lgl Regulates the Hippo Pathway Independently of Fat/Dachs, Kibra/Expanded/Merlin and dRASSF/dSTRIPAK

In both Drosophila and mammalian systems, the Hippo (Hpo) signalling pathway controls tissue growth by inhibiting cell proliferation and promoting apoptosis. The core pathway consists of a protein kinase Hpo (MST1/2 in mammals) that is regulated by a number of upstream inputs including Drosophila Ras Association Factor, dRASSF. We have previously shown in the developing Drosophila eye epithelium that loss of the apico-basal cell polarity regulator lethal-(2)-giant-larvae (lgl), and the concomitant increase in aPKC activity, results in ectopic proliferation and suppression of developmental cell death by blocking Hpo pathway signalling. Here, we further explore how Lgl/aPKC interacts with the Hpo pathway. Deregulation of the Hpo pathway by Lgl depletion is associated with the mislocalization of Hpo and dRASSF. We demonstrate that Lgl/aPKC regulate the Hpo pathway independently of upstream inputs from Fat/Dachs and the Kibra/Expanded/Merlin complex. We show depletion of Lgl also results in accumulation and mislocalization of components of the dSTRIPAK complex, a major phosphatase complex that directly binds to dRASSF and represses Hpo activity. However, depleting dSTRIPAK components, or removal of dRASSF did not rescue the lgl−/− or aPKC overexpression phenotypes. Thus, Lgl/aPKC regulate Hpo activity by a novel mechanism, independently of dRASSF and dSTRIPAK. Surprisingly, removal of dRASSF in tissue with increased aPKC activity results in mild tissue overgrowth, indicating that in this context dRASSF acts as a tumor suppressor. This effect was independent of the Hpo and Ras Mitogen Activated Protein Kinase (MAPK) pathways, suggesting that dRASSF regulates a novel pathway to control tissue growth.


Introduction
The Hpo pathway is conserved from flies to humans and plays a central role in development and disease. Hpo signalling controls organ size by coordinately promoting apoptosis and blocking cell proliferation (reviewed [1]). The core of the Hpo pathway is composed of two kinases: Hpo (orthologue of mammalian sterile 20-like kinase-1 and -2, MST1 and MST2) and Warts (Wts) (orthologue of human large tumor suppressor, LATS1 and LATS2). Hpo and Wts are found in a complex with the scaffolding proteins Salvador (Sav, an orthologue of human WW45) and Mob as tumor suppressor (MATS). Through various upstream pathways, such as the atypical cadherin Fat and/or the FERM domain-containing proteins Expanded (Ex) and Merlin (Mer) in association with Kibra, Hpo senses membrane associated signals and activates Wts through phosphorylation. Upon activation, Wts phosphorylates and inhibits Yorkie (Yki, an orthologue of human Yes-Associated Protein, YAP), a transcriptional coactivator that binds to transcription factors such as Scalloped (orthologue of human Tea Domain proteins, TEAD1-4) and activates expression of cell survival (e.g., diap1, Drosophila inhibitor of apoptosis-1) and cell proliferation (e.g., cyclin E) genes, as well as feedback pathway targets, expanded (ex) and four-jointed (fj).
The mammalian Ras Association domain family (RASSF) contains ten different members that fall into two groups, classical (RASSF1-6) and N-terminal (RASSF7-10) proteins (reviewed [2,3]). The defining feature of all RASSF proteins is a Ras association domain (RA) that can be found C-terminally (RASSF1-6) or N-terminally (RASSF7-10). RASSF proteins 1-6 also encode a SARAH protein-protein interaction domain (also present in Salvador, RASSF, Hippo). In Drosophila, there is a single, classical RASSF gene, dRASSF. Similar to its mammalian homologues (RASSF1-6) dRASSF encodes a C-terminal Ras Association and SARAH domain. In dRASSF, the SARAH domain mediates dimerization with Hpo and regulates tissue growth via the Hpo tumor suppressor pathway [4]. dRASSF competes with Sav for Hpo binding via the SARAH domain. In doing so, dRASSF inactivates the Hpo kinase cascade, thereby increasing Yki activity and promoting tissue growth [4]. Recent studies in Drosophila have identified a phosphatase complex, dSTRIPAK (Drosophila Striatin-interacting phophatase and kinase) that directly binds dRASSF, dephosphorylates Hpo and restricts the activity of the pathway to promote tissue growth [5]. Thus, Hpo activity is regulated by a variety of inputs from the cell membrane, and more directly via association with dRASSF and the dSTRIPAK complex.
It has long been recognized that loss-of-function mutations in the Drosophila apico-basal cell polarity gene lethal-giant-larvae (lgl) result in tissue overgrowth and neoplastic tumors [6]. We have shown in the developing Drosophila eye that loss of lgl and the concomitant increase in aPKC activity results in ectopic cell proliferation and suppression of developmental cell death (apoptosis) via impairment of the Hpo pathway, as assessed by decreased phospho-Yki and increased expression of target genes (cycE Diap1, ex and fj). Furthermore, when Yki levels were reduced in lgl −/− tissue ectopic expression of Hpo pathway targets were normalized and overgrowth was decreased, showing that Yki activation was rate-limiting for the tissue overgrowth phenotype conferred by Lgl depletion [7]. We also observed that Lgl/aPKC activity regulates the localization of Hpo/dRASSF [7]. Here we further dissect the relationship between Lgl/aPKC and Hpo pathway regulation. We show that Lgl/aPKC regulate the Hpo pathway independently of upstream components Fat and Kibra/Ex/Mer, the dSTRIPAK complex and dRASSF. Furthermore, dRASSF in the context of increased aPKC activity appears to have a tumor suppressor function.

