The Spectrin cytoskeleton regulates the Hippo signalling pathway

The Spectrin cytoskeleton is known to be polarised in epithelial cells, yet its role remains poorly understood. Here, we show that the Spectrin cytoskeleton controls Hippo signalling. In the developing Drosophila wing and eye, loss of apical Spectrins (alpha/beta-heavy dimers) produces tissue overgrowth and mis-regulation of Hippo target genes, similar to loss of Crumbs (Crb) or the FERM-domain protein Expanded (Ex). Apical beta-heavy Spectrin binds to Ex and co-localises with it at the apical membrane to antagonise Yki activity. Interestingly, in both the ovarian follicular epithelium and intestinal epithelium of Drosophila, apical Spectrins and Crb are dispensable for repression of Yki, while basolateral Spectrins (alpha/beta dimers) are essential. Finally, the Spectrin cytoskeleton is required to regulate the localisation of the Hippo pathway effector YAP in response to cell density human epithelial cells. Our findings identify both apical and basolateral Spectrins as regulators of Hippo signalling and suggest Spectrins as potential mechanosensors.


Introduction
The Hippo pathway transduces signals from the cell surface to the nucleus to control tissue growth and regeneration in animals (Pan, 2010;Halder & Johnson, 2011;Tapon & Harvey, 2012). Recent work implicates the Hippo pathway as a key regulator of stem cell proliferation (Camargo et al, 2007;Cai et al, 2010;Karpowicz et al, 2010;Shaw et al, 2010;Staley & Irvine, 2010;Zhang et al, 2010Zhang et al, , 2011Cordenonsi et al, 2011). The core of the pathway was discovered in Drosophila and includes the upstream kinase Hippo (MST1/2 in mammals) and the downstream kinase Warts (LATS1/2 in mammals), which acts to phosphorylate and inhibit the transcriptional activator Yorkie (Yki; YAP/TAZ in mammals) (Harvey et al, 2003;Udan et al, 2003;Wu et al, 2003;Huang et al, 2005). Yki then acts with the Mask (MASK1/2 in mammals) co-factor to switch the nuclear DNA-binding protein Scalloped (TEAD1-4 in mammals) from a transcriptional repressor to an activator (Wu et al, 2008;Koontz et al, 2013;Sansores-Garcia et al, 2013;Sidor et al, 2013).
Several proteins can act upstream of the core kinase cascade, including the apical FERM-domain proteins Expanded (Ex; similar to both FRMD6 and AMOT proteins in humans) and Merlin (Mer; the NF2 tumour suppressor in humans), which act in parallel with activate Hippo signalling (Hamaratoglu et al, 2006;Irvine, 2012). In Drosophila wing or eye epithelia, mutation of ex is sufficient to cause mild tissue overgrowth, but ex, mer double mutants cause a much stronger overgrowth phenotype, similar to hpo or wts mutants (Hamaratoglu et al, 2006). Ex is recruited to the membrane by the transmembrane protein Crumbs (Crb), an apical polarity determinant that can form apical cell-cell junctions in epithelia (Chen et al, 2010;Ling et al, 2010;Robinson et al, 2010). Mutants in crb therefore cause a mild overgrowth phenotype in wing and eye epithelia (Chen et al, 2010;Ling et al, 2010). Mer is also recruited to apical cell-cell junctions, where it binds to the Kibra (Kib) protein (Baumgartner et al, 2010;Genevet et al, 2010;Yu et al, 2010). ex, kib or kib, mer double mutants cause a strong hpo-like overgrowth phenotype, suggesting that the three proteins act together upstream of the core kinase cascade (Baumgartner et al, 2010;Genevet et al, 2010;Yu et al, 2010). In addition, ex, kib double mutants strongly affect polarisation of Crb in the ovarian follicular epithelium and polarisation of the actin cytoskeleton for border cell migration, functions that are independent of nuclear signalling via Yki (Fletcher et al, 2012;Lucas et al, 2013).
In mammalian cells in culture, the Hippo pathway responds to mechanical stimulation, being activated in densely confluent epithelial cells (such that YAP is cytoplasmic) and becoming inactivated when cells are sparse and stretched out across their substrate (such that YAP becomes nuclear) Aragona et al, 2013). A functioning F-actin cytoskeleton is required for this response to mechanical stimulation, but whether F-actin itself is the molecular mechanosensor or whether other molecules might mediate this function remains unclear Aragona et al, 2013). Interestingly, regulation of YAP by mechanical stretching and the F-actin cytoskeleton appears to be partly independent of LATS phosphorylation of YAP and likely involves a yet unidentified mechanism Aragona et al, 2013;Gaspar & Tapon, 2014).
