Oncogene (2012) 31, 238–250 & 2012 Macmillan Publishers Limited All rights reserved 0950-9232/12 www.nature.com/onc ORIGINAL ARTICLE Willin/FRMD6 expression activates the Hippo signaling pathway kinases in mammals and antagonizes oncogenic YAP

L Angus1, S Moleirinho1,2, L Herron1, A Sinha3,4, X Zhang3,4, M Niestrata2, K Dholakia5, MB Prystowsky6, KF Harvey3,4, PA Reynolds2,7 and FJ Gunn-Moore1,7

1School of Biology, University of St Andrews, St Andrews, UK; 2School of Medicine, University of St Andrews, St Andrews, UK; 3Cell Growth and Proliferation Laboratory, Peter MacCallum Cancer Centre, East Melbourne, Victoria, Australia; 4Department of Pathology, University of Melbourne, Parkville, Victoria, Australia; 5School of Physics and Astronomy, University of St Andrews, St Andrews, UK and 6Department of Pathology, Albert Einstein College of Medicine, Bronx, NY, USA

The Salvador/Warts/Hippo (Hippo) signaling pathway 2006; Harvey and Tapon, 2007). This pathway limits defines a novel signaling cascade regulating cell contact organ size by inhibiting cell proliferation and promoting inhibition, organ size control, cell growth, proliferation, apoptosis, and is therefore of particular importance apoptosis and cancer development in mammals. The during development and tissue size control (Tapon et al., Drosophila melanogaster protein Expanded acts in 2002; Harvey et al., 2003; Huang et al., 2005; Edgar, the Hippo signaling pathway to control organ size. 2006; Harvey and Tapon, 2007; Pan, 2007). In addition, Previously, willin/FRMD6 has been proposed as the several pieces of evidence suggest that the activities of human orthologue of Expanded. Willin lacks C-terminal multiple Hippo pathway components are deregulated in sequences that are present in Expanded and, to date, little human cancer (Tapon et al., 2002; Harvey and Tapon, is known about the functional role of willin in mammalian 2007; Pan, 2007; Zeng and Hong, 2008). cells. When willin is expressed in D. melanogaster The Hippo pathway consists of a series of kinases and epithelial tissues, it has the same subcellular localization adaptor proteins in which the Hippo kinase, in as Expanded, but cannot rescue growth defects associated association with an adaptor protein Salvador, phos- with expanded deficiency. However, we show that ectopic phorylates and activates Warts kinase, which is asso- willin expression causes an increase in phosphorylation of ciated with an activating subunit, Mats. This core kinase the core Hippo signaling pathway components MST1/2, unit phosphorylates and inactivates the transcriptional LATS1 and YAP, an effect that can be antagonized co-activator Yorkie, thereby suppressing expression of by ezrin. In MCF10A cells, loss of willin expression genes that promote cell survival, growth and prolifera- displays epithelial-to-mesenchymal transition features and tion (Huang et al., 2005). Inactivation of Yorkie in willin overexpression antagonizes YAP activity via the D. melanogaster imaginal disc tissues results in cell cycle N-terminal FERM domain of willin. Therefore, in arrest and apoptosis (Harvey and Tapon, 2007; Pan, mammalian cells willin influences Hippo signaling activity 2007). Ex, Mer and Kibra function upstream of the core by activating the core Hippo pathway kinase cassette. kinase cassette, activating the pathway by an unknown Oncogene (2012) 31, 238–250; doi:10.1038/onc.2011.224; mechanism via Hpo and Wts phosphorylation (Hamar- published online 13 June 2011 atoglu et al., 2006; Genevet et al., 2010; Yu et al., 2010). More recently, Ex has been shown to form a complex Keywords: Willin/FRMD6; Expanded; Hippo pathway; with Yorkie and is proposed to directly regulate its ezrin; merlin activity by the WW domains of YAP and the PPXY motifs of Ex (Badouel et al., 2009). Ex is also thought to function downstream of the Hippo pathway receptor protein, Fat, an atypical cadherin that represses growth Introduction of imaginal discs (Bennett and Harvey, 2006; Silva et al., 2006; Willecke et al., 2006; Tyler and Baker, 2007). Loss Recent studies utilizing Drosophila melanogaster genet- of ex results in the development of hyperplastic imaginal ics have indicated that merlin (Mer) and a second discs and overgrown adult structures such as the wing FERM protein called Expanded (Ex) has an important (Boedigheimer and Laughon, 1993). In addition, over- role in controlling the Salvador/Warts/Hippo (Hippo) expression of ex in the D. melanogaster wing and eye signaling cascade (Edgar 2006; Hamaratoglu et al., leads to a decrease in the number of cells in these tissues (Boedigheimer et al., 1997). Correspondence: Dr FJ Gunn-Moore, Medical and Biological Sciences Less is known about the functioning of the Hippo Building, School of Biology, University of St Andrews, North Haugh, signaling pathway in mammals, particularly the proteins St Andrews, Fife KY16 9TF, UK. that regulate the mammalian Hippo pathway upstream E-mail: [email protected] or [email protected] 7These authors contributed equally to this work. of the core kinase cassette. Components of the Hippo Received 8 March 2011; revised and accepted 6 May 2011; published pathway are conserved in mammals and consist of online 13 June 2011 MST1/2 (Hpo orthologues), WW45/Sav (Sav orthologue), Salvador/Warts/Hippo signaling pathway in mammals L Angus et al 239 LATS1/2 (Wts orthologues), MOB1 (Mats orthologue) (Figure 1b). We noted that YAP phosphorylation could and YAP (Yki orthologue) (Harvey and Tapon, 2007). only be studied over a period of 2 days post-induction, as At least four upstream regulatory branches of the Hippo high cell density caused YAP phosphorylation in unin- pathway exist in Drosophila: Fat/Dachsous, Kibra/Ex/ duced cells as was observed previously (Zhao et al., 2007), Mer, Lethal Giant Larvae/atypical protein kinase C and all experiments were therefore performed at a low cell and Crumbs (Grusche et al., 2010). The mammalian density. No MST1/2, LATS1 and YAP phosphorylation orthologues of Mer (also called NF2, Mer) and Kibra was observed when an empty GFP plasmid was transiently have been shown to function as upstream members of transfected into the stable inducible cell line or when 1 mg/ the Hippo pathway, but the identity of other upstream ml tetracycline was added to cells that did not contain the members of the mammalian Hippo pathway has pcDNA4/TO/myc-his willin-GFP plasmid, but only a remained enigmatic. For example, recent studies suggest plasmid expressing the pcDNA6-TR tetracycline repressor that the Fat/Dachsous pathway does not control organ (data not shown). size and activity of the Yki orthologues, YAP and TAZ, in mice (Saburi et al., 2008; Mao et al., 2011). Ezrin modulates the ability of willin to phosphorylate Interestingly, repression of YAP by AMOT via direct MST1/2 interaction of the WW domains of YAP with the PPXY To investigate the effect of the expression of other motifs of AMOT has been reported (Zhao et al., 2011), a FERM domain-containing proteins on MST1/2 phos- role fulfilled by the C terminus of Ex in D. melanogaster. phorylation, we expressed combinations of Mer and We sought to determine whether the mammalian ezrin in TRex-willin-GFP cells and induced willin-GFP homologue of D. melanogaster Ex, willin/Ex1/FRMD6, expression with 1 mg/ml tetracycline. A hierarchy in the functioned as an upstream member of the mammalian ability to activate MST1/2 phosphorylation was obser- Hippo pathway (Gunn-Moore et al., 2005; Hamaratoglu ved, in which the expression of either Mer or willin et al., 2006). Previously, willin overexpression was shown was sufficient to result in MST1/2 phosphorylation to influence YAP in a luciferase activity, but a mechanistic (Figure 1c). When Mer and willin were co-expressed, a link to the Hippo pathway was not explored (Zhao et al., synergistic effect was observed on MST1/2 phosphor- 2007). We show that increased willin expression causes ylation, in which willin expression enhances Mer’s phosphorylation of MST1/2, LATS1 and YAP. Loss of ability to phosphorylate MST1/2 (Po0.01). Interest- willin expression results in features of an epithelial-to- ingly, ezrin expression was found to have an inhibitory mesenchymal transition (EMT) in MCF10A cells and effect on MST1/2 phosphorylation when co-expressed willin overexpression antagonizes YAP activity by the with Mer, willin or willin and Mer (Po0.01) (Figure 1c). N-terminal FERM domain of willin. The majority of willin localizes to the apical junction of D. melanogaster Willin expression sensitizes HEK-293 cells to apoptotic imaginal disc cells in a manner similar to Ex, although stimuli willin could not rescue growth defects associated with Previous studies have shown that MST1/2 expression ex deficiency. Therefore, willin/Ex1/FRMD6 has both sensitizes cells to death upon addition of tumor necrosis similarities and differences to Ex, but in mammalian cells factor-a (TNFa) and cycloheximide (Lee et al., 2001). willin influences Hippo signaling activity by activating the Therefore, we explored whether willin-GFP expression core Hippo pathway kinase cassette. in HEK-293 cells could also sensitize these cells to cell death signals. Interestingly, MTT (3-(4,5-dimethyl- thiazol-2-yl)-2,5-diphenyltetrazolium bromide) assays Results showed that cell viability dropped by 20% in cells treated with TNFa that expressed willin-GFP as Expression of willin influences phosphorylation of compared with uninduced cells (Figure 2a). Further- MST1/2, LATS1 and YAP more, immunoblot analysis showed that caspase-3 was Mammalian willin and D. melanogaster Ex show 60% cleaved in these cells only upon willin-GFP expression similarity in their N-terminal FERM domain, but the (Figure 2b). A colorimetric assay supported these C-terminal domains are highly divergent with Ex being observations as active caspase-3 was detected in cells much longer and containing additional functional motifs expressing willin-GFP treated with TNFa, whereas no (Figure 1a). Previously, the expression of both willin and caspase-3 activity was observed in the control (TRex) Mer were reported to inactivate the transcriptional co- cells (Figure 2b). activator YAP by an unknown mechanism (Zhao et al., Other studies have shown that both phosphorylation 2007). Therefore, we created a stable tetracycline-induci- and cleavage of MST1 are required for apoptotic ble HEK-293 cell line to investigate the effect of willin downstream effects (Graves et al., 2001). Therefore, expression upon the kinases upstream of YAP. Cells were MST1 cleavage was investigated in the absence and seeded so that the cell density remained constant, presence of TNFa. MST1 cleavage was never observed tetracycline was added to induce willin-green fluorescent when cells were induced for willin-GFP expression. protein (GFP) expression and cells were harvested at 24, However, addition of TNFa did result in MST1 48 and 72 h. Immunoblot and subsequent image J analysis cleavage, which was enhanced upon willin-GFP expres- showed a significant increase in MST1/2, LATS1 and sion (Figure 2c). YAP phosphorylation in induced cells expressing willin- Previous studies have observed nuclear-to-cyto- GFP when compared with uninduced HEK-293 cells plasmic translocation of YAP upon phosphorylation

