dMyc Functions Downstream of Yorkie to Promote the Supercompetitive Behavior of Hippo Pathway Mutant Cells

Genetic analyses in Drosophila epithelia have suggested that the phenomenon of “cell competition” could participate in organ homeostasis. It has been speculated that competition between different cell populations within a growing organ might play a role as either tumor promoter or tumor suppressor, depending on the cellular context. The evolutionarily conserved Hippo (Hpo) signaling pathway regulates organ size and prevents hyperplastic disease from flies to humans by restricting the activity of the transcriptional cofactor Yorkie (yki). Recent data indicate also that mutations in several Hpo pathway members provide cells with a competitive advantage by unknown mechanisms. Here we provide insight into the mechanism by which the Hpo pathway is linked to cell competition, by identifying dMyc as a target gene of the Hpo pathway, transcriptionally upregulated by the activity of Yki with different binding partners. We show that the cell-autonomous upregulation of dMyc is required for the supercompetitive behavior of Yki-expressing cells and Hpo pathway mutant cells, whereas the relative levels of dMyc between Hpo pathway mutant cells and wild-type neighboring cells are critical for determining whether cell competition promotes a tumor-suppressing or tumor-inducing behavior. All together, these data provide a paradigmatic example of cooperation between tumor suppressor genes and oncogenes in tumorigenesis and suggest a dual role for cell competition during tumor progression depending on the output of the genetic interactions occurring between confronted cells.


Introduction
Growth regulation requires the fine tuning between the rate of cell death and cell proliferation in developing organs. Studies in Drosophila have revealed that somatic cells within a growing epithelium compete with one another for contribution to the adult organ and this phenomenon, known as ''cell competition'' [1], is possibly conserved among organisms, for a review [2]. Cell competition was discovered several decades ago comparing the clonal growth parameters of Drosophila wild type cells (+/+) and slow-dividing Minute/+ cells [1]. From those analyses and recent data [3], it has been concluded that the contact between wild type and slow-growing cells, in genetic mosaics, favors the positive selection and clonal expansion of faster cells (winners) at the expense of slow-dividing ones (losers), although eventually the final number of cells in the organs is unaffected [3]. The biological function of cell competition remains unclear but it is thought to contribute to tissue homeostasis by coordinating the rate of cell proliferation and cell death [4,5]. One of the best examples illustrating cell competition was obtained from the analysis of Drosophila myc [4,5], opening to the speculation that this phenomenon might play a role in tumorigenesis [2,6], however the basis of cell competition in tumorous situations has just begun to be investigated [7]. dmyc is an evolutionarily conserved protooncogene associated with different cellular processes, including cell cycle progression, cell growth and apoptosis [8][9][10][11]. The function of dMyc protein is both necessary and sufficient to control rRNA synthesis and ribosome biogenesis [12]. In Drosophila, cells carrying hypomorphic alleles of dmyc are viable in a homotypic context, but they are outcompeted and excluded from the epithelium when surrounded by wild type cells [5]. By contrast, dmyc overexpressing cells become ''supercompetitors'' able to kill wild type surrounding cells [4,5]. Remarkably, dMyc upregulation is related with many types of human cancers [13] and it favors the clonal expansion of cells carrying additional oncogenic mutations [14,15].
During the last years, the Hippo (Hpo) tumor suppressor pathway has emerged as a safeguard system restricting organ growth and preventing hyperplastic disease in metazoans [16,17]. Mutations in several members of this pathway have been associated with tumor formation both in Drosophila and in humans [18]. It has also been reported that mutations in many members of the Hpo pathway can rescue the viability of heterozygous M/+ cells in genetic mosaics [19], suggesting that these mutant cells behave as ''supercompetitors''. Therefore the detailed analysis of Hpo pathway members appears to be an attractive model in which to evaluate the relationship between cell competition and tumor growth, as well as the molecular mechanisms required for this crosstalk. Hpo, Salvador (Sav) and Warts (Wts) constitute the core of the Hpo pathway that regulates by phosphorylation the downstream transcriptional co-activator Yorkie (Yki) [18,20]. The hyperphosphorylated form of Yki is retained in the cytoplasm [21,22], thereby preventing the expression of several target genes involved in cell proliferation control (Cyclin E, E2F1, bantam miRNA) [16,[23][24][25], cell death (dIAP1) [16] and cell signaling regulation (dally and dally-like) [26]. It has been demonstrated that Yki regulates its target genes by binding to Scalloped (Sd), a TEAD/TEF family transcription factor [27][28][29][30]. In addition, recent data indicate that Yki is also able to bind to the homeoprotein Homothorax (Hth) forming a complex which regulates the transcription of bantam in the eye disc [31]. The atypical cadherins Fat (Ft) [26,[32][33][34][35][36][37] and Dachsous (Ds) [20,26,33,38], as well as the FERM-domain proteins Expanded (Ex) and Merlin (Mer) [39], have also been implicated in the pathway as upstream components. Although their biochemical functions are still uncertain, it is assumed that they converge on Wts to regulate Yki activity [40,41].