Here, we identify the Spectrin cytoskeleton as crucial upstream regulator of Yki in the developing wing, eye, follicular epithelium and border cells. We further show that human Spectrins are essential for regulation of YAP in response to cell density in human cells.

Results
To identify novel regulators of Hippo signalling, we performed an in vivo RNAi screen in the Drosophila wing for novel genes controlling tissue growth (M. Campos & B. J. Thompson, manuscript in preparation). In this screen, we identified the apical Spectrin cytoskeleton components a-Spectrin (a-Spec) and b-heavy Spectrin (b H Spec)-also known as Karst (Kst)-as producing moderate wing and eye overgrowth phenotypes, similar to RNAi knock-down of Crb . Spectrins are large cytoskeletal proteins that form hexagonal networks at the intracellular surface of the plasma membrane in all animal cells and have been reported to have mechanosensory properties (Bennett & Baines, 2001;Johnson et al, 2007;Bennett & Healy, 2009;Stabach et al, 2009;Meng & Sachs, 2012;Krieg et al, 2014). The Spectrin cytoskeleton is polarised in Drosophila epithelia, with dimers of aand b H -Spec/Kst localising to the apical domain and dimers of aand b-Spec localising to the basolateral domain (Thomas & Kiehart, 1994;Lee et al, 1997;Thomas et al, 1998;Thomas & Williams, 1999;Zarnescu & Thomas, 1999;Medina et al, 2002). Notably, RNAi knock-down of the basolateral b-Spec did not have a consistent effect on eye or wing size (Fig 1G and H and Supplementary Figs S3 and S4). Furthermore, just as crb and ex mutants are known to genetically interact with kib, knock-down of a-Spec or b H -Spec/Kst strongly enhanced the overgrowth caused by a kib null mutant in the eye (Fig 1I-R).
Despite previous reports that apical b H -Spec/Kst interacts physically with Crb, genetic analysis of b H -spec/kst mutants indicated that it is dispensable for polarisation of Crb and for epithelial polarity in general (Thomas et al, 1998;Zarnescu & Thomas, 1999;Medina et al, 2002;Pellikka et al, 2002). Since Crb is known to regulate Hippo signalling (Chen et al, 2010;Ling et al, 2010;Robinson et al, 2010), we tested whether loss of apical Spectrins also affects Hippo signalling outputs. We examined interommatidial cells in the pupal retina and found that loss of a-Spec or b H -Spec/Kst increases cell number and this effect is magnified by concominant mutation of kib (Fig 2A-F). We also examined the expression of the key Hippo reporter gene, ex.lacZ, in wing discs expressing a-spec RNAi in the posterior compartment with hh.Gal4. We found that, compared to controls, wing discs expressing a-spec RNAi exhibit a slightly elevated level of ex.lacZ expression in the posterior compartment ( Fig 2G and H). This elevation of ex.lacZ expression is similar in magnitude to that caused by kib RNAi and becomes stronger in a-spec, kib double RNAi wing discs, similar to hpo RNAi (Fig 2I-K).
These results show that apical Spectrins regulate Yki activity in the Drosophila wing and eye. They also show that Spectrins act in parallel with Kibra, in the same manner as Ex (Baumgartner et al, 2010), as the double-mutant spectrin kibra or expanded kibra each cause a stronger phenotype than the single mutants alone (Baumgartner et al, 2010).
We next investigated the mechanism by which apical Spectrins regulate Yki activity. Since Spectrins act in parallel with Kibrasimilar to Ex (Baumgartner et al, 2010)-and have been shown to physically associate with Crb (Medina et al, 2002;Pellikka et al, 2002)-again similar to Ex (Chen et al, 2010;Ling et al, 2010;Robinson et al, 2010)-we tested whether Ex might interact with Spectrins. We performed co-immunoprecipitation experiments from Drosophila S2 cells expressing V5-tagged Ex and a series of constructs expressing portions of the very large b H -Spec/Kst protein.
We found that Ex interacts strongly with the N-terminal region of b H -Spec/Kst ( Fig 3A). Pulling down the N-terminal region of b H -Spec/Kst with Ex also co-immunoprecipitated endogenous a-Spec, which is known to form dimers with b H -Spec/Kst ( Fig 3A).
