Glutamine‐utilizing transaminases are a metabolic vulnerability of TAZ/YAP‐activated cancer cells

Abstract

The transcriptional regulators TAZ and YAP (TAZ/YAP) have emerged as pro‐tumorigenic factors that drive many oncogenic traits, including induction of cell growth, resistance to cell death, and activation of processes that promote migration and invasion. Here, we report that TAZ/YAP reprogram cellular energetics to promote the dependence of breast cancer cell growth on exogenous glutamine. Rescue experiments with glutamine‐derived metabolites suggest an essential role for glutamate and α‐ketoglutarate (AKG) in TAZ/YAP‐driven cell growth in the absence of glutamine. Analysis of enzymes that mediate the conversion of glutamate to AKG shows that TAZ/YAP induce glutamic–oxaloacetic transaminase (GOT1) and phosphoserine aminotransferase (PSAT1) expression and that TAZ/YAP activity positively correlates with transaminase expression in breast cancer patients. Notably, we find that the transaminase inhibitor aminooxyacetate (AOA) represses cell growth in a TAZ/YAP‐dependent manner, identifying transamination as a potential vulnerable metabolic requirement for TAZ/YAP‐driven breast cancer.

Introduction

Altered cellular energetics is an established hallmark of cancer 1, a premise supported by observations that oncogenes directly modulate metabolic circuits required for tumorigenesis 2, 3, 4, 5. Oncogenic alterations in metabolic enzymes contribute to a variety of cancer‐associated traits, including uncontrolled cell proliferation, alterations in cell polarity, increased metastatic ability, and evasion from cell death, indicating that aberrant cellular metabolism acts as a tumorigenic “driver” 2, 6, 7, 8, 9. Cancer‐specific metabolic reprogramming therefore represents an exciting avenue for the development of novel diagnostic tools and targeted cancer therapy 10, 11.

The paralogous transcriptional regulators TAZ and YAP (herein referred to together as TAZ/YAP) have emerged as central factors in cancer biology. Nuclear TAZ/YAP activity has been shown to drive cell proliferation, survival, and mobility, and has key roles in directing cell fate 12. Increased levels and activity of TAZ/YAP have also been shown to correspond to high‐grade tumors that generally lack effective therapeutics 12, 13. Advanced breast cancers in particular exhibit high nuclear TAZ/YAP levels and rely on nuclear TAZ/YAP transcriptional activity for driving breast cancer cell growth and aggressiveness 14, 15. Thus, in‐depth understanding of the processes regulated by TAZ/YAP may provide important insight into the etiology of cancer and offer therapeutic opportunities.

Here, we report that TAZ/YAP promote glutamine dependence in breast cancer cells and activate the expression of glutamine‐utilizing transaminases to support cell growth. We found that glutamine deprivation greatly reduces the growth of breast cancer cells with high TAZ/YAP levels and that knockdown of TAZ/YAP mitigates cell death caused by glutamine depletion. Interestingly, we found that TAZ/YAP promote the expression of the transaminases GOT1 and PSAT1, and blockade of transamination with aminooxyacetate (AOA) suppresses the growth of breast cancer cells in a TAZ/YAP‐dependent manner. Collectively, our data indicate that glutamine addiction of breast cancer cells is mediated by TAZ/YAP and suggest that targeting of transamination could be exploited for breast cancer therapies to attenuate TAZ/YAP‐driven tumor growth.

Results and Discussion

TAZ/YAP control glutamine dependency of breast cancer cells

Given the emerging oncogenic roles of TAZ/YAP 12, 13, we thought to analyze the nutrient requirements associated with TAZ/YAP activation to uncover susceptible metabolic processes for potential targeting of cancer cells with hyper‐activated TAZ/YAP. To this end, we identified gene expression changes resulting from siRNA‐mediated depletion of TAZ/YAP in MDA‐MB‐231 breast cancer cells (data from Enzo et al 16), which is a cell line that exhibits high nuclear TAZ/YAP activity 17, and performed Gene Set Enrichment Analysis (GSEA) to determine whether any relationships exist with gene sets representing distinct metabolic features 8 (Dataset EV1). Among 38 molecular signatures examined, 13 were downregulated upon TAZ/YAP deficiency (Table EV1), suggesting a role for TAZ/YAP in stimulating these metabolic processes. Notably, the gene set representing amino acid metabolism showed the strongest statistical association with TAZ/YAP activity (Fig 1A and Table EV1).

Figure 1. Breast cancer cell lines with elevated levels of TAZ/YAP cancer exhibit glutamine dependence.

• A
  GSEA shows enrichment of amino acid metabolism‐associated genes in gene expression signatures induced by TAZ/YAP.
• B
  Lysates were isolated from density‐matched mammary cells cultured in complete medium and examined for endogenous TAZ/YAP protein levels by immunoblotting. GAPDH levels were used to control loading.
• C
  Cell growth of a panel of human breast cancer cell lines and a non‐malignant human mammary epithelial cell line (HMEC) HMT‐3522 S1 was measured in glutamine (Q)‐free medium and then normalized to growth in complete medium. The reduction in culture size following glutamine starvation is defined as glutamine‐dependent growth (red), while the remaining growth following glutamine starvation is defined as glutamine‐independent growth (blue). The average growth from three independent experiments is shown (±SD).
• D, E
  For the indicated cell lines, glutamine dependence measured in (C) was plotted against the expression levels of TAZ/YAP measured in (B) (shown in D), or the relative expression levels of the TAZ/YAP targets CTGF/CYR61/ANKRD1/EDN1 (determined from data available from the CCLE) (shown in E) to examine the correlation (A.U., arbitrary units) between these two biological features (see Materials and Methods for details).