Oncogene Salvador/Warts/Hippo signaling pathway in mammals L Angus et al 240

FERM Willin- 614aa PPXY PPXY

FERM Expanded- 1427aa

Hr 02448 72 Tet- ++ + 3.5 0 hr tet 24 hr tet 48 hr tet 72 hr tet pMST1/2 3 2.5 MST1 2 pLATS1 1.5 1 LATS1 0.5 Relative phosphorylation 0 pYAP MST1/2 LATS1 YAP YAP

pMST1/2

MST1

2.5 2 1.5 1

of MST1/2 0.5 0 Relative phosphorylation GFP + −−− −−−− Willin − + − − + + − + Merlin − + − + − − + + Ezrin − −+−−+++ Figure 1 Increased willin expression activates the Hippo signaling pathway. (a) Schematic representation of willin and Ex proteins. (b) Immunoblots showing increased MST1/2, LATS1 and YAP phosphorylation upon willin expression for 24–72 h in TRex-willin-GFP HEK-293 cells. Quantitation using Image J analysis shows a significant difference (n ¼ 3, Po0.01) of MST1/2, LATS1 and YAP phosphorylation. Error bars represent±s.e. (c) Expression of GFP-ezrin and/or merlin-GFP in TRex-willin-GFP HEK-293 cells. Relative pMST1/2 decreased upon ezrin-GFP co-expression (Po0.01), whereas willin expression increases merlin’s ability to phosphorylate MST (Po0.01), whereas merlin’s expression did not reach significant changes on willin’s ability to phosphorylate MST (P40.05). Relative pMST1/2 vs MST1 was analyzed using the Image J software. Error bars represent±s.e. (n ¼ 3).

at Ser127 through binding to 14-3-3 proteins (Zhao homologue (Ex) in D. melanogaster tissues, we assessed et al., 2007). Therefore, we investigated the subcellular willin subcellular localization, the effect of willin localization of YAP following willin-GFP expression. misexpression and whether willin could substitute for A cytoplasmic/nuclear fractionation assay showed that Ex. To perform these experiments, we generated a YAP is mainly located in the cytoplasmic fraction of transgenic D. melanogaster strain that afforded induci- willin-GFP-expressing cells and TNFa treatment does ble expression of the willin protein with GFP fused to its not further influence this localization, whereas YAP carboxyl terminus. Ex is localized predominantly to the remained mostly within the nucleus in uninduced cells apical junction of D. melanogaster epithelial imaginal (Figure 2d). disc cells, along with other Hippo pathway proteins such as Fat (Bennett and Harvey, 2006; Silva et al., 2006; Willecke et al., 2006). Ex also localizes to punctuate Willin shares similar subcellular localization as Ex, cytoplasmic structures thought to be vesicles. Using the but cannot rescue the growth defect of Ex clone wing-specific 32B-Gal4 driver, we expressed willin-GFP in D. melanogaster in developing wing imaginal discs and also stained these To assess whether willin could function in a tissues with Ex- and Fat-specific antibodies. The manner similar to its closest D. melanogaster sequence majority of willin-GFP was localized at the subapical

