YAP and β-Catenin Cooperate to Drive Oncogenesis in Basal Breast Cancer

Breast cancer is a heterogeneous human disease that has been classified into various subgroups (1–3). The majority of triple-negative (ER⁻, PR⁻, and HER2⁻) breast cancers (∼70%) are basal like (2, 4). Basal-like breast cancers are known to display high frequencies of gene alterations as TP53, RB, and PTEN, overexpression of WNT components, and activating mutations of receptor tyrosine kinases (RTK) such as the EGFR, MET, and others (5).

Previously, we generated a compound mutant mouse model that mimics the key features of human basal-like breast cancers (6). It combines the activation of β-catenin, a principal downstream component of Wnt signaling (7) with the expression of hepatocyte growth factor (HGF), which activates the RTK Met (8) under control of the whey acidic protein (WAP) promoter (Wnt-Met mice). WAP is expressed in late pregnancy, thus pregnancy stimulates the rapid growth of aggressive basal-like mammary gland tumors approximately 2 weeks postpartum (6). Single mutants also develop tumors, but usually over a period of months. Wnt-Met tumors exhibit basal-like characteristics, that is, high levels of K5, K14, and smooth muscle actin, whereas luminal markers K8, K18 are low (6). Gene expression analysis showed that Wnt-Met mutant tumors grouped closely with BRCA1+; p53− basal-like (triple-negative) breast cancers but not with luminal breast cancers (6). Moreover, the expression of Wnt target genes Lrp6, Lrp5, and Axin2 were increased and several metastasis-associated genes such as Twist1, Cxcr4, and Postn (6) were upregulated. Gene and protein expression studies have shown that Met is an essential protein in basal-like breast cancer progression and metastasis (9). In addition, Met controls the differentiation state of Wnt-Met tumor cells, while Wnt/β-catenin controls the stem cell property of self-renewal (6).

Basal-like breast cancers exhibit a heterogeneous cellular composition that includes tumor cells with stem cell-like properties (5, 6, 10). Cancer stem cells (CSC) have the unique capacity to initiate, maintain, and replenish tumors, in contrast to other tumor epithelial cells (TEC), which makes them promising targets for cancer therapy (11). A lack of targeted treatments leads to poor 5-year survival for patients presenting to the clinic with basal-like breast cancer (12). Current therapeutic options are primarily limited to chemotherapeutic agents such as doxorubicin, which are nonspecific, toxic and can reduce a patient's overall quality of life. Although these drugs are highly effective against proliferating cells, they have little effect on CSCs due to their intrinsically low proliferation rate (2). Increasing the survival rate for basal-like breast cancer patients will likely require rational therapies that target CSCs. Here we used Wnt-Met mice to dissect the biochemical pathways of these tumors in search of components that could be targeted in therapies directed at CSCs. We elucidated the role of YAP and its regulation in the CSCs of basal-like (triple-negative) breast cancers.

Animal experiments were conducted in accordance with European, National, and MDC regulations. Wnt-Met mutant mice were described previously (6, 13, 14). Yap1tm1.1Dupa/J flox mice were purchased from Jackson Laboratories (stock no.: 027929). Animal experiments were approved by the Ethical Board of the “Landesamt für Gesundheit und Soziales (LaGeSo),” Berlin. Mice were induced via pregnancy between 8 and 12 weeks old. Tumors were harvested between 1 and 2 weeks postpartum and when they reached a maximum size of 1 cm³.