Here we provide a detailed analysis of the autonomous and nonautonomous effects on growth of yki-expressing cells and mutations of members of the Hpo pathway. In addition we show that dmyc is a transcriptional target of Yki, able to confer competitive properties to the Hpo pathway mutant cells in the Drosophila wing. Furthermore, dmyc upregulation is essential to sustain the high rate of cell proliferation of Hpo mutant cells and to protect them from being eliminated in a competitive background. Finally, we show that the relative levels of dMyc protein between neighboring cells are critical in order to define the role of cell competition during tumor progression.

Hpo pathway mutant cells display supercompetitive properties
In order to analyze the competitive properties of Hpo pathway mutant cells, we used mosaic analysis to compare the size of yki overexpressing clones (hereafter referred to as yki over ) with their wild type twins. While clones and twins showed a comparable size in the wild type control ( Figure 1A, 1F, and 1G, and Figure S1A, S1C, S1D), yki over clones were notably larger than their wild type twins in wing discs dissected either 60h ( Figure 1D, 1H, and 1I) or 48h ( Figure S1B, S1E, S1F) after heat-shock induction. Furthermore, yki over wild type twins were almost disappeared from the epithelium at 120h after egg laying (AEL) ( Figure 1B and 1C). These differences in size were also prominent when discs were dissected at 96h AEL ( Figure 1D and 1E). Interestingly, the clonal expansion of yki over cells was also correlated with non-autonomous apoptosis, as revealed by active Caspase 3 immunoreactivity of a subset of surrounding wild type cells ( Figure 1B-1E). The size advantage of yki over clones and the induction of apoptosis in wild type cells is consistent with the broadly assumed definition of cell competition, which implies that the clonal expansion of the winner cells occurs at the expense of the juxtaposed losers, that are eliminated by apoptotic death [2,42,43]. The pattern of cell death in wild type and yki over cells ( Figure 1B-1E) was not confined to the interface between the two cell types; as can be seen in Figure 1E, cell death extends several cell diameters away and wild type cells tend to die massively when enclosed between nearby mutant clones ( Figure 1E, yellow arrowhead). A similar pattern of nonautonomous cell death was observed in wild type cells nearby mutant clones for other members of the Hpo pathway, such as ft and ex ( Figure S2A, S2B, S2C). Strikingly, yki over clones and wts mutant clones grown for a longer period presented autonomous cell death ( Figure 1C, see active Caspase 3 staining, and Figure  S2D), despite the upregulation of anti-apoptotic molecules such as dIAP1 [16]; this might be possibly due to either developmental constraints compensating for excessive proliferation of the entire organ or toxicity caused by high and constant levels of Yki. Altogether, these results confirm the previously suggested supercompetitive properties of the Hpo pathway mutant clones [19] by revealing their ability to overgrow and eliminate surrounding wild type cells.