Oncogene Salvador/Warts/Hippo signaling pathway in mammals L Angus et al 241 TRex-wGFP TRex 0 1 2 3 4 0 1 2 3 4 120 TRex TRex-wGFP caspase-3 100 80 cleaved 60 caspase-3 40 30 Cell Viability 20 25 20 0 050 15 10 TNFα concentration (ng/ml)

Activity (SA) 5

Relative Specific 0 TRex-wGFP TRex

Cytoplasmic Nuclear tet −++ −++ Tet − +−+ TNFα −−+ −−+ TNFα −−+ + YAP YAP MST1 cleaved β-actin β-actin MST1 coilin coilin

Figure 2 Controlled willin-GFP expression sensitizes HEK-293 cells to apoptotic stress. In all, 50 ng/ml TNFa was added to both TRex and TRex-willin-GFP HEK-293 cells incubated with 1 mg/ml tetracycline 2 days before the experiment and whole-cell lysates were collected 0–4 h after TNFa treatment (a) TNFa decreased cell viability in TRex-willin-GFP, but not in TRex cells as measured by an MTT assay. Percentage cell viability was calculated from absorbance treated/untreated cells. Error bars represent±s.d. (n ¼ 24). (b) Cleaved caspase-3 was detected by immunoblot analysis upon willin expression and TNFa treatment. CaspACE colorimetric assay confirmed active caspase-3 in TRex-willin-GFP, but not TRex cells treated with TNFa.(c) Immunoblot showing MST1/2 cleavage occurred upon TNFa treatment and was enhanced upon willin expression. (d) Willin expression results in cytoplasmic YAP translocation. Immunoblot analysis showing YAP protein translocated from nucleus to cytoplasm upon willin-GFP expression. b-Actin and coilin were used as loading controls.

junction, and overlapped with Fat and Ex expression function, using the mosaic analysis with a repressible cell (Figures 3a–f). Some willin was also found in the marker technique (Lee and Luo, 1999). Compared with cytoplasm, as evident in both planar sections and wild-type control clones, exMGH1 clones were larger and cross-sections of wing imaginal discs (Figures 3a–f). rounded in shape, and occupied more tissue relative to When overexpressed in developing imaginal discs ex GFP-negative control tissue (Figures 3l and m). exMGH1 causes a substantial reduction in organ size (Figure 3h). clones displayed high Yki activity as determined by To determine whether willin could limit size of imaginal assessing expression of the Hippo pathway target disc-derived tissues, we misexpressed willin-GFP with proteins, Ex (Figure 3m) and DIAP1 (data not shown). 32B-Gal4. Unlike overexpression of ex, misexpression of When willin-GFP was misexpressed in exMGH1 clones, no willin-GFP did not reduce the total area of adult wings substantial changes in size or shape of the clones were as compared with wings expressing GFP under control observed, and the Hippo pathway target proteins, Ex of the 32B promoter, or flies harboring 32B-Gal4 or and DIAP1, were still expressed at high levels (Figure 3n UAS-willin-GFP alone (Figures 3g–j). Similar results and data not shown). These data show that the closest were observed for other wing Gal4 drivers such as C5, human sequence homolog of ex, willin, cannot function- MS1096 and 71B (data not shown). We did observe a ally substitute for the ability of ex to control Hippo slight change in the shape of wings misexpressing willin- pathway-dependent tissue growth in D. melanogaster. GFP; however, wings were slightly broadened across the anterior/posterior axis relative to the proximal–distal axis, although the biological relevance of this result is Willin/FRMD6 knockdown induces features of EMT in unclear (Figures 3j and k). mammary epithelial cells Several mammalian Hippo pathway proteins can As deregulation of Hippo pathway components such as compensate for loss of their D. melanogaster counter- YAP induces EMT in MCF10A cells (Overholtzer et al., parts with respect to the ability to modulate organ size. 2006; Zhang et al., 2008), we knocked down the To determine whether willin could compensate for expression of willin in these cells using two independent ex loss, we overexpressed willin-GFP in clones of third lentiviral short hairpin RNA (shRNA) constructs. instar larval eye imaginal disc tissue lacking ex gene Although MCF10A cells expressing a non-targeting

Oncogene Salvador/Warts/Hippo signaling pathway in mammals L Angus et al 242 control (shScramble (shScr)) grew in epithelial-type abnormal features resembling YAP-induced EMT islands on monolayer cultures, MCF10A cells expressing (Figure 4a). Importantly, these changes accompanied a shRNA targeting willin (MCF10A-shWillin) displayed decrease in MST1/2, LATS1 and YAP phosphorylation

Oncogene Salvador/Warts/Hippo signaling pathway in mammals L Angus et al 243 in MCF10A-shWillin cells (Figure 4b). Efficient knock- The FERM domain of willin is sufficient to antagonize a down of willin mRNA was assayed by quantitative YAP-induced EMT phenotype in MCF10A cells polymerase chain reaction (PCR) (Figure 4e). MCF10A- To test whether the conserved FERM domain of willin shWillin cells produced threefold more colonies in was sufficient to illicit the observed effects of willin in soft agar than MCF10A-shScr cells, after 21 days in MCF10A cells, we cloned the FERM and C-terminal culture (Figure 4c). MCF10A-shWillin cells also showed domains of willin and expressed these in MCF10A cells a threefold increased migration compared with overexpressing YAP. In anchorage-independent growth MCF10A-shScr cells (Figure 4d). Consistent with the and cell migration assays, expression of the FERM EMT-like phenotype, expression of the mesenchymal domain alone was sufficient to antagonize a YAP- markers N-cadherin and vimentin was upregulated by induced EMT phenotype in MCF10A-YAP cells, YAP overexpression or by willin knockdown, whereas whereas the C-terminal domain of willin had no the epithelial markers, E-cadherin and occludin, antagonizing effect on YAP, as assay outputs remained were downregulated by both of these manipulations unchanged when compared with an empty vector (Figure 4e). Knockdown of Mer (shMerlin) also control (Figures 6a–c). Hao et al. (2008) described displayed a similar phenotype. When small interfering downstream YAP targets that were affected by LATS1 RNAs (siRNAs) targeting YAP were introduced into expression; therefore, we investigated some of these MCF10A-shWillin cells, they abrogated the willin targets to see whether willin and the FERM domain knockdown migration phenotype (Figure 4f). could inhibit YAP downstream targets. Quantitative PCR analyses showed that willin and the FERM domain could indeed antagonize YAP targets by Willin expression antagonizes a YAP-induced EMT upregulating BMP2 and RASSF8 and downregulating phenotype in MCF10A cells PRL and IGFBP3 (Figure 6d). Next, we investigated whether expression of willin could influence the activity of YAP in MCF10A cells. These cells were infected with a retrovirus-expressing willin or empty vector and an increase in MST1/2, LATS1 and Discussion YAP phosphorylation were all observed when MCF10A cells overexpressed willin (Figure 5a). Then, we explored Although the core kinase cassette of the Hippo pathway whether willin expression could antagonize a YAP- has been well characterized in mammalian cells, events induced EMT phenotype in MCF10A cells overexpres- upstream are far less understood. We have shown for sing YAP (MCF10A-YAP) that were infected with a the first time that willin expression can activate retrovirus containing an empty vector or willin cDNA. components of the Hippo signaling pathway in mam- Notably, MCF10A-YAP cells expressing willin pro- malian cells. We find that an increase in willin duced approximately 50% fewer colonies in soft agar expression causes phosphorylation of MST1/2, LATS1 than MCF10A-YAP cells containing empty vector, after and YAP. Ezrin, which is absent in D. melanogaster 21 days in culture (Figure 5b). Moreover, willin was but present in mammals, inhibits the effect of both Mer unable to antagonize a YAP(S127A) mutant under the and willin on MST1/2 phosphorylation in HEK-293 same conditions (Figure 5b). MCF10A-YAP cells cells. Therefore, it is possible that ezrin is an important expressing willin also showed attenuated migration by controlling mechanism for upstream activation 50% compared with MCF10A-YAP cells containing of the Hippo pathway in mammals. Furthermore, empty vector, whereas MCF10A-YAP(S127A) cells nuclear-to-cytoplasmic YAP translocation was observed were refractory to the effects of willin expression in both HEK-293 and MCF10A cells expressing willin, (Figure 5c). Mesenchymal and epithelial markers also which may explain the observation that willin expression changed: with a decrease of the mesenchymal markers results in sensitizing cells to TNFa-induced cell vimentin and N-cadherin in MCF10A-YAP cells death. Although phosphorylation of MST1/2 (and also expressing willin and an increase in the epithelial LATS1 and YAP) occurred upon willin expression, markers occludin and E-cadherin in these cells, but no cleaved MST1 was never observed in this scenario. Only changes were observed in MCF10A-YAP(S127A) cells upon TNFa stimulation was MST1 cleaved and an (Figure 5d). apoptotic downstream effect observed. Interestingly,

