Fat and Expanded act in parallel to regulategrowth through WartsYongqiang Feng and Kenneth D. Irvine*

Howard Hughes Medical Institute, Waksman Institute, and Department of Molecular Biology and Biochemistry, Rutgers, The State Universityof New Jersey, Piscataway, NJ 08854

Edited by Kathryn V. Anderson, Sloan–Kettering Institute, New York, NY, and approved November 7, 2007 (received for review July 17, 2007)

The conserved Drosophila tumor suppressors Fat and Expandedhave both recently been implicated in regulating the activity of theWarts tumor suppressor. However, there has been disagreement asto the nature of the links among Fat, Expanded, and Warts and thesignificance of these links to growth control. We report here thatmutations in either expanded or fat can be rescued to viabilitysimply by overexpressing Warts, indicating that their essentialfunction is their influence on Warts rather than reported effects onendocytosis or other pathways. These rescue experiments alsoseparate the transcriptional from the planar cell polarity branchesof Fat signaling and reveal that Expanded does not directly affectpolarity. We also investigate the relationship between expandedand fat and show, contrary to prior reports, that they have additiveeffects on imaginal disk growth and development. Although mu-tation of fat can cause partial loss of Expanded protein from themembrane, mutation of fat promotes growth even when Ex-panded is overexpressed and accumulates at its normal subapicallocation. These observations argue against recent proposals thatFat acts simply as a receptor for the Hippo signaling pathway andinstead support the proposal that Fat and Expanded can act inparallel to regulate Warts through distinct mechanisms.

Drosophila � Hippo � protocadherin

The regulation of growth is of fundamental importance toboth normal development and tumor formation. Studies over

the last several years have identified a cassette of genes, referredto as the Hippo signaling pathway, that act together to regulategrowth (1, 2). Most components were first identified by geneticstudies in Drosophila, although there is increasing evidence thathomologous genes regulate growth in mammals and that theirdysregulation is linked to cancer (1, 2). The eponymous hippogene (hpo) encodes a kinase that promotes the activation ofanother kinase, Warts (Wts) (1, 2). Activation of Wts alsodepends on the Hpo-interacting protein Salvador (Sav) and theWts-interacting protein Mob as Tumor Suppressor (Mats). Hpo,Wts, Sav, and Mats all act as tumor suppressors. A transcrip-tional coactivator protein, Yorkie, is negatively regulated byWts-dependent phosphorylation, and this regulation of Yorkieappears to be sufficient to explain the influence of Wts ongrowth (3).

Another Drosophila tumor suppressor, expanded (ex), has alsobeen linked to Hippo signaling (4–9). Ex and a structurallyrelated protein, Merlin (Mer), both contain FERM domains,which places them within a family of proteins that link membraneproteins to the cytoskeleton (10). Although Mer has only minoreffects on growth in Drosophila, its human homolog acts as atumor suppressor (10), and genetic studies suggest that Mer andex are partially redundant (11). Cells doubly mutant for ex andMer exhibit phenotypes similar to that of other components ofthe Hippo pathway, and Ex and Mer can influence Hpo and Wtsphosphorylation in cultured cells (4–8). Although the recentlinkage of Ex and Mer to Hippo signaling in Drosophila suggeststhat this might account for their influence on growth, it has alsobeen suggested that Ex and Mer influence growth in Drosophila

by affecting the endocytosis of multiple cell surface receptorsand the activity of multiple signaling pathways (9, 12).

Because Ex is a cytoplasmic protein, the linkage of Ex toHippo signaling left unanswered the question of whether thereare extracellular signals that modulate Hippo pathway activityand, if so, how those signals are received and transduced.Recently it was proposed that the protocadherin Fat acts as areceptor for Hippo signaling (4, 5, 8, 9). Fat is also a Drosophilatumor suppressor and regulates the same downstream genes asare regulated by Hippo signaling (4, 5, 7, 8). Although this couldbe consistent with Fat acting as a receptor for Hippo signaling,our own studies have identified a distinct, parallel signal trans-duction pathway in which Fat acts through the unconventionalmyosin Dachs to regulate the levels of Wts (7).