dmyc is a Hpo pathway target gene regulated by the activity of Yki It is well documented that the confrontation of different levels of dMyc protein between two populations of cells either in vivo [4,5] or in cell culture [44] can trigger cell competition, however the molecular mechanism by which this occurs is unknown. In addition, myc family oncogenes are frequently overexpressed in human cancers and it contributes to tumor progression of YAPexpressing cells (mammalian orthologue of yki) [17]. We have previously shown that a transcriptional activation of dmyc occurs in ft mutant tissues and that ft clones fail to grow in a dmyc hypomorphic background [45], indicating a possible regulation of this oncogene by the Hpo pathway. Moreover, the expression pattern of dMyc is complementary to that of Ds in the wing imaginal disc (Figure 2A), suggesting a possible functional interaction. To validate this hypothesis, we analyzed dMyc

Author Summary
One of the major challenges of developmental biology and cancer research is to get a better understanding of how different signals regulate proper organ growth and prevent tumor formation. Even though there is a strong correlation between tumor progression and Myc family misexpression or Hippo signaling pathway malfunction, the relationship between these organ growth regulators remains unclear. Here, we demonstrate that the Hippo signaling pathway controls the transcription of Drosophila dmyc. Furthermore, we show that the misregulated expression of dMyc in Hippo mutant cells elicits their proliferative expansion at the expense of normal surrounding cells. These findings reveal a molecular mechanism of cooperation between oncogenes and tumor suppressor genes that favors both tumor progression and wild-type tissue elimination. Additionally, our findings indicate a dual role for cell competition during the tumour progression depending on the cellular context. expression in mutant clones for several members of the Hpo pathway and in yki over cells by immunofluorescence. Noticeably, we found that dMyc was upregulated in a cell-autonomous manner in yki over clones throughout the wing disc ( Figure 2B and Figure S3), with the weakest activation in the lateral regions, and in a subset of clones mutant for several Hpo pathway members ( Figure 2C-2F). These differences in dMyc activation between yki over clones and clones mutant for other members of the Hpo signaling pathway might be due to additional levels of regulation of the Hpo cascade operating on upstream members. According to our previous observations, we would predict a repression of dMyc upon Hpo pathway hyperactivation. To investigate this hypothesis, we expressed Hpo in the spalt expression domain of the developing wing disc. Since Hpo overexpressing cells die massively by apoptosis during development [25], we coexpressed the anti-apoptotic factor p35. As expected, cells coexpressing Hpo and p35 show reduced levels of dMyc with respect to the control ( Figure S4A) in both late ( Figure S4B) and early ( Figure S4C) wing discs. Thus dMyc levels can be regulated by the Hpo pathway activity. dmyc is transcriptionally regulated by Yki dmyc was observed upregulated in RT-PCRs performed on ft mutant imaginal discs [45], suggesting that it could be a transcriptional target of the Hpo pathway. In order to investigate this, we first performed an in situ hybridization in Drosophila wing discs expressing yki under the control of the decapentaplegic (dpp) promoter. As expected, dmyc transcript is detectable in the dpp domain both in yki and control dmyc-expressing discs ( Figure 3A). No signal within the dpp domain was detected in dpp.GFP control discs (not shown). We were able to reproduce these data using a dmyc.lacZ line [46] which recapitulates accurately the dmyc pattern throughout the wing disc during development [7,47]. As can be seen in Figure 3B, the ßGal expression is increased in the dpp domain upon yki expression, indicating that Yki acts upon dmyc transcription. This result was supported using clonal analysis, both in yki over cells, as shown in Figure 3C, and in cells mutant for ft ( Figure S5). Altogether, these data demonstrate the ability of the Hpo pathway to regulate dmyc transcription in the imaginal wing disc.
Yki transcriptional activity depends on the formation of tissuespecific complexes with different partners such as Scalloped and Homothorax [27][28][29][30][31]. In order to study the contribution of Sd to dmyc upregulation by Yki in the wing disc, we generated yki over clones coexpressing either a UAS-sd or a UAS-sd-RNAi construct (see Figure S6A for validation). As can be seen in Figure 3D, sd over ; yki over clones overgrew relative to yki over clones (compare with Figure 2B, 68% increase on average, n = 27, P,0,005) confirming previous data [29], but we were not able to detect significant differences in dMyc protein levels compared to yki over clones (n = 22, P = 0,43). As expected, control sd over clones did not overgrow and did not deregulate dMyc ( Figure 3E), demonstrating that Yki is required for dMyc upregulation. We were not able to recover sd-RNAi; yki over clones in the wing pouch region, but clones generated in other territories of the wing disc, although large, did not upregulate dMyc ( Figure 3F), nor showed the same degree of hyperplasia as Yki expression alone ( Figure 1B-1E). sd-RNAi control clones were very small and did not deregulate dMyc (not shown). These data indicate a key role for Sd in vivo in upregulating dMyc in yki over clones, and in contributing to the yki over tumorous phenotype.