Fly Culture, Overexpression, and Clonal Analysis
Mitotic eye clones and mosaic analysis with a repressible marker (MARCM) clones were generated as previously described [8]. Clonal and MARCM crosses were raised at 25 °C. Overexpression crosses with GMR > aPKC CAAXWT were undertaken at 18 °C.

Immunohistochemistry, Imaging and Antibodies
Larval and pupal discs were dissected and fixed as previously described [7]. For Hpo, dRASSF, Cka and Mob4 staining, tissues were fixed in paraformaldehyde lysine periodate (PLP). Labeled samples were cleared through 80% glycerol and mounted. For Figure 1, Figure 2E, Figure 3A,C, Appendix Figure A1B,C, and Figure 4A,B images were collected on Biorad MRC1000 (Bio-Rad Laboratories, Hercules, CA, USA). Figure 2B, Figure 3B and Appendix Figure A1A were collected on an Olympus FV1000 (Olympus, Center Valley, PA, USA). Images in Figure 2C,D and Figure 4G-I were taken on a Nikon Eclipse 90i (Nikon, New York, NY, USA). Images were processed using Fiji, and assembled with Adobe Photoshop CS6 and Adobe Illustrator CS6. Adult eyes were imaged with a Scitec Infinity1 camera (Lumenera, Ottawa, Canada).

Quantification of Interommatidial Cell (IOC) Number
The number of IOCs (secondary and tertiary pigment cells) for a single ommatidium was quantified by drawing a hexagon connecting the centers of the six surrounding neighboring ommatidia. IOCs within the hexagon were then counted. IOCs that straddled the boundary were counted as half a cell.

Quantification of Adult Eye Size
The perimeter of adult eyes was outlined in Photoshop CS3 and the total area in pixels calculated. Data was analysed using unpaired t-test (Welch corrected).

Statistical Analysis of Signal Intensity
To determine the relative diap1-GFP1.8 expression in eye discs, images of each disc were taken with identical confocal settings. Using Image J, a ratio of average pixel intensity was determined for diap1-GFP1.8 immunofluorescence in an area posterior and anterior to the morphogenetic furrow that colocalized with DAPI staining. Data was analyzed using unpaired t-test (Welch corrected).

