The Hippo pathway acts via p53 and microRNAs to control proliferation and proapoptotic gene expression during tissue growth

Summary The Hippo pathway has a central role in coordinating tissue growth and apoptosis. Mutations that compromise Hippo pathway activity cause tissue overgrowth and have been causally linked to cancer. In Drosophila, the transcriptional coactivator Yorkie mediates Hippo pathway activity to control the expression of cyclin E and Myc to promote cell proliferation, as well as the expression of bantam miRNA and DIAP1 to inhibit cell death. Here we present evidence that the Hippo pathway acts via Yorkie and p53 to control the expression of the proapoptotic gene reaper. Yorkie further mediates reaper levels post-transcriptionally through regulation of members of the miR-2 microRNA family to prevent apoptosis. These findings provide evidence that the Hippo pathway acts via several distinct routes to limit proliferation-induced apoptosis.


Introduction
Cell proliferation is intimately linked with cell death. Cues that drive cell growth and division also induce apoptosis (Pelengaris et al., 2002). In an abnormal cell proliferation scenario, such as cancer, cells adopt a variety of strategies to overcome cell death (Hanahan and Weinberg, 2011). Many signaling pathways that drive tissue growth have been found to coordinate cell proliferation and apoptosis during animal development. Defects in these pathways quite often cause tissue overgrowth or cancer.
The Hippo pathway is one such signaling pathway, acting as a negative growth regulator. Mutations in several members of the pathway lead to tumorigenesis, implicating them as tumor suppressors (Cai et al., 2010;Pan, 2010). The core pathway comprises a kinase cascade including the Hippo and Warts kinases with their adaptors Salvador (Sav) and Mats (Moberg et al., 2001;Tapon et al., 2002;Harvey et al., 2003;Udan et al., 2003;Wu et al., 2003). Several proteins have been implicated as upstream regulators of this kinase cascade by genetic studies, including Merlin/NF2, Expanded and the atypical cadherin, Fat (Hamaratoglu et al., 2006;Silva et al., 2006;Yu et al., 2010). Downstream, the Warts kinase directly phosphorylates and inactivates transcriptional coactivators including YAP, TAZ, and in Drosophila, Yorkie (Yki) (Huang et al., 2005;Zhao et al., 2007;Zhao et al., 2008;Zhao et al., 2010). YAP/TAZ and Yki function to promote cell proliferation and inhibit apoptosis. These proteins possess no DNA binding activity and therefore bind to transcription factors including Scalloped/TEAD to activate their targets.
Genetic studies have identified functional targets of Yki with positive roles in cell proliferation, including cycE and Myc, as well as negative regulators of apoptosis, including the Drosophila Inhibitor of Apoptosis Protein (DIAP1) and the antiapoptotic microRNA bantam (Moberg et al., 2001;Moberg et al., 2004;Huang et al., 2005;Nolo et al., 2006;Thompson and Cohen, 2006;Neto-Silva et al., 2010). These studies have suggested that the Hippo pathway can balance proliferative drive with limitation of proliferation-induced apoptosis. This combination of roles may also explain the potency of mammalian YAP in control of tissue growth and its ability to induce cancer when overexpressed (Dong et al., 2007;Pan, 2010).
The tumor suppressor p53 is another key regulator coordinating cell division and cell death. Activation of p53 by the DNA damage checkpoint or other cell cycle abnormalities, leads to growth arrest, and initiates apoptosis. Activated p53 binds DNA and directs expression of downstream genes including p21, which inhibits the activity of cyclin-CDK complexes and activates cell cycle checkpoints to halt cell division (Gartel and Radhakrishnan, 2005). In addition, p53 promotes transcription of the proapoptotic genes Bax, PUMA and Apaf-1 to induce cell death. Recent studies in human cells have identified the ASPP1 protein (apoptosis-stimulating protein of p53-1) as a key mediator of p53-induced apoptosis (Aylon et al., 2010;Vigneron et al., 2010). The Hippo pathway kinase Lats, the mammalian homolog of Warts, phosphorylates ASPP1 and forms a complex with ASPP1 and p53 to activate the proapoptotic transcription program. Phosphorylation of ASPP1, however, can be antagonized by another Lats substrate YAP.
The upstream control of the apoptosis program is conserved in Drosophila, with p53 serving as a mediator of the DNA damage checkpoint. However, the effector program involves a set of insect-specific proapoptotic genes: reaper, head involution defective (hid), grim and sickle (skl) (Steller, 2008). The proapoptotic activity of these four proteins results from their ability to bind and inactivate DIAP, which in turn inhibits caspases. In mammals the corresponding functions are provided by Apaf-1 to cleave and activate caspases instead of derepression of caspases (Pop et al., 2006). Previous studies in Drosophila have shown that the bantam microRNA acts to repress hid to limit proliferation induced apoptosis . bantam mediates interaction between the EGFR and Hippo growth control pathways (Herranz et al., 2012). microRNAs of the miR-2 seed family have also been shown to regulate the expression of the proapoptotic genes reaper, grim and skl Brennecke et al., 2005;Leaman et al., 2005;Thermann and Hentze, 2007) and to limit apoptosis in the developing nervous system (Ge et al., 2012).
In view of the importance of the Hippo pathway in regulating proliferation-induced apoptosis, we have examined other modes of action for Yki. Here we provide evidence for additional parallel pathways involving Yki, p53 and the miR-2 family of microRNAs in controlling the expression of reaper another key proapoptotic gene. Yki acts via regulation of p53 on reaper transcription. In some tissues, Yki acts independently via members of the miR-2 family to regulate expression of reaper post-transcriptionally. Our findings place Yki at the center of a network of regulatory relationships balancing cell proliferation, p53-dependent checkpoints, proapoptotic genes and miRNAs in control of tissue growth.

