Abstract
Tissue growth is regulated by many signals, including polarity cues. The Hippo signalling pathway restricts tissue growth and receives inputs from the planar cell polarity-controlling Fat signalling pathway. The atypical cadherin Fat restricts growth via several mechanisms that ultimately control the activity of the pro-growth transcriptional co-activator Yorkie. Fat signalling activates the Yorkie inhibitory kinase Warts, and modulates the function of the FERM protein Expanded, which promotes Hippo signalling and also directly inhibits Yorkie. Although several Fat pathway activity modulators are known to be involved in ubiquitylation, the role of this post-translational modification in the pathway remains unclear. Moreover, no deubiquitylating enzymes have been described in this pathway. Here, using in vivo RNAi screening, we identify the deubiquitylating enzyme Fat facets as a positive regulator of Fat signalling with roles in tissue growth control. Fat facets interacts genetically and physically with Fat signalling components and regulates Yorkie target gene expression. Thus, we uncover a role for reversible ubiquitylation in the control of Fat signalling and tissue growth regulation.
Similar content being viewed by others
Introduction
Developmental tissue growth and morphogenesis are controlled by a plethora of molecular mechanisms that must be tightly regulated to achieve reproducible organ and body size. In epithelia, one of the most important pathways involved in tissue growth regulation is the Hippo (Hpo) pathway, an evolutionarily conserved signalling cascade that integrates multiple signals that report on epithelial integrity1,2,3,4. Hpo signalling culminates in the inhibition of Yorkie (Yki; mammalian YAP), a transcriptional co-activator that associates with transcription factors such as Scalloped (Sd; mammalian TEAD1-4) to promote the expression of genes involved in cell proliferation and inhibition of apoptosis1,5,6. Yki activity is restrained by a kinase cascade consisting of the kinases Hpo and Warts (Wts; mammalian LATS1/2). The latter directly phosphorylates Yki, inhibiting its nuclear translocation, primarily by promoting interaction with 14-3-3 proteins5,7. Given its crucial role in tissue growth control, homoeostasis is maintained by tight regulation of Hpo signalling, including a negative feedback loop in which Yki/YAP promote the expression of upstream activators of the kinase cascade8,9,10.
Among the signals that regulate Hpo signalling are inputs from the cellular polarity machinery3,4,9. Hpo signalling is regulated both by proteins involved in establishing and maintaining apico-basal polarity (e.g., Crumbs (Crb), Scribble (Scrib), among others)4,9, and planar cell polarity (PCP), such as the members of the Fat (Ft; mammalian FAT1-4) signalling pathway11,12. Ft is an atypical cadherin that localises to the sub-apical domain of epithelial cells and forms an heterotypic adhesion complex with the atypical cadherin Dachsous (Ds; mammalian DCHS1/2)11, which is regulated by the Golgi resident kinase Four-jointed (Fj; mammalian FJX1)13. The combination of the opposing Ds and Fj expression patterns in tissues, and the differential effect of Fj-mediated phosphorylation on the affinity of Ft and Ds to each other results in a gradient of Ft signalling that contributes to the regulation of PCP and tissue growth11,14,15,16.
Ft-mediated regulation of growth involves several mechanisms. Ft inhibits the function of the atypical myosin Dachs (D)17,18,19, a known negative regulator of Wts function, albeit the precise molecular mechanisms remain unclear17,20,21. Ft also limits the activity of the zDHHC9-like transmembrane palmitoyltransferase Approximated (App)22,23 and the D-interacting protein Dachs ligand with SH3s (Dlish), which control D sub-cellular localisation and function24,25. Moreover, Dlish also inhibits Hpo signalling in a D-independent manner, via the regulation of the upstream activator Expanded (Ex)26. Dlish interacts with Ex and promotes its degradation via the recruitment of Skp-Cullin-F-box (SCF) E3 ubiquitin ligase complexes containing the F-box protein Slimb, a known regulator of Ex function27,28,29.