Targeting amino acid metabolic enzymes in cancer cells has shown promise as a therapeutic strategy. In particular, enzymes important for metabolizing the “non‐essential” amino acid glutamine have emerged as important mediators of cancer cell growth (i.e., proliferation and survival) 10, 11, 18, 19, 20, 21, particularly in aggressive breast cancer cells 21. We observed that several genes that encode important regulators of glutamine metabolism were reduced in expression following TAZ/YAP knockdown in MDA‐MB‐231 cells (Fig EV1), which encouraged us to test the importance of glutamine in TAZ/YAP expressing cells. To start, we tested whether TAZ/YAP levels correlate with glutamine dependence in a panel of human mammary cells, including eight breast cancer cell lines and a non‐malignant human mammary epithelial cell (HMEC) line HMT‐3522 S1 22. Immunoblotting for TAZ and YAP showed variable protein levels among these cells lines, ranging from very high levels in the more aggressive breast cancer cells (such as in MDA‐MB‐231 and HCC38) to very low levels in normal mammary epithelial cells (HMT‐3522 S1; Fig 1B). The removal of glutamine from culture medium revealed that many cells exhibited glutamine‐dependent growth (Fig 1C, red bars), whereas others grew robustly in the absence of exogenous glutamine (Fig 1C, blue bars). A strong positive correlation was observed between TAZ/YAP levels and glutamine dependence across these cells, with the growth of cells with pronounced levels of TAZ/YAP showing very strong glutamine dependence (Fig 1D). By examining gene expression data available from the Cancer Cell Line Encyclopedia (CCLE) project 23, we also observed a strong positive correlation between glutamine dependence and the expression of the YAP/TAZ target genes CTGF, CYR61, ANKRD1, and EDN1 (Fig 1E). Taken together, these observations suggested that TAZ/YAP activity may alter metabolic processes that drive exogenous glutamine reliance in breast cancer cells.

Figure EV1. The expression of several genes encoding regulators of glutamine metabolism is reduced following TAZ/YAP knockdown.

The relative change in the expression of genes encoding glutamine regulators was examined in microarray data available from Enzo et al 16 in MDA‐MB‐231 cells following TAZ/YAP knockdown.

Glutamine depletion has been shown to trigger cancer cell death 18, 19. To further examine the dependence on glutamine in cells with different TAZ/YAP levels, we monitored the growth of two “high” TAZ/YAP cell lines (MDA‐MB‐231 and HCC38) and two “low” TAZ/YAP cell lines (BT474 and HMT‐3522 S1) in glutamine replete and depleted media. Cells were seeded at comparable numbers (Fig 2A, Day 0), and after a 4‐day incubation in complete medium, cell confluency increased in all cells examined [Fig 2A, Day 0 vs. Day 4 + Glutamine (Q)]. Removal of glutamine slightly reduced the confluence of low TAZ/YAP cells compared to that in complete medium (Fig 2A, left panel, Day 4 ± Q). By contrast, only few cells with high levels of TAZ/YAP survived glutamine‐free conditions as evidenced by a significant drop of cell numbers (Fig 2A, right panel, Day 0 vs. Day 4‐Q), suggesting that glutamine deprivation induces the death of cells with high levels of TAZ/YAP. We then performed cell counting and trypan blue exclusion assays to directly monitor cell death in MDA‐MB‐231 and HCC38 cells transfected with control siRNA or siRNA targeting TAZ, YAP, or both TAZ and YAP together. The efficacies of the siRNAs were validated by immunoblotting for TAZ and YAP levels (Fig EV2). In glutamine‐free medium, the numbers of both MDA‐MB‐231 and HCC38 cells that received control siRNA profoundly declined after being switched to glutamine‐free medium (Fig 2B, black lines), which was accompanied by an increase in cell death (Fig 2C, black bars). Interestingly, TAZ/YAP deficiency prevented the decline of the cell population (Fig 2B, red lines) and protected both cell lines against cell death (Fig 2C, gray bars) caused by glutamine deprivation. TAZ depletion alone similarly protected the decline of the cell population (Fig 2B, blue lines) following glutamine deprivation, albeit at a lower level than depletion of both TAZ and YAP, whereas YAP depletion alone had minimal effects. This predominant reliance on TAZ for mediating glutamine dependence is consistent with a more dominant role for TAZ in mediating tumorigenic phenotypes and clinical outcomes in breast cancers 13.

Figure 2. Downregulation of TAZ/YAP alleviates glutamine dependence of breast cancer cells.