Tumors were minced and digested in DMEM/F12 HAM (Invitrogen) supplemented with 5% FBS (Invitrogen), 5 μg/mL insulin (Sigma-Aldrich), 0.5 μg/mL hydrocortisone (Sigma-Aldrich), 10 ng/mL EGF (Sigma-Aldrich; Digestion medium) containing 300 U/mL collagenase type III (Worthington), 100 U/mL Hyaluronidase (Worthington) and 20 μg/mL Liberase (Sigma) at 37°C for 1.5 hours shaking. Resulting organoids were resuspended in 0.25% trypsin-EDTA (Invitrogen) at 37°C for 1 minute and dissociated in digestion medium containing 2 mg/mL Dispase (Invitrogen) and 0.1 mg/mL DNase I (Worthington) at 37°C for 45 minutes while shaking. Samples were filtered with 40 μm cell strainers (BD Biosciences) and incubated with 0.8% NH4Cl solution on ice for 3 minutes (RBC lysis). Lysis was stopped by washing in 30 mL Dulbecco's Phosphate-Buffered Saline (DPBS) w/o Ca²⁺, Mg²⁺ (Gibco, catalog no. 14190169). Resulting pellets were used for downstream applications outlined below.

Single-cell suspensions from dissociated tumors were resuspended at 10,000 cells/μL and incubated with conjugated primary antibodies at 4°C for 15 minutes. Cells were then washed three times in DPBS and incubated with 7AAD (5 μL/10⁶ cells) at room temperature for 5 minutes to stain dead/dying cells. Cells were sorted using the FACSAria II or III (BD Biosciences) or analyzed using LSRFortessa (BD Biosciences). Compensation and unstained controls were carried out for every FACS experiment as required. Data were analyzed using FLowJo Analysis Software.

PanCancer Pathways panel was purchased from Nanostring including additional custom gene set for selected YAP and β-catenin target genes. A total of 70 ng of RNA was hybridized for 16 hours at 65°C. Hybridized RNA was loaded onto and analyzed using nCounter SPRINT Profiler. Data were analyzed using nSolver software. Heatmaps were generated using nSolver software from normalized data using Pearson correlation. Genes included were manually filtered using available gene lists (15, 16).

For Western blotting, protein was extracted using RIPA buffer with added cOMPLETE Mini Protease Inhibitor Cocktail (#11836153001) and phosphatase inhibitor cocktail 2 and 3 (Sigma). For two-dimensional cell culture, cells were scraped into RIPA buffer [150 mmol/L NaCl, 1% Triton-X-100, 0.5% sodium deoxycholate, 0.1% SDS, 50 mmol/L Tris pH 8.0 and cOMPLETE Mini Protease Inhibitor Cocktail (#11836153001) and phosphatase inhibitor cocktail 2 and 3 (Sigma)] for lysis. Lysis was carried out on ice for 20 minutes, samples were then centrifuged at full speed and supernatant was transferred to fresh tubes, and protein concentration was measured using Bio-Rad Bradford reagent.

For coimmunoprecipitation, cells were lysed as above in coimmunoprecipitation buffer [150 mmol/L NaCl, 50 mmol/L Tris pH 7.5, 1% IGPAL-CA-630 (Sigma #I8896), 5% glycerol, 0.5% deoxycholate, 0.1% SDS with cOMPLETE Mini Protease Inhibitor Cocktail (#11836153001), and phosphatase inhibitor cocktail 2 and 3 (Sigma)]. Equal amounts of protein were incubated with primary antibodies overnight at 4°C rotating. Immunocomplexes were then captured on Dynabeads Protein G Invitrogen catalog no. 10003D for 1 hour at 4°C rotating. Samples were washed and boiled in sample buffer and resolved on 10% SDS-PAGE or gradient gels (4%–20%, Bio-Rad) and electrotransferred to polyvinylidene difluoride membrane. Membranes were blocked with 5% BSA for 1 hour and incubated with primary antibodies (see antibody list).

Breast cancer cell lines were purchased from ATCC (MCF7, BT474, T47D, MDA-MB-231, and BT-549) and Asterand Bioscience (SUM1315). Experiments were conducted between passage 3 and 10 after thawing. Cells were not routinely tested for Mycoplasma. ATCC cells were authenticated by Multiplexion by SNP profiling. SUM1315 were purchased directly from Asterand.