Interestingly, besides Dlish-mediated regulation of Ex stability, several steps of the Ft signalling pathway appear to be regulated by post-translational modifications such as ubiquitylation. D is thought to regulate Wts protein levels by an unknown mechanism that is likely to involve ubiquitylation17,20. In addition, Ft-mediated regulation of D function is at least partly dependent on the F-box protein Fbxl7, though it is still unclear whether D itself is ubiquitylated and degraded30,31. Finally, a recent report identified the E3 ligase Early girl (Elgi) as a new Ft signalling component involved in tissue growth regulation32. Elgi is a D-interacting protein that controls D protein levels and, along with App, is proposed to control D and Dlish localisation to the apical membrane32. Despite these observations, the precise molecular mechanisms by which ubiquitylation regulates Ft signalling and, by extension, tissue growth remain incompletely characterised. Importantly, to date there have been no reports of deubiquitylating enzymes (DUBs) as potential regulators of Ft signalling components.
To address this, we performed an in vivo RNAi modifier screen to uncover DUBs involved in Ft signalling and identified Fat facets (Faf; mammalian USP9X) as a regulator of tissue growth. In Drosophila, faf has not previously been connected to Hippo signalling. However, various studies have shown that Faf regulates Drosophila eye and embryonic development via the deubiquitylation and/or stabilisation of targets such as Liquid facets (Lqf)33, D-Jun34, Medea35, and Dscam136. Here, we show that Faf genetically and physically interacts with Ft signalling components, and controls expression of Yki target genes. Therefore, the function of Faf illustrates the crucial role of ubiquitylation in the regulation of Ft signalling and tissue growth.
Results
Identification of Faf as a deubiquitylating enzyme involved in Ft signalling
Previous studies have identified E3 ubiquitin ligases involved in the regulation of Ft signalling, such as Elgi and a Fbxl7-containing SCF complex30,31,32. However, to date, a role for deubiquitylating enzymes (DUBs) in Ft signalling regulation has not been described. To identify DUBs that modulate Ft function, we performed an in vivo RNAi modifier screen using the Drosophila adult wing as a model (Figure S1a). We used the wing driver nub-Gal4 (nub > ) to target the full complement of Drosophila DUBs with UAS-RNAi transgenes. Each DUBRNAi line was crossed to nub-Gal4, UAS-ft (nub > ft) or nub-Gal4 (nub > ) as a control. Figures S1b and S1c show the in vivo screening results and quantification of the relative wing size of the different genotypes tested. In agreement with its effect on Hpo and PCP signalling, UAS-ft expression in the wing pouch resulted in smaller and rounder wings (Figs. 1d, j, k and s2d) compared with controls (Figs. 1a, j, k,S1d, S2a, S2c and S2d).
a–i Modulation of Faf expression affects tissue growth during normal development or in conditions when Ft is over-expressed or depleted. Shown are adult wings from flies raised at 25 °C expressing the indicated transgenes in the wing pouch under the control of nub-Gal4 (nub > ). Compared to control adult wings expressing GFP (a), Ft-expressing wings were smaller (d), while ft depletion caused increased growth (g). Depletion of faf mildly enhanced tissue growth (b), while faf over-expression resulted in undergrowth (c). faf depletion resulted in a partial rescue of the undergrowth phenotype caused by UAS-ft (e), while it enhanced the overgrowth phenotype of ftRNAi flies (h). In contrast, UAS-faf enhanced the growth impairment of UAS-ft flies (f) and mildly suppressed the overgrowth phenotype of ftRNAi flies (i). j Quantification of relative adult wing sizes from flies expressing the indicated transgenes under the control of nub-Gal4. Data are represented as % of the average wing area of the respective controls (nub > GFP, average set to 100%). Data are shown as average ± standard deviation, with all data points depicted. (n = 22, 25, 29, 20, 31, 26, 27, 29, 25, 23, 15, 47 and 19). Significance was assessed using Brown-Forsythe and Welch one-way ANOVA analysis comparing all genotypes to their respective controls (nub > GFP, nub > ft or nub > ftRNAi; black, blue or red asterisks, respectively) with Dunnett’s multiple comparisons test. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ns, non-significant. Scale bar: 500 µm. k Quantification of wing shape. Data is represented as the ratio between the length of the PD axis and the length of the AP axis. All data is represented as average ± standard deviation, with all data points depicted. Vertical dashed lines separate different genotype conditions (nub > , nub > ft or nub > ftRNAi). Significance was assessed using Kruskal-Wallis ANOVA analyses comparing all genotypes to the respective control (nub > GFP, nub>ft or nub>ftRNAi; black, blue or red asterisks, respectively), with Dunn’s multiple comparisons test. *, p < 0.05; **, p < 0.01; ***, p < 0.001. ns, non-significant. (n = 22, 25, 29, 20, 31, 26, 27, 29, 25, 23, 25, 47 and 19).