lgl Regulates the Hpo Pathway Independently of Fat and Dachs
To determine how Lgl regulates the Hpo pathway we asked if loss of lgl function affected the phenotypes of known upstream regulators of the Hpo pathway. First, we assessed if Lgl genetically interacted with the atypical cadherin Fat [9][10][11]. A defining characteristic of Hpo pathway mutants is an increase in cell number in the pupal retina [12]. The Drosophila eye develops from the eye imaginal disc, a single-cell-layered epithelial sheet that grows during larval stages. After photoreceptor and cone cells are determined, interommatidial cells (IOC's) are specified to become pigment cells, with excess IOC's are eliminated by apoptosis. Interommatidial (or pigment cells) surround the centrally located photoreceptor and cone cells, generating a precise, repeating hexagonal structure (ommatidium), which can be visualized during the pupal stage ( Figure 1A; white arrowheads mark photoreceptor, cone and primary pigment cells, yellow arrowheads indicate secondary and tertiary pigment cells). In mutants such as hpo −/− , where developmental cell death is inhibited, IOC's accumulate between the ommatidia of the pupal retina ( Figure 1C; yellow arrowhead). lgl −/− tissue also fails to undergo programmed cell death and a few extra cells can be observed in lgl −/− retinae ( Figure 1E; yellow arrowhead), however this phenotype is not as strong as that displayed by hpo −/− tissue (compare Figure 1C,E). In contrast to hpo −/− retinae, fat −/− tissue showed only a few extra cells ( Figure 1B yellow arrowhead). Notably, lgl, fat double mutant tissue showed a greater number of IOC's than either lgl or fat single mutant retinae (compare Figure 1D with Figure 1B,E). Quantification of the number of IOC's per ommatidia in fat or lgl single mutants revealed an average of two or eight additional cells respectively. Strikingly the average number of IOC's per ommatidia in lgl, fat double mutant tissue was thritythree (quantified in Figure 1H). This data shows that lgl and fat act synergistically to regulate IOC number during eye development, and therefore might function redundantly within the same pathway or in parallel pathways.
We then wished to determine if Lgl and Fat acted together within the same complex or independently of each other to regulate Hpo signalling. Therefore we analyzed if Wts protein levels were reduced in lgl −/− tissue. Fat regulates the Hpo pathway, in part by regulating the stability of Wts kinase. Wts protein levels are reduced in fat −/− tissue and this decrease in Wts is dependent on the activity of an unconventional myosin, Dachs [13,14]. In fat, dachs double mutant tissue, the loss of dachs restores Wts protein levels, thereby increasing the levels of phospho-Yki and downregulating target gene expression [13].
We previously used the ey-FLP-actin-FLP-out system combined with a lgl-RNAi transgene to specifically knockdown Lgl and showed that when Lgl is depleted in the eye-antennal imaginal disc phospho-Yki levels decrease and Yki target gene expression is increased [7]. When Lgl was knocked down by 80% using ey-FLP-actin-FLP-out (Figure 2A), Western blot analysis showed that there was no statistically significant change (unpaired t-test Welch's correction, p > 0.05, n = 4) in Wts protein levels between control eye-antennal tissue and lgl-RNAi depleted tissue ( Figure 2A). This data shows that lgl does not mediate Hpo pathway activation by decreasing Wts protein levels.
We have shown that lgl −/− tissue upregulated the diap1 transcriptional reporter diap1-lacZ [7]. To test the requirement of dachs in this phenotype, we investigated whether removing dachs would reduce the elevated level of diap1-lacZ expression in lgl −/− tissue to wildtype. dachs GC13/GC13 clones grow poorly when surrounded by wildtype tissue [14], therefore we generated lgl, dachs GC13/GC13 clones in a dachs 1 background. This enabled us to recover lgl, dachs GC13/GC13 clones in larval eye discs, however lgl, dachs GC13/GC13 clones were not observed in adult eyes (data not shown). We found that in lgl, dachs GC13/GC13 tissue upregulation of diap1-lacZ persisted (arrowheads Figure 2B-B'''). Thus, activation of Yki target genes upon depletion of Lgl is independent of dachs function.
Taken together, the synergistic effects on increased cell number observed in lgl, fat double mutant pupal retinas, maintenance of elevated diap1-lacZ gene expression in lgl, dachs double mutant tissue, and the lack of change in Wts protein levels in lgl-RNAi depleted eye-antennal discs, strongly supports the notion that lgl acts separately and in parallel with Fat/Dachs to regulate the Hpo pathway.