Thus, prior studies have raised fundamental questions aboutboth the contribution of Ex to Fat signaling and the contributionof Hippo signaling to Ex function. Results we describe heresupport a model in which Ex acts through the regulation of Wtsactivity and in so doing acts in parallel to a regulation of Wtsprotein levels exerted by Fat.

Resultsfat or ex Mutants Can Be Partially Rescued by Wts Overexpression.Although it has been proposed that fat and ex both act throughWts to regulate growth and gene expression, other models havebeen put forward. For example, it has been suggested that Ex andMer influence endocytosis and thereby influence growth bymodulating the EGFR, Notch, Hedgehog, Wg, and/or Fat sig-naling pathways (12). It has also been suggested that Fat and Exinfluence membrane sterol composition and thereby affectmultiple pathways including Wg, Dpp, and Hippo (9).

To the extent that Fat signaling involves an influence on Wtsstability, we reasoned that it might be possible to saturate thismechanism by overexpressing Wts. Overexpression of Wts didnot preclude observation of the influence of Fat signaling on Wtsprotein levels in prior experiments (7), implying that levels ofWts were not high enough to saturate effects on Wts stability.However, expression of Wts at a higher level, using a differentUAS-Myc:Wts transgene insertion [supporting information (SI)Fig. 5], was sufficient to rescue fat mutant animals to viability(Fig. 1B). This same level of Wts expression in wild-type or fatheterozygous animals results in a mild wing vein phenotype andslightly smaller wings, but the flies otherwise appear normal(Fig. 1E and data not shown).

Wts-rescued fat mutants are similar to wild-type animals interms of their overall size and most aspects of their morphology(Fig. 1B). Their most dramatic phenotype is a planar cell polarity

Author contributions: Y.F. and K.D.I. designed research; Y.F. performed research; Y.F. andK.D.I. analyzed data; and Y.F. and K.D.I. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

*To whom correspondence should be addressed. E-mail: irvine@waksman.rutgers.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0706722105/DC1.

© 2007 by The National Academy of Sciences of the USA

20362–20367 � PNAS � December 18, 2007 � vol. 104 � no. 51 www.pnas.org�cgi�doi�10.1073�pnas.0706722105

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

2, 2

021

(PCP) phenotype, evident in the misalignment of hairs andbristles. In the abdomen (Fig. 1I), this PCP phenotype wassimilar to that of unrescued fat mutants (13, 14). However, in thewing (SI Fig. 6), the PCP phenotype is weaker than unrescuedfat mutants and instead appears similar to the PCP phenotype offat mutants that have been partially rescued by expression ofthe fat intracellular domain (15). The persistence of a PCPphenotype is consistent with studies that have implied theexistence of two distinct branches of Fat signaling, one affectingtranscription and one dedicated to PCP (14, 16). At the sametime, it appears, both from these Warts rescue experiments andfrom prior studies (14, 15), that at least in the wing a transcrip-tional branch of Fat signaling contributes to the PCP phenotypeof fat mutants.

Wts-rescued fat mutants also exhibit abnormal legs, smallerwings, and wing vein phenotypes (Fig. 1 B and F). When the crossveins are visible, the spacing between them is reduced (Fig. 1F),which is a classic Fat pathway phenotype. It is not yet clearwhether these phenotypes result from imperfect rescue of Wts-dependent Fat signaling or a contribution of Wts-independentFat signaling to wing patterning. Nonetheless, the relativelynormal growth and development of these animals confirms thatcritical functions of Fat are mediated by its regulation of Wts.

Ex appears to affect Wts through Hippo signaling (6) andhence is expected to regulate Wts phosphorylation and/or itsassociation with Mats. If the mechanisms that promote Wtsactivation were completely eliminated in ex mutants, then over-expressing Wts would not be expected to have any effect.However, if Wts activation were reduced but not abolished in exmutants, then it is conceivable that elevation of Wts might atleast partially rescue ex mutants, because by the law of mass

action one could expect increased Wts to result in increasedproduction of active Wts. Indeed, ex-null mutant animals couldbe partially rescued by overexpression of Wts, resulting in therecovery of adult f lies (Fig. 1C). The rescue is incomplete, in thatthe wing discs and adult wings of these animals are enlarged (Fig.1G), but, aside from this, and mild wing vein phenotypes, theseWts-rescued ex mutants appear remarkably normal (Fig. 1C).They differ most obviously from Wts-rescued fat mutants in theirenlarged wings and in the absence of any visible PCP phenotype(Fig. 1J and SI Fig. 6), which suggests that previously reportedinfluences of ex on PCP (17) reflect the influence of Ex onWts-dependent transcription.