Interestingly, examination of dmyc locus revealed the existence of several CATTCCA repeats in non-coding regions of the gene, which perfectly match the mammalian [48,49] and Drosophila [28,29] TEAD/TEF family transcription factor consensus binding motifs (mammaliam orthologues of Scalloped). In addition, these putative binding motifs for Yki/Sd complexes are evolutionarily conserved in D. simulans ( Figure 3G) and relatively close to the insertion point of P elements that recapitulate the endogenous expression of the gene (dm PL35 LacZ [50,51] and dm BG02383 Gal4 insertions -http://flybase.org/reports/FBti0018138. html).
To test the significance of these sequences in dmyc regulation, we generated a dmyc-firefly reporter containing the putative responsive elements for Yki/Sd complexes ( Figure 3H) and performed a transient dual luciferase assay in S2 cells. As can be seen in Figure 3I, the reporter was specifically activated upon Sd and Yki cotransfection but, unexpectedly, the transfection of Yki alone was able to activate the reporter as efficiently as the cotransfection Yki/Sd ( Figure 3I). This result suggests that in presence of high levels of Yki alone, additional partners such as Hth [31] could bind it and co-regulate dmyc expression.
Indeed, complementarily to Yki/Sd complexes, Yki/Hth complexes seemed to play the same role in the presumptive thoracic region of the wing disc. Supporting this conclusion, hth-RNAi; yki over clones down-regulated dMyc in the notum (30% reduction on average, n = 15, P,0,05, Figure S6B, yellow arrows) and did not grow as tumors in that region. By contrast, they were undistinguishable from yki over clones in the wing pouch region ( Figure S6B, white arrowhead), where Hth expression is almost undetectable ( Figure S6C). Altogether, these latter results indicate that Sd and Hth play a role in Yki-induced tumorigenesis by regulating dmyc expression in the wing disc, with Sd playing a more critical role in the pouch and Hth acting in the presumptive thorax.

dMyc upregulation enhances cell proliferation of the Hpo pathway mutant cells in an autonomous manner
With the aim to investigate the cell-autonomous contribution of dMyc overexpression to yki over phenotypes, we first compared the size of yki over clones with that of yki over ; dmyc-RNAi clones (Figure 4, see also Figure S7A, S7A9, and [7] for RNAi construct validation ). As expected, dmyc-RNAi clones showed a reduced number of cells with respect to that observed in wild type clones (21% reduction on average, compare Figure 4B and 4B9 with Figure 4A and 4A9, P,0.01). The reduction in cell number displayed by the yki over ; dmyc-RNAi clones with respect to the yki over clones was even more evident (43% reduction on average, compare Figure 4D and 4D9 with Figure 4C and 4C9, P,0.01), and this percentage raised up to 65% (n = 87, P,0,001) when these clones were induced earlier in development (42-54h AEL), indicating a strong cell-autonomous requirement of dMyc protein for the expansion of yki over clones. We also observed that the non-autonomous apoptosis induced by yki overexpression was reduced upon dmyc deprivation (32% on In situ hybridization on imaginal wing discs with a full-length dmyc RNA probe. The dmyc transcript is robustly upregulated in the dpp.yki disc on the left. A dpp.dmyc disc is shown on the right as a control. (B) ßGal staining (red) of dmyc.lacZ G0354 /+; dpp.Gal4, UAS-GFP/UAS-yki imaginal wing discs. dmyc expression is upregulated in the dpp domain (GFP + ). (C) ßGal staining (red) of dmyc.lacZ G0354 /hs-Flp; actFRTy + FRTGal4, UAS-GFP/UAS-yki imaginal wing discs. yki over clones (GFP + ) were induced at 42-54h AEL. As can be observed, a robust activation of dmyc regulatory sequences is visible within the mutant clones. (D,E) dMyc staining of yw, hs-Flp/+; UAS-sd/+; actFRTy + FRTGal4, UAS-GFP/UAS-yki and yw, hs-Flp/+; UAS-sd/+; actFRTy + FRTGal4, UAS-GFP/+ imaginal wing discs respectively. Clones (GFP + ) were induced at 42-54 h AEL. As can be observed in E, dMyc levels are comparable inside and outside the mutant clones. (F) dMyc staining of yw, hs-Flp/+; UAS-sd-RNAi/+; actFRTy + FRTGal4, UAS-GFP/UAS-yki imaginal wing discs. Clones (GFP + ) were induced at 42-54 h AEL. No clones in the wing pouch region were recovered and those outside the wing pouch never took the tumorous shape typical of yki over clones. All larvae were dissected at 120h AEL. (G) dIAP1 Hpo Responsive Elements (HRE) for Yki/Sd complexes are indicated in red [28]. average, n = 28, P,0,01, Figure S7B). These data suggest that dMyc upregulation promotes cell proliferation of yki over clones in an autonomous manner, and also promotes their competitive behavior.