Hippo pathway controls apoptosis by limiting reaper expression
The transcription coactivator Yorkie mediates Hippo pathway activity to control gene expression in Drosophila. We used RNAi to deplete yorkie (yki) mRNA from S2 cells, to assess the contribution of the Hippo pathway to expression of genes involved in regulation of apoptosis. Depletion of yki mRNA was effective, and resulted in increased expression of reaper mRNA and a smaller increase in hid mRNA (Fig. 1A, **P,0.01). To test this relationship in a growing tissue yki was overexpressed in the wing imaginal disc under control of nubbin-Gal4. yki overexpression decreased the level of reaper mRNA ( Fig. 1B, P,0.01; control for yki mRNA level in supplementary material Fig. S1A). Thus Yki appears to negatively regulate expression of reaper.
Does regulation of reaper contribute to the growth regulatory activity of the Hippo pathway in vivo? We made use of patched-Gal4 (ptc-Gal4) to direct depletion of yki in a defined region of the wing (shaded in Fig. 1C). Expression of a UAS-yki RNAi transgene under ptc-Gal4 control reduced the area of the relevant region of the wing (Fig. 1C). This effect was quantified by measuring the ratio of the width of the vein 3-4 region to that of the vein 4-5 region (indicated by solid and dashed red lines, upper left panel of Fig. 1C). Depletion of yki reduced the relative size of the region where the Gal4 driver was expressed (Fig. 1D).
This effect was partially offset by concurrently limiting reaper expression using a chromosomal deletion, Df(3L)XR38, which removes reaper and skl, but not the adjacent grim and hid genes (Peterson et al., 2002). Df(3L)XR38 on its own showed no effect on growth, but limited the undergrowth caused by yki depletion (Fig. 1C,D; P,0.001). These observations suggest that increased expression of reaper contributes to the effects of yki depletion in vivo.

