Research ArticleCancer Metabolism

The receptor tyrosine kinase EphA2 promotes glutamine metabolism in tumors by activating the transcriptional coactivators YAP and TAZ

See allHide authors and affiliations

Science Signaling  05 Dec 2017:
Vol. 10, Issue 508, eaan4667
DOI: 10.1126/scisignal.aan4667

EphA2 exposes an Achilles’ heel in breast cancer

Cancer cells alter their metabolism to adapt to the tumor microenvironment, for example, by switching from a reliance on glucose to glutamine. Edwards et al. found that a high abundance of the receptor tyrosine kinase EphA2 drives a switch to glutamine dependency in HER2-positive breast cancers. In cells and mouse models, EphA2 activated Rho-ROCK signaling that mediated the nuclear accumulation of the transcriptional coactivators YAP and TAZ, where they induced the expression of genes encoding an amino acid transporter that promoted glutamine uptake and an enzyme that promoted glutamine metabolism. EphA2 may be a biomarker for sensitivity to glutaminase inhibitors in patient tumors. However, because inhibiting glutamine metabolism may have adverse effects elsewhere, such as in the immune system, inhibiting EphA2 signaling upstream of YAP and TAZ may be a way to exploit the glutamine dependency of HER2-positive breast cancers.


Malignant tumors reprogram cellular metabolism to support cancer cell proliferation and survival. Although most cancers depend on a high rate of aerobic glycolysis, many cancer cells also display addiction to glutamine. Glutamine transporters and glutaminase activity are critical for glutamine metabolism in tumor cells. We found that the receptor tyrosine kinase EphA2 activated the TEAD family transcriptional coactivators YAP and TAZ (YAP/TAZ), likely in a ligand-independent manner, to promote glutamine metabolism in cells and mouse models of HER2-positive breast cancer. Overexpression of EphA2 induced the nuclear accumulation of YAP and TAZ and increased the expression of YAP/TAZ target genes. Inhibition of the GTPase Rho or the kinase ROCK abolished EphA2-dependent YAP/TAZ nuclear localization. Silencing YAP or TAZ substantially reduced the amount of intracellular glutamate through decreased expression of SLC1A5 and GLS, respectively, genes that encode proteins that promote glutamine uptake and metabolism. The regulatory DNA elements of both SLC1A5 and GLS contain TEAD binding sites and were bound by TEAD4 in an EphA2-dependent manner. In patient breast cancer tissues, EphA2 expression positively correlated with that of YAP and TAZ, as well as that of GLS and SLC1A5. Although high expression of EphA2 predicted enhanced metastatic potential and poor patient survival, it also rendered HER2-positive breast cancer cells more sensitive to glutaminase inhibition. The findings define a previously unknown mechanism of EphA2-mediated glutaminolysis through YAP/TAZ activation in HER2-positive breast cancer and identify potential therapeutic targets in patients.


Receptor tyrosine kinases (RTKs) serve as regulators of key signaling pathways that promote cell growth, proliferation, and migration and are commonly increased in abundance and/or activity in human cancers (1). As a member of the largest family of RTKs, the Eph receptors, ephrin type-A receptor 2 (EphA2) is frequently overexpressed in human cancer, with enhanced EphA2 activity correlating with tumor progression and reduced survival (25). However, unlike other RTKs, cumulative evidence supports two modes of EphA2 signaling. In the first mode, ligand-dependent EphA2 forward signaling in normal cells suppresses cell proliferation and invasiveness, often resulting from cell-cell interactions. In the second mode, because of reduced ligand binding or transactivation through another RTK, such as the epidermal growth factor (EGF) receptor or human EGF receptor 2 (HER2), ligand-independent EphA2 signaling promotes tumor malignancy, in part, through activation of the Rho–guanosine triphosphatase (GTPase) (3, 59). EphA2 is overexpressed, whereas its ligands, including ephrin-A1, are often less abundant in malignant human breast cancer specimens, correlating with poor survival and resistance to therapeutic drugs (3, 4, 10, 11). In parallel to the role of EphA2 in human breast cancer, ablation of EphA2 inhibits metastatic progression in the murine HER2-dependent mammary tumor model MMTV-Neu (5).

Similar to many RTK signaling pathways, the Hippo tumor suppressor pathway also controls cell proliferation, with dysregulation often contributing to breast tumorigenesis. Like EphA2, the Hippo pathway represents a complex signaling cascade, differentially responding to various signals. Whereas cell-cell contact activates Hippo signaling to suppress cell growth and proliferation, growth factor signaling deactivates the Hippo pathway, supporting tumor growth and progression (1215). Cell growth and proliferation result from increased transcription due to nuclear accumulation and activation of the Yes-associated protein (YAP) and WW domain–containing transcription regulator 1 (WWTR1 or TAZ), transcriptional coactivators that act as the primary downstream effectors of the Hippo pathway. In the nucleus, YAP and TAZ primarily increase pro-proliferative gene expression through interaction with the TEA domain (TEAD) family of transcription factors (1619). Phosphorylation by the primary mediators of Hippo signaling, the large tumor suppressor kinases 1 and 2 (LATS1/2), or additional unidentified kinases suppress YAP and TAZ activity through 14-3-3–dependent nuclear export (2022), with YAP phosphorylation at Ser127 and Ser397 resulting in cytoplasmic retention and proteasomal degradation, respectively (2325), whereas TAZ is primarily degraded upon phosphorylation at Ser89 (26). Growth factors and defects in Hippo signaling decrease phosphorylation and promote nuclear accumulation of YAP and TAZ in human breast cancer (2730), supporting RTK signaling as potential regulators of Hippo and YAP/TAZ activity. Nonetheless, few specific RTKs have been identified as upstream mediators of YAP/TAZ activity.

Accumulating evidence supports a role of RTK signaling in regulating metabolic processes. Although many tumors primarily depend on glycolysis to generate adenosine 5′-triphosphate and other biomolecules necessary for an enhanced proliferative phenotype under hypoxic conditions, breast cancer cells also use glutamine for biogenesis of nucleotides, lipids, and amino acids; for generation of the cellular antioxidant glutathione; and as a mitochondrial substrate (2, 31, 32). Increased glutamate concentrations in breast tumors are associated with invasiveness and drug resistance and correlate with increased risk of recurrence (33, 34), suggesting that enhanced glutamine metabolism may support tumor survival and therapeutic evasion. The dependence of breast tumors on glutamine metabolism relies on glutamine uptake, primarily from the neutral amino acid transporter solute carrier family 1 member 5 (SLC1A5) (35, 36), as well as utilization of glutamine by glutaminase (GLS) through conversion to the glutamate metabolite (37). Among breast cancer subtypes, the HER2-positive subtype is one of the most reliant on glutamine metabolism, likely resulting from HER2-dependent up-regulation of GLS and SLC1A5 (38, 39). We have previously shown that EphA2 promotes glutamine metabolism to support tumor growth in a HER2-positive breast cancer model (2); however, the mechanism of EphA2-dependent metabolism has not been elucidated.

Here, we found that the RTK EphA2 promoted glutamine metabolism by activating the transcriptional coactivators YAP and TAZ. EphA2 overexpression increased nuclear localization and activation of YAP and TAZ in cells overexpressing HER2 or the rat ortholog Neu, as well as a HER2-positive breast cancer mouse model (MMTV-Neu). Active Rho and its downstream Rho-associated protein kinase (ROCK) were required for EphA2-mediated activation of YAP and TAZ (YAP/TAZ). YAP and TAZ enhanced glutamine metabolism through differential transcriptional regulation of the glutaminolysis-related genes SLC1A5 and GLS, respectively. Analysis of patient breast cancer samples revealed that the expression of YAP, TAZ, SLC1A5, and GLS correlated with that of EphA2, and greater activation of the EphA2-YAP/TAZ pathway correlated with reduced patient survival, greater metastatic potential, and increased sensitivity of the tumor to GLS inhibition. Together, these data define a previously unknown mechanism of YAP and TAZ activation by the RTK EphA2 and describe a new transcription-dependent role of YAP and TAZ to enhance glutamine metabolism in a HER2-dependent breast cancer model.