Figure 3 Willin displays similar subcellular localization as Expanded, but cannot functionally replace it. Planar (a–c) and optical cross (d–f) sections of third instar larval wing imaginal discs expressing willin-GFP under control of the 32B promoter. Willin-GFP (green in a, d and the merged images, c and f) was detected at the apical plasma membrane, similar to Fat expression (red in b and c) and Expanded (red in e and f). (g–i) Adult female D. melanogaster wings of the indicated genotypes. (j) Quantification of the area of adult wings of the indicated genotypes (n ¼ 32 for each genotype). Error bars represent (±s.e.m.). (k) Quantification of the shape of adult wings of the indicated genotypes. Shape was assessed by determining the ratio of the distance across the anterior/posterior axis divided by the distance across the proximal/distal axis (n ¼ 32 for each genotype). Error bars represent±s.e.m. and *Po0.0001. Wing shape, but not area was altered when willin-GFP was misexpressed under the control of the 32B promoter. (l–n00) Third instar larval eye imaginal discs, anterior is to the right. Clones are positively marked with GFP (green) and are of the following genotypes: wild-type (l), exMGH1 (m) and exMGH1 also misexpressing willin-GFP (n). Expression of Ex (gray in l0, m0 and n0 and blue in the merged images, l00, m00 and n00) and ELAV (red in the merged images, l00, m00 and n00) was determined. exMGH1 clones were larger than control clones and displayed elevated Ex protein, regardless of whether willin-GFP was misexpressed.

Oncogene Salvador/Warts/Hippo signaling pathway in mammals L Angus et al 244 cells undergoing TNFa treatment and overexpressing agree with data from Song et al. (2010), which showed willin had more cleaved MST1 compared with cells that TNFa-induced cell death in hepatocytes requires undergoing TNFa treatment alone. Our findings MST1/2.

MCF10A MCF10A MCF10A MCF10A MCF10A MCF10A shScr shWillin-A shWillin-B shMerlin Vector YAP

MCF10A 1.4 shScr shWillin-A shScr shW-A 1.2 pMST1/2 1 MST1/2 0.8

pLATS1 0.6

LATS1 0.4 0.2 pYAP Relative phosphorylation 0 YAP MST1/2 LATS1 YAP

120 400 100 300 80 60 200 40 100 20 Number of cells Number of colonies 0 0

YAP YAP shScr Vector shScr Vector shMerlin shMerlin shWillin-B(late) shWillin-A MCF10A MCF10A

MCF10A MCF10A shW-B shScr shW-A shMer 1.0 Vector YAP (early) 0.8 E-cad E-cad 0.6 Occludin 0.4

Occludin mRNA 0.2 Vimentin Vimentin Relative Willin 0 N-cad N-cad shScr YAP Merlin shWillin-AshWillin-B(early) shWillin-B(late) β -actin β-actin MCF10A

120

shScr shWillin-A 80 siConsiCon siYAP1 siYAP2

40 YAP Number of cells β-actin 0 siControl siControl siYAP-1 siYAP-2

shScr shWillin-A MCF10A

Oncogene Salvador/Warts/Hippo signaling pathway in mammals L Angus et al 245 Many Hippo pathway proteins can rescue the importantly, other Hippo pathway components have function of their orthologues in D. melanogaster tissues, been shown to induce EMT (Overholtzer et al., 2006; emphasizing the high degree of conservation inherent in Zhang et al., 2008). Moreover, the conserved FERM this pathway. When misexpressed in D. melanogaster domain of willin is sufficient to antagonize a YAP- larval imaginal discs, willin displayed an overlapping induced EMT phenotype in MCF10A cells and is able localization, that is, predominantly localized at the to control the downstream targets of YAP activity. apical junction of epithelial cells. Despite this, willin did We propose a model (Figure 6e) whereby upstream not display growth-suppressive activity in D. melanoga- signals acting on willin and Mer are antagonized by ster tissues. In addition, willin mis-expression did not ezrin. As the FERM domain is a known protein-binding limit the size of D. melanogaster wings and it failed to motif, we speculate that the FERM domain of willin rescue the overgrowth defect and elevated Yki activity acts as a scaffold and binds other proteins that elicit the associated with ex deficiency. The ex gene is significantly observed effects on the core Hippo pathway. Impor- larger than the willin gene, and may have split its tantly, not all FERM domain-containing proteins can functions during vertebrate evolution. The Ex and willin activate the Hippo Pathway, as we have shown that proteins are especially divergent in the carboxyl termini, ezrin has an inhibitory effect. Indeed, the FERM which is the region of Ex that is proposed to bind to and domain of willin may act to block the inhibitory effects inhibit Yki (Badouel et al., 2009). In D. melanogaster, of ezrin. Yki is believed to contact Ex via WW domain–PY motif The reduction of endogenous willin expression was interactions, but none of the three PPXY motifs in the also able to induce an EMT in MCF10A cells with a Ex C terminus are conserved in willin. Thus, it appears similar phenotypic trend to YAP overexpression both in that the direct mode of inhibition of Yki function by Ex terms of morphological changes, but also cell marker is not conserved between willin and YAP in mammals. expression. Specifically, epithelial markers E-cadherin Interestingly, repression of YAP by AMOT via direct and occludin were downregulated and the mesenchymal interaction of the WW domains of YAP with the PPXY markers N-cadherin and vimentin were upregulated. motifs of AMOT has been reported (Zhao et al., 2011), These effects of willin knockdown were YAP dependent. and AMOT may fulfill the role of the C terminus. Ex has The ability of willin knockdown to induce an EMT was also been proposed to function upstream of the core further confirmed through an increase in cell migration components of the Hippo pathway and influence and anchorage-independent growth in soft agar. This all activity of the Warts kinase (Hamaratoglu et al., 2006; suggests that willin is involved, together with other Genevet et al., 2010; Yu et al., 2010). Our data support members of the mammalian Hippo pathway, in the this proposed mode of regulation. negative regulation of YAP. However, although other We found that ectopic willin expression antagonized a members of the pathway (Mer, LATS, MOB1, YAP) YAP-induced EMT phenotype in MCF10A cells, both have been linked to tumorigenesis (Harvey and Tapon, in terms of changes in functional outputs of anchorage- 2007), the question of whether willin is a tumor independent growth and migration, but also changes in suppressor gene in human cancer awaits further marker expression. Furthermore, willin could not analysis, as no studies have addressed if this gene is antagonize a constitutively active YAP mutant (YAP- mutated or silenced in cancer. Interestingly, other S127A), showing that the action of willin on YAP is by FERM domain-containing proteins have been impli- Ser127 phosphorylation. These cells are a non-tumori- cated in cancer, for example, FRMD3 is silenced in genic, human mammary epithelial cell line in which, non-small-cell lung carcinoma (Haase et al., 2007).