fat and ex Act in Parallel. Previously, we reported that Fat affectsWts protein levels and proposed this as a basis for Fat signaling(7). However, others reported an effect of Fat on Ex levels ormembrane localization and proposed this as a basis for Fatsignaling (4, 5, 8). Analysis of double mutant animals providesa test of whether fat and ex act in a single linear pathway or inparallel pathways, because, if Fat signaled solely through Ex,then ex fat double mutants would be expected to be identical tofat or ex single mutants. Indeed, a critical piece of evidence infavor of the hypothesis that Fat signals through Ex was the claimthat ex mutants, fat mutants, and ex fat double mutants haveidentical phenotypes (4, 5, 8). However, this claim was based onanalysis of a single phenotype, the number of interommatidialcells in pupal eyes, which is normally reduced by Hippo pathway-dependent apoptosis (2). One report did acknowledge that ex fatdouble mutants have stronger overgrowth phenotypes in thehead (4), but this phenotype was not well characterized, and itssignificance was discounted.

Fig. 1. Rescue of fat and ex by Wts overexpression. Null mutations in fat or ex are lethal but can be rescued to viability by overexpression of Wts. (A–C) Adultflies of wild type (A), fat8 UAS-Myc:Wts.2/ftG-rv; tub-Gal4 (B), and exe1 UAS-Myc:Wts.2/exe1; tub-Gal4 (C). (D–G) Adult wings from wild type (D), UAS-Myc:Wts.2/tub-Gal4 (E), fat8 UAS-Myc:Wts.2/ftG-rv; tub-Gal4 (F), and exe1 UAS-Myc:Wts.2/exe1; tub-Gal4 (G). The arrows point to extra vein material. (H–J) Portion ofabdomens from wild type (H), fat8 UAS-Myc:Wts.2/ftG-rv; tub-Gal4 (I), and exe1 UAS-Myc:Wts.2/exe1; tub-Gal4 (J). In H and J all hairs and bristles point posteriorly(down); in I bristles and hairs are misoriented and swirling patterns of hairs are visible.

Feng and Irvine PNAS � December 18, 2007 � vol. 104 � no. 51 � 20363

DEV

ELO

PMEN

TAL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

2, 2

021

In one assay, the relative areas of mutant clones were mea-sured. For these experiments we used a null allele of ex, exe1, andtwo different alleles of fat, fat8 and fatGrv. Both of these encodeproteins that are truncated in the extracellular domains (15) andlack detectable expression using antibodies directed against theintracellular domain (data not shown). Clones of cells mutant forfat or ex are larger than control clones in both wing and eye discs(Fig. 2 A and B) (14, 18). ex fat double mutant clones were largerthan single mutant clones (Fig. 2 A and B). This indicates that fatand ex have additive effects on growth, consistent with thehypothesis that they act in parallel pathways. Unexpectedly,fatGrv clones were larger than fat8 clones. The reason for thisdifference is not clear, but, because there was a correspondinglygreater enhancement of growth in both ex fat double mutantcombinations, it does not affect the conclusion that fat and exhave additive effects. As an alternative growth assay, we exam-ined the relative size of the entire wing disk in animals transh-

eterozygous for ex fatGrv and ex fat8 and compared it to therespective single mutants. This confirmed that fat and ex haveadditive effects, because double mutants have larger discs (Fig.2 C–E). Many Drosophila tumor suppressors delay pupariation,and the additive influence of fat and ex mutants on disk growthwas especially pronounced during this extended larval period.The additive effects of ex and fat mutants on growth confirm thatat least some Fat signaling occurs independent of ex.

Although mutation of ex results in overgrowth, the developingeye is actually reduced in size (19). This reduction in the eye fieldis visible in wandering third-instar larvae stained with a pan-neural antibody, anti-ELAV (Fig. 2H). fat mutants appear tohave a modest loss of eye development (Fig. 2G). fat ex doublemutants have a strong additive phenotype, because in most cases(28 of 30 discs) they completely failed to initiate eye develop-ment, as monitored by ELAV expression (Fig. 2I). Thus fat andex mutations are additive not only for growth, but also for otherphenotypes.