• A
  HMT‐3522 S1 and BT474 (low TAZ/YAP) as well as MDA‐MB‐231 and HCC38 (high TAZ/YAP) cells seeded at similar density were cultured in complete or Q‐deprived conditions. Representative photographs were taken before (Day 0) and 4 days after media change (Day 4). Scale bars = 100 μm.
• B, C
  Cells transfected with control siRNA (black line) or siRNA targeting either TAZ (blue line), YAP (green line), or both TAZ and YAP (red line) were cultured in glutamine‐free medium. (B) Cell numbers and (C) cell death were monitored at the indicated time points after medium switch. The average from three independent experiments +SEM is shown (*P < 0.05; **P < 0.01; ***P < 0.001; unpaired two‐tailed t‐test).
• D, E
  MDA‐MB‐231 cells were engineered to constitutively express mouse TAZ (mTAZ), which is insensitive to siRNA targeting human TAZ. Cells expressing mTAZ along with control cells were transfected with control siRNA or siRNA targeting TAZ/YAP, and then: (D) the levels of endogenous human TAZ, ectopically expressed mTAZ, and GAPDH (as a loading control) were examined by immunoblotting lysates from respective cell lines; or (E) were cultured in Q‐deprived conditions, and the relative growth of the cells was determined as a percentage of growth relative to complete media. The average from three independent experiments +SEM is shown (**P < 0.01; unpaired two‐tailed t‐test).

Figure EV2. Efficacy of TAZ and YAP siRNA .

MDA‐MB‐231 or HCC38 cells were transfected with indicated siRNA. Lysates were analyzed by immunoblotting with a TAZ/YAP antibody (CST, #8418) to examine antibody specificity and TAZ and YAP knockdown efficiency.

The more pronounced effects of TAZ depletion on cell growth following glutamine deprivation prompted us to next test the consequences of TAZ expression. For this, we generated MDA‐MB‐231 cells stably expressing Mus musculus TAZ (mTAZ) that is not targeted by the human siRNA we used for endogenous TAZ depletion (Fig 2D). We found that expression of mTAZ was sufficient to reverse the growth protective effects of TAZ/YAP depletion following glutamine withdrawal, leading to a marked decline in cell growth after being switched to glutamine‐free medium, similar to what was observed in control cells normally expressing high levels of TAZ/YAP (Fig 2E). Together, these data implicate TAZ as an essential mediator of glutamine addiction of breast cancer cells, with YAP playing a redundant role.

TAZ/YAP promote anaplerotic entry of glutamine through transamination

Glutamine serves as a precursor to provide carbon and nitrogen for the biosynthesis of metabolites that are involved in cancer survival and proliferation 10 (Fig 3A). To evaluate how different glutamine‐metabolizing pathways mediate the growth of cancer cells with high TAZ/YAP levels, we tested the ability of several glutamine‐derived metabolites to rescue the growth of MDA‐MB‐231 and HCC38 cells under glutamine‐deprived conditions. Supplement with dimethyl glutamic acid, a cell‐permeable analog of glutamate, rescued the growth of both breast cancer cell lines in a dosage‐dependent manner (Fig 3B, red bars), consistent with the role of glutamate being the predominant metabolic fate of glutamine in proliferating cells 10, 24. Glutamate has been shown to promote cancer cell growth by maintaining the TCA cycle as an anaplerotic substrate, balancing cell redox status though glutathione synthesis and/or providing non‐essential amino acids for protein synthesis 10 (Fig 3A). Notably, the addition of dimethyl 2‐oxoglutarate, a membrane permeable analog that elevates mitochondrial α‐ketoglutarate (AKG) levels to promote the TCA cycle, greatly restored culture size in both cell lines following glutamine withdrawal (Fig 3B, blue bars). The addition of the glutathione precursor N‐acetylcysteine (NAC) (Fig 3B, green bars) or non‐essential amino acids (Fig 3C, purple bars) had limited effects. These results suggest that the TCA cycle is a main metabolic destiny of glutamine in supporting the growth of cancer cells expressing high levels of TAZ/YAP.

Figure 3. TAZ/YAP are required for glutamine‐utilizing transaminase expression in breast cancer cells.