Figure 4 Knockdown of Willin results in features of EMT. Drug-selected pools of infected cells were used in these analyses to avoid clonal selection effects. (a) Willin loss induces a morphology change in MCF10A cells that resembles YAP-induced EMT. Representative phase-contrast images of cells growing in monolayer cultures and transfected with either non-targeting shRNA (shScr) or shRNA targeting willin (shWillin), merlin (shMerlin) or MCF10A cells expressing empty vector (MCF10A-vector) or YAP (MCF10A-YAP). (b) Willin knockdown decreases MST1/2, LATS1 and YAP phosphorylation in MCF10A cells. Immunoblot analysis of retroviral-infected MCF10A cells with either shScr or shWillin-A shows that loss of Willin expression resulted in decreased phosphorylation of MST1/2, LATS1 and YAP. Cells were treated with 50 ng/ml TNFa for 4 h before harvest. Relative phosphorylation to total protein levels (MST1/2, LATS1 or YAP) is shown and background phosphorylation (in MCF10A-shScr) is set to 1. Error bars represent±s.d. (n ¼ 3). (c) Willin knockdown in MCF10A cells promotes anchorage-independent growth in soft agar. MCF10A-shScr, MCF10A-shWillin-B (late passage), MCF10A-shMerlin, MCF10A-YAP, MCF10A-vector and cells were cultured in soft agar for 21 days. Error bars represent±s.e. (n ¼ 9). (d) MCF10A-shWillin-A, MCF10A-shScr, MCF10A-YAP or MCF10A-vector cells were cultured in Boyden chambers and migration assessed after 24 h. Error bars represent±s.e. (n ¼ 9). (e) Immunoblot analysis of E-cadherin and occludin (epithelial markers) and N-cadherin and vimentin (mesenchymal markers) shows loss of the epithelial markers and gain of mesenchymal markers in MCF10A-shWillin cells. b-Actin was used as a loading control. Quantitative real-time PCR analysis of the expression of willin mRNA in knockdown cells confirming knockdown of willin mRNA expression by 80% for shWillin-A and by 20% for shWillin-B in early passage cells, which increased to 50% knockdown in late passage cells. b-Actin was used to normalize for variances in input cDNA. Error bars±s.d. (n ¼ 3, except Willin-B late where n ¼ 2). (f) Effect of shWillin is YAP dependent. MCF10A-shScr or MCF10A-shWillin-A cells with listed siRNA addition were cultured in Boyden chambers and migration assessed after 24 h. Error bars represent±s.e. (n ¼ 9). Immunoblot analysis of efficient knockdown of YAP was carried out using two independent siRNA duplexes (siYAP1 and siYAP2). siControl refers to a non-targeting siRNA duplex. b-Actin was used as a loading control.

Oncogene Salvador/Warts/Hippo signaling pathway in mammals L Angus et al 246 MCF10A 4.5 Vector Willin 4 pMST1/2 3.5 3 MST1/2 2.5 pLATS1 2 1.5 LATS1 1 pYAP 0.5 Relative phosphorylation 0 YAP control MST1/2 LATS1 YAP Willin-HA

80 300

60 200 40 100 20 Number of cells

Number of colonies 0 0 Vector Willin Vector Willin Vector Willin Vector Willin

MCF10A-YAP MCF10A-YAP MCF10A-YAP MCF10A-YAP (S127A) (S127A)

MCF10A-YAP MCF10A-YAP (S127A) 60 Vector Willin Vector Willin

E-cad 40

Occludin 20 Vimentin 0 Relative Willin mRNA N-cad Vector Willin Vector Willin MCF10A-YAP MCF10A-YAP Flag-YAP (S127A) β-actin

Figure 5 Willin expression antagonizes YAP-induced EMT in MCF10A cells. Drug-selected pools of infected cells were used in these analyses to avoid clonal selection effects. (a) Immunoblot analysis of retroviral-infected MCF10A cells with either willin or empty vector shows that willin expression resulted in phosphorylation of MST1/2, LATS1 and YAP. Relative phosphorylation to total protein levels (MST1, LATS1 or YAP) and background phosphorylation in MCF10A vector (control) is set to 1. Error bars represent±s.d. (n ¼ 3). (b) Willin expression in MCF10A-YAP cells suppresses anchorage-independent growth in soft agar. MCF10A- YAP-willin, MCF10A-YAP-vector, MCF10A-YAP(S127A)-willin or MCF10A-YAP(S127A)-vector cells were cultured in soft agar for 21 days. Error bars represent±s.e. (n ¼ 9). (c) MCF10A-YAP-willin, MCF10A-YAP-vector, MCF10A-YAP(S127A)-willin or MCF10A-YAP(S127A)-vector cells were cultured in Boyden chambers and migration assessed after 24 h. Error bars represent±s.e. (n ¼ 9). (d) Immunoblots show increased E-cadherin, occludin and decreased vimentin, N-cadherin in MCF10A-YAP cells expressing willin, but no changes in MCF10A-YAP(S127A) cells expressing willin compared with vector control cells. Quantitative real-time PCR analysis of the expression of willin mRNA in MCF10A-YAP and MCF10A-YAP(S127A) expressing willin and vector control cells. b-Actin was used to normalize for variances in input cDNA. Error bars represent±s.d. (n ¼ 9).