Yki acts via induction of p53 activity
Previous reports have shown that p53 can directly regulate reaper expression in Drosophila (Brodsky et al., 2000;Peterson et al., 2002;Zhou and Steller, 2003). This raised the possibility that Yki might act via p53 to control reaper during tissue growth in vivo.
To test this we used the ptc-Gal4 UAS-yki RNAi undergrowth assay. Coexpression of a dominant negative form of p53 (p53DN) partially suppressed the tissue undergrowth caused by depletion of yki ( Fig. 2A; P,0.001). Expression of p53DN on its own had no effect on growth. Similarly, reducing p53 activity by introducing a null allele of the p53 gene also partially suppressed the effects of engrailed-Gal4 UAS-yki RNAi on tissue growth ( Fig. 2B; P,0.05). The p53 mutant had no effect on its own. p53 can also be activated through the caspase Dronc (Nedd2like caspase, Nc (Wells et al., 2006;Shlevkov and Morata, 2012)). This raised the possibility that depletion of Yki by RNAi could lead to reduced DIAP1 expression and thereby trigger Dronc-mediated activation of p53. To address this possibility, we depleted both Yki and Dronc from S2 cells and found that the increase in reaper mRNA levels was not reduced compared to cells depleted of Yki only, as might have been expected if the effects of Yki depletion were mediated through this feedback loop (Fig. 2C). Furthermore, reaper mRNA levels were higher in wing discs coexpressing UAS-DIAP1 and UAS-Yki RNAi compared to UAS-Yki RNAi alone ( Fig. 2D, P,0.05; DIAP1 overexpression quantified in supplementary material Fig. S1B). The increase in reaper levels may reflect improved survival of Yki-depleted cells when expressing DIAP1. Caspase activation due to low DIAP1 levels also seems unlikely to explain the effects of Yki depletion on reaper mRNA levels.
In mammalian cells expressing the oncogenic form of H-Ras, Lats, a component of the Hippo pathway, has been shown to phosphorylate ASPP1 and form a complex with ASPP1 and p53 to direct expression of pro-apoptotic genes (Aylon et al., 2010;Vigneron et al., 2010). To ask whether Drosophila ASPP (CG18375) might also be involved in the context of Yki regulation of p53 activity in normal tissue growth, we assessed the effects of removing one copy of the ASPP gene on reaper mRNA levels in wing discs. Quantitative RT-PCR showed that reaper mRNA was reduced by ,25%, when ASPP mRNA was reduced to ,50% in these discs ( Fig. 2E; **P,0.01). Next, we assessed the effects of depleting ASPP by RNAi and the effects of removing one copy of the ASPP gene in the ptc-Gal4 UASyki RNAi undergrowth assay. In both scenarios reduced ASPP activity partially restored growth of the yki-depleted tissue (Fig. 2F,G; ***P,0.001).
Taken together, these observations suggest that the Hippo pathway acts through Yki and p53 to control reaper expression. The involvement of ASPP, suggests that this regulation is likely to be mediated through Yki binding to Lats/Wts and competing for ASPP1 phosphorylation, as described in mammalian cell culture models (Aylon et al., 2010;Vigneron et al., 2010). Here we present evidence that limiting reaper levels by manipulating p53-ASPP1 activity contributes to suppressing the tissue growth effects of the Hippo pathway. This observation is consistent with a model in which the Hippo pathway regulates p53 activity to control proliferation-induced apoptosis.
Yki regulates miRNA expression to control reaper level Previous reports have shown that microRNAs of the miR-2 seed family (Fig. 3A) can regulate reaper, grim and skl Leaman et al., 2005;Brennecke et al., 2005;Thermann and Hentze, 2007). This prompted us to ask whether there might be a miRNA-based mechanism by which the Hippo pathway controls reaper expression. As a first step we asked which of the miR-2 family miRNAs is subject to regulation by the Hippo pathway in S2 cells. Depletion of yki in S2 cells by RNAi led to a significant reduction in the levels of expression of miR-2a and b (P,0.05, Fig. 3B; miR-13a/b were on average lower, but the effect was variable and so not statistically significant). miR-11 was not significantly changed. miR-6 is expressed at very low levels in S2 cells.
As a first step to address how Yki might regulate miR-2 expression, we sought to identify cis-regulatory control elements that direct expression of miR-2 loci in S2 cells. miR-2a-1, miR-2a-2 and miR-2b-2 are expressed as a cluster of 3 miRNAs located in an intron of the spitz gene (Fig. 3C). A 1.9 Kb DNA fragment covering the intronic sequences upstream of the miRNA cluster and spanning the next upstream exon proved sufficient to direct expression of a luciferase reporter gene in S2 cells ( Fig. 3C; supplementary material Fig. S2A). We then used this luciferase reporter to assess the effects of depleting yki by RNAi. Expression of the miR-2a cluster reporter decreased significantly in ykidepleted cells (Fig. 3D), suggesting that Yorkie regulates transcription of the miR-2a cluster.
To further assess this regulation in vivo, we first asked whether overexpressing members of the miR-2 family could rescue the ptc-Gal4 UAS-yki RNAi undergrowth assay. Coexpression of a miR-2a/2b cluster transgene or a miR-11 transgene suppressed the undergrowth of yki-depleted tissue caused by elevated reaper mRNA (Fig. 4A,B; P,0.001). Expression of miR-2a/2b or miR-11 on their own had no effect on growth. Next we introduced a In normal flies this ratio is ,1.2:1. Note that there was no effect of ptc-Gal4 driven Yki RNAi on the size of the region between veins 4-5 (not shown), so the ratio reflects reduction of the L3-4 region. *** indicates statistically significant increase in the width of the ptc-Gal4 expression domain when one copy of the rpr and skl genes were removed (P,0.001).
miR-2a sensor into the ptc-Gal4 UAS-yki RNAi assay to report miR-2a activity in vivo. The sensor transgene expresses GFP under control of the ubiquitously-expressed tubulin promoter and carries two miR-2a sites in its 39 UTR (as described (Brennecke et al., 2005)). However, depletion of yki had no effect on the expression of the miR-2a reporter in wing imaginal discs (supplementary material Fig. S3). Although ectopically expressing members of miR-2 family could suppress undergrowth of yki RNAi tissue, the negative result with the miR-2a reporter suggests that the effects of Yki on reaper are not mediated by regulation of miR-2a expression in the wing discs. To ask whether this regulation occurred in other tissues in vivo, in addition to S2 cells, we expressed UAS-yki RNAi ubiquitously under tubulin-Gal4 control and found a significant reduction of miR-2a and b in the whole 3 rd instar larvae ( Fig. 4C; P,0.05). These findings suggest that the Hippo pathway contributes to control of apoptosis through regulation of miR-2 expression in some but not all tissues.  activity is sufficient, more cycE binds to CDK2 to promote the transition from G1 to S phase promoting cell division (Hinds et al., 1992). Meanwhile, elevated Myc activates numerous target genes for ribosome assembly and cellular growth (Grewal et al., 2005). Myc activation is sufficient to induce apoptosis (Pelengaris et al., 2002), and YAP/TAZ/Yki act in parallel to limit apoptosis to ensure balance in the coordinated drive for cells to grow and divide.