In vivo, we found that b H -Spec/Kst co-localises with Ex at the apical domain of wing disc epithelial cells (Fig 3B and C). a-Spec or b H -Spec/Kst is not required to localise Ex or Crb apically (Supplementary Fig S5 and data not shown). Instead, they appear to be required for normal activation of Ex signalling to Hpo and Wts, as the tissue overgrowth phenotype of a-Spec or b H -Spec/Kst knockdown is completely suppressed by overexpression of Ex (Fig 3D-I).
These results show that apical Spectrins bind to and co-localise with Ex and act genetically upstream or at the level of Ex to regulate signalling to Yki.
Hippo signalling has been proposed to have a possible mechanosensory role in the Drosophila wing disc, where a pattern of stretching and compression of cells at their apical surfaces correlates with the pattern of Yki activity as measured with ex.lacZ (Fig 4A and B;Aegerter-Wilmsen et al, 2007;Legoff et al, 2013;Mao et al, 2013;Schluck et al, 2013). We found that this pattern of stretching and compression influences the intensity of Crb and apical Spectrin staining in cells (Fig 4A-F). Thus, Crb and b H -Spec/Kst are concentrated at the junctions of small compressed cells and diluted at the junctions of stretched cells in a manner that inversely correlates with ex.lacZ expression. This correlation suggests a potential model of mechanosensory regulation of Yki activity via Spectrin-dependent clustering of Crb complexes ( Fig 4G). According to this model, stretching of cells would exert force upon the apical Spectrin cytoskeleton that would de-cluster Crb complexes and therefore reduce Hpo and Wts activation and increase Yki activity (Fig 4G; see also Discussion).
To test this model, we aimed to induce clustering of Crb complexes by overexpression of a form of Crb whose intracellular domain was replaced with GFP (Crb ExTM -GFP; Fig 5A) (Pellikka et al, 2002;Thompson et al, 2013). Overexpression of Crb ExTM -GFP during wing development with nub.Gal4 resulted in a small wing phenotype, highly similar to overexpression of Wts (Fig 5B-D). Furthermore, the overgrowth phenotype caused by RNAi knockdown of apical Spectrins was completely suppressed by coexpression of either Crb ExTM -GFP or Wts (Fig 5E-J). Crb ExTM -GFP appears to act upstream of Wts, because the tissue undergrowth phenotype induced by Crb ExTM -GFP is suppressed by RNAi knockdown of Wts (Fig 5K and L). These results are consistent with the notion that clustering of Crb complexes induces Hippo signalling to inhibit tissue growth, lending support to a model of mechanosensation involving clustering of Crb complexes. However, other interpretations and models of mechanosensation are also possible.
One alternative model of mechanosensation involves recruitment of Warts to E-cadherin via the Ajuba protein (Rauskolb et al, 2014). This mechanism appears to be distinct from and to act in parallel with the Spectrin/Crumbs-mediated version we involves the actin cytoskeleton, which can directly influence the nuclear localisation of the Yki homologues YAP and TAZ in mammalian cell culture independently of MST and LATS kinases Aragona et al, 2013). Current evidence suggests that all effects of the actin cytoskeleton on Yki activity appear to be mediated via Hpo and Wts (Sansores-Garcia et al, 2011), although further work is needed to rigorously test whether Wts-independent regulation of Yki also occurs in Drosophila (Gaspar & Tapon, 2014). Thus, we currently favour the view that apical Spectrins act with Crb complexes to help sense forces by activating Hpo-Wts signalling during Drosophila wing and eye development.