To conclude, we have shown that willin expression willin, Mer and other effectors acting upstream of the can activate the Hippo signaling pathway resulting in core components of the Hippo pathway core kinase the sensitization of mammalian cells to apoptotic cassette. stimuli, and willin acts antagonistically to YAP. As with D. melanogaster Ex, the precise mechanism by which willin regulates phosphorylation and activation Materials and methods of the Hippo pathway core kinase cassette is unknown. Our findings place willin in the Hippo signaling Cell culture and transfection pathway in mammals, and raise the possibility that HEK-293 cells were grown in minimum essential medium YAP activity is modulated by the combined effects of supplemented with 10 mM non-essential amino acids, 10%

Oncogene Salvador/Warts/Hippo signaling pathway in mammals L Angus et al 247 Vector C-term FERM willin pMST1/2 Vector C-term FERM willin MST1 YAP-Flag pLATS1 Willin-HA LATS1

pYAP Cterm/ FERM-HA YAP

30 0.6 25 0.5 20 0.4 15 0.3 10 0.2 5 0.1

number of colonies 0 0 Absorbance at 570nm Vector Willin C-term FERM Vector Willin C-term FERM

PRL IGFBP3 1.5 ? Plasma Membrane 1 ? ? Ezrin 0.5 Willin Merlin FERM

Relative RNA level 0 Vector Willin C-term FERM P P WW45 MST1/2 MST

50 BMP2 RASSF8 P MOB1 LATS 40 Cell death P 30 YAP

20 YAP 14-3-3

10 Nucleus Cytoplasm Relative RNA level 0 cell proliferation & survival VectorWillin C-term FERM Figure 6 The FERM domain of willin is sufficient for its activity. Stable MCF10A-YAP cell lines were made overexpressing full- length willin-HA, FERM-HA, Cterm-HA or empty vector. (a) Immunoblot analysis shows that expression of both the FERM domain of willin and full-length willin in MCF10A cells results in phosphorylation of the core Hippo pathway components MST1/2, LATS1 and YAP. (b) Willin and FERM domain expression in MCF10A-YAP cells suppresses anchorage-independent growth in soft agar. Cells were cultured in soft agar for 21 days. Error bars represent±s.d. (n ¼ 6). (c) Willin and FERM domain expression in MCF10- YAP cells inhibits migration through a Boyden chamber. Error bars represent±s.d. (n ¼ 3). (d) Quantitative PCR analysis shows that willin and FERM domain expression increases PRL and IGFBP3 mRNA levels and decreases BMP2 and RASSF8 mRNA levels. Error bars represent±s.d. (n ¼ 5). (e) Schematic representation of a model for the action of willin, merlin and ezrin on Hippo signaling.

fotal bovine serum, 2 mM glutamine, 100 mg/ml streptomycin, addition with 5 mg/ml blasticidin and 250 mg/ml zeocin. Willin- 100 U of penicillin per ml at 37 1C and atmosphere of 5% CO2. GFP expression was induced with 1 mg/ml tetracycline MCF10A cells were obtained from American Type Culture (Invitrogen). Cells and were grown according to Debnath et al. (2003) in Expression vectors were obtained as follows: pGFP-ezrin standard cell culture flasks (Nunc Nunclon) in a humidified was a generous gift from Dr R Lamb, Cancer Research UK incubator at 37 1C with 5% CO2. Centre of Cell and Molecular Biology, Institute of Cancer A stable tetracycline-inducible system was created using a Research, London, UK; pmerlin-GFP and pcDNA3-merlin- TRex-inducible plasmid pcDNA4/TO/myc-his (Invitrogen, FLAG from Dr W IP, University of Cincinnati, Cincinnati, Paisley, UK), cloned to contain willin-GFP, which was then OH, USA; willin-HA (pBabe-puro; Addgene, Cambridge, transfected into stable HEK cells containing a plasmid MA, USA); pBabe-Flag-YAP (Addgene plasmid 15682) has expressing tetracycline repressors, pcDNA6/TR, a kind been previously described (Overholtzer et al., 2006; Zhang gift from Dr C Tate (MRC, Laboratory of Molecular et al., 2008); pBabe-Flag-YAP(S127A) was a kind gift from Dr Biology, Cambridge, UK). Stable cells were selected by the D Haber (Harvard University, Harvard, MA, USA). Willin-