1. A depiction of the metabolic fates of glutamine.
2. MDA‐MB‐231 and HCC38 cells were cultured in glutamine starvation medium supplemented with indicated metabolites for 48 h. The addition of dimethyl glutamate and dimethyl 2‐oxoglutarate (AKG) resumed cell growth in glutamine‐free medium in a dosage‐dependent manner, while the supplement of other glutamine‐derived downstream metabolites showed no, or marginal effects. The average from three independent experiments +SEM is shown (*P < 0.05; **P < 0.01; ***P < 0.001; unpaired two‐tailed t‐test).
3. The relative expression levels of 588 genes altered in expression by knockdown of TAZ/YAP (i.e., TAZ/YAP gene expression signature) were examined in gene expression data from 1,088 breast cancer biopsy samples curated in TCGA BRCA dataset and 112 paired normal samples. The analysis of this data is shown as a heat map with each row representing a TAZ/YAP‐regulated gene and each column representing a sample that was ranked by TAZ/YAP activity score (purple bar on the top) calculated by the ASSIGN analysis 26. Breast cancer samples (red) and normal samples (black) separated into two clusters, with the cancer samples aligning with higher TAZ/YAP activity.
4. TAZ/YAP activity positively correlated with the expression of several transaminases and anti‐correlated with the expression of both isoforms of glutamate dehydrogenase in 1,200 mammary biopsy samples analyzed. The Spearman's correlation coefficient [R] between TAZ/YAP activity and indicated metabolic enzymes is shown by the length and color code of each bar. The corresponding P‐values are shown next to each bar.
5. MDA‐MB‐231 cells were transfected with control siRNA (siCTL) or siRNAs targeting TAZ and YAP (siTAZ + siYAP), or a separate siRNA targeting both TAZ and YAP (siTAZ/YAP), and the protein levels of the transaminases GOT1 and PSAT1, as well as GAPDH (as a loading control) were assessed by immunoblotting. The changes in GOT1 and PSAT1 protein levels from three separate experiments were quantified relative to GAPDH, and the average +SEM is shown (***P < 0.001; unpaired two‐tailed t‐test).
6. RT–qPCR was performed to determine the relative expression of GOT1 and PSAT1 mRNA in MDA‐MB‐231 cells treated with control siRNA, siRNAs targeting TAZ and YAP, or a separate siRNA targeting both TAZ and YAP. The average expression levels from three independent experiments +SEM are shown (*P < 0.05; ***P < 0.001; unpaired two‐tailed t‐test).
7. MDA‐MB‐231 cells were subjected to ChIP analysis using control rabbit IgG, TAZ, or TEAD4 antibodies. Samples were analyzed by RT–qPCR using primers recognizing an enhancer for the GOT1 gene, which is illustrated at the top of the panel. Normalized % input values are shown as the average of three independent experiments +SEM (*P < 0.05; **P < 0.01; unpaired two‐tailed t‐test).
8. Control cells and cells ectopically expressing mTAZ or MYC were transfected with control siRNA or siRNA targeting TAZ/YAP, and then, the relative expression levels of GOT1, CTGF, MYC, YAP, and human TAZ mRNA were assessed by RT–qPCR. Note the significant rescue of GOT1 and CTGF reduction resulting from TAZ/YAP depletion following the expression of mTAZ. The average from three independent experiments +SEM is shown (**P < 0.01; unpaired two‐tailed t‐test).

Glutamate can be converted to AKG through transamination or oxidative deamination. The former is mediated by transaminases, including the glutamic–oxaloacetic transaminases GOT1 and GOT2 (GOT1/2), the glutamate pyruvate transaminases GPT1 and GPT2 (GPT1/2), and the phosphoserine aminotransferase PSAT1, while the latter is controlled by the glutamate dehydrogenases GLUD1 and GLUD2 (GLUD1/2). To gain more insights on how TAZ/YAP control this metabolic step in a pathophysiological setting, we analyzed gene expression profiles from 1,088 breast cancer biopsies available from The Cancer Genome Atlas (TCGA) BRCA dataset 25, and compared them to 112 paired normal samples for statistical associations between TAZ/YAP activity and the expression levels of aforementioned enzymes. To this end, we examined the expression pattern of a group of TAZ/YAP target genes as a proxy for TAZ/YAP activity and performed Adaptive Signature Selection and InteGratioN (ASSIGN) analysis to calculate the TAZ/YAP activity score for each biopsy sample (See Shen et al 26 and Materials and Methods). When ranked by their TAZ/YAP activity score (Fig 3C, purple bars), most tumor samples (Fig 3C, red bars) were clustered together and showed a higher score than their surrounding benign tissues (Fig 3C, black bars), consistent with reported elevation of TAZ/YAP activity in breast cancer 14, 17. Similar to what previous studies have suggested 14, 27, 28, basal and triple‐negative breast cancer samples exhibited higher TAZ/YAP activity than other subtypes (Fig EV3). Interestingly, by examining the correlation of TAZ/YAP activity with genes encoding glutamine metabolism regulators, we found a strong positive correlation with GOT1 and PSAT1 expression and a negative correlation with glutamate dehydrogenase (GLUD1/2; Fig 3D), suggesting that TAZ/YAP promote transamination over oxidative deamination.

Figure EV3. TAZ/YAP activity in breast cancer subtypes.

Patient samples curated in TCGA were classified into different subtypes by PAM50. ASSIGN analysis was performed (see Materials and Methods) to calculate TAZ/YAP activity score and is shown as a box plot in each subtype. ANOVA showed a significant difference in TAZ/YAP activity scores between tumor subtypes (P < 2.2e‐16 for both classification methods). The Mann–Whitney U‐test was performed to evaluate whether triple‐negative breast cancer samples have a higher TAZ/YAP ASSIGN score compared with ER/Her2 samples, and whether basal breast cancer samples have a higher TAZ/YAP ASSIGN score compared with other subtypes. In both cases, triple‐negative and basal breast cancer samples show significantly higher TAZ/YAP activity (P < 2.2e‐16).