lgl Regulates the Hpo Pathway Independently of the Apical Kibra/Expanded/Merlin Complex
The discovery of the apical FERM domain containing complex Kibra/Expanded/Merlin (KEM) as a regulator of Hpo signalling [15][16][17], and that in mammalian systems, Kibra has been shown to be a direct target of aPKC phosphorylation [18], raised the possibility that lgl may regulate Hpo pathway activity via the KEM complex.
Initially we investigated tissue growth phenotypes of lgl −/− ; ex-RNAi or lgl −/− ; mer-RNAi depleted larval eye discs. lgl −/− ; ex-RNAi or lgl −/− ; mer-RNAi depleted larval eye discs showed excess tissue growth, greater than single mutants alone, and these animals mostly failed to pupate (data not shown). In one lgl −/− , mer-RNAi depleted pupal disc that we were able to recover, we observed a significant increase in IOC's, (compare Figure 1E-G), similar to that observed for lgl, fat double mutants (compare Figure 1D,G). Consistent with the breakdown of tissue observed in the lgl −/− ;mer-RNAi pupal retina ( Figure 1G), rare adult male escapers double mutant for mer and lgl showed a dramatic loss of differentiated adult eye tissue (white arrowhead) and an increase in head capsule tissue (yellow arrowhead, compare Figure 1I-L). The increase in IOC's in lgl −/− ; mer-RNAi retina and increased tissue growth of lgl −/− ; ex-RNAi larval eye discs shows that Lgl and the KEM complex act synergistically which might indicate that they act redundantly in the same pathway or in parallel pathways to regulate the Hpo pathway.
In our analysis of the effect of Lgl depletion/aPKC activation on the localization of Hpo pathway components, we observed co-mislocalization of Hpo and dRASSF, and proposed that Lgl/aPKC regulate the Hpo pathway through Hpo/dRASSF apical localization [7]. To determine if components of the KEM complex and Lgl regulate the Hpo pathway via redundant or independent mechanisms we investigated whether the localization of Hpo and dRASSF was altered by depletion of the KEM complex. As members of the KEM complex are functionally redundant with one another we chose to undertake these studies in a stronger double mutant background of kibra −/− , mer-RNAi (the efficacy of the mer-RNAi is demonstrated in Figure 1G and discussed above). We have previously shown that Hpo and dRASSF proteins accumulate apically and are mislocalized basolaterally in lgl −/− tissue [7], (demonstrated here again in Figure 3A,C, showing apical accumulation of Hpo in lgl −/− tissue). In contrast, in kibra −/− , mer-RNAi tissue, Hpo and dRASSF showed normal apical levels ( Figure 2C,D, RFP positive tissue, arrowheads). Furthermore, a sideview of a kibra −/− , mer-RNAi eye disc ( Figure 2D) showed no basolateral mislocalization of Hpo. Thus, kibra and merlin activity do not regulate the localization of Hpo and dRASSF.
To further investigate the genetic interaction between lgl and kibra, we generated adult eyes comprised entirely of kibra mutant tissue and asked what happened to tissue growth if we generated lgl mutant tissue, or overexpressed a constitutively membrane-bound (active) version of aPKC (aPKC CAAXWT hereafter referred to as aPKC CA ) in this genetic background. lgl −/− ; kibra −/− adult eyes have increased tissue growth compared to kibra −/− or lgl −/− alone (compare Figure 2H,J,K). This is particularly noticeable in head capsule tissue that shows excessive tissue growth ( Figure 2K, arrowheads). As previously reported [17], adult eyes comprised of entirely kibra −/− tissue showed a weak increase in eye size (3%) when compared with control eyes (compare Figure 2F,H, quantified in Figure 2L). GMR > aPKC CA adult eyes showed a 12% increase in eye size compared with control eyes (compare Figure 2F,G, quantified in Figure 2L). Consistent with the phenotype of adult eyes double mutant for lgl and kibra ( Figure 2K), adult eyes mutant for kibra and overexpressing aPKC CA (GMR > aPKC CA ) have a synergistic increase in eye size (38% compared to control eyes), greater than kibra −/− or GMR > aPKC CA alone ( Figure 2I, quantified in Figure 2L). Taken together, the normal levels and localization of Hpo and dRASSF in kibra −/− , mer-RNAi tissue, and genetic interaction data showing that kibra −/− and mer-RNAi act synergistically with lgl −/− or aPKC CA overexpression to regulate tissue growth, supports the notion that Lgl/aPKC regulate the Hpo pathway independently of the apical KEM complex.