We next tested whether loss of Spectrins can produce an ex-like phenotype in other tissue contexts. In the ovarian follicular epithelium, Ex is known to act in parallel with Kibra to regulate polarisation of Crb, such that ex, kib double-mutant cells accumulate Crb in vesicles (Fletcher et al, 2012). This role of Ex and Kibra does not require more downstream signalling components such as wts (Fletcher et al, 2012). We found that b H -Spec/Kst co-localises with Ex at the apical domain of follicle cells, while Kibra is present both apically and in the cytoplasm in a punctate pattern that co-localises with the Exocyst complex-with which Kibra has been shown to interact (Rosse et al, 2009) ( Supplementary Fig S6A-C). We found that loss of apical Spectrins alone does not affect Crb, but loss of We noted that a-spec RNAi in kib mutant clones caused a strong overproliferation and multilayering phenotype in follicle cells ( Supplementary Fig S6M). We therefore tested whether Spectrins are required for regulation of Yki activity in the ovarian follicular epithelium. Yki is known to promote early proliferation and suppress apoptosis in follicle cells until stage 6, during which time it expresses the transcription factor Cut (Huang & Kalderon, 2014). After stage 6, Cut expression is silenced and cells arrest proliferation and switch to an endocycle. When Hippo signalling is disrupted after stage 6, Cut is known to be re-expressed in a group of posterior follicle cells that continue to proliferate (Meignin et al, 2007;Polesello & Tapon, 2007;Genevet et al, 2010;Yu et al, 2010). Surprisingly, we found that mutation of either b H -spec/kst or crb is not sufficient to drive early overproliferation (Fig 6A-D) or to induce Cut in late-stage posterior follicle cells (Fig 6G-I). Multilayering only occurs in around 30% of crb mutant clones, while mutation of a-spec or b-spec leads to early overproliferation and multilayering in 95% of clones (Fig 6E and F) and also leads to re-expression of Cut in the posterior follicle cells (Fig 6J and K). Furthermore, another Hippo target gene, ex.lacZ, is also ectopically expressed in posterior follicle cells mutant for a-spec or b-spec but not in b H -spec/ kst or crb mutant clones (Fig 6L-P). Notably, mutation of b-spec causes a disruption of membrane tension such that the normally regular hexagonal shape of follicle cells becomes disorganised (Fig 6Q-S). These findings indicate that the basolateral a-b Spectrin cytoskeleton controls membrane tension and Hippo signalling to the nucleus in the follicular epithelium. A small group of anterior follicle cells, called border cells, are known to delaminate from the epithelium at stage 8 of development and migrate across the egg chamber to reach the oocyte at the posterior by stage 10. We previously showed that upstream Hippo pathway components have a key role in promoting border cell migration by regulating the actin cytoskeleton (Lucas et al, 2013). We found that a-spec RNAi knock-down in kib mutant border cell clusters causes a strong delay in reaching the oocyte by stage 10 of oogenesis ( Supplementary Fig S7A-E). a-spec RNAi knock-down alone, or mutation of kib alone, causes a milder phenotype (Supplementary Fig S7E). Interestingly, RNAi knockdown of b H -spec/kst did not cause a phenotype, alone or in combination with kib ( Supplementary Fig S7E). Accordingly, we found that b H -Spec/Kst is not actually expressed in border cells, while b-Spec localises around the entire plasma membrane with a-Spec ( Supplementary Fig S7F-I). These results indicate that a-b Spectrin dimers regulate Hippo signalling during border cell migration.
The Hippo pathway was recently shown to have a key role in regulating stem cell proliferation in the Drosophila intestine (Karpowicz et al, 2010;Shaw et al, 2010;Staley & Irvine, 2010). RNAi knock-down of wts or overexpression of yki in enterocytes with the myo1A.G4 driver leads to induction of stem cell proliferation (Karpowicz et al, 2010;Shaw et al, 2010;Staley & Irvine, 2010). However, the upstream regulators that control Hippo signalling in the intestine have not been identified. We therefore examined the requirement for Crb and Spectrins in the adult fly intestine. We found that RNAi of crb or apical b H -spec/kst did not lead to stem cell overproliferation (Fig 7A-C). However, RNAi silencing of a-spec or b-spec does produce a strong stem cell proliferation response, similar to overexpression of Yki (Fig 7D-G). Furthermore, a Hippo pathway reporter gene DIAP1-HRE-GFP is activated in the overproliferating stem cells (Fig 7H-K). These results indicate that the basolateral a-b Spectrin cytoskeleton, rather than the apical a-b H -Spectrin cytoskeleton, is crucial to promote Yki activity in enterocytes.