EphA2 activates YAP and TAZ in HER2-positive MCF10A cells and MMTV-NeuT mammary tumors

We previously showed that EphA2 promotes glutamine metabolism through a Ras homolog family member A (RhoA)–dependent pathway (2). Although Rho activation can lead to increased GLS activity (2, 40), the mechanism has not been elucidated. Because RhoA can activate the transcriptional cofactors YAP and TAZ (41), we reasoned that increased glutaminolysis induced by EphA2 could be mediated by RhoA-dependent YAP and TAZ activation. To determine whether the RTK EphA2 activates YAP and/or TAZ, we transduced MMTV-Neu cells with Ad-EphA2 or control Ad–GFP (green fluorescent protein), and YAP or TAZ subcellular localization was assessed by immunofluorescence. Compared to control cells, both YAP and TAZ were more frequently colocalized with nuclear 4′,6-diamidino-2-phenylindole (DAPI) staining (Fig. 1, A and B) in cells infected with Ad-EphA2, corresponding with increased EphA2 (Fig. 1C). To confirm that the colocalization with DAPI was indicative of intranuclear accumulation, we imaged cells by super-resolution microscopy, which collects multiple z-stack images of an individual cell. Intranuclear localization of YAP was observed in z-stack images upon EphA2 overexpression (Fig. 1D and movie S1). In addition, EphA2 was found to increase YAP and TAZ nuclear localization in the human MCF10A cell line overexpressing HER2 (MCF10A-HER2; fig. S1, A and B), suggesting that YAP and TAZ are accumulated in the nucleus in response to EphA2 overexpression.

Fig. 1 EphA2 activates the transcriptional coactivators YAP and TAZ in vitro.

(A and B) Immunofluorescence of YAP (A) or TAZ (B) (both red) in MMTV-Neu cells infected with Ad-GFP or Ad-EphA2 for 72 hours. DAPI-stained nuclei (blue) were counted in ImageJ, and YAP- or TAZ-positive nuclei were counted using three fields of view in three independent experiments. Error bars are SEM. Scale bars, 10 μm. ***P < 0.005, Student’s t test. (C) Western blot of EphA2, YAP, and TAZ in MMTV-Neu cells infected with Ad-GFP or Ad-EphA2. (D) Super-resolution microscopy of YAP immunofluorescence from (A). Displayed image was compiled from 10 consecutive z-stack images. Signal intensity of YAP (red) or DAPI (blue) was determined per pixel moving across the y axis using ImageJ software. Scale bar, 10 μm. a.u., arbitrary units. (E) Relative mRNA expression was measured in MMTV-Neu cells retrovirally transduced with pBABE (“Control”) or pBABE-EphA2 (“EphA2”) by qRT-PCR from three independent experiments. Data are means ± SEM. *P < 0.05 and ***P < 0.005, Student’s t test.

Next, we sought to determine whether EphA2-mediated nuclear localization of YAP and TAZ reflected a functional activation for the transcriptional coactivators. We examined the ability of EphA2 to enhance transcription of known YAP/TAZ-target genes, Cyr61, Ctgf, and Inhba (16, 19, 4244). Control or EphA2-overexpressing MMTV-Neu cells were subjected to quantitative real-time polymerase chain reaction (qRT-PCR) (Fig. 1E). EphA2 overexpression increased the expression of Cyr61, Ctgf, and Inhba (Fig. 1E). Similarly, EphA2 overexpression in MCF10A-HER2 cells (fig. S1C) also increased the expression of the YAP/TAZ target genes CYR61 and AXL (fig. S1D). Together, these findings support our hypothesis that EphA2 acts as a positive regulator of the transcriptional coactivators YAP and TAZ.

Given the unique bidirectional nature of EphA2 signaling, we next assessed whether EphA2-mediated YAP and TAZ nuclear accumulation was due to EphA2–ephrin-A1 interactions. Control or EphA2-overexpressing MMTV-Neu cells were treated with the control Fc or the soluble chimeric EphA2-Fc competitor, which binds ephrin-A1 to suppress ligand-dependent signaling (45). Whereas EphA2 overexpression increased YAP and TAZ nuclear accumulation, we did not observe any significant changes in the presence of soluble EphA2-Fc (fig. S2), suggesting a primarily receptor-driven role of EphA2 in YAP/TAZ activation.

To determine whether EphA2 increases YAP and TAZ activity in vivo, we used an ephrin-A1 (EFNA1) knockout MMTV-NeuT breast cancer mouse model that overexpresses EphA2 (2). Mammary tumors were collected from wild-type (EFNA1+/+) or ephrin-A1 knockout (EFNA1−/−) mice, and YAP and TAZ subcellular localization was determined by immunohistochemistry. As we expected, EphA2 expression was significantly increased in ephrin-A1 knockout tumors (2). Whereas YAP and TAZ were primarily observed in the cytoplasm of wild-type tumors, nuclear localization was significantly increased in ephrin-A1 knockout tumors (Fig. 2A). Analysis of tumor lysates by Western blot (Fig. 2B) indicated that ephrin-A1 knockout tumors exhibited significantly lower phosphorylation of YAP at Ser381 (corresponding to Ser397 in humans) (Fig. 2, B and C). Phosphorylation at Ser381 reportedly marks YAP for degradation (2325). A similar loss of TAZ phosphorylation at Ser89 was also observed (Fig. 2, B and C). A robust inverse correlation between ephrin-A1 and EphA2 protein abundance was confirmed, with greater EphA2 protein detected in ephrin-A1 knockout tumors (Fig. 2, B and D). Not surprisingly, loss of EFNA1 also reduced Tyr588 phosphorylation and increased Ser897 phosphorylation (Fig. 2, B and E), indicating the activation of a ligand-independent mode of EphA2 signaling (6, 46, 47). Although we cannot rule out the possibility that the phenotype observed could be due to loss of ephrin-A1, increased EphA2, or both, together, our results strongly support the finding that EphA2 overexpression and ligand-independent EphA2 signaling promote YAP and TAZ activation.

Fig. 2 EphA2 activates YAP and TAZ in an MMTV-Neu mouse model.

(A) Immunohistochemistry of YAP, TAZ, and EphA2 (brown staining) in tumors collected from wild-type (EFNA1+/+) and EFNA1 knockout (EFNA1−/−) mice. Scale bar, 10 μm. Percentages of YAP- or TAZ-positive nuclei and EphA2 intensity/cell were calculated on the basis of hematoxylin-stained nuclei (blue) using CellProfiler software from four fields per tumor sample (n = 4). *P < 0.05 and ***P < 0.005, Student’s t test. (B) Western blot of tumors collected from wild-type (EFNA1+/+) and EFNA1 knockout (EFNA1−/−) mice. The dotted line denotes noncontiguous lanes within a single blot. (C to E) Quantitation of Western blot data shown in (B) from EFNA1+/+ (n = 3) and EFNA1−/− (n = 3) tumors. (C) Phospho-YAP (Ser127 or Ser381) to total YAP, phospho-TAZ (Ser89) to total TAZ, (D) relative EFNA1 and EphA2 protein, and (E) phospho-EphA2 (Ser897 or Tyr588) are shown. Data are means ± SEM. *P < 0.05 and **P < 0.01, Student’s t test.

