioportal.org) was used to compare the mRNA expression (RNA Seq V2 RSEM) of TAZ, VEGFA, NRP2, and β2-chimaerin using the TCGA invasive breast carcinoma provisional data set (68, 69). In addition, the mRNA expression (RNA Seq V2 RSEM) of TAZ and β2-chimaerin was compared in the TCGA glioblastoma provisional data set and the TCGA colorectal adenocarcinoma Nature 2012 data set. The CCLE (Novartis/Broad, Nature 2012) was used to compare the expression of TAZ and β2-chimaerin across various cell lines using the mRNA expression z-scores microarray. To determine whether the expression of two genes is inversely correlated, we performed the mutual exclusivity analysis with a z-score threshold of ±1.5 as expressed, and calculated the log odds ratio between the two genes and P value using Fisher’s exact t test. In addition, we stratified breast cancer patients in the TCGA Cell 2015 database based on their ER status, and compared β2-chimaerin expression in the ER+ and ER subgroups using Welch t test.

Microarray data analysis

The microarray data set from Gene Expression Omnibus (GEO) (GSE48204) (72) was downloaded using the Bioconductor package GEOquery (version 2.41.0). Moderated t test was used to identify differentially expressed genes between TGF-β–induced EMT cells and NMuMG cells treated with vehicle. Genes with an adjusted P value of ≤0.05 using Benjamini-Hochberg method were considered significant (84).

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/11/528/eaao6897/DC1

Fig. S1. VEGF-NRP2 signaling contributes to YAP activation.

Fig. S2. VEGF-NRP2 activation of Rac1 is independent of VEGFR.

Fig. S3. VEGF-NRP2-Rac1 regulation of Merlin phosphorylation (pSer518) is LATS-dependent.

Fig. S4. Expression of wild-type and S89A TAZ constructs.

Fig. S5. β2-chimaerin is repressed by the EMT.

REFERENCES AND NOTES

Acknowledgments: We thank C.-C. Hsieh for consulting on statistical analysis, and D. McCollum, J. Mao, D. Kim, and T. Fazzio for helpful discussions. Funding: This work was supported by NIH grants CA168464 and CA203439 (A.M.M.). A.L.E. was supported by a Ruth L. Kirschstein National Research Service Award from the National Cancer Institute (F30CA206271) and an American Medical Association Foundation Medical Student Seed Grant. Author contributions: A.L.E. and A.M.M. designed experiments and wrote the manuscript. A.L.E. performed experiments. S.G. performed immunofluorescence confocal microscopy. C.C. cloned wild-type and S89A TAZ plasmids and generated puromycin-sensitive EPTH and MES cells. J.O. and L.J.Z. performed the ENCODE, cBioPortal, and microarray analyses. J.J.A. provided technical support. H.L.G. contributed to ChIP experiments, provided feedback on data, and helped write the manuscript. All authors read and approved the final manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.
View Abstract

Navigate This Article