Influence of Fat Signaling on Ex Localization. Central to the hypoth-esis that Fat signals through Ex was the observation thatmutation of fat causes a decrease in the levels of Ex at the apicalmembrane. There are, however, discrepancies among priorreports in terms of the strength of this effect, ranging from amodest decrease (5) to virtually complete loss (4, 8). We haveassessed the influence of Fat on Ex by carefully examining Exprotein staining in fat mutant clones throughout the third larvalinstar, examining both fat8 and fatG-rv clones, in both wing andeye imaginal discs, and throughout the apical-to-basal axis ofthese discs. We do see in most instances some decrease in thelevels of Ex membrane staining within fat mutant clones, al-though in no case is Ex staining completely lost (Fig. 3 A and B).Confirmation of the specificity of Ex staining in these experi-ments was provided by examining ex mutant clones, in which lossof Ex is readily detectable (data not shown).

In addition to reducing Ex levels at the membrane, fat clonesare sometimes associated with a shift in distribution of Exstaining to a more basal focal plane, especially at late third instar.In single horizontal sections (SI Fig. 7), this can give theimpression of a substantial loss of Ex staining, which might havecontributed to prior reports of almost complete loss of Exstaining. However, in vertical sections (Fig. 3B) or more basalhorizontal sections (SI Fig. 7) it is clear that Ex staining is shiftedbasally rather than lost. To visualize Ex staining in different focalplanes within a single horizontal image, we used maximumprojection, which reveals a modest decrease in Ex staining (Fig.3A and SI Fig. 7). Although the basal shift could be a conse-quence of a specific relocalization of Ex in response to loss of Fat,it could also derive from altered cell shape. To investigate this,fat mutant clones were stained for E-cadherin, which normallylocalizes near Ex in the subapical membrane. E-cadherin levelsoften appear slightly elevated within fat mutant clones (20).E-cadherin staining was also shifted basally, in what appears tobe a consequence of a change in cell shape rather than a specificeffect on the subcellular localization of Ex or E-cadherin (Fig. 3and SI Fig. 7).

Previously, we described a pathway in which signaling down-stream of Fat is mediated by its antagonism of the unconven-tional myosin Dachs (7, 14, 21). We also identified anotherDrosophila tumor suppressor, discs overgrown (dco) (22), as akinase that acts genetically upstream of dachs within the Fatpathway (7). To investigate how the influence of Fat on Exrelates to this branch of Fat signaling, we analyzed their influ-ence on Ex. Ex staining appears slightly reduced in dco3 clones(SI Fig. 8). The influence of dco on Ex is weaker than that of fat,but it also has weaker effects on the expression of downstreamtarget genes (7). The influence of Fat on growth, cell affinity,and gene expression depends completely on dachs (14, 21). The

Fig. 2. fat and ex have additive effects on growth. (A and B) The average sizes(in arbitrary units) of GFP-expressing clones of the indicated genotypes weremeasured. Error bars show SEM. For wild-type (�), n (number of clonesmeasured) � 166 for wings and 95 for eyes; for exe1 n � 132 for wings and 154for eyes; for ft8 n � 114 for wings and 152 for eyes; for ftG-rv n � 133 for wingsand 133 for eyes; for exe1 ft8 n � 46 for wings and 70 for eyes; for exe1 ftG-rv n �184 for wings and 169 for eyes. The increased size of ex fat double mutantclones compared with single mutant clones is significant in all cases (P � 0.01).(C–E) Representative wing imaginal discs of the indicated genotypes areshown from larvae dissected 7 days after egg laying. (F–I) Eye imaginal discs ofthe indicated genotypes from wandering third-instar larvae stained for ELAV.Arrows point to posterior regions of the eye disk that lack ELAV staining inmutants. All discs are shown at the same magnification.

20364 � www.pnas.org�cgi�doi�10.1073�pnas.0706722105 Feng and Irvine

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

2, 2

021

reduction in Ex levels observed in fat mutant clones is similarlydachs-dependent, because in fat dachs double mutant clones Exstaining was indistinguishable from that in surrounding wild-type cells (Fig. 3 C and D).