Members of the dSTRIPAK Complex Are Mislocalized in lgl Mutant Tissue
The discovery of a PP2A phosphatase complex, dSTRIPAK, which exclusively interacts with dRASSF/Hpo and not Hpo/Sav [5], raised the possibility that Lgl/aPKC regulates the Hpo pathway by modulating dSTRIPAK activity. In human cells, PP2A complex formation is achieved through the combinatorial assembly of two PP2A catalytic subunits, a regulatory B subunit and two scaffolding subunits (reviewed [19]). The dSTRIPAK complex associated with Hpo also contains two catalytic subunits, a regulatory subunit Cka (Connector of kinase to AP-1) and a kinase Mob4 (Mps one binder kinase activator-like 4). We used immunohistochemistry to determine, if similar to Hpo and dRASSF, Cka and Mob4 are mislocalized in lgl −/− tissue. In the developing Drosophila larval eye disc, Hpo, Cka and Mob4 localized apically and associated with cell membranes to outline the developing ommatidial clusters ( Figure 3A,C). Like Hpo, in lgl −/− tissue, Cka accumulated apically compared with the surrounding wildtype tissue (Figure 3A arrowheads). Cross section of the eye disc revealed that in lgl −/− clones posterior to the morphogenetic furrow there was apical accumulation and basolateral mislocalization of Cka ( Figure 3B, lgl −/− tissue denoted by arrowheads). The change in Cka localization in lgl −/− tissue is independent of changes in apico-basal cell polarity, because we have previously shown that cell polarity is maintained in lgl −/− larval eye disc clones [7]. Mob4 and Hpo staining partially overlapped in lgl −/− tissue. Although there were some regions of the eye disc that showed some accumulation of Mob4 ( Figure 3C, arrowhead) Mob4 was less affected than Hpo ( Figure 3C arrowhead). Taken together, these results show that Lgl is required for the correct localization of dRASSF and Hpo, and members of the phosphatase dSTRIPAK complex, Cka and Mob4.
We predicted that mislocalization of negative regulators of the Hpo pathway such as dRASSF and members of the dSTRIPAK complex, lead to the observed decrease in Hpo pathway activity and the subsequent upregulation of Yki pathway targets in lgl −/− tissue. Therefore, reduction of negative regulators of the Hpo pathway, such as dSTRIPAK [5], might restore Hpo signalling and reduce tissue growth. We used the GMR > aPKC CA phenotype, which shows mislocalization of Hpo proteins, upregulation of Hpo pathway targets and a weak adult eye overgrowth phenotype ( Figure 2G, quantified 2L), to determine if depletion of dSTRIPAK complex members could suppress aPKC mediated Hpo pathway activity. To achieve this, we depleted each dSTRIPAK member (CCM3 (CG5073), SLMAP (CG17494), CTTNBP2 (CG10915), FAM40A (CG11526), FGOP2 (CG10158), cka and mob4) using RNAi in the developing Drosophila eye via the GMR driver. Knockdown of dSTRIPAK members alone via the GMR driver produced no observable effects. However, as predicted [5], all RNAi's enhanced the GMR-GAL4, UAS-hpo phenotype at 18 °C and were late pupal lethal displaying severe eye and head defects (data not shown). Moreover, examination of GMR-GAL4 induced knockdown of Cka at 18 °C, or in clones at 18 °C, revealed a significant depletion of apically-localized Cka protein (compare Cka staining in control disc to cka-RNAi depleted disc Supplementary Figure 1A,B, arrowheads and in clones 1C). However, when tested for their interaction with GMR > aPKC CA , none of the dSTRIPAK complex component depletions suppressed the GMR > aPKC CA phenotype ( Figure 3D-F; and data not shown). Conversely, mob4 depletion in conjunction with GMR > aPKC CA , resulted in a slight increase in adult eye size and an accumulation of glassy tissue in the posterior region of the adult eye ( Figure 3G) relative to GMR > aPKC CA alone. These data show that only one dSTRIPAK complex member, Mob4, showed a genetic interaction with Lgl/aPKC. However, this interaction is opposite to what would be expected if dSTRIPAK is responsible for the decreased Hpo activity in aPKC CA over-expressing tissue. Thus, depletion of the dSTRIPAK complex components does not reduce tissue growth in GMR > aPKC CA adult eye, and therefore the dSTRIPAK complex is not a critical factor in the impairment of Hpo activity by activated aPKC.

lgl Mutant Tissue Mislocalizes and Inactivates Hpo Pathway Signalling Independently of dRASSF Function
Previous studies have shown that Hpo and dRASSF form a protein complex that antagonizes Hpo pathway activity [3]. We wished to determine if dRASSF was required to mislocalize Hpo proteins in lgl −/− tissue. To achieve this, we generated dRASSF homozygous mutant animals that also contained mosaic lgl −/− tissue in the developing eye and examined Hpo localization. Hpo proteins still accumulated apically in lgl −/− ; dRASSF −/− tissue ( Figure 4A arrowheads) and in cross sections of the developing eye disc, Hpo was clearly mislocalized basolaterally (arrowheads Figure 4B). The basolateral changes were not associated with cell polarity changes in lgl −/− tissue because aPKC localization remains unchanged ( Figure 4B). These data show that although Hpo and dRASSF directly bind to each other, this interaction is not required for the localization of Hpo in either dRASSF single mutant, or lgl; dRASSF double mutant tissue.