To further characterize this proliferation-promoting effect of dMyc, we compared the clonal behavior of various mutations in members of the Hpo pathway grown in two different genetic backgrounds: a wild type context and a genetic background overexpressing dmyc under the control of a hedgehog promoter in the posterior (P) compartment of the wing disc. We found that ft, ex and ds mutant clones were consistently larger in those territories expressing uniform levels of dMyc than in the wild-type background ( Figure 5 and Figure S8). It is however described that the overexpression of dMyc is able to autonomously increase apoptosis [8][9][10][11]. In fact, the wild type tissue expressing high amounts of dMyc tends to die and does not overgrow (see active Caspase 3 stainings in Figure 5A and 5D). Noticeably, the apoptosis mediated by dMyc overexpression seems to be extremely reduced inside ft and ex clones ( Figure 5A and 5D) with respect to the wild type surrounding territories, likely due to the upregulation of antiapoptotic genes such as dIAP1, a target of the Hpo pthway [20]. In addition, the dying cells in this genetic background might induce morphogens to promote compensatory proliferation [52] that may contribute to the extra-growth of ftor ex-UAS-dmyc expressing clones. To circumvent this problem, we repeated the same experiment coexpressing dmyc and dIAP1. As can be seen in Figure S9, both ft ( Figure S9A) and ex ( Figure S9D) mutant clones grown in the P compartment were still consistently larger than those originated in the A compartment, thus confirming a specific cooperation of dmyc and Hpo pathway mutants in clonal expansion.
Hpo mutant cells therefore seem to show the ability to take advantage of the cell mass accumulation boosted by dMyc overexpression to proliferate faster.
dMyc upregulation prevents the Hpo pathway mutant clones from being restrained in a competitive background To address the non-autonomous relevance of dmyc upregulation in providing yki over cells with a supercompetitive behavior, we compared the size of yki over clones generated in a wild type background to that of yki over clones generated in a background ubiquitously overexpressing dmyc under the control of a tubulin (tub) promoter (cell competition assay, [4,5]). In this assay cells express the endogenous dmyc gene plus an extra copy of the gene under the control of a tub promoter that ensures two-to-threefold increase of dmyc transcript [5]. This extra copy of dmyc is located in a removable cassette between the tub promoter and a Gal4 cDNA. Upon dmyc cassette excision, the tub promoter drives Gal4 expression in the clones and, as a result, those cells express lower levels of dmyc relative to the background and are rapidly eliminated from the tissue by cell competition. Only few genes have so far been found whose overexpression rescues cell viability in this context [5]. The relative difference in dMyc levels between ykiexpressing cells and the surrounding tub.dmyc cells was minimized in a competitive background compared to a wild type context (compare Figure S7C and S7C9 with Figure 2B). In this competitive background, yki over clones showed a diminished ability to overgrow compared to a wild type background (44% reduction on average, compare Figure 6C and 6C9 and Figure 6B and 6B9; P,0,01). Besides the reduction in size, yki over clones showed an important reduction in clone number both in discs ( Figure 6C) and adult wings (compare Figure S7E to Figure S7D). Moreover, yki over clones induced earlier in development (42-54h AEL) were never recovered at the end of larval development (not shown). These data indicate that the competitive properties of yki over cells are extremely reduced when they are surrounded by cells expressing very high amounts of dMyc.