Patient-derived xenograft (PDX) models were established from patients with triple-negative breast cancer with their informed consent as described previously (17). The experimental protocol and animal housing were in accordance with institutional guidelines as proposed by the French Ethics Committee (agreement no. B75-05-18). PDXs were obtained from patients with their informed written consent.

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD021113. Nanostring log₂ data can be found in Supplementary Table S1.

Mice with single Wnt or Met mutations developed mammary gland tumors in 30–40 weeks postpartum, while double mutants exhibited tumors as early as 2 weeks postpartum (Wnt-Met tumors; ref. 6). Proteomic analysis of Wnt-Met mammary glands at early stages of tumorigenesis revealed strong upregulation of the YAP signature (15), indicating that YAP is activated in double but not in single mutants (Fig. 1A). Wnt-Met double mutant mammary glands contained 6-fold more YAP-positive nuclei than mice with only Wnt mutations (Fig. 1B and C). This was confirmed by Western blotting for active YAP (Fig. 1D; Supplementary Fig. S1A and S1B). Analysis of the phosphoproteome revealed higher levels of YAP phosphorylation at S46 and T48, suggesting that these sites play a role and nuclear activation of YAP (Supplementary Fig. S1C).

The expression of YAP and Wnt target genes Birc5, Ctgf, Cd44, and Lgr5 was higher in the mammary glands of Wnt-Met mice (Fig. 1E; refs. 16, 18). This was validated by Western blotting for CTGF (Supplementary Fig. S1D). We examined whether Met signaling is responsible for nuclear translocation of β-catenin. Double mutants exhibited 10-fold increase of nuclear β-catenin, compared with mammary glands with Wnt mutation (Supplementary Fig. S1E and S1F). We also examined the expression of YAP and Wnt target genes in Wnt-Met mammospheres (19) treated with the Met inhibitor PHA665752 (6, 15, 16). Remarkably, this revealed significantly lower levels in the expression of Ctgf, Ccnd1 (CyclinD1) and other genes, confirmed by qPCR (Fig. 1F; Supplementary Fig. S1G; refs. 15, 16). This shows that the RTK Met is required for YAP and β-catenin nuclear translocation in the mammary glands of Wnt-Met mice.

We examined the functional role YAP through genetic and pharmacologic interference. We crossed floxed YAP alleles (13) into Wnt-Met mice. Upon stimulation via pregnancy, this led to homozygous YAP ablation in Wnt-Met mice (denoted Wnt-Met-YAP^KO) and produced an increase in the number of tumor-free mice from 11 days in controls (Wnt-Met-YAP^Ctrl) to 17days (Fig. 2A). The average weight of Wnt-Met-YAP^KO mammary gland tumors was also decreased (Fig. 2B). We confirmed YAP ablation by qPCR (Fig. 2C). Remarkably, hematoxylin and eosin (H&E) staining and IHC of Wnt-Met-YAP^KO mammary glands revealed large alveoli, which were mostly YAP-free and exhibited healthy one-layered acini. Their proliferation was lower, as determined by Ki-67 staining (Fig. 2D, bottom, quantification in Fig. 2E). In contrast, Wnt-Met-YAP^Control mammary glands showed large tumorous areas, which were strongly YAP stained, without large empty alveoli (Fig. 2D, top). Of note, tumors that developed in Wnt-Met-YAP^KO mammary glands showed expression of YAP; however, YAP-free areas remained as single layered, healthy acini (Supplementary Fig. S2A). Fluorescent images of YFP confirmed the successful Cre-recombination of the mammary epithelial cells; YFP-positive (YFP⁺)cells were found in single-layered acini in Wnt-Met-YAP^KO glands, in contrast to filled tumors in the controls (Fig. 2F). YAP has been reported to be important in the mammary gland during pregnancy (20); qPCR analysis confirmed that our tumor-free phenotype was not a result of reduced Wap expression in Wnt-Met-YAP^KO mammary glands (Supplementary Fig. S2B). Furthermore, YAP ablation under the control of the WAP promoter did not alter the normal mammary glands during pregnancy (Supplementary Fig. S2C).