Fat Signaling Can Occur Independent of Effects on Ex Levels. Theresults described above confirm prior observations that Fat caninfluence Ex levels at the membrane but do not address thesignificance of decreased Ex staining to the regulation of growthand gene expression. Thus, to further investigate the relationshipbetween Ex levels and Fat signaling, we examined the influenceof fat on Ex overexpression. When the MARCM method (23)was used to overexpress Ex within fat mutant cells, high-level Exstaining was readily detected and appeared at its normal sub-apical location (Fig. 4 A and B). Thus, although Fat can modulateEx membrane localization, it is not required for it.

The observation of elevated Ex staining within these clonesprovided an opportunity to further explore the relationshipbetween fat and ex. Prior studies have indicated that Ex over-expression appears to mimic Ex gain of function, resulting in aWts-dependent induction of apoptosis and repression of down-stream targets of Hippo signaling (6, 7, 17). Consequently,Ex-expressing clones are greatly reduced in size and number(Fig. 4 C and D) (7, 9). Conversely, fat mutant clones overgrowcompared with wild-type clones. Clones of cells mutant for fatand overexpressing Ex exhibited an intermediate phenotype(Fig. 4). These clones are slightly overgrown compared with

wild-type control clones and substantially overgrown comparedwith Ex-expressing clones (Fig. 4 C and D). The observation thatmutation of fat dramatically enhances the growth of Ex-expressing clones, even though Ex staining remains strong at thesubapical membrane, clearly argues against models in which Fatsignals primarily through modulation of Ex levels or localization.By contrast, it is consistent with the hypothesis that fat and ex actin parallel to regulate growth. As a further test, we examined theexpression of downstream target of Wts and Yorkie activity. Wgexpression in the proximal wing is up-regulated by mutation offat or by mutation of any of the tumor suppressors in the Hippopathway (Fig. 4E) (7, 21). Conversely, overexpression of Exrepresses Wg expression in the proximal wing (7). Clones of cellsmutant for fat and overexpressing Ex exhibit an elevation of Wgexpression (Fig. 4F), further demonstrating that Fat signalingcan occur independent of an effect on Ex levels or localization.

DiscussionFat and Ex Act Through Wts. The observation that fat and exmutants are rescued to viability simply by overexpressing Wts(Fig. 1) provides a powerful argument that regulation of Wts istheir most critical function and against the hypothesis that theinfluence of Fat and Ex on growth stems from combinatorialeffects on many pathways, due to influences on endocytosis ormembrane composition (9, 12). Instead, reported effects of ex onother pathways or processes are likely mediated downstream ofWts. Indeed, whereas an increase in Fat protein staining in Mer

A A’ A’’B’’

D’’D’’

B’

C’D’

B

CD

Fig. 3. Influence of Fat signaling on Ex. Shown are wing imaginal discs stained for Ex (red) and E-cadherin (green). (A and C) Projections through horizontalsections. (B and D) Vertical sections. A�–D� and A�–D� show individual channels of the image. Mutant clones (outlined by dashes) were marked by the absenceof GFP (blue). GFP staining does not overlap Ex and E-cadherin and is shown from more basal focal planes in the horizontal images. Endogenous Ex staining islower along the dorsal–ventral boundary (D-V, yellow arrows) and anterior–posterior boundary (A-P, green arrows). (A and B) fat8 mutant clones. Ex stainingis decreased. E-cadherin staining sometimes appears increased (20). (C and D) dGC13 fat8 mutant clones. No consistent difference in Ex staining between wild-typeand mutant tissue was observed.

Feng and Irvine PNAS � December 18, 2007 � vol. 104 � no. 51 � 20365

DEV

ELO

PMEN

TAL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

2, 2

021

ex double mutants was interpreted as supporting a generalinfluence of Mer and Ex on endocytosis (12), we have observeda similar effect on Fat staining in clones of cells mutant for wtsor overexpressing Yorkie (C. Rauskolb and K.I., unpublishedobservations), which implies that the influence of Mer and ex onFat is actually mediated through their influence on transcription.Although the incomplete rescue of ex mutants might be taken asevidence that Ex regulates growth in part independent of Wts,it is also possible that it is simply not possible to generatewild-type levels of active Wts in ex mutants by overexpression.