EphA2 activates YAP and TAZ in a Rho-dependent mechanism

Overexpression of EphA2 has previously been shown to enhance RhoA signaling to support tumor malignancy (2, 5, 9, 48), and RhoA has also been implicated in the activation of YAP/TAZ (41), suggesting that the Rho GTPase may act as an intermediary in the EphA2-YAP/TAZ signaling axis. To examine whether Rho is necessary to activate YAP and TAZ upon overexpression of EphA2, we treated MMTV-Neu cells overexpressing EphA2 with vehicle or the selective Rho inhibitor CT04, and YAP and TAZ subcellular localization was determined by immunofluorescence (Fig. 3, A to D). In control cells, YAP and TAZ were localized to the nuclei, but CT04 treatment resulted in greater cytoplasmic localization (Fig. 3, A to D, and movie S2), suggesting that Rho inhibition blocks YAP and TAZ activity. CT04-treated cells also exhibited greater phosphorylation of YAP at Ser381 (but not at Ser127), as well as reduced total YAP and TAZ protein (Fig. 3E), consistent with cytoplasmic retention and degradation of YAP and TAZ upon Rho inhibition. However, Rho inhibition did not increase LATS1 Ser909 phosphorylation (fig. S3A), a marker of LATS1 kinase activation, suggesting that Rho may increase YAP and TAZ activity independent of the Hippo pathway.

Fig. 3 EphA2 depends on Rho signaling to activate YAP and TAZ.

(A to C) Immunofluorescence analysis of YAP (A) or TAZ (B) (red) and DAPI (blue) in MMTV-Neu cells infected with Ad-EphA2 and treated with PBS (control) or Rho inhibitor (CT04; 3 μg/ml for 4 hours). Scale bars, 10 μm. DAPI-stained nuclei were counted in ImageJ, and YAP- or TAZ-positive nuclei were counted using three fields of view in three independent experiments. Data (C) are means ± SEM. *P < 0.05 and **P < 0.01, Student’s t test. (D) Super-resolution microscopy of YAP immunofluorescence in control or CT04-treated cells from (A), compiled from 10 consecutive z-stacks. Scale bar, 10 μm. (E) Western blot analysis of cells in (A) and (B), treated with CT04 (3 μg/ml) for 6 hours. (F to H) As described for (A) to (C) with ROCK inhibitor (Y-27632; 10 μM for 4 hours). (I) Western blot analysis of cells in (F) and (G), treated with Y-27632 (10 μM) for 6 hours. (J) Western blot analysis of MMTV-Neu cells demonstrating ROCK inhibition by Y-27632.

Next, we assessed how the downstream effector of Rho, ROCK, contributes to YAP/TAZ activation. Similar to the treatment with CT04, a reduction in YAP and TAZ nuclear localization was observed in cells treated with the ROCK kinase inhibitor Y-27632 compared to control (Fig. 3, F to H). Likewise, inhibition of the ROCK kinase increased phosphorylation of YAP (Ser381 and Ser127) and TAZ (Ser89) (Fig. 3I) without affecting LATS1 Ser909 phosphorylation (fig. S3B). To confirm ROCK inhibition, we examined phosphorylation of myosin light chain 2 (MLC2), a known substrate of the ROCK kinase (49). Compared to control cells, a substantial loss of MLC2 phosphorylation was observed in cells treated with Y-27632 (Fig. 3J), confirming ROCK inhibition (50). Collectively, these results support the mechanism that YAP and TAZ are activated by EphA2 in a Rho-dependent manner.

Using a complementary genetic approach, YAP or TAZ subcellular localization was assessed by immunofluorescence in EphA2-overexpressed MMTV-Neu cells that were transduced with adenovirus to overexpress the control GFP, the dominant-negative Rho-T19N mutant, or the constitutively active Rho-Q63L mutant (fig. S4). Compared to control cells, overexpression of the dominant-negative Rho mutant reduced the abundance of YAP and TAZ localizing to the nuclei, consistent with pharmacological Rho inhibition. Conversely, overexpression of the constitutively active Rho-Q63L mutant maintained YAP and TAZ nuclear accumulation. Together, these data support the role of Rho catalytic activity in promoting EphA2-mediated nuclear accumulation of YAP and TAZ.

YAP and TAZ enhance EphA2-mediated glutamine metabolism by regulating GLS and SLC1A5 expression

Overexpression of HER2 has been previously shown to promote “glutamine addiction” of breast tumors (38, 39). We reported that concurrent EphA2 overexpression in models of HER2-positive breast cancer promotes a similar increase of glutamine metabolism (2). To determine whether cross-talk between EphA2 and HER2 is necessary to increase glutaminolysis, we examined the ability of MCF10A and MCF10A-HER2 cells with or without EphA2 overexpression to convert extracellular glutamine to intracellular glutamate after a period of glutamine stimulation of starved cells. Compared to parental cells, overexpression of EphA2 alone did not significantly increase intracellular glutamate concentration. However, concurrent overexpression of EphA2 along with HER2 significantly enhanced intracellular glutamate concentrations (Fig. 4A). A smaller increase in glutaminolysis was also observed with HER2 overexpression alone (Fig. 4B). Together, these results suggest that cross-talk with HER2 may promote EphA2-mediated glutamine metabolism in HER2-positive breast tumors.

Fig. 4 YAP and TAZ promote glutamine metabolism in a HER2-positive breast cancer model.

(A) Intracellular glutamate concentration was measured 20 min after addition of EGF (20 ng/ml) + l-glutamine (2.5 mM) in MCF10A (“Parental”), MCF10A-EphA2 (“+EphA2”), MCF10A-HER2 (“+HER2”), or MCF10A-HER2-EphA2 (“+HER2/EphA2”) cells. Data were calculated as fold change from parental cells. Data are means ± SEM from three independent experiments. *P < 0.05 and **P < 0.01, one-way analysis of variance (ANOVA) and Tukey’s post hoc test. (B) Western blot of cells described in (A). (C) Intracellular glutamate concentration was determined as described in (B) in MCF10A-HER2 (“Control”) or MCF10A-HER2-EphA2 (“EphA2”) cells with YAP or TAZ knockdown and calculated as fold change from Control/siCtrl cells. Data are means ± SEM from three independent experiments. *P < 0.05 and **P < 0.01, two-way ANOVA and Tukey’s post hoc test. (D) Growth assay of YAP or TAZ knockdown MCF10A-HER2-EphA2 cells treated with vehicle (−DKG) or DKG (+DKG) for 3 days. Fold change in cell number was calculated on the basis of controls in each respective treatment group. Data are means ± SEM from three independent experiments. ***P < 0.005, two-way ANOVA and Tukey’s post hoc test.

To determine whether YAP and TAZ can mediate EphA2-dependent glutaminolysis, YAP or TAZ was knocked down in MCF10A-HER2 or MCF10A-HER2-EphA2 cells by small interfering RNA (siRNA)–mediated gene silencing (Fig. 4C), and intracellular glutamate concentrations were measured after a glutamine pulse under serum-starvation conditions. Consistent with aforementioned results (Fig. 4A) (2), EphA2 overexpression significantly increased intracellular glutamate concentrations in MCF10A-HER2 cells. Knockdown of YAP or TAZ in control cells shows no significant changes in glutamate concentrations (Fig. 4C). However, loss of YAP or TAZ significantly reduced intracellular glutamate concentrations in cells overexpressing EphA2 (Fig. 4C), indicating a role for YAP and TAZ in regulating EphA2-mediated glutamine metabolism. We observed a similar requirement for both YAP and TAZ in cells stimulated with both glutamine and serum (fig. S5), with loss of YAP (fig. S5, A and B) or TAZ (fig. S5, C and D) resulting in significantly reduced intracellular glutamate concentrations over time. The impact of YAP and TAZ on glutamine metabolism did not stem from off-target effects because three independent siRNAs targeting YAP (fig. S5, E and F) and TAZ (fig. S5, G and H) resulted in significantly reduced intracellular glutamate compared to the control.