Given the established oncogenic roles of GOT1 and PSAT1 in breast cancer and their strong associations with TAZ/YAP activity in our ASSIGN analysis, we investigated whether TAZ/YAP control the levels of these transaminases. For this, we examined GOT1 and PSAT1 mRNA levels by quantitative real‐time PCR (qPCR) or protein levels by immunoblotting in cell lysates isolated from MDA‐MB‐231 cells transfected with control siRNA or two different sets of siRNA targeting TAZ and YAP. We found that depletion of TAZ/YAP reduced both the protein (Fig 3E) and mRNA (Fig 3F) levels of GOT1 and PSAT1. To investigate whether TAZ/YAP directly mediate the control of GOT1 and PSAT1 gene expression, we analyzed recent chromatin conformation capture (3C) and chromatin immunoprecipitation (ChIP)‐sequencing data for YAP/TAZ from MDA‐MB‐231 cells 29 searching for binding to promoter or enhancer regions associated with these genes. We identified a potential binding site for YAP/TAZ and the TEAD transcription factors, which are binding partners of YAP/TAZ that are known to direct the pro‐tumorigenic functions of YAP/TAZ 30, in a GOT1 enhancer (Fig 3G). Binding sites associated with the PSAT1 gene were less clear. The potential association with a GOT1 enhancer prompted us to perform our own ChIP experiment using TAZ and TEAD antibodies, which revealed that indeed TAZ and the TEAD transcription factors are bound to the GOT1 enhancer (Fig 3G).

Given the direct association of TAZ with the GOT1 gene, we next investigated whether TAZ expression was sufficient to induce GOT1 expression. For this, we used the MDA‐MB‐231 cells we generated that stably express mTAZ (Fig EV4), which is not targeted by the siRNA that recognizes endogenous human TAZ. We found that expression of mTAZ was sufficient to increase the expression of GOT1 following the knockdown of human TAZ/YAP, which was similar to what we observed for the TAZ/YAP target CTGF (Fig 3H). GOT1 expression is known to be affected by the levels of the transcription factor MYC 31, an important mediator of glutamine metabolism in cancer cells 32. Given the association between MYC and TAZ/YAP activity in directing cell growth 33, we considered the possibility that changes in MYC levels contribute to the reduced levels of GOT1 following TAZ/YAP knockdown in MDA‐MB‐231 cells. To investigate the contribution of MYC to GOT1 expression, we generated MDA‐MB‐231 cells stably expressing MYC (Fig EV4). MYC expression was insufficient to rescue the reduced expression of GOT1 following TAZ/YAP knockdown (Fig 3I), suggesting that TAZ/YAP directly induce the expression of GOT1. We did, however, observe changes in MYC mRNA and protein levels following TAZ/YAP knockdown, even for MYC expressed from a CMV promoter (Fig 3H), indicating that TAZ/YAP control MYC levels. Given the importance of MYC in controlling glutamine metabolism and dependence 32, it is thus likely that TAZ/YAP‐mediated control of MYC levels contributes to the metabolic phenotypes observed in breast cancer cells with high TAZ/YAP levels.

Figure EV4. MDA‐MB‐231 cells expressing MYC or mTAZ .

MDA‐MB‐231 cells were engineered to constitutively express mouse TAZ (mTAZ), which is insensitive to siRNA targeting human TAZ. Control cells and cells expressing MYC or mTAZ were transfected with control siRNA or siRNA targeting TAZ/YAP, and the levels of endogenous human TAZ and ectopically expressed mTAZ, MYC, and GAPDH (as a loading control) were examined by immunoblotting lysates from respective cell lines.

Inhibition of transamination preferentially suppresses breast cancer cells expressing high levels of TAZ/YAP

Given that glutamine‐utilizing transaminases have been shown to support cell proliferation and stem cell renewal 20, 34, we hypothesized that elevated levels of these enzymes might play key roles in cancer cells with high TAZ/YAP activity. To test this idea, we examined how the glutamate‐dependent transaminase inhibitor AOA modulates the growth of cells expressing high (MDA‐MB‐231 and HCC38) or low levels of TAZ/YAP (BT474 and HMT‐3522 S1). The growth of cells expressing “low” TAZ/YAP levels was refractory to the inhibition of transamination, indicating their growth was mostly independent of this metabolic reaction (Fig 4A). By contrast, the growth of two “high” TAZ/YAP cells was profoundly suppressed by AOA, resulting in more than 60% reduction in relative growth at highest dose tested (Fig 4A). Notably, the growth deficiency in the “high” TAZ/YAP MDA‐MB‐231 cells following AOA treatment was partially rescued by increased levels of exogenous AKG or aspartate (Fig EV5), consistent with a central role of the GOT transaminases in these cells, as suggested by prior observations 31.

Figure 4. Blockade of transamination suppresses the growth of breast cancer cells in a TAZ/YAP‐dependent manner.