Parallel Action of Fat and Ex. The nature of the processes thatregulate the Hippo pathway has been a major question in thefield—hence the critical importance of the claim that Fat acts asthe Hippo receptor. This claim was based largely on threeobservations. First, it was reported that mutation of fat leads toa substantial loss of Ex protein from its normal subapicallocation. In fact, however, Ex levels are only partially reduced(Fig. 3 and SI Fig. 7). Moreover, even when Ex levels areartificially elevated, Fat has strong effects, because mutation offat completely reverses the consequences of Ex overexpressionand even leads to some elevation of growth and downstreamgene expression (Fig. 4). This indicates that Fat can regulate Wtsindependent of any effects on Ex levels or localization. Thisresult contradicts that of Tyler and Baker (9), but we confirmedboth the overexpression of Ex (Fig. 4) and the absence of Fat(data not shown) by antibody staining and quantified the effectsof �400 clones in two different discs using two different fatalleles. Others have claimed that mosaic loss of fat had no effecton Ex overexpression phenotypes during late eye development(4, 8), but close examination of their figures suggests that anintermediate phenotype is actually observed, which would beconsistent with our results.

A second argument that Fat acts through Ex to regulate Hipposignaling was provided by the claim that ex fat double mutantsexhibit the same phenotype as fat or ex single mutants (4, 5, 8).Although this may be the case in terms of their influence on thenumber of interommatidial cells, it is not true for other fat andex phenotypes (Fig. 2). Instead, analysis of ex fat double mutants,together with Ex overexpression in fat mutants, confirms that Fatand Ex can act in parallel. We note that our results do not excludethe possibility that Fat can also act through Ex, whether throughits effect on Ex levels or an effect on Ex activity, but they doclearly indicate that at least some Fat signaling occurs indepen-dent of Ex.

A third argument was a reported influence of Fat on Wts andHpo phosphorylation (4, 5). However, this effect has beenobserved only in experiments in which an intracellular fragmentof Fat is expressed at high levels within cultured cells. It couldnot be observed when full-length Fat was expressed, and it hasnot been observed in vivo. Because fat can affect Ex staining indiscs (Fig. 3) (4, 5, 8), it is plausible that high-level expression ofthe Fat intracellular domain might influence Ex in cultured cellsand thereby influence Hippo signaling. However, the relevanceof this mechanism to normal Fat and Hippo signaling in vivoremains to be determined. By contrast, an influence of fat on Wtsprotein levels has been reported in vivo using simple loss-of-function mutations for fat and examining endogenous Wtsprotein (7). Moreover, the observation that genetically fat canact independent of ex clearly supports the conclusion that effectsof Fat that are independent of Ex, such as its influence on Wtslevels, are important.

What then is the functional significance of the influence of Faton Ex? One possibility is that Fat signals both through an effecton Wts levels (independent of Ex) and through an effect on Wtsactivity (via Ex). A dual pathway mechanism like this couldensure a robust response to Fat signaling. Resolving whether Fatnormally signals through Ex will require reagents for monitoringWts activity in vivo, which do not yet exist. This dual pathwayhypothesis also raises the interesting possibility that the respec-tive contributions of these pathways to Fat signaling could varyin different developmental contexts.

An alternative explanation for the effect of Fat on Ex issuggested by the realization that the discernible effect of fat onEx protein staining underestimates the actual effect, because extranscription is up-regulated within fat mutant clones (4, 5, 7).Mutations in components of the Hippo pathway (e.g., wts)elevate both ex mRNA levels and Ex protein staining (6). Theseobservations suggest that a negative influence of Fat signaling on

Fig. 4. Fat signals even when Ex is overexpressed. (A and B) Portions of wingimaginal discs stained for Ex (red) and E-cadherin (green). A shows horizontalsections, and B shows vertical sections. MARCM clones mutant for fat8 andoverexpressing Ex were marked by the presence of GFP (blue). Ex staining wascaptured with intensity settings different from those in Fig. 3 to illustrate thedifference in expression levels between endogenous Ex (barely visible) andectopic Ex (bright red); the localization of ectopic Ex appears normal. (C andD) The average sizes of GFP-expressing clones of the indicated genotypes.Error bars show SEM. For wild-type, n � 166 for wings and 95 for eyes; for ft8

n � 114 for wings and 152 for eyes; for UAS-ex n � 61 for wings and 82 for eyes;for ft8 UAS-ex n � 68 for wings and 102 for eyes. (E and F) Wing imaginal discsstained for WG (magenta) and containing clones (arrows, marked by GFP,green) mutant for fat8 (E) or mutant for fat8 and overexpressing Ex (F).