We then performed the same competition assay as before while reducing dmyc activity inside the clones. We used the pupal lethal dmyc PL35 allele [49] and, taking advantage of dmyc locus association to chromosome X, we were able to analyze both female (heterozygous condition, the expression of dmyc is halved) and male (hemizygous condition, the expression of dmyc is completely removed) larvae. In dmyc PL35 /+; tub.dmyc females, yki over clones were smaller than those described in the previous assay (28% reduction on average, compare Figure 6D and 6D9 to Figure 6C and 6C9, P,0,05), whereas they were completely outcompeted by 48h after the heat shock in males (not shown). Since it has been observed that a dmyc PL35 heterozygous condition does not impair cell growth or proliferation rate [49], our results reveal an important role for dmyc-induced cell competition in controlling the clonal expansion of yki over cells, which may occur via their nonautonomous capabilities to compete with neighboring wild type cells.
dMyc expression alone is not sufficient to prevent the elimination of yki mutant cells yki LOF clones generated in a wild type background are not able to grow [16,25] and the ectopic expression of the antiapoptotic proteins dIAP1 [25] or p35 ( Figure S10A) poorly rescues their viability, whereas a Minute background [53] or bantam overexpression within yki clones has been shown to partially rescue their growth [25]. Since our results have indicated that dmyc participates in tumor growth of the Hpo pathway mutant cells, we therefore analyzed if the expression of dMyc was sufficient to prevent the death of yki mutant cells. The overexpression of dMyc failed to rescue the viability of yki 2/2 cells ( Figure S10B). Since yki mutant cells express low levels of the apoptosis inhibitor dIAP1 (not shown), this result is not surprising, considering the autonomous cell death described for cells overexpressing dMyc [11]. However, yki mutant cells coexpressing dMyc and p35 also failed to grow ( Figure S10C). The lack of expression of additional antiapoptotic genes and cell cycle regulators [18] possibly impedes the clonal growth of yki mutant cells even though they overexpress dMyc. This result suggests that dmyc expression is able to enhance the ability of Hpo pathway mutant cells to grow, but it is not sufficient to rescue tissue growth of yki 2/2 clones.

Discussion
Cells within a tissue coordinate and execute complex genetic programs in order to succeed in completing a variety of processes during development. In this context, the phenomenon of cell competition may be part of the developmental plan that ensures removal and replacement of defective cells in growing organs, thus keeping their size invariant. In this work, we have evaluated in details the relationships between the phenomenon of cell competition and the clonal expansion of tumorous cells, using for that purpose mutants in components of the evolutionarily conserved Hpo pathway. From our studies we reveal that the Hpo pathway regulates dMyc expression, and show that this is critical for the tissue growth and competitive behavior of Hpo pathway mutant clones.
dMyc is a Hpo pathway transcriptional target dmyc upregulation has been demonstrated in many studies to provide cells with supercompetitive properties [4,5,7]. The model explaining how dMyc can confer competitive properties to cells is based on the relative levels of this protein in neighboring cell populations, transforming those cells expressing higher levels of dMyc into supercompetitors [4,5]. dmyc overexpression is nevertheless insufficient to drive tumorous growth; dmyc over clones fail to overproliferate and show strong autonomous apoptosis [9]. Interestingly, we found that dMyc protein is overexpressed in Hpo pathway mutant clones, indicating an involvement for this cascade in dmyc regulation ( Figure 2). Furthermore, the upregulation of dMyc in Yki-expressing cells correlates with an increase in the amount of mRNA, observed by in situ hybridization ( Figure 3A) and using a dmyc.lacZ line ( Figure 3B and 3C). Finally, we have identified a regulatory region in the second intron of dmyc that is sensitive to Yki abundance; importantly, this regulatory region includes predicted consensus-binding motifs for Sd ( Figure 3H). Clonal experiments in the wing disc indicate that Sd is necessary for Yki function in vivo, since upon Sd downregulation Yki is no longer able to induce tumorous growth and does not upregulate dMyc ( Figure 3F). All these findings support the notion that there is a transcriptional regulation of dMyc mediated by Yki/Sd complexes in the wing pouch. Importantly, similar results were observed for dMyc regulation in the notum by Yki/Hth complexes, suggesting that tumor growth and dmyc regulation are tissue-specific.
What is the contribution of dMyc to the Hpo pathway mutant phenotypes?