Studies conducted in mammals and
Yki acts at multiple levels to control apoptosis. Yki directs expression of the Drosophila Inhibitor of Apoptosis Protein, DIAP1 (Huang et al., 2005). We have provided evidence that Yki acts via regulation of p53 activity to regulate reaper transcription. Yki acts in parallel in some tissues via regulation of miR-2 family miRNAs to regulate reaper activity. miR-2 has been shown to regulate translation of repaer mRNA (Thermann and Hentze, 2007). In addition, Yki mediated regulation of bantam miRNA expression Thompson and Cohen, 2006) controls hid transcript levels . Thus, Yki is at the center of a network of regulatory relationships involving p53-dependent checkpoints, proapoptotic genes and expression of multiple miRNAs in control of proliferation induced apoptosis (illustrated in Fig. 4D).
Why use a variety of parallel effector mechanisms? Our findings suggest that there may be tissue-specific differences in pathway use. As well, use of multiple pathways allows for the possibility that their activity may be regulated in a manner that depends on physiological context. This may be advantageous in adapting control of growth and apoptosis to the needs of different tissue types during development and in the adult for homeostasis and tissue repair. Diverse modes of regulation may also reflect the importance of having adequate checkpoints to limit proliferation. Bypassing apoptosis and negative growth regulatory signals are important steps along the path to cancer (Hanahan and Weinberg, 2011).

Quantitative RT-PCR
Total RNA was extracted from S2 cells or wing imaginal discs and treated with DNase-1 to eliminate genomic DNA contamination. Reverse transcription to synthesize the first strand used oligo-dT primers and Superscript RT-III (Invitrogen). PCR was performed and analyzed on Applied Biosystems 7500 fast real-time PCR system. The following primers were used: yki-f,  (Zhang et al., 2011). Data were normalized at least to the two having similar behaviors.
For microRNA quantification, reverse transcription and PCR were performed using TaqManH MicroRNA Assays from Applied Biosystems.

Immunostaining and microscopy
Wandering 3rd instar larvae were collected and dissected. Tissues were fixed in PBS with 4% paraformaldehyde at room temperature for 20 min, then rinsed and washed in PBST (PBS+0.05% Triton X-100) before blocked in PBST+5% BSA. Anti-GFP and Anti-Gal4 were incubated at 4˚C overnight. Secondary antibodies were incubated at room temperature for 2 hrs with DAPI. Wing imaginal discs were mounted and imaged using a Zeiss LSM700 confocal microscope. levels. Histogram at right shows quantification of the effects on growth of the ptc-Gal4 expression domain. Error bars represent standard deviation from measurement of at least 8 wings for each genotype. *** indicates statistically significant increase in the width of the ptc-Gal4 expression domain when miR-2 cluster level was increased (P,0.001). (B) Photomicrographs of adult wings of the indicated genotype, as in panel A. Lower panels: flies expressing a UAS-miR-11 transgene to increase miR-11 level in the ptc-Gal4 domain. Histogram shows quantification of the effects of the UAS-miR-11 transgene alone and together with UAS-yki RNAi . Error bars represent standard deviation from at least 6 wings for each genotype. *** indicates statistically significant increase in the width of the ptc-Gal4 expression domain when miR-11 level was increased (P,0.001). As a member of the miR-2 seed family, miR-11 is expected to regulate the same targets as miR-2. Regulation of reaper by miR-11 has been confirmed in the embryo by Leaman et al. and Ge et al. (Leaman et al., 2005;Ge et al., 2012). (C) Histogram showing the levels of miR-2 and miR-13 measured by quantitative RT-PCR. Wandering 3 rd instar larvae expressed UAS-yki RNAi under ubiquitous tubulin-Gal4 control. Control (white bars) expressed tubulin-Gal4 without the UAS-RNAi transgene. Total RNA from whole larvae was extracted and normalized for cDNA synthesis. RNA levels were normalized to small RNA U14. Error bars represent standard deviation from 3 independent experiments. Student's t-test for miR-2 vs U14: *P,0.05, **P,0.01. (D) Yki acts at multiple levels to control apoptosis. Arrows indicate activation, whereas bar-ended lines indicate inhibitory interactions.