20366 � www.pnas.org�cgi�doi�10.1073�pnas.0706722105 Feng and Irvine

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

2, 2

021

Ex protein accumulation might act as a homeostatic mechanism,maintaining low amounts of Ex at the subapical membranedespite increases in ex mRNA. This could, for example, facilitatethe continued regulation of Hippo signaling through Ex inde-pendent of Fat.

The Fat–Hippo Signaling Network. Currently we can describe atleast three signal transduction processes downstream of Fat: aninfluence on Wts levels, an influence on Ex levels, and aninfluence on PCP (SI Fig. 9). The influence of Fat on PCP isseparable from its influences on transcription, because overex-pression of Wts only partially rescues fat PCP phenotypes (Fig.1) and mutation of dachs only partially suppresses fat PCPphenotypes (14). However, the persistence of a some PCPphenotype in fat dachs double mutants, together with thepolarized distribution of Dachs protein (14), raises the possibilitythat Dachs, along with Atrophin (16), might have an input intoPCP (SI Fig. 9).

Although we have found that the influence of Fat on both Wtsand Ex requires Dachs, these processes are distinct because wtsmutation does not decrease Ex levels (4–6) and ex does notinfluence Wts levels (7). Instead, our results imply that twoparallel pathways converge on a common target, Wts, butregulate it in distinct ways, with Hippo signaling influencing Wtsactivity (1, 2) and Fat signaling influencing Wts levels (7) (SI Fig.9). Although our results show that these pathways can act inparallel, it is clear that they intersect at multiple points, formingwhat might be better described as a Fat–Hippo signaling network(SI Fig. 9). Points of intersection include regulation of Wts,regulation of Ex, and transcriptional feedback regulation ofnetwork components including four-jointed and ex. A fullerunderstanding of the relationship between these pathways awaitsthe identification of additional regulatory inputs into Ex and Meractivity and of reagents to monitor Wts and Yorkie activityin vivo.

Materials and MethodsDrosophila Stocks and Crosses. For examination of Myc:Wts expression, yw;UAS-Myc:Wts.2/CyO and yw; UAS-Myc:Wts.1/TM6b were crossed to ptc-Gal4UAS-GFP.

Negatively marked clones were generated by crossing yw; ft8 FRT40A/CyO-GFP, yw; ftGrv FRT40A/CyO-GFP, or yw; dGC13 ft8 FRT40A/CyO-GFP to yw hs-FLP;Ubi-GFP FRT40A/CyO or yw hs-FLP;M(2) Ubi-GFP FRT40A/CyO, and yw; dco3

FRT82B/TM6b to yw hs-FLP; Ubi-GFP FRT82B/CyO.Positively marked (MARCM) clones were generated by crossing w; exe1

FRT40A/CyO-GFP, yw; exe1 ft8 FRT40A/CyO-GFP, yw; exe1 ftGrv FRT40A/CyO-GFP, yw; ft8 FRT40A/CyO-GFP, yw; ftGrv FRT40A/CyO-GFP, yw; UAS-ex/TM6b, oryw; ft8 FRT40A/CyO-GFP; UAS-ex/TM6b to yw hs-FLP tub-Gal4 UAS-GFP;FRT40A tub-Gal80.

For Wts rescue experiments, we crossed fat8 UAS-Myc:Wts.2/CyO to ftG-rv

tub-Gal4[LL7]/TM6b, UAS-Myc:Wts.2 to tub-Gal4[LL7]/TM6b, and exe1 UAS-Myc:Wts.2/CyO to exe1/CyO; tub-Gal4[LL7]/TM6b.