As a result of our screening approach, we identified the DUB Fat facets (Faf, encoded by faf, CG1945) as a potential regulator of Ft signalling. faf depletion using several independent RNAi lines resulted in a partial suppression of the Ft undergrowth phenotype (Figs. 1e, j and S1c). Interestingly, Faf seems to primarily affect the tissue growth function of Ft, but not its PCP function, based on the ratio of the lengths of the anterior-posterior (AP) and proximal-distal (PD) axes of the adult wing, wing circularity and the orientation of adult wing hairs (Figs. 1k, S2b-d and S2i; see Materials and Methods for details). Next, we sought to determine if modulation of faf expression alone affects tissue growth and to validate its interaction with Ft signalling. faf depletion in the developing wing using nub-Gal4 resulted in a mild increase in wing size, when compared to controls (Figs. 1j and S1e–g). Conversely, over-expression of faf (Fafisoform C, fafisoC37; Fig. 2a) reduced wing size (Figs. 1c, j and S1h). When combined with UAS-ft, depletion of faf suppressed the Ft phenotype (Fig. 1e and j), while faf over-expression enhanced it (Fig. 1f and j). In contrast, simultaneous depletion of ft and faf caused an enhancement of the ftRNAi phenotype (Fig. 1g, h and j), while over-expression of faf in the context of ftRNAi partially suppressed the wing overgrowth phenotype (Fig. 1g, i and j). Interestingly, we observed that modulation of Faf levels resulted in a very specific defect of the L2 wing vein, with the appearance of extra wing material in fafRNAi wings (Figs. 1b and S2f-h). Notably, this phenotype has previously been associated with changes in Ft levels38,39 and was also observed when Ft levels were modulated in isolation (Fig. 1d and g). Importantly, modulating Faf levels enhanced the L2 wing vein phenotypes (Figure S2h). Our results suggest that Faf promotes Ft activity in tissue growth control.
a Schematic representation of the faf locus. mRNA transcripts encoded in the faf locus are represented (boxes depict exons, while introns are shown as black lines. Coding sequences are shown as blue boxes, and UTRs are represented as grey boxes). Mutant alleles used are mapped onto the faf locus. faf B3 corresponds to a nonsense mutation leading to a truncated Faf protein (Q71 > STOP). The faf BX4 lesion is the site of an inversion breakpoint. faf F08 is a point mutation in the DUB catalytic domain (H1986 > Y). faf BX3 corresponds to a 15 bp deletion in the vicinity of an exon-intron boundary. b–g Loss of one copy of faf enhances the overgrowth of ft mutants. Shown are tiled confocal images of wing imaginal discs. Compared to control L3 imaginal discs (wiso, b), the trans-heterozygous combination of ftG-rv and ft8 leads to a dramatic increase in tissue size (c), which is further enhanced by the presence of a faf mutant allele (faf F08 (d), faf BX4 (e), faf B3 (f), faf BX3 (g)). h Quantification of relative wing imaginal disc size. Data are represented as % of the average imaginal wing disc area of controls (wiso, which was set to 100% and indicated by the black dashed line). Red dashed line indicates average wing disc size of ft trans-heterozygous mutants. Data are shown as average ± standard deviation, with all data points depicted. (n = 11, 17, 20, 8, 13 and 9). Significance was assessed using Brown-Forsythe and Welch one-way ANOVA analyses comparing all genotypes to the respective control (wiso or ftG-rv/ft8; black or blue asterisks, respectively), with Dunnett’s multiple comparisons test. *, p < 0.05; **, p < 0.01; ***, p < 0.001. Scale bar: 100 µm.