1. The indicated “TAZ/YAP‐low” and “TAZ/YAP‐high” cells were treated with AOA (0, 0.25, 0.5, 1.0 mM) in complete medium. The culture size of AOA‐ and mock‐treated cells was examined 48 h post‐treatment, and the relative growth of AOA/mock was determined as a percentage. Notably, the breast cancer cells expressing elevated levels of TAZ/YAP were more sensitive to AOA. The average from three independent experiments +SEM is shown (in comparison with HMT‐3522 S1, ^### P < 0.001; or BT474, *P < 0.05; ***P < 0.001; unpaired two‐tailed t‐test).
2. MDA‐MB‐231 (left) and HCC38 (right) cells transfected with control siRNA or siRNA targeting TAZ/YAP showed that growth suppression induced by AOA was significantly alleviated in cells depleted of TAZ/YAP. The average from three independent experiments +SEM is shown (*P < 0.05; **P < 0.01; unpaired two‐tailed t‐test).
3. AOA‐induced MDA‐MB‐231 cell death was significantly alleviated following siRNA‐mediated TAZ/YAP depletion compared to cells treated with control siRNA. The average from three independent experiments +SEM is shown (**P < 0.01; unpaired two‐tailed t‐test).
4. Control cells and cells ectopically expressing mTAZ were transfected with control siRNA or siRNA targeting TAZ/YAP, and the growth of cells was assessed in the presence or absence of 1 mM AOA. The average relative growth +SEM of AOA/mock conditions from three independent experiments is shown (**P < 0.01; unpaired two‐tailed t‐test).
5. MDA‐MB cells transfected with control siRNA or siRNA targeting TAZ/YAP were examined for the growth suppression effects induced by the glutaminase inhibitor BPTES. Notably, TAZ/YAP depletion significantly alleviated the relative growth of cells following BPTES treatment. The average from three independent experiments +SEM is shown (*P < 0.05; **P < 0.01; unpaired two‐tailed t‐test).

Figure EV5. AOA‐mediated repression of MDA‐MB‐231 cell growth was partially rescued by the addition of exogenous aspartate or dimethyl 2‐oxoglutarate (AKG).

MDA‐MB‐231 cell growth was assessed in the presence or absence of 1 mM AOA in media supplemented with or without Asp or dimethyl 2‐oxoglutarate (AKG). The average relative growth +SEM of supplemented/mock conditions from three independent experiments is shown (**P < 0.01, ***P < 0.001; unpaired two‐tailed t‐test).

To elucidate whether TAZ/YAP control the sensitivity of cells to the inhibition of transamination, AOA sensitivity was examined in MDA‐MB‐231 and HCC38 cells transfected with control siRNA or siRNA targeting TAZ and YAP. As shown in Fig 4B and C, cells transfected with control siRNA remained sensitive to AOA, whereas depletion of TAZ/YAP significantly alleviated the growth suppression induced by AOA. We observed that AOA treatment induced cell death in MDA‐MB‐231 cells, contributing to the growth suppression phenotypes, and that the AOA‐induced cell death phenotype was partially rescued following TAZ/YAP knockdown (Fig 4C). We further observed that mTAZ expression, which is insensitive to knockdown by the human siRNA we used, was sufficient to overcome the protective effects of TAZ/YAP knockdown following AOA treatment in MDA‐MB‐231 cells (Fig 4D). Therefore, our data indicate a role for TAZ/YAP in mediating transaminase inhibitor sensitivity in breast cancer cells.

Given the effects of transaminase inhibition on cells expressing high levels of TAZ/YAP, we decided to investigate whether inhibition of other enzymes important for glutamine metabolism might offer similar targeting potential. Inhibition of glutaminase activity has been shown to be effective in reducing tumorigenic phenotypes in various tumor models 35, 36, 37, and given the importance of glutamate in cells expressing high levels of TAZ/YAP, we speculated that inhibition of glutaminase activity might offer an additional vulnerability point in these cancer cells. Indeed, we observed that MDA‐MB‐231 cells exhibited growth sensitivity to the glutaminase inhibitor Bis‐2‐[5‐phenylacetamido‐1,3,4‐thiadiazol‐2‐yl]ethyl sulfide (BPTES) and that this growth suppression was reversed by depletion of TAZ/YAP (Fig 4E). Thus, these observations indicate that TAZ/YAP promote the sensitivity of cells to both transaminase and glutaminase inhibitions in breast cancer cells, indicating a central role for TAZ/YAP in altering the glutamine metabolism network.

Glutamine‐utilizing enzymes as potential therapeutic targets for TAZ/YAP‐driven breast cancer

Although the activation of the Hippo pathway effectors TAZ/YAP has been shown to accelerate cell proliferation in a variety of cancers, including breast cancers, little is known about how TAZ/YAP mobilize metabolism to support cell growth. Our analysis of TAZ/YAP‐driven gene expression profiles revealed that amino acid metabolism signatures are repressed upon TAZ/YAP knockdown, suggesting that amino acid metabolism in breast cancer cells is stimulated by TAZ/YAP activity. This directed us to find a strong positive correlation between TAZ/YAP levels and the glutamine dependency of breast cancer cells. Knockdown of TAZ/YAP mitigated the decrease in cell numbers and the corresponding cell death in breast cancer cell lines with elevated levels of TAZ/YAP, demonstrating an essential role for TAZ/YAP in stimulating glutamine dependency.