Histology and Imaging. Discs were fixed and stained as described previously(21) using mouse anti-Wg (1:800, 4D4; Developmental Studies HybridomaBank), guinea pig anti-Ex (1:5,000, R. Fehon, University of Chicago, Chicago),rat anti-E-cadherin (1:40, DCAD2; Developmental Studies Hybridoma Bank),mouse anti-Myc (9E10, 1:800; Babco), rat anti-ELAV (Developmental StudiesHybridoma Bank), and rat anti-Fat. Fluorescent stains were captured on a LeicaTCS SP5. For horizontal sections, maximum projection using Leica softwarewas used to allow visualization of staining in different focal planes. Thismethod takes the brightest pixel at any given xy position in each of a series ofz sections being projected.

For adult tissues, combineZM software was used to allow visualization offeatures in different focal planes within a single image.

Size Measurements. To measure clone sizes, embryos were collected fromcorresponding crosses for 12 h. Forty-eight hours later, larvae were heat-shocked at 36°C for 10 min, and an additional 72 h later animals were dissectedand fixed. Clone sizes were measured by tracing in NIH Image J. Samples fromdifferent genotypes were collected and analyzed in parallel.

To compare disk sizes, embryos were collected from corresponding stocksor crosses for 8 h and animals were dissected and fixed at appropriate stagesas mentioned in the text.

ACKNOWLEDGMENTS. We thank R. Fehon, J. Jiang, H. McNeill, Z. C. Lai, M.Mlodzik, K. Harvey, G. Halder, the Developmental Studies Hybridoma Bank,and the Bloomington Stock Center for antibodies and Drosophila stocks andC. Rauskolb for comments on the manuscript. This research was supported bythe Howard Hughes Medical Institute and National Institutes of Health GrantGM078620.

1. Pan D (2007) Genes Dev 21:886–897.2. Edgar BA (2006) Cell 124:267–273.3. Huang J, Wu S, Barrera J, Matthews K, Pan D (2005) Cell 122:421–434.4. Willecke M, Hamaratoglu F, Kango-Singh M, Udan R, Chen CL, Tao C, Zhang X, Halder

G (2006) Curr Biol 16:2090–2100.5. Silva E, Tsatskis Y, Gardano L, Tapon N, McNeill H (2006) Curr Biol 16:2081–2089.6. Hamaratoglu F, Willecke M, Kango-Singh M, Nolo R, Hyun E, Tao C, Jafar-Nejad H,

Halder G (2006) Nat Cell Biol 8:27–36.7. Cho E, Feng Y, Rauskolb C, Maitra S, Fehon R, Irvine KD (2006) Nat Genet 38:1142–

1150.8. Bennett FC, Harvey KF (2006) Curr Biol 16:2101–2110.9. Tyler DM, Baker NE (2007) Dev Biol 305:187–201.

10. McClatchey AI, Giovannini M (2005) Genes Dev 19:2265–2277.11. McCartney BM, Kulikauskas RM, LaJeunesse DR, Fehon RG (2000) Development

127:1315–1324.

12. Maitra S, Kulikauskas RM, Gavilan H, Fehon RG (2006) Curr Biol 16:702–709.13. Lawrence PA, Casal J, Struhl G (2004) Development 131:4651–4664.14. Mao Y, Rauskolb C, Cho E, Hu WL, Hayter H, Minihan G, Katz FN, Irvine KD (2006)

Development 133:2539–2551.15. Matakatsu H, Blair SS (2006) Development 133:2315–2324.16. Fanto M, Clayton L, Meredith J, Hardiman K, Charroux B, Kerridge S, McNeill H (2003)

Development 130:763–774.17. Blaumueller CM, Mlodzik M (2000) Mech Dev 92:251–262.18. Boedigheimer MJ, Nguyen KP, Bryant PJ (1997) Dev Genet 20:103–110.19. Pellock BJ, Buff E, White K, Hariharan IK (2007) Dev Biol 304:102–115.20. Jaiswal M, Agrawal N, Sinha P (2006) Development 133:925–935.21. Cho E, Irvine KD (2004) Development 131:4489–4500.22. Zilian O, Frei E, Burke R, Brentrup D, Gutjahr T, Bryant PJ, Noll M (1999) Development

126:5409–5420.23. Lee T, Luo L (2001) Trends Neurosci 24:251–254.

Feng and Irvine PNAS � December 18, 2007 � vol. 104 � no. 51 � 20367

DEV

ELO

PMEN

TAL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

2, 2

021