To validate the results obtained with fafRNAi, we next assessed if faf genetically interacts with ft mutations in the regulation of tissue growth. Due to its critical role in controlling Hpo signalling, ft homozygous mutations are lethal and associated with extreme tissue overgrowth phenotypes40,41. However, certain ft mutations allow the analysis of tissues from late L3 larvae in a trans-heterozygous situation, such as the ftG-rv/ft8 combination, allowing them to be studied alongside other genetic alterations15,42. ftG-rv/ft8 trans-heterozygous mutant wing discs displayed extreme tissue overgrowth, compared to wild-type (wiso) wing discs (~170% larger than controls; compare Fig. 2b and c). Remarkably, when we combined ftG-rv/ft8 mutations with various faf mutant alleles (Fig. 2a), we observed an enhancement of the tissue overgrowth phenotype (Fig. 2d-h). This indicates that faf genetically interacts with ft and that Faf function is important to restrict tissue growth. Given that faf mutation enhances ft mutant phenotypes, it is possible that Faf controls growth in both Ft-dependent and Ft-independent manners. Alternatively, Faf may act on residual Ft protein in ftG-rv/ft8 trans-heterozygotes. We assessed this latter possibility by monitoring Ft protein levels in wing discs from control flies (Figure S3a), ftG-rv/ft8 trans-heterozygotes (Figure S3b) and ftG-rv/ft8 trans-heterozygotes with loss of a copy of faf (fafB3, Figure S3c). ft mutation leads to a dramatic reduction in Ft protein levels and, in the absence of faf, Ft localisation at the cell periphery appears further reduced. Taken together, our data in the Drosophila wing are consistent with Faf promoting the tissue growth suppressing function of Ft.
Faf genetically interacts with core Ft signalling proteins
Having observed that Faf genetically interacts with Ft and plays an important role in the regulation of tissue growth, we extended our analysis to other members of the Ft signalling pathway that directly interact with Ft, such as Ds, Dlish and Fbxl7 (Fig. 3). Ft and Ds regulate tissue growth at least in part via their physical interaction across cell boundaries11. This interaction enhances the activity of Ft, thereby promoting its growth-suppressing function11. Accordingly, we found that Ds over-expression caused a reduction in wing size (Fig. 3b and p). Similarly to what was observed with Ft, fafRNAi reversed this Ds-induced phenotype (Fig. 3c–e and p). This is consistent with a positive role for Faf in Ft signalling.
a–o Modulation of Faf expression levels affects the tissue growth phenotypes of ds- (b–f), DlishRNAi- (g–k) and Fbxl7-expressing flies (l–o). Shown are adult wings from flies raised at 25°C expressing the indicated transgenes in the wing pouch under the control of nub-Gal4 (nub > ). Compared to control adult wings (a), wings expressing ds displayed significant tissue undergrowth (b), which was partially rescued by simultaneous depletion of faf (c, d and e) and enhanced by faf over-expression (f). RNAi-mediated depletion of Dlish (g) caused a mild reduction in wing size, which was partially suppressed by co-depletion of faf (h), and enhanced by faf over-expression (i). Expression of ft resulted in an enhancement of the undergrowth of DlishRNAi wings (j), and this was in part prevented by fafRNAi (k). Fbxl7 over-expression caused reduced wing size compared to controls (l) and this was not significantly affected by fafRNAi (m) or ft expression (n), but this was partially suppressed when ft was combined with fafRNAi (o). p–r Quantification of relative adult wing sizes in genetic interactions with ds (p), Dlish (q) or Fbxl7 (r). Data are represented as % of the average wing area of control wings (nub > GFP, which were set as 100%). Data are shown as average ± standard deviation, with all data points represented. (n = 23, 25, 18, 30, 30 and 18 for (p); n = 20, 28, 26, 19, 29 and 28 for (q); and n = 23, 28, 30, 25 and 27 for (r). Black dashed lines represent average size of controls (100%) whilst red dashed lines indicate average size of adult wings from flies expressing ds, DlishRNAi or Fbxl7 under the control of nub-Gal4. Significance was assessed using a one-way ANOVA comparing all genotypes to their respective controls (nub > GFP, nub > GFP + UAS-Ds, nub > GFP+DlishRNAi or nub > GFP + UAS-Fbxl7; black, green, brown or blue asterisks, respectively), with Dunnett’s multiple comparisons test. **, p < 0.01; ***, p < 0.001. n.s. non-significant. Scale bar: 500 µm.
We also assessed if Faf could modulate phenotypes caused by downstream effectors of Ft signalling, such as Dlish and Fbxl724,25,30,31. Dlish negatively regulates Hippo signalling and, therefore, promotes tissue growth24,25,26. Accordingly, depletion of Dlish (DlishRNAi) in the developing wing resulted in reduced tissue growth (Fig. 3g and q). Co-depletion of faf and Dlish suppressed this phenotype (Fig. 3h and q), whilst Faf over-expression enhanced the undergrowth (Fig. 3i and q). As Dlish function is influenced by Ft, we combined DlishRNAi with Ft over-expression, which resulted in an enhancement of the DlishRNAi-induced undergrowth (Fig. 3j and q). To address if this is dependent on Faf activity, we co-depleted faf in these conditions (DlishRNAi + ft over-expression), which led to a rescue of the phenotype (Fig. 3k and q), suggesting that Faf does indeed modulate Dlish phenotypes via Ft. As previously observed, fafRNAi mostly affected wing size rather than shape, as wings remained rounder than controls (Fig. 3k), further indicating that Faf has a minor role in Ft-mediated regulation of tissue shape.