We treated Wnt-Met mice systemically with two inhibitors of YAP activity: simvastatin (SIM) inhibits YAP via the mevalonate-Rho kinase pathway (21), and verteporfin (VP) inhibits the interaction of YAP with its transcriptional activator TEAD (22). Both inhibitors reduced tumor volumes and weight, although animals eventually developed small tumors (Supplementary Fig. S2D–S2G). H&E and IHC of SIM- and VP-treated tumors also revealed small, empty alveoli and a strong reduction in Ki-67–positive cells and Ctgf expression compared with controls (Supplementary Fig. S2H–S2O). TAZ has been reported to be important for regulating mammary CSCs (23). We show that although TAZ is highly expressed in Wnt-Met-YAP^Control and Wnt-Met-YAP^KO tumors, Wnt-Met-YAP^KO glands remained as healthy, single-layered acini, suggesting that TAZ is unable to compensate for YAP ablation (Supplementary Fig. S2P and S2Q). These data demonstrate that YAP is crucial for tumor proliferation and growth in Wnt-Met tumors.

Met signaling controls the differentiation of Wnt-Met tumor cells (6). To identify whether Wnt-Met tumor cells arise from luminal or basal cells, we investigated WAP-cre activity in Wap-cre;YFP^rosa26 mice at various stages of induction. WAP-cre activity marked by YFP expression was found almost exclusively in CK8-positive and CD24⁺CD49f⁺ luminal cells throughout WAP-cre induction, as seen by immunofluorescence and FACS (Supplementary Fig. S3A and S3B). However, upon activation of Wnt and Met signaling, YFP+ luminal cells gradually transitioned from CK8-positive, CD24^hiCD49f⁺ to P63-positive, CD24^hi, CD49f^hi basal cells between 0 and 2 weeks postpartum (Fig. 3A and B). This was also accompanied by a gradual increase in the expression of the EMT genes Slug and Twist, which have been shown to be important for luminal and basal cell differentiation (Fig. 3C; ref. 24). To examine whether luminal-to-basal transition depends on YAP, we examined Wnt-Met-YAP^Control and Wnt-Met-YAP^KO mammary glands for the expression of the luminal cell maker CK8 and the basal-like cell marker P63. We observed strong expression of P63 in control glands, which exhibited a minimal expression of CK8 (Fig. 3D and E, left). Strikingly, Wnt-Met-YAP^KO mammary glands showed negligible P63-positive cells but a high number of CK8-positive cells (Fig. 3D and E, right). As confirmation, in Wnt-Met tumors we found strong expression of cytokeratin 14 (CK14) and low expression of CK8, a pattern that was reversed in SIM-treated tumors (Supplementary Fig. S3C). We also observed that YAP deletion resulted in a significant decrease in the expression of the EMT genes Slug and Twist, suggesting that YAP regulates luminal-to-basal cell transdifferentiation through regulation of EMT (Fig. 3F). Overall, these data provide genetic and pharmacologic evidence that YAP is required for the acquisition of basal-like characteristics in Wnt-Met tumors.

We have previously shown that Wnt-Met tumors induced by WAP-Cre harbor CD24^hi and CD49f^hi/CD29^hi CSCs that produce aggressive basal-like cancers when transplanted into immune-deficient mice (6). We examined the contribution of YAP to Wnt-Met CSCs. Using an antibody specific for the active state of YAP, we found high levels of nuclear YAP on the outer rim of Wnt-Met tumor acini, which correlated to the increase of expression of specific stem cell genes such as CD44 and Sox10 (Fig. 4A; Supplementary Fig. S4A and S4B; refs. 10, 14, 25). We confirmed high levels of YAP by Western blotting and mRNA expression (Fig. 4B and C). Moreover, the expression of target genes of Wnt/β-catenin and YAP including Axin2, Ccnd1, Ctgf, and Igfbp3 increased (Fig. 4D and E; refs. 15, 16).