We found that dMyc upregulation is a common feature of Hpo pathway mutant cells. Since dmyc has been repeatedly associated with tumor progression and cell competition, we analyzed its role in the clonal expansion of Hpo pathway mutant cells. We observed that the reduction of dMyc expression restricts the ability of Hpo pathway mutant cells to proliferate (Figure 4), whereas its uniform overexpression strongly promotes their proliferation ( Figure 5). Furhermore, while dMyc-expressing wild type cells surrounding mutant clones are rapidly eliminated by autonomous apoptosis, Hpo pathway mutant cells are able to take advantage of dMyc role in protein biosynthesis and cellular growth to divide rapidly. This is a clear example of functional cooperation between different genes in order to favor tumor progression, but it also indicates a specific role of dMyc in promoting the clonal expansion of Hpo pathway mutant cells. According to these data, we conclude that dMyc behaves as a growth-promoting factor which sustains the hyperplastic phenotype of Hpo pathway mutant cells. Importantly, this specific cooperation might be evolutionarily conserved, since c-myc appears to be upregulated in a murine model of YAPinduced carcinoma [17].

Relative levels of dMyc in neighboring cells restrict/ promote clonal expansion of hyperplastic cells, likely through cell competition
It has been suggested that cell competition may be a mechanism potentially restricting the clonal expansion of tumorous cells [7], but it might also help faster proliferation of transformed cells. Our data indicate that Hpo pathway mutant cells are able to use high levels of dMyc to proliferate rapidly ( Figure 5), but in a competitive context, where neighboring cells express high levels of dMyc, clonal expansion of yki over cells is restrained (Figure 6), therefore suggesting a tumor suppressor role for cell competition. Conversely, dMyc upregulation in yki over clones grown in a wild type background favors their clonal expansion promoting cell autonomous proliferation and also conferring the ability to outcompete sourrounding cells in a non-autonomous manner. These findings suggest that the phenomenon of cell competition may play a dual role in tumor progression depending on the output of the genetic interactions occurring between adjacent cells.
In summary, we have shown a tumor-braking gene network in Drosophila epithelia which tightly controls cell proliferation, apoptosis and cell competition via the Hpo pathway and dMyc expression. Importantly, YAP deregulation has been reported in several types of human cancers [54][55][56], therefore the mechanism of clonal expansion of Hpo pathway mutant cells in Drosophila might be relevant to understand tumor progression in mammals.
All experiments were carried out at 25uC unless otherwise indicated.
MARCM UAS-yki twin-spot clones were induced at different stages of development by a 35-minutes heat shock at 37uC and larvae of the following genotype were dissected at either 84-100h AEL or 120h AEL: yw, hs-Flp, tub-Gal4, UAS-GFP; FRT42D, tub-Gal80/FRT42D, Ubi-GFP; UAS-yki/+. Clones of the same genotype were induced 54-66 h AEL and dissected 48h after a 20-minutes heat shock ( Figure S1). For FRT-Flp twin analysis, the following hypomorphic or null alleles were used: ds D36 , ft G-rv , ex E1 , wts X1 , yki B5 . Loss-of-function clones of ds, ft, ex and wts in either wild-type or mutant backgrounds overexpressing different transgenes in the posterior compartment were induced at 48-72h AEL by 1 hour heat shock at 37uC. Larvae of the following genotype were dissected at 120h AEL: yw, hs-Flp; FRT40A, Ubi-GFP/FRT40A, ds D36 or ft G-rv or ex E1 yw, hs-Flp; FRT82B, Ubi-GFP/FRT82B, wts X1 yw, hs-Flp; FRT40A, Ubi-GFP/FRT40A, ds D36 or ft G-rv or ex E1 ; hh-Gal4/UAS-dmyc yw, hs-Flp; FRT40A, Ubi-GFP/FRT40A, ft G-rv or ex E1 ; hh-Gal4/ UAS-dmyc, UAS-dIAP1 The size of non-confluent clones was measured drawing each Zstack of the confocal images using ImageJ software (http://rsbweb. nih.gov/ij). Afterwards the area of the clones was normalized dividing by the area of the wing pouch, considered as the territory encircled by the first outer folding of the wing. In Figure S1, the narrower window of clonal induction allowed us to compare clonal size without size normalization respect to the wing pouch. Statistical analysis was performed with Microsoft Excel and R (www.r-project.org). Statistical significance was determined by two tailed Student's t test and reported as the associated probability value (P).