Because loss of YAP or TAZ has been associated with cellular growth defects (1215), we next determined whether restoration of α-ketoglutarate (α-KG), the tricarboxylic acid (TCA) cycle intermediate and downstream metabolite of glutamine uptake and conversion to glutamate, would rescue cell growth. MCF10A-HER2-EphA2 cells transfected with individual siRNAs to knockdown YAP or TAZ were treated with vehicle or dimethyl α-KG (DKG), a cell-penetrable version of α-KG (51), and the number of cells was counted (Fig. 4D). Consistent with YAP and TAZ promoting cell growth, loss of YAP or TAZ significantly reduced cell number compared to the nontargeting control. However, the addition of DKG partially rescued cell growth in cells with YAP or TAZ knockdown, supporting YAP and TAZ as regulators of glutamine metabolism.

Next, we examined how glutamine metabolism is stimulated by YAP and TAZ. As regulators of transcription, we expected that YAP and TAZ would promote expression of genes critical in glutamine metabolism, including the glutamine transporter SLC1A5 and glutaminase GLS, both of which are commonly overexpressed in human breast cancer, especially in the HER2-positive subtype (38, 39). We used qRT-PCR to determine whether EphA2, YAP, and TAZ promote SLC1A5 and GLS expression (Fig. 5). There are two alternatively spliced isoforms of GLS encoded by GLS1, KGA, and GAC (52). We focused on the more catalytically active GAC isoform that is frequently overexpressed in human cancers, including breast cancer (5254). Overexpression of EphA2 significantly increased SLC1A5 expression (Fig. 5A), an effect that was confirmed at the protein level (Fig. 5B). Surprisingly, expression of GLS was unaffected (Fig. 5, A and B), possibly due to compensatory mechanisms of GLS regulation, such as MYC or NFB (39, 40, 55, 56). Notably, YAP and TAZ appear to differentially regulate glutamine metabolism genes. Loss of YAP significantly decreased expression of SLC1A5, but not of GLS, both at the mRNA (Fig. 5C) and protein (Fig. 5D) levels. In contrast, loss of TAZ reduced both GLS and SLC1A5 expression (Fig. 5, E and F).

Fig. 5 GLS and SLC1A5 gene expression is enhanced by EphA2-YAP/TAZ-TEAD4 signaling.

(A to F) Abundance of the indicated mRNA (by qRT-PCR) and protein (by Western blotting) in MCF10A-HER2 cells after overexpression of EphA2 (A and B), overexpression of EphA2 and knockdown of YAP (C and D), or overexpression of EphA2 and knockdown of TAZ (E and F) relative to control cells. Data are means ± SEM from three independent experiments. **P < 0.01 and ***P < 0.005, Student’s t test (A) or one-way ANOVA (C and E) and Dunnett’s post hoc test. (G) ChIP-seq for TEAD4 at GLS (top) and SLC1A5 (bottom), downloaded from the University of California, Santa Cruz (UCSC) ENCODE Genome Browser. TEAD4-associated regions (hatched box) correlate with H3K4Me3 and H3K27Ac near exon 1 (black). (H and I) ChIP of MCF10A-HER2 cells transduced with shCtrl, shEphA2#1, or shEphA2#2. Relative immunoprecipitated genomic DNA was determined by qRT-PCR and normalized to IgG controls. EphA2 knockdown was confirmed by Western blotting (I). Data are means ± SEM from four independent experiments. #P < 0.005, two-way ANOVA and Tukey’s post hoc test. n.s., nonsignificant.

To identify the mechanism by which YAP and TAZ promote GLS and SLC1A5 expression, we searched the Encyclopedia of DNA Elements (ENCODE) database (57), which consolidates submitted chromatin immunoprecipitation–sequencing (ChIP-seq) data, to identify transcription factor binding sites near the promoters of genes critical in glutaminolysis. The TEAD family of transcription factors is necessary for YAP/TAZ functionality as a transcriptional regulator, supporting transcription of proliferative genes while suppressing pro-apoptotic gene expression (1619). ChIP-seq data from ENCODE indicated that TEAD4 localizes to the promoters of both GLS and SLC1A5 (Fig. 5G), as demonstrated by enhanced histone H3 methylation (H3K4Me3) and acetylation (H3K27Ac) defining promoter areas and active enhancer elements, respectively (58, 59). ChIP-seq data for other members of the TEAD family of transcription factors were unavailable (60, 61).

To investigate whether YAP and TAZ may bind to GLS and SLC1A5 promoters through TEAD4 in HER2-positive cancer cells, we performed ChIP in control or EphA2 knockdown cells using antibodies targeting immunoglobulin G (IgG) or TEAD4 (Fig. 5, H and I). Specificity for TEAD4 was assessed using the positive controls CYR61 and CTGF, as well as the negative control FAT3, whose expression occurs either dependently (CYR61 and CTGF) or independently (FAT3) of the YAP/TAZ/TEAD4 complex (16). Immunoprecipitated DNA for CYR61 and CTGF, but not for FAT3, was enriched in TEAD4 immunocomplexes from control cells (Fig. 5H). In addition, TEAD4-immunoprecipitated GLS and SLC1A5 DNA were significantly enriched over IgG controls in control cells. EphA2 knockdown significantly reduced this enrichment for CYR61, CTGF, GLS, and SLC1A5, suggesting that EphA2 is required to induce TEAD4 binding to GLS and SLC1A5 promoters. Like TEAD4, YAP was also found to be associated with GLS and SLC1A5 promoters (fig. S6). Together, these results support the finding that EphA2 activates the YAP/TAZ/TEAD4 complex to promote glutamine metabolism through the enhancement of GLS and SLC1A5 expression.

Increased activation of the EphA2-YAP/TAZ signaling axis correlates with increased glutamine metabolism and reduced survival in breast cancer patients

To determine whether the EphA2-YAP/TAZ signaling axis is relevant to glutamine metabolism in human breast cancer, expression of EphA2, YAP, TAZ, TEAD4, GLS, and SLC1A5 from patient samples was analyzed in a previously published breast cancer database (62). Consistent with EphA2 overexpression promoting enhanced YAP/TAZ activity, strong correlations exist between expression of EphA2 and YAP, as well as EphA2 and TAZ (Fig. 6A). Additional positive correlations between EphA2 and TEAD4, SLC1A5, and GLS were also observed (Fig. 6A). Furthermore, YAP/TAZ expression is positively correlative with GLS and SLC1A5 (Fig. 6B), supporting the finding that YAP and TAZ regulate EphA2-mediated glutamine metabolism. To determine whether the EphA2-YAP/TAZ-glutaminolysis signaling axis affects clinical outcomes, breast cancer databases from the Gene Expression Omnibus (GEO) (63) were examined for overall survival (OS) in patients with low or high combined expression of EphA2, YAP, TAZ, TEAD4, GLS, and SLC1A5. High expression of genes involved in the EphA2-YAP/TAZ signaling axis significantly correlated with decreased OS in HER2-positive breast cancer patients (Fig. 6C). A similar trend was also observed with glutaminolysis genes, with high expression of GLS and SLC1A5 also strongly associated with decreased OS (Fig. 6D), strongly supporting the importance of EphA2, YAP/TAZ, and GLS/SLC1A5 in promoting tumor progression. Individually, recurrence-free survival (RFS) in patients with high expression of YAP, TAZ, or EphA2 (fig. S7, A to C) significantly correlated with decreased RFS in all breast cancer subtypes. However, stratification for HER2-positive breast cancer patient data resulted in greater hazard ratios (HRs) for high expression of YAP, TAZ, or EphA2 (fig. S7, D to F), suggesting that EphA2 and YAP/TAZ may play a more clinically relevant role in HER2-positive breast cancer compared to cases lacking HER2 overexpression. Still, further bioinformatic analysis is necessary to completely explore the link between EphA2 and YAP/TAZ.