In agreement with its previously reported role, over-expression of Fbxl7 leads to a significant reduction in tissue size30,31 (Fig. 3l and r). Interestingly, depletion of faf did not significantly affect the Fbxl7 phenotype, suggesting that Faf may act at the level of Fbxl7 (Fig. 3m and r). Moreover, Ft over-expression failed to enhance the Fbxl7 undergrowth phenotype (Fig. 3n and r), which is consistent with Fbxl7 playing a crucial role downstream of Ft. fafRNAi only had a slight effect on tissue growth in these conditions (Fig. 3o and r). We then extended our observations to additional Ft-associated proteins, such as Ex, Elgi and App (Figure S4). Our results suggest that Faf modifies phenotypes associated with Ex (Figure S4b-e and S4p) and Elgi (Figure S4f-j and S3q), but not with App (Figure S3k-o and S3r). Taken together, these results suggest that Faf genetically interacts with multiple Ft signalling components and may work at the level of Fbxl7 and App, and potentially antagonistically to Dlish.
To position Faf within the Hpo pathway, we additionally tested whether modulation of Faf levels could enhance or suppress phenotypes elicited by the Hpo and Wts kinases. We combined Faf expression and fafRNAi with depletion of hpo (hpoRNAi, Figure S5a–d) or over-expression of wts (Figure S5e–k). Faf did not alter the hpoRNAi overgrowth phenotype in the adult wing, suggesting that Hpo acts downstream of Faf (Figure S5d). Additionally, while expression of faf resulted in an enhancement of the undergrowth phenotype elicited by wts expression, faf depletion had no effect, suggesting that Faf acts upstream of Wts (Figure S5k), in agreement with our observations regarding the Ft signalling components.
Faf physically interacts with Ft and regulates its protein levels
Next, we assessed whether the genetic interactions between faf and members of the Ft signalling pathway could be the result of specific protein-protein interactions. For this, we expressed Faf in Drosophila S2 cells (Fig. 4a; FafLD, a protein encoded by the cDNA clone LD22582) and performed co-immunoprecipitation (co-IP) assays with the intracellular region of Ft (FtICD43). FtICD was readily detected in FafLD co-IPs, but not in the respective GFP controls (Figs. 4b and S6a). In agreement with the widespread role of DUBs as regulators of protein stability, we noticed that the protein levels of FtICD in cell lysates appeared higher when FtICD was co-expressed with FafLD (Fig. 4b). Therefore, we conducted further experiments using FafLD expression or faf RNAi-mediated depletion to validate this observation. Indeed, FtICD was stabilised in the presence of FafLD (Fig. 4c and S6b) and, conversely, FtICD levels were reduced when endogenous faf was depleted from S2 cells (Fig. 4d, S6c and S6d). This suggests that Faf may regulate Hippo signalling and tissue growth by modulating Ft protein levels.
a Schematic representation of the Faf constructs used in this study. Numbers denote amino acid position. USP DUB domain is represented in green. b Faf binds to Ft. HA-tagged FtICD was co-expressed with FLAG-tagged GFP or FafLD in Drosophila S2 cells. Cells were lysed and lysates were subjected to co-immunoprecipitation using FLAG agarose beads. Lysates were analysed by immunoblot using the indicated antibodies for detection of protein expression and co-purification. Tubulin (Tub) was used as loading control. (n = 3 independent experiments). c, d Faf regulates Ft protein levels in Drosophila S2 cells. c HA-tagged FtICD was expressed in S2 cells in the presence or absence of FafLD. 48 h after transfection, cells were lysed and lysates were analysed by immunoblot using the indicated antibodies. Ø represents expression of empty vector. d S2 cells were treated with the indicated dsRNAs 24 h before co-transfection with the indicated constructs. FtICD protein levels were analysed by Western blotting with the indicated antibodies 48 h after cell transfection. GFP and Tubulin (Tub) were used as transfection and loading control, respectively. (n = 3 independent experiments). e–h Faf regulates Ft protein levels in vivo. Shown are XY confocal micrographs of third instar wing imaginal discs expressing the indicated constructs under the control of hh-Gal4, showing Ft antibody staining (e-h; red in e”-h”), Arm antibody staining (e’-h’; cyan in e”-h”), and direct fluorescence from GFP (green in e”-h”). Compared to the controls (e, lacZRNAi), faf depletion (f) and faf over-expression (g) resulted in a decrease or increase in Ft protein levels, respectively. Shown is also ftRNAi control (h) to validate antibody specificity. Ventral is up in XY sections, whilst GFP marks the hh-Gal4-expressing posterior compartment (right). Dashed white line depicts boundary between anterior and posterior compartments. (n = 11, 14, 12, 8). Scale bar: 50 µm.