The addition of glutamate or AKG rescued the MDA‐MB‐231 and HCC38 growth under glutamine‐deprived conditions, arguing an anaplerotic role of glutamine in maintaining the growth of breast cancer cells with elevated levels of TAZ/YAP. Among enzymes that mediate the conversion of glutamate into AKG, we found that the expression of genes encoding the transaminases GOT1 and PSAT1 is positively associated with TAZ/YAP activity in breast cancer patient biopsies curated by TCGA. Moreover, our examination of protein and RNA levels following TAZ/YAP knockdown indicated that TAZ/YAP mediate GOT1 and PSAT1 expression. We were able to confirm that GOT1 is directly regulated by TAZ, which is likely recruited to GOT1 gene regulatory regions by the TEAD transcription factors. Interestingly, the converse relationship was observed with glutamate dehydrogenase levels, with an anti‐correlative relationship with TAZ/YAP activity being observed in TCGA gene expression data, suggesting that TAZ/YAP direct a switch in the metabolic circuit that produces AKG. Our observations, as well recent observations by others 6, 7, 38, 39, suggest an elaborate association between TAZ/YAP activity and enzymes directing glutamine metabolism. Of interest is the potential for TAZ/YAP‐mediated metabolic changes to impact stem cell fate. Glutamine metabolism has been shown to play a key role in maintaining stem cell properties, such as the pluripotency of embryonic stem cells 34, 40. In particular, PSAT1 has been shown to mediate conversion of glutamine into AKG playing a key role in maintaining the mouse embryonic stem cell state 34. Given the importance of TAZ/YAP in a variety of stem cell populations 41, including breast cancer stem cells 14, the link with glutamine metabolism in this context will be of interest to study. The connection between cell confluence, cell growth, and glutamine metabolism is also relevant. Aspartate transaminase activity has been shown to be reduced in confluent mammary epithelial cell culture 42, contributing to cell quiescence under cell contact‐mediated growth inhibition. Cell compaction under confluent conditions is well described to restrict nuclear TAZ/YAP localization and activity 43, 44, and signals that overcome compaction‐mediated inhibition of TAZ/YAP drive increased cell proliferation and survival 44. Thus, it is likely that compaction‐mediated cytoskeletal changes that impact TAZ/YAP activity are central in coordinating glutamine metabolism.

Notably, our studies with AOA show that breast cancer cells expressing higher levels of TAZ/YAP were sensitive to the blockade of transaminases, whereas cells with lower TAZ/YAP levels, including a non‐malignant HMEC line and a breast cancer line, were fairly refractory. We similarly observed that MDA‐MB‐231 cells, which have high levels of TAZ/YAP, are also sensitive to the glutaminase inhibitor BPTES. Further, the depletion of TAZ/YAP protected sensitive cell lines from AOA‐ or BPTES‐mediated growth suppression. These data therefore suggest that inhibitors of glutamine‐utilizing enzymes can repress the growth of breast cancer cells with elevated TAZ/YAP activity, arguing that the assessment of TAZ/YAP activity in breast cancers may offer a means for predicting response to such therapies.

Materials and Methods

Cell culture

The breast cancer cells used were purchased from ATCC. HMECs HMT‐3522 S1 and tumor‐derived T4‐2 cells were kindly provided by Dr. Mina Bissell (Lawrence Berkeley National Laboratory) and were cultured based on the protocol previously described 22, 45, 46. Cells were grown at 37°C in media specified in Table EV2. MDA‐MB‐231 stable cell lines were generated by integrating the respective construct using lentiviral transduction followed by selection in Puromycin (Invivogen). For stable MYC‐expressing cells, the pCDH‐puro‐cMyc lentiviral vector directing expression of cMyc (#46970 47, Addgene) was used, and for TAZ‐expressing cells, the pLVXpuro system was used as we have previously described 17. To measure glutamine dependence, cells were seeded at 1 × 10⁵ per well in a six‐well plate and switched to replete or glutamine‐free media supplemented with 5% Gibco‐dialyzed FBS (Thermo Fisher, 26400‐044). For metabolite addition experiments, glutamine starvation medium was further supplemented as indicated with l‐glutamic acid dimethyl ester hydrochloride (Sigma, 49560), l‐serine methyl ester hydrochloride (Sigma, 412201), l‐aspartic acid/aspartate (Fisher, BP374), dimethyl 2‐oxoglutarate (Sigma, 349631) or N‐acetyl‐l‐cysteine (Sigma, A7250) at a final concentration of 2, 4, or 8 mM, or non‐essential amino acid mixture (alanine, glycine, aspartate, asparagine, proline, serine) at a final concentration of 0.05, 0.1, or 0.2 mM, and then adjusted pH to 7.4 and filtered. To inhibit transaminases, cells were treated with O‐(carboxymethyl) hydroxylamine hemihydrochloride (AOA) from Sigma (C13408) at indicated concentrations in complete medium for 48 h. To inhibit glutaminase activity, cells were treated with BPTES (Sigma, # SML0601) for 48 h. For RNA interference, cells were transfected as previously described 9 with control siRNA (Qiagen, 1027310) or an equal molar mixture of siRNA targeting human TAZ (Thermo Scientific MQ‐016083‐00‐0002) and/or YAP (Thermo Scientific MQ‐012200‐00‐0002), or a single siRNA targeting both TAZ and YAP (UGUGGAUGAGAUGGAUACA). Knockdown efficiency was verified by immunoblotting.