High levels of YAP activity have been reported in the stem cell compartments of many tissues (26). YFP⁺CD24^hiCD49f^hi CSCs were found in the basal cell population of tumors but not the luminal population, which contained YFP⁻CD24⁺CD49f⁺ cells as shown by FACS (Fig. 4F; Supplementary Fig. S4C). CSCs have been reported among the basal-like cells marked by CK14 (6, 27). In agreement with this, we found that YAP was expressed in the nuclei of basal-like, CK14-positive cells but not in those that were luminal and CK8 positive (Fig. 4G). Moreover, YAP target genes were more highly expressed in CSCs than in other TECs (Fig. 4H and I; Supplementary Fig. 4D and E). These data show that Wnt-Met signaling generates basal-like tumors with high YAP activity in CSCs of the mammary gland.

To study the function of YAP in CSCs, we first established stem cell–enriched control and Wnt-Met spheres in mammary gland stem cell–promoting medium (Supplementary Fig. S5A; ref. 28). Control spheres (WAP-Cre; ROSA26^EYFP) formed single-layered epithelial structures with empty lumens, in contrast to the large, filled structures produced by Wnt-Met spheres (Fig 5A; Supplementary Fig. S5B). FACS for YFP showed that Wnt-Met spheres contained 1.4-fold (57%–76%) more YFP⁺ cells than controls (Supplementary Fig. S5C). They also exhibited a higher expression of transgenes, as confirmed by qPCR for Wap, Hgf, and Lgr5 (Supplementary Fig. S5D). Immunofluorescence demonstrated a decrease in the luminal marker CK8 and an increase in the basal-like marker CK14 (Supplementary Fig. S5E). Moreover, a strong nuclear β-catenin signal was observed in basal cells located at the outer rim of spheres (Fig. 5B). We isolated cells by FACS and seeded equal cell numbers to generate stem cell–enriched spheres. The CSC population underwent an 8-fold higher outgrowth than TECs (Supplementary Fig. S5F). This shows that spheres enrich and expand CSCs in Wnt-Met mammary gland tumors. We identified nuclear YAP in tumor-derived spheres through immunofluorescence (Supplementary Fig. S5G) and quantified the expression of YAP target genes using qPCR. We found a 2.5-fold increase in Cyr61 (15) and a 3.7-fold increase in Igfbp3 (Supplementary Fig. S5H; ref. 15).

To understand the functional role of YAP, we inhibited its activity using SIM and VP. Because VP may have off-target affects (29), we treated Wap-cre;YFP cells with increasing concentrations of VP and observed no change the cell number, suggesting that the reduction in cell proliferation in Wnt-Met tumor cells was not due to toxicity (Supplementary Fig. S5I). These treatments reduced proliferation of Wnt-Met spheres, as determined by bromodeoxyuridine (BrdU) and Ki-67 immunofluorescence (Fig. 5C, quantification on the right; Supplementary Fig. S5J). Expression analysis for the stem cell–associated genes Itga6, Kit (30), and Prom1 (31) suggested the inhibition of CSCs (Fig. 5D). Moreover, sphere initiation was repressed 2.8-fold by VP (Fig. 5E, quantification on the right). We also treated cells with YAP-directed short hairpin RNA and siRNA, which resulted in a 4-fold decrease of the size of spheres generated by Wnt-Met cells and a 2.5-fold decrease in the number of spheres by MDA-MB-231 cells, which exhibit high Wnt and YAP activity (Supplementary Fig. S5K–S5N). This shows that YAP is essential for β-catenin in both gain of function (GOF) and wild-type cells. FACS revealed a 3-fold decrease in the content of CD49f^hiCD24^hi CSCs (6) in Wnt-Met-YAP^KO mammary glands compared with controls (Fig. 5F and G). qRT-PCR revealed equal expression of Yfp in Wnt-Met-YAP^KO mammary glands compared with controls (Fig. 5H). Furthermore, spheres generated from Wnt-Met-YAP^KO mammary glands generated 6-fold fewer spheres than controls (Fig. 5I; Supplementary Fig. S5O). This demonstrates that YAP activity is essential for growth, initiation, and self-renewal of Wnt-Met CSCs.