Fig. 6 EphA2 and YAP/TAZ expression correlates with increased glutaminolysis gene expression in human breast cancer patient data.

(A and B) Log2 (mRNA) expression from the Minn Breast 2 database (n = 121). Pearson correlations (r) and t distribution (P) coefficients are shown. (C and D) Kaplan-Meier analysis of OS in HER2-positive breast cancer patients (C, n = 73; D, n = 117) exhibiting low (black) or high (red) (C) EPHA2, YAP, TAZ, and TEAD4 or (D) GLS and SLC1A5 expression. (E to G) Immunohistochemical analysis of YAP in HER2-positive samples from a human breast carcinoma TMA. Samples were categorized as metastatic or nonmetastatic, and images [represented in (E)] were assessed for YAP staining intensity per cell (F) and per sample (G). Scale bar, 10 μm. ***P < 0.005, Student’s t test. (G) Samples from (F) further stratified according to the amount of nuclear localization of YAP using the median percentage as the cutoff. χ2 analysis was calculated. (H and I) BPTES sensitivity in MMTV-Neu (H) or MCF10A-HER2 (I) cells retrovirally transduced with pBABE (“Control”) or pBABE-EphA2 (“EphA2”) and treated with vehicle or BPTES (10 μM) for 24 hours. Cell counts were calculated as fold change from vehicle controls from three independent experiments. ***P < 0.005, two-factor ANOVA (H) or one-way ANOVA (I) and Tukey’s post hoc test.

Survival data suggest that enhanced expression of YAP and TAZ may confer an advantageous environment for tumor progression in HER2-positive breast cancer. To examine the role of YAP in tumor progression, we examined YAP protein expression and YAP nuclear localization in nonmetastatic and metastatic HER2-positive breast cancer samples by performing immunohistochemistry (IHC) on a breast cancer tissue microarray (TMA) (Fig. 6, E to G). Compared to localized, nonmetastatic disease, YAP protein was significantly higher in tumors scored as metastatic (Fig. 6, E and F). Likewise, nuclear YAP was found to be more strongly associated with metastatic progression (Fig. 6, E and G). Our laboratory has previously shown that metastatic spread is also associated with enhanced EphA2 phosphorylation at Ser897, a measure of ligand-independent EphA2 activity, suggesting that EphA2 and YAP/TAZ activity correlate in advanced disease (2). Together, these data are strongly supportive of EphA2 and YAP/TAZ coordinating to promote tumor progression in HER2-positive breast cancer, resulting in metastatic spread and reduced survival of patients.

Finally, we examined whether EphA2-overexpressing cells may be more sensitive to GLS inhibition. MMTV-Neu or MMTV-Neu-EphA2 cells were treated with vehicle or the GLS inhibitor BPTES, and cells were counted as a measure of cellular growth (Fig. 6H). GLS inhibition reduced cellular growth, but this effect was significantly more remarkable in cells overexpressing EphA2 than in control cells (Fig. 6H). A similar effect was observed in the MCF10A-HER2 (Fig. 6I), strongly supporting the clinical relevance of EphA2-mediated glutamine metabolism.


The Hippo signaling pathway is an emerging growth control and tumor suppressor pathway that suppresses cell proliferation and stem cell function (27, 28). Genome-wide analyses identified increased expression and nuclear localization of the Hippo pathway downstream effectors and transcriptional coactivators YAP and TAZ in human cancer (16). Although few mutations in components of the Hippo pathway have been discovered, amplification and overexpression of YAP and TAZ are observed in human breast cancer (1215, 29). Although up-regulation of YAP/TAZ is linked to proliferation, stem cell function, and drug resistance in breast cancer, how the Hippo pathway is deregulated is poorly understood because of lacking evidence of oncogenic driver mutations in this pathway. Here, we identify EphA2 as an RTK that stimulates YAP/TAZ activity. Overexpression of EphA2 induced YAP and TAZ nuclear accumulation in cell lines and MMTV-Neu tumors and the expression of YAP/TAZ target genes (Cyr61, Ctgf, and Inhba), demonstrating that EphA2 functionally stimulates YAP/TAZ activity. Furthermore, the EphA2-YAP/TAZ signaling axis is clinically relevant because there is a strong correlation between EphA2 and YAP/TAZ expression in a human breast cancer data set, with hyperactivation of the pathway enhancing metastatic potential and reducing survival, particularly in HER2-positive patients.

Cell-cell communication has been identified as the primary mode of YAP/TAZ regulation, and active Rho signaling is required for full activation (41). YAP/TAZ act as sensors of cell density to limit proliferation in response to excessive cell contacts (24). However, dysregulation of the upstream communicative pipeline hyperactivates YAP/TAZ, thus eliminating their antitumor role (2730). The unique bidirectional nature of EphA2 signaling mirrors these intercellular communications that modulate YAP/TAZ activation. Whereas interactions between ephrin-A1 and EphA2 located on neighboring cells suppress cell proliferation, EphA2 ligand–independent signaling promotes tumor growth and cell migration (3, 4, 68). We show that high EphA2 expression is associated with poorer survival in lymph node–positive breast cancer patients, especially those with HER2-positive tumors, confirming the role of EphA2 in tumor progression. Although we cannot completely rule out how loss of EFNA1 may be contributing to this mechanism, in part, our data suggest that these tumor-promoting processes are initiated by EphA2 ligand–independent activation of Rho (2, 9, 48). Loss of EFNA1 results in increased EphA2 phosphorylation at Ser897, a marker for ligand-independent signaling (6). We have previously shown that kinase activity is also required for EphA2 to regulate glutamine metabolism (2). Although we observed that phosphorylation of Tyr588 (6466) is decreased in EFNA1 knockout tumors, the EphA2 kinase activity may still be important and can be regulated in a ligand-independent manner. Tyr588 was often dephosphorylated in tumor cells by phosphatases, such as low–molecular weight protein tyrosine phosphatases (LMW-PTP), whereas EphA2 kinase activity was intact in these cells (67). Collectively, our data support the finding that EphA2-mediated YAP/TAZ activation requires intact Rho signaling because Rho/ROCK inhibition prevented nuclear accumulation through protein destabilization. Although YAP and TAZ are commonly overexpressed and/or hyperactivated in human cancers, few mutations have been observed (27). Our data suggest that the RTK EphA2 may contribute to a previously unknown mechanism of YAP/TAZ dysregulation in human cancers.

Rho is known to promote GLS activity, as the GLS1 inhibitor 968 was first identified as an inhibitor for Rho-dependent transformation (40). However, the mechanism by which Rho regulates GLS activity was poorly understood. Our data support a role of TAZ in mediating Rho-dependent GLS activation. Previous studies show that Rho promotes YAP activity through actin dynamics and phosphorylation of LATS1/2 (68, 69), although Rho may also promote YAP/TAZ activity independently of Hippo signaling (41, 70). However, we were unable to detect changes of LATS1 phosphorylation in EphA2-overexpressing cells treated with Rho or ROCK inhibitors. It is possible that EphA2 induced YAP and TAZ activation through LATS2 or a LATS-independent mechanism, such as actin-binding angiomotin proteins that can bind to YAP directly (71). In addition to enhancement of YAP/TAZ nuclear translocation, EphA2 appears to also increase the expression of YAP/TAZ in human breast cancer. Analyses of an ONCOMINE data set revealed a positive correlation between EphA2 and YAP/TAZ mRNA expression. It is currently unclear how EphA2 may increase YAP/TAZ expression at the mRNA level.