The Spectrin cytoskeleton is conserved between Drosophila and humans, but is more complex in humans as it features many variant b-Spectrins. To test whether the Spectrin cytoskeleton also regulates Hippo signalling in human cells, we silenced expression of the sole human non-erythrocyte alpha-Spectrin protein, SPTAN1, and a major non-erythrocyte b-Spectrin, SPTBN1, by siRNA transfection of Caco-2 cells. The human homologue of Yki, called YAP, is normally nuclear in sparsely plated cells, but becomes cytoplasmic in densely confluent epithelial monolayers due to mechanical cues   (Fig 8A and B). We found that siRNA knock-down of The EMBO Journal Spectrins regulate Hippo signalling Georgina C Fletcher et al SPTAN1 or SPTBN1 in densely confluent epithelia is sufficient to send YAP to the nucleus and to reduce YAP phosphorylation and LATS1 phosphorylation (Fig 8C-G). These results show that the Spectrin cytoskeleton is crucial for regulation of YAP in response to density in human cell culture and that disruption of Spectrins has similar effects to mechanical stretching of cells by plating at low density.

Discussion
Our results identify the Spectrin cytoskeleton as an important upstream regulator of the Hippo signalling pathway in both Drosophila and human cells. In all fly tissues we have examined, Spectrins are essential to maintain full repression of Yki activity. However, different tissues exhibit different dependence on apical a-b H versus basolateral a-b Spectrins. The localisation of Spectrin proteins in each model tissue we have examined is summarised in Supplementary Fig S8. In the developing wing and eye, apical a-b H -Spectrins act with Crb and Ex to activate Hippo signalling and thereby restrict tissue growth (Figs 1 and 2). Since Crb binds to Ex (Chen et al, 2010;Ling et al, 2010), and Ex can bind to both Kibra (Genevet et al, 2010) and Spectrins (Fig 2), and Crb can also associate with Spectrins   (Genevet et al, 2010;Yu et al, 2010), and we are able to co-immunoprecipitate both Kibra and Mer with apical a-b H Spectrins (Supplementary Fig S9). In contrast, Hpo and Wts localise primarily to the cytoplasm, but must transiently associate with upstream pathway components at the plasma membrane in order to be activated (Lucas et al, 2013;Yin et al, 2013). Direct binding of Wts to Mer and indirect binding of Hpo via Sav to a network of WW domain-PPXY motif interactions appear to facilitate activation of Hpo and Wts at the plasma membrane, where molecular clustering may induce Hpo autophosphorylation and phosphorylation of Wts to drive signalling. In the wing and eye epithelium, apical a-b H Spectrins may assist this signal-activating process by forming a dense meshwork that promotes clustering of Crb-Ex-Kib-Mer complexes to induce Hpo and Wts kinase activity to restrict growth (Fig 4). In support of this model, inducing clustering of endogenous Crb by overexpressing a form of Crb lacking the intracellular domain is sufficient to overcome the loss of Spectrins and to suppress wing growth (Fig 5).
In the developing follicular epithelium and adult intestinal epithelium, basolaterally localised a-b Spectrins, rather than apical a-b H Spectrins, are crucial for repression of Yki activity (Figs 6 and 7). This finding may explain why apical Crb is surprisingly not required for Yki repression in these tissues (Figs 6 and 7). This finding is all the more remarkable because apical a-b H Spectrins do act in parallel with Kibra to polarise Crb in follicle cells (Supplementary Fig S6). Thus, the mere presence of Crb, Ex, Kib and a-b H Spectrins in a cell does not necessitate that these proteins will form the primary control mechanism for the regulation of Yki. How basolateral a-b Spectrins connect to Hippo signalling and/or regulation of Yki in these tissues remains to be explored. Nevertheless, in all Drosophila tissues we have examined, either the a-b or a-b H Spectrins are necessary for normal repression of Yki.
In human epithelial cells, polarisation of a (SPTAN1) and b (SPTBN1) Spectrins is much less apparent than in Drosophila (Supplementary Fig S8). Nevertheless, our data show that these Spectrins are essential for regulating the classic YAP localisation response to cell density in culture (Fig 8), as are homologues of Crb (CRB3), Ex (AMOT) and Mer (NF2) (Zhao et al, 2007(Zhao et al, , 2011Varelas et al, 2010). This response to cell density was shown to be a mechanosensory response that requires an intact F-actin cytoskeleton . However, the identity of the mechanosensory molecule(s) that mediate this response is not clear. Our findings suggest that the Spectrin cytoskeleton is a candidate for this Hippo pathway mechanosensor.