Oncogene Salvador/Warts/Hippo signaling pathway in mammals L Angus et al 248 GFP, TRex-willin-GFP (pcDNA4/TO/myc-his; Invitrogen), Colorimetric caspase-3 assay Flag-YAP, willin-FERM-HA and willin-Cterm-HA plasmids All materials for the caspase-3 assay came from the caspACE were cloned in our lab. colorimetric assay kit (Promega, Southampton, UK) and For microscopy studies, HEK-293 cells were plated onto protocol used was according to the manufacturer’s handbook. glass coverslips and transfected with 1 mg of DNA per A total of 25 ng of protein extract was used for the caspase-3 coverslip using GeneJammer (Agilent, Stockport, UK). Cells assay and absorbance was measured at 405 nm using a MRX were left for 24–48 h, after which they were washed twice with Microplate Reader (DYNEX Technologies). phosphate-buffered saline, fixed in neutral-buffered formalin (10% formalin, 0.4% NaH2PO4 Á H2O, 0.65% Na2HPO4), Nuclear/cytoplasmic fractionation washed again with phosphate-buffered saline and then Cells were grown in two 90 mm dishes, and when desired mounted with mowiol (Merck Chemicals, Nottingham, UK) density was reached, cells were pelleted at 300 g for 5 min at 0 and 4 ,6-diamidino-2-phenylindole (Sigma-Aldrich, Giling- 4 1C. Pellet was resuspended in 0.5 ml ice-cold cytoplasmic ham, UK). For biochemical studies, HEK-293 cells were extraction buffer (10 mM HEPES, pH 7.9, 1.5 mM MgCl2, plated onto 90 or 150 mm dishes and transfected with 6 or 10 mM KCl, 0.5 mM dithiothreitol) and incubated on ice for 20 mg of DNA per dish, respectively, using GeneJammer 5 min. Cells were lysed with 20 stokes on Dounce homogenizer (Agilent). and lysate was centrifuged at 228 g for 5 min at 4 1C to pellet pBabe constructs were packaged in Phoenix A cells for viral nuclei and other fragments. The supernatant, containing the propagation. Retroviral infection of MCF10A cells was cytoplasmic fraction, was stored at À20 1C. Nuclear pellet was carried out using standard methods. Puromycin selection resuspended in 0.5 ml ice-cold nuclear extraction buffer (2 mg/ml) or hygromycin selection (300 mg/ml) was applied 24 h (0.25 mM sucrose, 10 mM MgCl ) and layered over a sucrose after retroviral infection and continued for 5 days (puromycin) 2 bed (0.88 mM sucrose, 0.5 mM MgCl2). Nuclear fraction was or 8 days (hygromycin), after which cells were harvested or pelleted by centrifugation at 2800 g for 10 min at 4 1C. Protein used for further experiments. The shRNA duplex hairpins sample buffer was added to both nuclear and cytoplasmic against human willin (19mer target sequences: Willin-A, fractions and run on a sodium dodecyl sulfate gel. 50-ACAGAGCAGCAAGATACTA-30, Willin-B, 50-TTCATT CTCAGTGTGTGCT-30) and Mer (Okada et al., 2005) were synthesized and cloned into pLKO.1puro (Addgene; Soft agar assay 4 plasmid 8453). Control shScr was obtained from Addgene Cells were seeded at 2 Â 10 cells per well in 0.35% agarose/ (plasmid 1864). Constructs were packaged in HEK-293T for Dulbecco’s modified Eagle’s medium:F12 complete media viral propagation. Lentiviral infection of MCF10A cells was onto a lower layer of 0.5% agarose/Dulbecco’s modified carried out using standard methods and drug selection was Eagle’s medium:F12 complete media. Once the agarose was performed following standard protocols. siRNAs targeting set, a layer of Dulbecco’s modified Eagle’s medium:F12 YAP (siYAP1 and siYAP2) along with their siRNA control complete media were added above the agarose layers. This were purchased from Ambion, Warrington, UK. media were replaced every other day and the number of colonies (X5 cells represented a colony) in each well were counted after 21 days. Immunoblot analysis Cells were lysed in 10 mM Tris (pH 8.0), 150 mM NaCl, 1% Na deoxycholate, 1% Nonidet P-40, 1% sodium dodecyl sulfate, Migration assay 1mM ethylenediaminetetracetic acid and protease inhibitor. A measure of 0.5 ml aliquots of serum-free cell suspension 5 Proteins were separated on a sodium dodecyl sulfate gel and (5 Â 10 cells) were added to the top chamber of 24-well proteins were equally loaded. After transfer onto either chambers with 8.0 mm pores (BD Biosciences) and media polyvinylidene fluoride or nitrocellulose membrane (GE supplemented with 20% serum were added to the lower Healthcare, Amersham, UK), protein expression was detected chamber. After 24 h, the top of the insert membrane was by specific antibodies. Primary antibodies used were: MST1/2, scrubbed free of cells by using a cotton swab and phosphate- Phospho-MST1(Thr183)/MST2 (Thr180), LATS1, Phospho- buffered saline washes, and the bottom side was stained with LATS1(Ser909), YAP, Phospho-YAP(Ser127) (Cell Signaling 0.1% crystal violet (Sigma-Aldrich). The number of cells on Technology, Hitchin, UK), E-cadherin, N-cadherin, vimentin the lower surface of each chamber was either counted or dye (BD Biosciences, Oxford, UK), Mer (Abcam, Cambridge, was extracted with 500 ml 0.1 M sodium citrate in 50% ethanol UK), and caspase-3, coilin, occludin, FLAG and b-actin and absorbance was measured at 570 nm. (Sigma-Aldrich). Secondary antibodies used were anti-mouse horseradish peroxidase and anti-rabbit horseradish peroxidase RNA isolation and quantitative real time PCR (Santa Cruz, CA, USA). Protein expressions on immunoblots Isolation of RNA from cell pellets was carried out using were analyzed using image J analysis and a Student’s unpaired an RNeasy Mini Kit (Qiagen, Crawley, UK). Transcriptor t-test was performed to test significance. Reverse Transcriptase (Roche, Burgess Hill, UK) was used to synthesize cDNA from 2 ml of the DNase-digested RNA. TNFa treatment and MTT assay Quantitative PCR was performed using a SYBR green assay Cells were plated in 96-well plates in a total of 100 ml medium. and an Mx3005P machine (Stratagene) and primers corre- Both TRex and TRex-willin-GFP cells were incubated in sponding to willin or b-actin (amplifying 494-2338 of 1 mg/ml tetracycline 2 days before a 4 h 50 ng/ml TNFa NM_001042481 and 496-885 of NM_001101) or PRL treatment. After this treatment, MTT solution was added to (NM000948), IGFBP3 (NM001013398), BMP2 (NM001200) a final concentration of 1 mg/ml and left to incubate for 3 h in or RASSF8 (NM001164746). a sterile tissue culture incubator at 37 1C. All medium was removed and 100 ml dimethylsulfoxide was added into each D. melanogaster stocks well. Plates were shaken for 1 min and absorbance was Misexpression experiments used the following stocks: 32B- read at 570 nm using MRX Microplate Reader (DYNEX Gal4, 71B-Gal4, MS1096-Gal4, C5-Gal4, UAS-ex and UAS- Technologies, Worthing, UK). willin-GFP. The mosaic analysis with a repressible cell

Oncogene Salvador/Warts/Hippo signaling pathway in mammals L Angus et al 249 marker technique was used to express willin-GFP in exMGH1 measured using Adobe Photoshop by taking the number of mutant cells. Genotypes of mosaic analysis with a repressible pixels bound within the wing circumference. Wing shape was cell marker experiments were as follows: y w hs-FLP measured by taking the ratio of lengths along the anterior– UAS-GFP; Tub-GAL80 FRT40/exMGH1FRT40; Tub-GAL4/ posterior and proximal–distal axes. t-tests were used to UASwillin-GFPy w hs-FLP UAS-GFP; Tub-GAL80 FRT40/ determine whether changes in size or shape were statistically exMGH1FRT40; Tub-GAL4/ þ and y whs-FLPUAS-GFP; and significant. Tub-GAL80 FRT40/FRT40;Tub-GAL4/ þ . All fly stocks were grown at 24±1 1C. Conflict of interest D. melanogaster tissue immunohistochemistry Antibodies used were as follows: anti-Fat (Milton et al., 2010), The authors declare no conflict of interest. anti-Ex (from Allen Laughon, University of Wisconsin, Madison, WI, USA), anti-ELAV (DSHB), and anti-rabbit and anti-rat secondary antibodies (Invitrogen). Tissues were fixed in 4% paraformaldehyde, washed with phosphate- Acknowledgements buffered saline containing 0.3% Triton X-100, incubated with antibodies overnight at 4 1C in phosphate-buffered saline and We thank P Burke for embryo injections and N Chang for 0.1% Triton X-100 and mounted in 90% glycerol as in Zhang technical assistance. We also thank the BBSRC and EPSRC et al. (2009). Images were collected on an Olympus Confocal for funding to LA; Scottish University Life Science Alliance for microscope and processed using Olympus Confocal Fluoview funding to SM. KFH holds Career Development Awards from FV1000 software and Adobe Photoshop. the International Human Frontier Science Program Organiza- tion and the National Health and Medical Research Council of D. melanogaster wing measurements Australia and a Project Grant from the National Health and Adult wings were mounted in Canada Balsam with xylene and Medical Research Council of Australia. AS holds Melbourne imaged using an Olympus BX-51 microscope. Wing area was International Research and Fee Remission Scholarships.