Whereas the focus of the present study remained on GLS and SLC1A5, the possibility remains that EphA2 and YAP/TAZ may promote the expression of additional genes involved in glutamine metabolism. Numerous glutamine transporters, including SLC38A1 and LAT1, have been attributed to tumor promotion (72, 73), but strong evidence supports SLC1A5, also known as ASCT2, as an important contributor to tumor growth in breast cancer (74). Furthermore, two isozymes of glutaminase, GLS (kidney type) and GLS2 (liver type), have been identified in mammals. Whereas GLS overexpression is commonly observed in human cancer (38, 39, 75, 76), the role of GLS2 is more controversial, with both tumor-promoting and tumor-suppressive roles being described (7781). Although we are unable to rule out GLS2 regulation, GLS was previously identified as a transcriptional target of YAP and TAZ in pulmonary hypertension (82), suggesting that this more widely expressed form of GLS (37) may be targeted by YAP/TAZ in multiple disease contexts. Additional YAP targets in glutamine metabolism include glutamine synthetase and the glutamine transporter SLC38A1 in liver cancer (83, 84), implying that the EphA2-YAP/TAZ signaling axis may enhance glutamine metabolism through many gene targets. However, in the context of breast cancer, transcriptional regulation of the more catalytically active GLS isoform, GAC, by TAZ supports the tumor-promoting properties associated with EphA2 and YAP/TAZ overexpression. EphA2 overexpression increased sensitivity to GLS inhibition, suggesting that EphA2 and YAP/TAZ may be potential biomarkers for a favorable patient response to GLS inhibitors (CB-839) currently in clinical trials (85).

In addition to the YAP/TAZ transcriptional coactivators, the proto-oncogene MYC and the nuclear factor κB (NF-κB) pathway are known to increase glutamine metabolism in cancer (39, 40, 55, 56). Although we cannot rule out a MYC contribution to the observed changes in glutamine metabolism, we demonstrated that the TEAD4 transcription factor is bound to the GLS and SLC1A5 promoter regions, and these interactions were significantly reduced after EphA2 knockdown. Overexpression of MYC has been shown to only partially rescue the proliferation of YAP/TAZ knockdown in human breast cancer cells (16), possibly due to negative feedback on YAP/TAZ-TEAD activity (86). Furthermore, the TEAD family of transcription factors have been identified as the primary recruiters of YAP/TAZ to chromatin (16), supporting the TEAD family of transcription factors as a major contributor to EphA2-YAP/TAZ–dependent glutamine metabolism in HER2-positive breast cancer. Still, TEAD4 and other TEAD proteins, including TEAD1 to TEAD3, have been reported to exhibit redundancy (87, 88). How the remaining TEAD family members contribute to EphA2-mediated glutamine metabolism needs further investigation.

Our data indicated that the glutamine transporter SLC1A5 and glutaminase GLS are differentially regulated, as our results suggested that GLS was more weakly associated with the EphA2-YAP/TAZ-TEAD4 pathway than SLC1A5. Although indications of defined regulatory mechanisms contributing to SLC1A5 gene expression in cancer have been limited, previous evidence has demonstrated that GLS gene expression is enhanced by other transcription factors, most notably, NF-κB in HER2-positive breast cancer cells (39, 40). It was also noted NF-κB–mediated GLS gene expression is dependent on Rho-GTPase activity (40), consistent with our TAZ-TEAD4 data. Together, we speculate that upon activation of Rho-GTPase, both NF-κB and TAZ-TEAD may promote glutamine metabolism in HER2-positive breast cancer by up-regulating GLS expression. However, our data strongly support an EphA2- and YAP/TAZ-dependent up-regulation of SLC1A5, a novel and unique regulatory mechanism.

Overall, these findings define a novel role of the RTK EphA2 in the activation of the transcriptional coactivators YAP and TAZ to promote glutamine metabolism in HER2-positive breast cancer. Targeting the EphA2-YAP/TAZ signaling axis reveals great promise in reducing glutamine metabolism in HER2-positive breast cancer, possibly extending to other subtypes and additional forms of cancer. Therefore, our work describes the potential clinical benefit of using EphA2, YAP, and TAZ as future biomarkers in HER2-positive breast cancer to predict patient outlook and response to pharmacological agents that inhibit EphA2 or glutamine metabolism.


Cell culture

MMTV-Neu tumor cells, which were isolated and provided by R. Cook (Vanderbilt University), were cultured as reported previously (89, 90). MCF10A-HER2 cells were generated and cultured as previously described (89, 91, 92). All cells were maintained at low passages. Transient overexpression of EphA2 or Rho mutants was achieved by adenoviral infection with Ad-GFP, Ad-EphA2, Ad-Rho-T19N, or Ad-Rho-Q63L adenovirus for 72 hours. EphA2 was stably overexpressed by infection with pBABE or pBABE-EphA2 lentivirus, provided by W. Song (Vanderbilt University), and cells were selected with puromycin (2 μg/ml) for 5 days. Flag-YAP was overexpressed using pBABE-YAP1 (a gift from J. Brugge; plasmid #15682, Addgene) (12). To achieve transient knockdown, cells were transfected with ON-TARGETplus Non-Targeting Pool siRNA (D-001810-10), siGENOME YAP1 SMARTpool (M-012200-00) or individual siRNAs (siYAP#1, M-012200-01; siYAP#2, M-012200-02; siYAP#3, M-012200-03), and siGENOME WWTR1 SMARTpool (M-016083-00) or individual siRNAs (siTAZ#1, M-016083-01; siTAZ#2, M-016083-02; siTAZ#3, M-016083-04) (GE Healthcare) (all at 40 nM), as indicated, with Lipofectamine RNAiMAX reagent (Invitrogen), as per the manufacturer’s instructions, for 72 hours. To generate stable EphA2-knockdown cells, shEphA2#1 (sense, 5′-CGGACAGACATATAGGATATT-3′) or shEphA2#2 (sense, 5′-GCGTATCTTCATTGAGCTCAA-3′) was cloned into pLKO.1-blast (a gift from K. Mostov; plasmid #26655, Addgene) (93), and MCF10A-HER2 cells were transduced with lentivirus, followed by selection with blasticidin (8 μg/ml) for 2 days. To inhibit Rho or ROCK activity, MMTV-Neu cells were treated with phosphate-buffered saline (PBS) control, Rho inhibitor I (CT04; 3 μg/ml; #CT04, Cytoskeleton), or the ROCK inhibitor Y-27632 (10 μM; #1254, Tocris) for 4 to 6 hours, as indicated in serum-free medium. Cells were incubated with EphA2-Fc (R&D Systems) or Fc (1 μg/ml) for 16 hours.