We carried out Nanostring gene expression analysis (32) of control and Wnt-Met-YAP^KO mammary glands. Wnt-Met-YAP^KO mammary glands exhibited a strong decrease in YAP and β-catenin activity (Fig. 6A). This tissue exhibited downregulation of target genes such as Ctgf and Igfpb3 (YAP targets) and Axin2 and BMP4 (β-catenin targets); the same was found for VP-treated spheres (Fig. 6B and C; Supplementary Fig. S6A). Quantification showed that β-catenin–positive nuclei were decreased 11-fold by the ablation of YAP (Fig. 6D, quantification on the right) and a strong decrease by pharmacologic inhibition of YAP (Supplementary Fig. S6B). Similar results were observed upon siRNA knockdown of YAP in MDA-MB-231 cells (Supplementary Fig. S6C–S6G), suggesting that in the absence of YAP, β-catenin can no longer translocate to the nucleus and transcribe target genes.

In the absence of Wnt signaling, interaction of YAP and β-catenin has been implicated in the β-catenin destruction complex (33). Immunofluorescence revealed that YAP and β-catenin are colocalized in the nuclei of Wnt-Met spheres and tumors (Fig. 6E). Active YAP coimmunoprecipitated with β-catenin^GOF and TEAD4 in Wnt-Met spheres, but less with wild-type β-catenin (Fig. 6F). In addition, p-YAP (which is cytoplasmic) coimmunoprecipitated with β-catenin and LATS1 suggesting YAP and β-catenin interact in both the cytoplasm and nucleus (Supplementary Fig. S6H).

YAP has been reported to be required for the development of β-catenin–driven tumors (34). Because YAP inhibition resulted in a decrease in the expression of a number of Wnt target genes, we hypothesized that YAP/TEAD might bind with β-catenin at promoter and/or enhancer regions of common target genes to regulate their expression in CSCs. We obtained YAP1, TEAD4, and CTNNB1 (β-catenin) chromatin immunoprecipitation sequencing (ChIP-seq) data from ChIP-Atlas (35). Because YAP preferentially binds enhancer regions rather than promoters (36), we investigated whether YAP/TEAD4/CTNNB1 occupies common genomic regions. We found that YAP/TEAD4/CTNNB1 had overlapping peaks at 601 regions in the genome (Supplementary Fig. S6I). In depth analysis of Wnt and YAP target genes (13, 14) identified enhancers of AXIN2, CD44, and BCL2L2, in addition to enhancer and/or promoter regions of NUAK2 (a newly identified YAP target gene; ref. 37), AMOTL2 and CCN2, containing overlapping CTNNB1, TEAD4, and YAP1 ChIP-seq peaks (Fig. 6G). These data show that TEAD4, CTNNB1, and YAP1 interact and can colocate at gene promoters and enhancers of both Wnt and YAP target genes.

Analysis of human primary patient proteomic data (38) revealed high expression of the YAP signature (15) in basal-like breast cancer, in contrast to other subgroups of breast cancers (Fig. 7A). qPCR of YAP target genes CTGF, IGFBP3, and ANKRD1 (13) in spheres of human basal-like breast cancer cell lines BT549, MDA-MB-231, and SUM1315 confirmed elevated YAP activity, compared with the luminal cell lines MCF7, BT474, and T47D (Fig. 7B; Supplementary Fig. S7A). Online datasets available from Gene expression–based Outcome for Breast cancer Online (GOBO; ref. 39) confirmed higher expression of YAP in tumors of human patients with basal-like breast cancer than in tumors of patients with luminal breast cancer (Supplementary Fig. S7B). Immunofluorescence analysis of YAP in 10 basal-like breast cancer PDX models (17) exhibited cells with strong nuclear YAP and high β-catenin expression (Fig. 7C; Supplementary Fig. S7C).