References

Badouel C, Gardano L, Amin N, Garg A, Rosenfeld R, Le Bihan T Hamaratoglu F, Willecke M, Kango-Singh M, Nolo R, Hyun E, Tao et al. (2009). The FERM domain protein Expanded regulates Hippo C et al. (2006). The tumour-suppressor genes NF2/Merlin and pathway activity via direct interactions with the transcriptional Expanded act through Hippo signalling to regulate cell proliferation activator Yorkie. Dev Cell 16: 411–420. and apoptosis. Nat Cell Biol 8: 27–36. Bennett FC, Harvey KF. (2006). Fat cadherin modulates organ size in Harvey KF, Pfleger CM, Hariharan IK. (2003). The Drosophila Mst Drosophila via the Salvador/Warts/Hippo signaling pathway. Curr ortholog, Hippo, restricts growth and cell proliferation and Biol 16: 2101–2110. promotes apoptosis. Cell 114: 457–467. Boedigheimer MJ, Laughon A. (1993). Expanded: a gene involved in Harvey K, Tapon N. (2007). The Salvador-Warts-Hippo pathway: an the control of cell proliferation in imaginal discs. Development 118: emerging tumour-suppressor network. Nat Rev Cancer 7: 182–191. 1291–1301. Huang J, Wu S, Barrera J, Matthews K, Pan D. (2005). The Hippo Boedigheimer MJ, Nguyen KP, Bryant PJ. (1997). Expanded functions signaling pathway co-ordinately regulates cell proliferation and in the apical cell domain to regulate the growth rate of imaginal apoptosis by inactivating Yorkie, the Drosophila homolog of YAP. discs. Dev Genet 20: 103–110. Cell 122: 421–434. Debnath J, Muthuswamy SK, Brugge JS. (2003). Morphogenesis and Lee KK, Ohyama T, Yajima N, Tsubuki S, Yonehara S. (2001). MST, oncogenesis of MCF-10A mammary epithelial acini grown in three- a physiological caspase substrate, highly sensitizes apoptosis both dimensional basement membrane cultures. Methods 30: 256–268. upstream and downstream of caspase activation. J Biol Chem 276: Edgar BA. (2006). From cell structure to transcription: Hippo forges a 19276–19285. new path. Cell 124: 267–273. Lee T, Luo L. (1999). Mosaic analysis with a repressible cell marker for Genevet A, Wehr MC, Brain R, Thompson BJ, Tapon N. (2010). studies of gene function in neuronal morphogenesis. Neuron 22: 451–461. Kibra is a regulator of the Salvador/Warts/Hippo signaling Mao Y, Mulvaney J, Zakaria S, Yu T, Morgan KM, Allen S et al. network. Dev Cell 18: 300–308. (2011). Characterization of a Dchs1 mutant mouse reveals require- Graves JD, Draves KE, Gotoh Y, Krebs EG, Clark EA. (2001). Both ments for Dchs1-Fat4 signaling during mammalian development. phosphorylation and caspase-mediated cleavage contribute to Development 138: 947–957. regulation of the Ste20-like protein kinase Mst1 during CD95/ Milton CC, Zhang X, Albanese NO, Harvey KF. (2010). Differential Fas-induced apoptosis. J Biol Chem 276: 14909–14915. requirement of Salvador-Warts-Hippo pathway members for organ Grusche FA, Richardson HE, Harvey KF. (2010). Upstream size control in Drosophila melanogaster. Development 137: 735–743. regulation of the hippo size control pathway. Curr Biol 13: Okada T, Lopez-Lago M, Giancotti FG. (2005). Merlin/NF2 mediates R574–R582. contact inhibition of growth by suppressing recruitment of Rac to Gunn-Moore FJ, Welsh GI, Herron LR, Brannigan F, Venkateswarlu the plasma membrane. J Cell Biol 171: 361–371. K, Gillespie S et al. (2005). A novel 4.1 ezrin radixin Overholtzer M, Zhang J, Smolen GA, Muir B, Li W, Sgroi DC et al. moesin (FERM)-containing protein, ‘Willin’. FEBS Lett 579: (2006). Transforming properties of YAP, a candidate oncogene on the 5089–5094. chromosome 11q22 amplicon. Proc Natl Acad Sci 103: 12405–12410. Haase D, Meister M, Muley T, Hess J, Teurich S, Schnabel P et al. Pan D. (2007). Hippo signaling in organ size control. Genes Dev 21: (2007). FRMD3, a novel putative tumour suppressor in NSCLC. 886–897. Oncogene 26: 4464–4468. Saburi S, Hester I, Fischer E, Pontoglio M, Eremina V, Gessler M Hao Y, Chun A, Cheung K, Rashidi B, Yang X. (2008). Tumor et al. (2008). Loss of Fat4 disrupts PCP signaling and oriented suppressor LATS1 is a negative regulator of oncogene YAP. J Biol cell division and leads to cystic kidney disease. Nat Genet 40: Chem 283: 5496–5509. 1010–1015.

Oncogene Salvador/Warts/Hippo signaling pathway in mammals L Angus et al 250 Silva E, Tsatskis Y, Gardano L, Tapon N, McNeill H. (2006). The signaling in conjunction with Merlin and Expanded. Dev Cell 18: tumor-suppressor gene fat controls tissue growth upstream of 288–299. expanded in the hippo signaling pathway. Curr Biol 16: 2081–2089. Zeng Q, Hong W. (2008). The emerging role of the Hippo pathway in Song H, Mak KK, Topol L, Yun K, Hu J, Garrett L et al (2010). cell contact inhibition, organ size control, and cancer development Mammalian Mst1 and Mst2 kinases play essential roles in organ in mammals. Cancer Cell 13: 188–192. size control and tumor suppression. Proc Natl Acad Sci USA 107: Zhang J, Smolen GA, Haber DA. (2008). Negative regulation of YAP 1431–1436. by LATS1 underscores evolutionary conservation of the Drosophila Tapon N, Harvey KF, Bell DW, Wahrer DC, Schiripo TA, Haber DA Hippo pathway. Cancer Res 68: 2789–2794. et al. (2002). Salvador promotes both cell cycle exit and apoptosis Zhang X, Milton CC, Humbert PO, Harvey KF. (2009). Transcrip- in Drosophila and is mutated in human cancer cell lines. Cell 110: tional output of the Salvador/Warts/Hippo pathway is controlled in 467–478. distinct fashions in Drosophila melanogaster, and mammalian cell Tyler DM, Baker NE. (2007). Expanded and fat regulate growth and lines. Cancer Res 69: 6033–6041. differentiation in the Drosophila eye through multiple signaling Zhao B, Li L, Lu Q, Wang LH, Liu CY, Lei Q et al. (2011). pathways. Dev Biol 305: 187–201. Angiomotin is a novel Hippo pathway component that inhibits Willecke M, Hamaratoglu F, Kango-Singh M, Udan R, Chen CL, Tao YAP oncoprotein. Genes Dev 25: 51–63. C et al. (2006). The Fat cadherin acts through the Hippo tumor- Zhao B, Wei X, Li W, Udan RS, Yang Q, Kim J et al. (2007). suppressor pathway to regulate tissue size. Curr Biol 16: 2090–2100. Inactivation of YAP oncoprotein by the Hippo pathway is involved Yu J, Zheng Y, Dong J, Klusza S, Deng WM, Pan D. (2010). Kibra in cell contact inhibition and tissue growth control. Gene Dev 21: functions as a tumor suppressor protein that regulates Hippo 2747–2761.

Oncogene