Antibodies and immunoblotting

The following antibodies were used at 1:1000 unless otherwise indicated: EphA2 (Santa Cruz Biotechnology, #sc-924), EphA2 [#6997, Cell Signaling Technology (CST)], pEphA2-Tyr588 (#12677S, CST), pEphA2-Ser897 (1:250; #AP3722a, Abgent), EFNA1 (#AF702, R&D Systems), YAP/TAZ (#8418, CST), pYAP-Ser397 (#13619, CST), pYAP-Ser127 (#13008, CST), pTAZ-Ser89 (1:500; #75275, CST), MLC (#3672, CST), pMLC2-Ser19 (#3671, CST), LATS1 (#3477, CST), pLATS1-Ser909 (#9157, CST), ERBB2 (#MS-730-P0, Neomarkers), β-actin (#sc-47778, Santa Cruz Biotechnology), and β-tubulin (#T4026, Sigma-Aldrich). Secondary antibodies used were IRDye 680LT goat anti-mouse (1:20,000; #925-68020, LI-COR), IRDye 800CW goat anti-rabbit (1:10,000; #925-32211, LI-COR), horseradish peroxidase (HRP)–conjugated goat anti-rabbit (1:5000; #W401B, Promega), and HRP-conjugated goat anti-mouse (1:5000; #W402B, Promega). For immunoblotting, precleared lysates were electrophoresed by SDS–polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes, which were blocked for 1 hour in 5% nonfat dry milk or 5% bovine serum albumin (BSA). Membranes were incubated with primary antibodies overnight, followed by incubation with secondary antibodies for 1 hour at room temperature, and imaged using LI-COR Odyssey or enhanced chemiluminescence (Clarity ECL kit, Bio-Rad). Densitometry was performed using ImageJ software, and measured proteins were normalized using tubulin controls.


Tumors from EFNA1+/+ and EFNA1−/− MMTV-NeuT female mice, which were provided by V. Youngblood (Vanderbilt University) (2), were subjected to immunohistochemical staining (IHC), as described previously (2). Briefly, formalin-fixed paraffin-embedded tumor sections were rehydrated, and antigen retrieval was achieved using Retrievagen A (BD Biosciences), according to the manufacturer’s instructions. After blocking for endogenous peroxidases and several washes in PBS, tissue sections were probed with antibodies against EphA2 (10 μg/ml; #34-7400, Zymed Laboratories), YAP (1:400; #14074, CST), or TAZ (1:400; #ab84927, Abcam) overnight at 4°C. After washing, samples were incubated with biotinylated anti-rabbit IgG (#BA-1000, Vector Laboratories) for 1 hour at room temperature, and streptavidin peroxidase reagents (#51-75477E, BD Pharmingen), liquid diaminobenzidine (DAB) (#00-2014, Invitrogen), and hematoxylin were used for staining. Stained tissues were mounted with Cytoseal XYL, and images of at least four fields of view were obtained using an Olympus inverted fluorescence microscope (40×).

Nuclear localization and staining intensity were calculated using CellProfiler software. Briefly, to quantitate nuclear localization, original images were divided into red and blue channels, and nuclei were identified as the primary objects from the blue channel, as defined between 10 and 40 pixels in diameter. Thresholding was controlled using the Global Otzu method (three-class thresholding), with a threshold correction factor of 2 and lower and upper bounds of 0 and 1.0, respectively. Smoothing was automatically applied, and clumped objects were identified on the basis of intensity. DAB-stained nuclei were identified as secondary objects, using the propagation method with an automatic threshold and a regularization factor of 0.05. Positive nuclei were filtered on the basis of mean intensity values and classified as positive when the mean intensity value was greater than the threshold of 0.2. To detect EphA2 and YAP DAB staining intensity per cell, the original image was divided into red and blue channels, and overall DAB intensity was measured from the red channel image. Nuclei were identified as the primary objects, as described above, except for use of two-class thresholding with a correction factor of 1. To calculate intensity per cell, total DAB intensity was divided by the total number of nuclei per image.


MMTV-Neu or MCF10A-HER2 cells were infected with adenovirus, as described above. Subsequently, cells were fixed with 2% paraformaldehyde for 15 min, permeabilized with 1% Triton X-100 in PBS for 5 min, and blocked with 3% BSA in PBS for 1 hour, all at room temperature. Cells were probed with anti-YAP (1:100; #14074, CST) or anti-TAZ (1:50; #ab84927, Abcam) overnight at 4°C and washed multiple times with 0.5% Tween 20 in PBS, followed by incubation with Alexa Fluor 594 anti-rabbit secondary antibody (1:500; #A-11012, Invitrogen) for 1 hour at room temperature. After several washes, coverslips were mounted using SlowFade Diamond antifade reagent containing DAPI (#S36963, Molecular Probes). Images of at least two fields of view per sample, as indicated, were obtained using an Olympus inverted fluorescence microscope (40×) or a DeltaVision OMX Super-Resolution microscope (60×). ImageJ software was used to quantitate YAP- or TAZ-positive nuclei and generate z-stack videos.

Glutamate assay

Intracellular glutamate concentrations were determined in duplicate using the Glutamate Assay kit (Sigma-Aldrich), according to the manufacturer’s instructions and as previously described (2). MCF10A-HER2 or MCF10A-HER2-EphA2 cells were transfected with the designated siRNAs, as described above, for 48 hours and then cultured in serum-free and glutamine-free Dulbecco’s modified Eagle’s medium/F12 medium for 24 hours. Cells were stimulated with EGF (20 ng/ml) and l-glutamine (2.5 mM) with or without 5% fetal bovine serum for 0, 5, 10, or 20 min, as indicated. Glutamate concentrations were calculated from known standards, and all data are corrected according to baseline values.

Cell proliferation assays

MCF10A-HER2 cells were transfected to knockdown YAP or TAZ, as described above. After 24 hours, cells were supplemented with vehicle or DKG (dimethyl 2-oxoglutarate; 4 mM; Sigma-Aldrich), and cells were counted after 72 hours. Cell number was normalized to nontargeting controls in each treatment group. To assess sensitivity to BPTES, MMTV-Neu, MMTV-Neu-EphA2, MCF10A-HER2, or MCF10A-HER2-EphA2 cells were treated with vehicle or BPTES (10 μM) for 24 hours, and cells were counted. Cell number was normalized to untreated controls for each cell line.

Quantitative real-time polymerase chain reaction

Cells were prepared as indicated, and RNA was collected using the RNeasy kit (Qiagen) and converted to complementary DNA by reverse transcription using iScript (Bio-Rad) according to the manufacturer’s instructions. Using TaqMan Fast Advanced Master Mix, samples were amplified as previously described (2) in triplicate on the StepOnePlus (Applied Biosystems) using the following Taqman probe sets: EPHA2 (Hs00171656_m1), YAP (Hs00902712_g1), TAZ (WWTR1, Hs00210007_m1), Cyr61 (Mm00487498_m1), Ctgf (Mm01192933_g1), Inhba (Mm00434339_m1), CYR61 (Hs00998500_g1), AXL (Hs01064444_m1), GLS (GAC splice variant, Hs01022166_m1), SLC1A5 (Hs01056542_m1), Actb (Mm02619580_g1), or ACTB (Hs01060665_g1). Quantitation was performed using the ΔΔCt method.