We also found an opposing correlation between high YAP expression and patient survival in luminal (ER⁺, PR⁺) breast cancer and basal-like (triple-negative; HER2⁻, ER⁻, PR⁻) breast cancer, as shown by Kaplan–Meier analyses (Fig. 7D and E; ref. 40). The high expression of YAP in luminal breast cancer correlated with an increase in patient survival, while it correlated with a decrease in survival in patients with triple-negative breast cancer. The high expression of YAP target genes such as CTGF, CYR61, and IGFBP3 correlated with reduced survival of patients with triple-negative breast cancer (Fig. 7F). This shows that YAP is highly expressed in basal-like (triple-negative) breast cancers, which can be used to predict patient survival in a subtype-dependent manner.

We identified key mechanisms demonstrating the oncogenic cooperation of Met, YAP, and β-catenin in basal-like breast cancer. Met signaling controls the nuclear translocation and activation of YAP and β-catenin. YAP and β-catenin interact and bind to enhancer regions of Wnt target genes, leading to gene transcription. In the absence of YAP, Met signaling cannot induce nuclear translocation of β-catenin, sequestering it in the cytoplasm and preventing the transcription of Wnt target genes. Thus, YAP is a key bottleneck required for the oncogenic activity of Met and β-catenin in basal-like breast cancer.

YAP activity has been shown to be regulated by several upstream mechanisms (26). Utilizing our genetic mouse model as well as proteomic and phosphor-proteomic approaches, we found that Met signaling promotes YAP activity. We identified two phosphorylation sites that may aid YAP activation. Moreover, our analysis revealed that the regulation of basal-like characteristics by Met depends on YAP. This suggests that YAP might also play a major role in the dedifferentiation process of other models of basal-like breast cancers (41). Our findings demonstrate that Met signaling confers its oncogenic and metastatic abilities through activating YAP. It would be interesting to investigate if bypassing the Met receptor by overexpression of only YAP and β-catenin lead to tumors that are identical to those of Wnt-Met.

Wnt-Met activation generates cancer-propagating cells (6). Wnt/β-catenin activity is known to be enhanced in CSCs and confers breast cancer reoccurrence and progression (42). Several studies have suggested that Wnt/β-catenin is linked to Hippo/YAP activity (33, 34, 43). Sulaiman and colleagues reported that dual inhibition of Wnt and YAP activity is required to inhibit CSCs in basal breast cancer cell lines (44). In this study, we found both YAP and β-catenin target genes were increased in Wnt-Met tumors. Moreover, we identified active YAP in the CSCs of these tumors. Functionally, YAP controls CSC properties: through pharmacologic and genetic knockouts of YAP, we demonstrated that the absence of YAP strongly delayed Wnt-Met–driven tumorigenesis and decreased the expansion of CD24^hiCD49f^hi CSCs. Furthermore, we established that Wnt/β-catenin target gene expression was decreased upon YAP inhibition, demonstrating that YAP is required for Wnt/β-catenin–dependent CSC expansion in Wnt-Met tumors.

Wnt and Hippo signaling have been described as tightly intertwined. For example, in the mouse liver, the coactivation of β-catenin and YAP have been shown to lead to the rapid generation of hepatocellular carcinoma (45). It has been shown that the absence of Wnt signals results in YAP/TAZ sequestration in the cytoplasm by the β-catenin destruction complex, comprised of APC, Axin, and GSK3 (33). In this context, the knockout of YAP/TAZ in embryonic stem cells leads to the activation of β-catenin and compensates for loss of Wnt signaling. We found that genetic ablation of YAP controls β-catenin activity at two levels: (i) in the nucleus through cobinding the regulatory regions of β-catenin target genes such as Axin2, and (ii) by regulating β-catenin nuclear translocation, showing that basal-like breast cancers require YAP for nuclear β-catenin activity. It is likely that the master regulatory control of Wnt target gene expression by YAP is both context and tissue dependent. However, the mechanism of how the YAP-β-catenin physical interaction regulates β-catenin activity requires further experiments by means of mapping the interactions.