Cell viability and relative growth

To quantify relative growth, cells were cultured as indicated for 48 h and then fixed with 1% paraformaldehyde in PBS. Fixed cells were stained with 0.1% crystal violet (CV) in 10% EtOH for 10 min, then properly washed under running water, and air‐dried overnight. CV stain was extracted by gradually adding 10% acetic acid, allowing the OD590 reading of each control group close to 1.0 ± 0.1. Absorbance values of the experimental groups were then normalized by the value of the corresponding control group and shown as percentages. Living and dead cell numbers were assessed by direct counting and trypan blue exclusion assay with a hemocytometer.

Immunoblotting

Cell lysates were collected and resolved by a 10% gel for SDS–PAGE as previously described 17. Antibodies used included those recognizing GOT1 from Novus Biologicals (NBP154778), PSAT1 antibody (Abnova, H00029968‐A01), TAZ/YAP antibody (Cell Signaling, 8418), MYC D84C12 antibody (Cell signaling, 5605), and GAPDH antibody (Genscript, A00192‐100).

Bioinformatics and statistical analysis

Gene Set Enrichment Analysis (GSEA) was performed as previously described 48, using published gene expression data from NCBI GEO (GSE59230) 16. The metabolic gene sets utilized for GSEA were compiled from Possemato et al 8 and available in the Dataset EV1. The TAZ/YAP activity in patient biopsy samples was calculated by Adaptive Signature Selection and InteGratioN (ASSIGN) 26, based on the expression pattern of 588 TAZ/YAP targets that were defined as the TAZ/YAP signature. Those TAZ/YAP signature genes were selected from the aforementioned expression data 16 through a differential analysis of the samples transfected with control or two different TAZ/YAP shRNA. Corrected multiple hypothesis testing by the FDR Benjamini–Hochberg (BH) method 49 was performed for identifying differentially expressed genes that change at least twofold, and the FDR cutoff was set at 0.05. A list of 588 genes included in the signature is available in the Dataset EV2.

Data from cell‐based and qPCR experiments were shown as mean ± SEM and analyzed by unpaired two‐tailed Student's t‐test. The differences in the ASSIGN scores for different subtypes of breast cancer were analyzed by ANOVA followed by the Mann–Whitney U‐test as indicated in the figure legends. Spearman correlation analysis was performed to assess the correlation between the TAZ and YAP activity score derived from ASSIGN and the expression levels of indicated metabolic enzymes.

To assess the combined expression of the TAZ/YAP targets CTGF, CYR61, ANKRD1, and EDN1, we performed a z‐score transformation for the four genes (the expression of each gene was normalized by subtracting the mean expression and dividing by the standard deviation), and the summed z‐score across the four targets for each cell line was then used for correlation calculation.

Chromatin immunoprecipitation (ChIP)

MDA‐MB‐231 cells were used for ChIP experiments, using methods we have previously described 17. The antibodies used included a rabbit TAZ antibody (Cell Signaling Technologies, 4883) a rabbit TEAD4 antibody (Aviva, ARP38276‐P050. Note that this antibody also recognizes TEAD1 and TEAD3 50), and a control Rabbit IgG (Cell Signaling Technologies, 2729). Samples were analyzed by RT–qPCR to determine the percent input using the following primers that amplify a 127‐bp region within a region in chromosome 10 (Chr10: 101155844‐101155950) bound by the TAZ/YAP and TEAD transcription factors that functions an enhancer for GOT1 gene regulation 29: GTTTTCCAACTGCTCCCTGC; AAGGAGCTTCCCTACCCCAT.

Quantitative real‐time PCR (qPCR)

Extraction of total RNA and the synthesis of cDNA were performed as previously described 17. Quantitative PCR was performed using Fast SYBR green mix (Applied Biosystems, 4385610) and measured on ViiA 7 real‐time PCR system. Expression levels of each gene were calculated using the ΔΔC _t method and normalized to cyclophilin. Sequences of primers are TAZ (CCATCACTAATAATAGCTCAGATC, GTGATTACAGCCAGGTTAGAAAG); YAP (CTCGAACCCCAGATGACTTC, CCAGGAATGGCTTCAAGGTA); GOT1 (CAACTGGGATTGACCCAACT, GGAACAGAAACCGGTGCTT); PSAT1 (ATTGTCCGTGATGACCTGCT, CGGCACCTCCATTGTTTTTA); MYC (AAGACAGCGGCAGCCCGAAC, TGGGCGAGCTGCTGTCGTTG); CTGF (GCAGAGCCGCCTGTGCATGG, GGTATGTCTTCATGCTGG); and cyclophilin (AGCACTGGGGAGAAAGGATT, CATGCCTTCTTTCACCTTCC).

Author contributions

C‐SY and XV developed the concept and wrote the manuscript. C‐SY, ES, and NMK designed and performed experiments, with help from LZ and SM for bioinformatics analysis. C‐SY, ES, and NMK analyzed and visualized data. XV and SM supervised the project.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Expanded View Figures PDF

Table EV1

Table EV2

Dataset EV1

Dataset EV2

Review Process File

EMBO Reports (2018) 19: e43577