Immunofluorescence
Immunostainings were performed using standard protocols. The following primary antibodies were used: mouse anti-dMyc (1:5, P Gallant), mouse anti-En (1:50, DSHB), rabbit anti-active Caspase 3 (1:100, Cell Signaling Technology), rabbit anti-p35 (1:1000, Stratagene), rabbit anti-Ds (1:100, D Strutt), rabbit anti-Hth (1:400, A Salzberg, [57]), mouse anti-dIAP1 (1:100, B Hay) and rabbit anti-ßGal (1:400, F Graziani). Anti-mouse and antirabbit Alexa Fluor 555 (1:200) (Molecular Probes) and anti-mouse Cy5 (1:200) (Jackson Laboratories) against corresponding primary antibodies were used as secondary antibodies. Imaginal discs were mounted in Vectashield (Vector Laboratories) for confocal imaging. Single Z stacks were acquired with Leica SP2 and SP5 confocal microscopes. Images for Figure 4 and Figure 6 were captured with an epifluorescence Nikon 90i microscope. Entire images were elaborated with Photoshop CS2 (Adobe) and the projections along the Z axis were rebuilt starting from 35-55 Z stacks using the ImageJ public software (NIH). For measurements of dMyc abundance, fluorescence intensity was calculated using the ImageJ public software (NIH) as the average gray value within selectioned portions of confocal Z stacks. For measurement of active Caspase 3 signal outside UAS-dmyc-RNAi; UAS-yki and UAS-yki clones, staged wing discs were chosen containing as few clones as possible and single cells positive to active Caspase 3 observed at a maximum distance of five nuclei (counterstained with DAPI) from the border of the clone were counted on confocal Z stacks. In situ hybridization was performed with a full length dmyc probe [9] on wing imaginal discs of L3 larvae expressing UAS-GFP, UAS-dmyc or UAS-yki under the control of dpp-Gal4. RNA in situ hybridization was carried out using digoxigenin-labeled RNA probes [58].

Luciferase transient expression assays
Drosophila S2 cells were grown at 25uC in Schneider medium (GIBCO) supplemented with 10% heat-inactivated FCS and 100 units of penicillin.
The PCR product was first cloned in pCR 2.1 TOPO-TA (Sigma) and then subcloned in KpnI/XhoI sites of pGL3 Promoter vector. For luciferase transient expression assays, 2610 4 cells were plated in 96-well dishes. Cells were harvested at 48 hours after transfection and luciferase activity was measured using the Dual-Luciferase reporter assay system (Promega). Dual-Luciferase measurements were performed using a FLUOstar Optima luminometer (BMG Labtech) and normalized to the Renilla luciferase activity using pAct5C-seapansy as an internal control. All transient expression data reported in this paper represent the means from three parallel experiments, each performed in triplicate. Average relative luciferase activity was graphed and statistically analyzed by the Student's t-test. (B-C) dMyc staining of late (B) and early (C) w; sal.Gal4/+; UAS-Hpo/+; UAS-p35/+ imaginal wing discs. p35 is shown in the green channel and dMyc in red. As can be observed in the Z-sections dMyc abundance is lower inside the sal domain. The position of Zsection is indicated by white bars in the surface view of the wing discs. Found at: doi:10.1371/journal.pgen.1001140.s004 (2.76 MB TIF) Figure S5 dmyc is transcriptionally upregulated in ft mutant clones. ßGal staining (red) of dmyc.lacZ G0354 /hs-Flp; ft G-rv , FRT40/UbiGFPnls, imaginal wing discs. As can be observed, a robust activation of dmyc regulatory sequences is visible within the mutant clones (arrows). Larvae were dissected at 120h AEL.  Figure  A is shown. The percentage of dMyc abundance reduction inside the mutant clones calculated as the abatement of fluorescence intensity (see Methods -Immunofluorescence) with respect to the neighboring tissue was 62% on average (n = 12). (B) dmyc-RNAi; yki over clones display a reduced non-autonomous apoptotic activity (yellow arrows, see Methods -Immunofluorescence -for calculation) compared to yki over clones (see Figure 1). (C) yki over clones can compete in a high dmyc level background, where wild type clones fail to grow; clones were induced at 66-78h AEL and allowed to grow until 120h AEL. In red, staining for dMyc indicates that dMyc levels are quite similar inside the yki over clone and in the tub.dmyc background. (C9) A projection along the Z axis of the clone presented in figure C is shown. (D-E) In adult wings, tub.yki over clones generated in a wild type background (D) are bigger than tub.yki over clones generated in a tub.dmyc background (red arrows, E) confirming the results illustrated in Figure 6. Clones were induced at 66-78h AEL and survived up to the adult stage.