Several lines of evidence support a mechanosensory function for Spectrins. Firstly, electron microscopy studies show that Spectrins form a dense meshwork at the plasma membrane that can stretch out under force, separating the N-terminal nodes at which FERMdomain proteins bind (Hirokawa et al, 1983;Byers & Branton, 1985;Ursitti et al, 1991). Secondly, individual Spectrin molecules can deform under force via conformational change (Johnson et al, 2007;Stabach et al, 2009). Thirdly, construction of a FRET-based Spectrin sensor reveals that Spectrins are under tension and can sense force in human cells in culture (Meng & Sachs, 2012) and in C. elegans neurons (Krieg et al, 2014). Finally, loss of Spectrins causes a loss of membrane tension in C. elegans neurons (Krieg et al, 2014) as well as red blood cells (Stokke et al, 1986), and our results are consistent with a similar role in Drosophila epithelial cells (Figs 2A-F and 6Q-S). Thus, it is likely that force may cause alterations in the Spectrin cytoskeleton that modulate Hippo signalling, possibly by spatially separating the N-terminal nodes of the network that bind to FERM-domain proteins, thereby inducing de-clustering of upstream pathway components and signal inhibition (Fig 4G).
Since Spectrins can bind to F-actin, it is also possible that force upon Spectrins may influence the actin cytoskeleton and therefore potentially influence Yki activation independently of Hpo and Wts activation, as has been shown for the mammalian homologues YAP/TAZ Aragona et al, 2013). Furthermore, Wts has also been shown to influence F-actin polymerisation via regulation of Ena/VASP proteins (Lucas et al, 2013), raising the possibility that canonical Hpo-Wts signalling may also act via the regulation of F-actin dynamics to control Yki, in addition to direct phosphorylation of the Yki protein by Wts. Further work is necessary to test whether direct F-actin regulation of Yki localisation occurs in Drosophila as it does in mammalian cells (Gaspar & Tapon, 2014).
Regardless of which signalling mechanisms Spectrins use to control Yki activity, their molecular nature makes them excellent candidate mechanosensors and Spectrins are clearly able to influence Yki-dependent tissue growth. Furthermore, there is good evidence that forces can and do contribute to promoting cell proliferation in Drosophila wing epithelia (Aegerter-Wilmsen et al, 2007;Legoff et al, 2013;Mao et al, 2013;Schluck et al, 2013), and it has been proposed that forces may also drive proliferation in the follicular epithelium (Wang & Riechmann, 2007). In addition, input from the Dachsous-Fat pathway to Hippo signalling is mediated via the atypical myosin Dachs, which has been shown to promote tension at apical cell-cell junctions (Mao et al, 2011). Thus, it is not implausible that forces could be sensed by Drosophila Spectrins as part of the physiological control of tissue growth during development and that this function could help explain how disparate upstream regulators of Hippo signalling, such as Crb-Crb junctions and the Dachsous-Fat cadherin system, converge to control Yki activity.
A physiological role for Spectrins as sensors of tension that regulate Hippo signalling may well be conserved across metazoa. For example, the dramatic effect of forces on human skin growth is well known, and the Hippo pathway has been shown to regulate skin proliferation in mice (Schlegelmilch et al, 2011;Silvis et al, 2011;Zhang et al, 2011). Notably, a recently proposed alternative mechanism for mechanosensation involving recruitment of Wts to a-catenin via the Ajuba protein is less likely to be conserved in mammals (Das Thakur et al, 2010;Rauskolb et al, 2014). In ajuba mutants, Wts recruitment to a-catenin is disrupted, leading to stronger Hippo signalling and tissue undergrowth in Drosophila (Rauskolb et al, 2014). However, in mammals, loss of a-catenin leads to weaker Hippo signalling and tissue overgrowth in mouse skin (Schlegelmilch et al, 2011;Silvis et al, 2011) and loss of adherens junctions also weakens Hippo signalling to activate YAP in cultured human cells (Kim et al, 2011). Thus, our identification of Spectrins as potential mechanosensors in the Hippo pathway provides an important novel mechanism that is more likely to be conserved in all metazoans and may be a critical control mechanism in both normal development and cancer.

Materials and Methods
Mitotic clones in follicle cells were generated using the FLP/FRT system and were marked either negatively (absence of GFP) or positively (presence of GFP; MARCM). Third instar larvae were heatshocked once at 37°C for 1 h and dissected 4 days after eclosion. Expression of UAS-driven transgenes in follicle cells was achieved with the MARCM system and in wing imaginal disc cells with the posterior compartment driver hh.Gal4 and in enterocytes with Myo1A.G4.

Drosophila genotypes
See Supplementary Materials and Methods for all Drosophila genotypes used in this study.