YAP was recently shown to be inactive and even downregulated in ductal carcinoma in situ (46). In addition, patients with breast cancer displaying high YAP activity appeared to have a higher overall positive prognosis, leading to the belief that YAP might act as a tumor suppressor in breast cancer (47, 48). In this scenario, keeping YAP activity under tight control is believed to help tumor cells escape immunosurveillance (49). However, other oncogenic transcriptional systems are likely to stimulate a strong immune response in the host (50). Activation of a strong host immune response may also account for contradictory observations by Britschgi and colleagues, who showed that upregulation of the ER and YAP activity in response to LATS1/2 deletion reduces anti-ER therapy efficacy (51). These studies did not investigate the functional role of YAP in a context-dependent manner such as breast cancer or tumor cell subtypes.

Our study addresses these issues by using genetic, siRNA and pharmacologic interference with YAP in basal-like breast cancer to demonstrate its function specifically in CSCs. In basal-like breast cancer, YAP's genetic ablation strongly delayed Wnt-Met–driven tumorigenesis. Segregation of luminal (ER⁺, PR⁺) breast cancer patients and basal-like, triple-negative patients revealed that YAP expression correlated with patient survival in opposing ways. High YAP activity was associated with prolonged survival in patients with luminal breast, but decreased survival in patients with basal-like (triple-negative) breast cancer. Thus, YAP appears to function as an oncogene in certain cancer subtypes and a tumor suppressor gene in others. Further studies will be required to understand the mechanisms responsible for the context-dependent function of YAP and whether its inhibition in breast cancer has converse therapeutic benefits in basal-like and luminal breast cancers.

Y. Fuchs reports grants from Israel Science Research Foundation and Israel Cancer Research Fund during the conduct of the study; grants from Rainin Foundation, Cronos Group, and EMBO outside the submitted work; in addition, Y. Fuchs has a patent for use of caspase-3 inhibitors and caspase-3 activators in the manufacture of medicament for treating cancer and wound healing pending. No disclosures were reported by the other authors.

H.M. Quinn: Conceptualization, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. R. Vogel: Data curation, technical assistance. O. Popp: Data curation, formal analysis. P. Mertins: Resources, supervision. L. Lan: Data curation, formal analysis, investigation. C. Messerschmidt: Formal analysis. A. Landshammer: Resources, formal analysis. K. Lisek: Initiated collaboration and obtained samples from E. Marangoni. S. Chateau-Joubert: Resources. E. Marangoni: Resources, provided PDX models. E. Koren: Writing–review and editing. Y. Fuchs: Supervision, writing–review and editing. W. Birchmeier: Supervision, writing–review and editing.

We thank Russ Hodge of the MDC for improving our article style, Hans-Peter Rahn and the FACS Core Facility of the MDC for their expertise, Anca Margineanu and the Confocal Microscopy Core Facility of the MDC for the support with microscopy, and Diana Behrens at EPO-Buch for conducting the in vivo drug treatments. H.M. Quinn was a recipient of an Intl. PhD fellowship of the MDC and the Humboldt University of Berlin. The project was supported by the German-Israeli Helmholtz Research School SignGene, funded by the German Federal Ministry of Education and Research and the Initiative and the Networking Fund of the Helmholtz Association grant no. VH-KO-612. This work was supported by MDC central funding. Y. Fuchs received ICRF (2028184) and ISF (2027619).

H.M. Quinn was a recipient of an Intl. PhD fellowship of the MDC and the Humboldt University of Berlin.

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