dRASSF Acts as a Tumor Suppressor in lgl Mutant and GMR>aPKC CA Tissue
As stated previously, we predicted that mislocalization of negative regulators of the Hpo pathway, such as dRASSF, led to the observed upregulation of Hpo pathway targets in lgl −/− tissue. Therefore, reduction in dRASSF activity might restore Hpo signalling and reduce tissue growth in lgl −/− or GMR>aPKC CA tissue. To achieve this, we investigated the genetic interaction between lgl −/− , GMR>aPKC CA and dRASSF. Adult eyes from homozygous dRASSF single mutants develop and differentiate normally, but showed slightly reduced adult eye size consistent with an overall decrease in body size [4], ( Figure 4C,C'). When dRASSF activity was removed in lgl −/− mosaic or GMR > aPKC CA animals the resulting adult eyes were larger and showed tissue folds, consistent with increased tissue growth ( Figure 4D,D',E,E' respectively). These data were opposite to that expected if dRASSF was acting to restrict Hpo function in lgl −/− or GMR > aPKC CA eyes, and instead demonstrate that dRASSF is required to restrict growth when aPKC activity is increased. These data suggest that dRASSF may act a tumor suppressor in this context.
We predicted that loss of dRASSF would restore Hpo pathway signalling and decrease Yki target gene expression in GMR > aPKC CA tissue. Therefore, we determined if Hpo pathway activation was suppressed and Yki target gene expression decreased by determining the levels of diap1-GFP reporter gene expression, (diap1-GFP1.8, which contains enhancer regions critical for the activation of diap1 expression by Yki [20]). GMR > aPKC CA and GMR > aPKC CA ; dRASSF −/− eye discs show upregulation of diap1-GFP1.8 reporter gene expression (compare Figure 4G-J) when compared to control dRASSF single mutant eye discs (quantified in Figure 4J) but there was no statistically significant change of diap1-GFP1. 8 reporter gene expression between GMR > aPKC CA and GMR > aPKC CA ; dRASSF −/− eye discs ( Figure  4H,I, quantified in 4J) indicating that Yki target gene expression remains elevated in these eye discs. This result indicates that Yki target gene expression remains elevated when dRASSF function is depleted in GMR > aPKC CA eye discs. This supports the observation that tissue growth was not reduced in GMR > aPKC CA ; dRASSF −/− adult eyes and demonstrates that increased aPKC activity can lead to the activation of Yki target gene expression independently of dRASSF.
In mammalian systems classical RASSF proteins, such as RASSF5A/Nore1A, are known effectors of Ras signalling [21][22][23]. In Drosophila, loss of dRASSF is able to rescue growth defects in Ras mutant tissue [4], suggesting that dRASSF may be a negative regulator of Ras signalling. Activated Ras leads to increased phosphorylation of MAPK (ERK) therefore, we isolated protein extracts from larval eye discs mutant for dRASSF alone, expressing GMR > aPKC CA , or GMR > aPKC CA ; dRASSF −/− tissue and determined the ratio of pERK versus total ERK levels by western blot analysis. As shown in Figure 4F, and quantified in Figure 4F', although a two fold increase in the relative ratio of pERK/ERK was detected in GMR > aPKC CA discs compared to control wildtype eye discs, the removal of dRASSF had no effect on the ratio of pERK/ERK. This result demonstrates that MAPK signalling is not changed in dRASSF −/− or GMR > aPKC CA ; dRASSF −/− tissue.