Immunostaining of ovaries, imaginal discs and pupal retinas
Ovaries and imaginal discs were dissected in PBS, fixed for 20 min in Secondary antibodies (all from Molecular Probes; Invitrogen) were used at 1:500 for 2-4 h prior to multiple washes in PBST and staining with DAPI at 1 lg/ml for 10-30 min prior to mounting on slides in Vectashield (Vector labs). Images were taken with a Leica SP5 confocal microscope and processed with Adobe Photoshop.

Immunostaining of intestinal epithelia
Flies were maintained on standard media, which was changed every 3 days. Crosses were set up and maintained at 18°C until adulthood. Three-to four-day-old adults were shifted to 29°C for 10 days for induction of transgene expression. Female adult flies were dissected in 1× PBS. The gastrointestinal tract was removed and fixed in 0.5× PBS with 4% paraformaldehyde for 30 min. Samples were washed in 0.1% Triton X-100 (PBST), permeabilised for 30 min in 0.3% PBST and pre-blocked for 1 h in 10% NGS before incubation with primary antibody overnight at 4°C in 5% NGS. Samples were washed in PBST and pre-blocked for 1 h in 10% NGS before incubation with secondary antibody in 5% NGS for 2 h at room temperature. Samples were stained with DAPI and washed in PBST, followed by PBS and mounted in Vectashield.
The following primary antibody was used: rabbit anti-phosphohistone H3 (Millipore) 1/1,000. Secondary antibodies (all from Molecular Probes; Invitrogen) were used at 1:500. Images were acquired on a Zeiss LSM710 confocal microscope and were processed using Adobe Photoshop.

Molecular biology
The Expanded expression plasmid was generated previously (Genevet et al, 2010). b-heavy Spectrin expression plasmids were generated using Gateway technology. Constructs detailed below were PCR-amplified and cloned into the pDONR Zeo Entry vector. For Destination vectors, the pAWF and pAWH plasmids from the Drosophila Gateway Vector Collection were used.

Cell culture and transfection
Human Caco-2 adenocarcinoma colon cells were grown in DMEM (Gibco: 41966) containing L-glutamine and sodium pyruvate supplemented with 10% heat-inactivated FCS and 100 lg/ml streptomycin and 100 lg/ml penicillin. Cells were maintained in a 37°C incubator at 5% atmospheric CO 2. All siRNA transfections were performed with interferin transfection reagent (Polyplus) using antibiotic-free media. Caco-2 cells were seeded upon 10 mm coverslips coated with 20 lg/ml fibronectin in a 48-well plate at either low density or high density, and 2 h after seeding treated with the siRNA/transfection mix. A final concentration of 100 nM siRNA was used. The following day, the media was changed and another round of siRNA transfection was performed. Cells were left for a total of 72 h before being processed for either immunofluorescence or immunoblotting. siRNA oligonucleotides used for RNAi of SPTAN1 and SPTBN1 were as a siGENOME SMARTpool (Thermo scientific). For fixation, cells were washed with 1× PBS and either fixed in 4% paraformaldehyde in PBS for 15 min, before permeabilising with 0.3% Triton X-100 in blocking buffer (PBS containing 0.5% BSA, 10 mM glycine, and 0.1% sodium azide), or À20°C methanol for 5 min, before rehydrating with PBS. Samples were then washed and incubated in blocking buffer for 1 h before staining. Primary and secondary antibodies were incubated in blocking buffer. Coverslips were mounted using prolong anti-fade reagent (Invitrogen).

Antibodies, image acquisition and quantification for cell culture
Antibodies used in the paper were as follows: mouse and rabbit anti-SPTAN1 (Santa Cruz and CST); mouse anti-SPTBN1 (Santa Cruz); rabbit anti-YAP (Santa Cruz), rabbit anti-pYAP (CST); rabbit anti-pLATS1 and anti-LATS1 (CST); mouse a-tubulin (Sigma). Secondary antibodies used were either from Jackson immunoresearch or from Invitrogen. DNA was stained with DAPI 1:1,000 and samples imaged with a Zeiss 710 or a Leica SP5 confocal microscope. Quantification of YAP localisation was scored in three categories: N = nuclear; N/C = nuclear and cytoplasmic and C = cytoplasmic. Cells were assessed over three independent experiments counting 350-500 cells per condition.
Supplementary information for this article is available online: http://emboj.embopress.org