Chromatin immunoprecipitation

MCF10A-HER2/Flag-YAP and stable EphA2 knockdown cells were generated as described above and subjected to ChIP, as previously described (94), with modification. Briefly, cells were fixed in 1% formaldehyde solution [50 mM Hepes (pH 7.5), 100 mM NaCl, 1 mM EDTA, 0.5 mM EGTA, and 1% formaldehyde] for 10 min at room temperature, followed by quenching using 125 mM glycine. Cross-linked cells were washed and collected in PBS, and nuclear enrichment was achieved by resuspension in LB1 [50 mM Hepes (pH 7.5), 140 mM NaCl, 1 mM EDTA, 10% glycerol, 0.5% NP-40, and 0.25% Triton X-100] for 10 min at 4°C and LB2 [10 mM tris (pH 8), 200 mM NaCl, 1 mM EDTA, and 0.5 mM EGTA] for 5 min at 4°C, both with rotation. The collected nuclear pellets were lysed in LB3 [10 mM tris (pH 8), 100 mM NaCl, 1 mM EDTA, 0.5 mM EGTA, 0.1% Na-deoxycholate, and 0.5% N-lauroylsarcosine], and samples were sonicated using a Bioruptor XL to generate DNA fragments of 500 to 1000 base pairs, with Triton X-100 (1% final) and protease inhibitor cocktail (Roche) added afterward. Diluted nuclear extracts were precleared with salmon sperm DNA/protein A agarose slurry (EMD Millipore) for 30 min at 4°C, with rotation, and 50 μl was saved as the input control. The beads were pelleted, and the supernatant was immunoprecipitated with 3 μg of normal rabbit IgG (#sc-3888, Santa Cruz Biotechnology), TEAD4 (#ab58310, Abcam), or FLAG M2 (#F2426, Sigma-Aldrich) at 4°C overnight with rotation. Immunocomplexes were collected with salmon sperm DNA/protein A agarose slurry for 1 hour at 4°C, with rotation, followed by multiple rounds of washing and elution twice in 200 μl of elution buffer (1% SDS and 0.1 M NaHCO3) for 15 min at room temperature, with agitation. Cross-links were reversed by addition of 16 μl of 5 M NaCl and 1 μg of ribonuclease A at 65°C overnight, followed by treatment with proteinase K (20 μg) for 1 hour at 50°C. Immunoprecipitated DNA was collected by phenol-chloroform extraction, followed by ethanol precipitation with 20 μg of glycogen and resuspended in 50 μl of 1× TE buffer [10 mM tris (pH 8) and 1 mM EDTA]. For quantitation, 1/50 of the DNA, 1 mM forward primer and 1 mM reverse primer (table S1), and SYBR Green PCR Master Mix (Applied Biosystems) were combined, and qRT-PCR was performed in triplicate using the StepOnePlus (Applied Biosystems) system.

Human breast cancer TMA and gene expression data set analysis

YAP immunohistochemical staining was performed on a commercially available human breast cancer TMA (#CC08-10-001, Cybrdi). Using two fields of view per sample, YAP staining intensity and nuclear localization were determined using CellProfiler software. HER2-overexpressing tumors were further stratified for metastases based on the TNM classification and staging system, with patients staged with lymph node (N > 0) and/or distant metastases (M = 1) being classified as having metastatic disease. For YAP activation, the percentage of cells with YAP nuclear localization was calculated using the CellProfiler software, as described above, and samples were classified as having low or high activation when below or above the median percentage of YAP-positive nuclei (27.5%), respectively.

To assess TEAD4 binding sites, TEAD4 ChIP-seq data acquired from multiple cell lines by the Myers Laboratory were examined using the ENCODE database ( (57). H3K4Me3 (GEO GSM733680 and GSM733720) and H3K27Ac (GEO GSM733656 and GSM733674) data submitted by the Bernstein Laboratory were collected to indicate promoter regions and active enhancer elements, respectively.

To determine whether gene expression is correlative in human breast tissue, the Minn Breast 2 database (62) was searched on ONCOMINE ( (62, 95) for mRNA expression of EPHA2 (gene ID 203499_at), YAP (gene ID 213342_at), WWTR1 (TAZ; gene ID 202133_at), TEAD4 (gene ID 204281_at), GLS (GAC isoform–specific; gene ID 221510_s_at), and SLC1A5 (gene ID 208916). The reported data were collected and reported on a log2 scale.

EPHA2 (203499_at), YAP (213342_at and 224894_at), WWTR1 (TAZ; 202133_at), TEAD4 (41037_at), GLS (GAC isoform–specific; 221510_s_at), and SLC1A5 (208916_at) OS and RFS were determined using Kaplan-Meier Plotter online ( (63), using automatic cutoff. Data were analyzed from 1764 breast cancer patients and 73 to 156 patients exhibiting tumors with the HER2-positive subtype. EPHA2 RFS was examined in lymph node–positive patients. Patient samples were classified according to gene expression as high or low expressers of each designated gene or gene set.

Statistical analysis

Correlation was determined using Pearson’s correlation coefficient and tested using Student’s t distribution. For statistical analysis of Kaplan-Meier plots, HR with 95% confidence intervals and log-rank P values were calculated from stratified patient data. Categorical data were analyzed by χ2 analysis, and for comparisons between two groups, unpaired Student’s t test was performed. Statistical analysis for multiple comparisons was performed using one- or two-way ANOVA for one or more characteristics, respectively. To identify where differences existed among groups, post hoc analyses consisting of Dunnett’s or Tukey’s methods were completed, as indicated. All statistical tests performed were two-tailed, and P < 0.05 was considered statistically significant.


Fig. S1. EphA2 activates YAP/TAZ in MCF10A-HER2 cells.

Fig. S2. EphA2–ephrin-A1 interactions do not significantly contribute to YAP/TAZ activation.

Fig. S3. Rho and ROCK inhibition did not enhance LATS1 phosphorylation in MMTV-Neu cells.

Fig. S4. EphA2-mediated YAP/TAZ activation requires Rho catalytic activity.

Fig. S5. YAP and TAZ promote glutamine metabolism in MCF10A-HER2-EphA2 cells.

Fig. S6. YAP and TEAD4 are associated with GLS and SLC1A5 promoters.

Fig. S7. YAP/TAZ and EphA2 expression strongly correlates with decreased patient survival in HER2-positive breast cancer.

Table S1. ChIP primers.

Movie S1. EphA2 overexpression leads to intranuclear accumulation of YAP.

Movie S2. Nuclear exclusion of YAP upon Rho inhibition.


Acknowledgments: We thank V. Youngblood and R. Cook (Vanderbilt University) for providing MMTV-Neu tumor samples, W. Song (Vanderbilt University) for providing pBABE-EphA2 vector, J. Brugge (Harvard Medical School) for the pBABE-YAP1 plasmid, and K. Mostov (University of California, San Francisco) for the pLKO.1-blast plasmid. In addition, we express our gratitude to the ENCODE Consortium, as well as the Bradley Bernstein laboratory (Broad Institute) and the Richard Myers laboratory (Hudson Alpha Institute for Biotechnology), for providing a platform to examine ChIP-seq data. Funding: Super-resolution microscopy experiments were performed, in part, through the use of the Vanderbilt Cell Imaging Shared Resource (supported by NIH grants CA68485, DK20593, DK58404, DK59637, and EY08126). Sample preparation was supported by the Vanderbilt Translational Pathology Shared Resource (National Cancer Institute/NIH grant 2P30 CA068485-14). This work was supported by NIH grants R01 CA177681 (to J.C.), R01 CA95004 (to J.C.), T32 CA009592 (to D.N.E. and V.M.N.), F30 CA216891-01 (to E.S.), and F31 CA220804-01 (to L.C.K.); a VA Merit Award 5101BX000134 (to J.C.); a pilot project from the Vanderbilt Breast SPORE grant CA098131 (to J.C.); and a Susan G. Komen Postdoctoral Fellowship Award (#17480733 to D.N.E.). Author contributions: D.N.E., A.B.R., and J.C. conceptualized the project. D.N.E., V.M.N., S.W., E.S., D.M.B.-S., and L.C.K. performed the experiments and analyzed the data. D.N.E. and J.C. wrote the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All unique plasmids, cell lines, and mouse models not commercially available require a materials transfer agreement from the Vanderbilt University Medical Center.

Stay Connected to Science Signaling

Navigate This Article