Discussion and Conclusions
Here we demonstrate that the apico-basal cell polarity regulators Lgl/aPKC regulate the Hpo pathway independently of upstream inputs Fat/Dachs and the Kibra/Ex/Mer complex. We also show that in addition to Hpo and dRASSF, members of the dSTRIPAK complex Cka and Mob4 are mislocalized in lgl −/− tissue. We predicted that comislocalization of Hpo and dRASSF observed in lgl −/− tissue, or eye discs overexpressing aPKC CA , was critical for deregulation of the Hpo pathway and upregulation of Yki target genes such as diap1 and cycE. However, we found that dRASSF was not required for Hpo mislocalization. Surprisingly, removal of dRASSF in lgl −/− tissue, or eye discs overexpressing aPKC CA , did not suppress tissue overgrowth phenotypes, or reduce diap1 levels as predicted, but resulted in increased tissue growth. Moreover, knockdown of dSTRIPAK components did not rescue the eye overgrowth defects due to overexpression of aPKC CA . These data suggest that dRASSF or dSTRIPAK are not required for inhibition of Hpo by aPKC upregulation, and therefore Lgl/aPKC regulates the Hpo pathway by a novel mechanism ( Figure 5).
Previous studies in Drosophila, have proposed that dRASSF antagonizes Hpo pathway activity by directly competing with Sav for binding to the SARAH domain within Hpo [4]. In addition, dRASSF forms a multimeric protein complex with Hpo and a phosphatase complex (dSTRIPAK) that dephoporylates Hpo, thereby reducing Hpo activity, and promoting Yki target gene expression [5]. We expected that the comislocalization of Hpo and dRASSF, in lgl −/− tissue, could account for the decreased Hpo activity and increased Yki target gene expression observed [7]. However, in this study, we show that in the developing eye disc, dRASSF is not required to mislocalize Hpo in lgl mutant tissue. Furthermore, removal of dRASSF in GMR > aPKC CA eye discs failed to reduce elevated levels of diap1 gene expression. These data indicate that the removal of dRASSF in lgl −/− or GMR > aPKC CA tissue does not simply lead to the reestablishment of Hpo/Sav protein complexes and subsequent inhibition of Yki target gene expression.
In this study we also demonstrated that the regulatory subunits of the dSTRIPAK complex, Cka and Mob4 colocalized with Hpo, and similar to Hpo, accumulated apically and Cka was mislocalized basolaterally in lgl −/− tissue. Since dSTRIPAK is a negative regulator of Hpo, its accumulation with Hpo in lgl −/− tissue might account for the decrease in Hpo activity. However, our data showing that depletion of members of the dSTRIPAK complex did not suppress the overgrown eye phenotype of GMR > aPKC CA , argues against this. This suggests that Lgl/aPKC regulate Hpo activity by a novel mechanism, although it remains formally possible that dRASSF and dSTRIPAK act redundantly to inactivate Hpo, and that removal of both will be necessary to restore Hpo activity upon lgl depletion/aPKC CA overexpression. Thus, the effect of Lgl/aPKC on the biochemical and functional interactions between dRASSF/dSTRIPAK/Hpo and Hpo/Sav protein complexes warrants further investigation.
We observed a mild increase in tissue growth in lgl; dRASSF −/− and GMR > aPKC CA ; dRASSF −/− adult eyes suggesting that dRASSF acts as a tumor suppressor when Lgl/aPKC activity is altered. Our data, showing that diap1-GFP1.8 levels are not further elevated in GMR > aPKC CA ; dRASSF −/− compared to GMR > aPKC CA eye discs, suggests that increased tissue growth is not due to further inactivation of the Hpo signalling pathway and upregulation of Yki target genes. This raises the possibility that dRASSF regulates additional growth signalling pathways. Consistent with previous data in mammalian and zebrafish systems, we observed an increase in MAPK signalling when the activity of aPKC cell polarity complexes was altered [24][25][26]. Although dRASSF has been previously linked to negative regulation of the Ras signalling pathway, we did not observe any change in pERK levels in dRASSF or GMR > aPKC CA ; dRASSF −/− tissue relative to GMR > aPKC CA , demonstrating that MAPK signalling is not altered by removal of dRASSF. Taken together our data clearly demonstrate that the molecular mechanisms utilized by dRASSF to regulate tissue growth, and the cellular contexts in which they are employed remain to be elucidated. The further exploration of the role of dRASSF in Drosophila development, and cancer models, should provide exciting insights into the complexities of cancer biology. . The Hpo pathway controls tissue growth by regulating the subcellular localization of Yki, which when phosphorylated is cytoplasmic and unphosphorylated is nuclear. Yki associates with cognate transcription factors to promote cell proliferation and inhibit apoptosis. Lgl/aPKC regulate Hpo signalling independently of Fat/Dachs, Kibra/Expanded/Merlin and dRASSF/dSTRIPAK complexes.