To validate these observations in vivo, we modulated Ft or Faf expression in the posterior compartment of the wing disc (using hh-Gal4) and assessed the effect on Ft protein levels using a specific antibody19,43,44 (Fig. 4h). As a control for general effects on the levels of proteins localised at the apical cell surface, we monitored Armadillo (Arm; Drosophila β-catenin) protein levels. Altering Faf levels in vivo recapitulated the results observed in S2 cells, and fafRNAi expression resulted in a reduction in Ft levels (Figs. 4e, f, S7a, S7b, S7d and S7e), whilst over-expression of Faf increased Ft protein levels (Figs. 4e, g and S7d and S7e). Importantly, the effect of Faf on Ft levels was specific as Arm levels were generally unaffected (Fig. 4e’-h’, Figure S7c). Using mirr-Gal4 to control gene expression, we also observed that Ft levels were regulated by Faf in the developing eye imaginal disc (Figure S7f–i). Expression of faf resulted in increased levels of Ft as assessed using the Ft-specific antibody (Figure S7f and S7h). In this tissue, Ft levels appear to be more susceptible to increased levels of Faf, as faf depletion had no effect (Figure S7f and S7i). Together, these data suggest that Faf regulates Ft protein levels both in Drosophila S2 cells and in vivo.
Effect of Faf on Dachs subcellular localisation
Next, we tested whether Faf regulates events downstream of Ft and focused on a potential modulation of D function. As extensively documented, D is one of the most important effectors of Ft signalling15,17,18, and Ft regulates D by controlling its subcellular localisation18. To assess D localisation, we used a D::GFP knock-in allele in combination with en-Gal4, UAS-RFP. This allowed us to monitor D subcellular localisation in the posterior compartment of the developing wing disc (marked by RFP) and use the anterior compartment as a control (Figure S8a). Additionally, Arm was used as a control for global effects on apical protein localisation (Figure S8h-m). As seen in Figure S8b, in control flies (lacZRNAi), D is localised at the membrane, similarly to Arm, and the stereotypical D polarisation toward the distal side of the cell can be observed with no major differences seen between the anterior and posterior compartments. We also assessed the effects on D localisation by monitoring the plot profile of both D and Arm (Figure S8b”). In agreement with published data18,20,22,30,31,45,46, over-expression of Ft resulted in D mislocalisation, which appeared both more cytoplasmic and less polarised at the membrane, compared to controls (Figure S8c). In agreement with our previous results, combining Ft over-expression with faf depletion (UAS-ft, UAS-fafRNAi) resulted in a return to control conditions (Figure S8d). In contrast, fafisoC expression resulted in increased levels of cytoplasmic D and, in some cases, we observed patches of tissue where D was not localised at the membrane, a phenotype similar to Ft over-expression (Figure S8e)18. Depletion of ft or faf resulted in less apparent D polarisation (Figure S8f and S8g). Importantly, no changes were observed at the level of Arm localisation in any condition (Figure S8h-m). Together, these results suggest that Faf is, at least partly, required for regulation of signalling events downstream of Ft.
Faf regulates Yki target genes in vivo
Our data suggest that Faf regulates tissue growth by modulating Ft function. To confirm that the effects of Faf on tissue growth were indeed due to changes in signalling activity downstream of Ft, we tested whether Faf could affect gene expression modulated by Yki, the main effector protein regulated by the Hippo pathway. For this, we monitored Yki-mediated transcription using as readouts two widely used Hippo signalling in vivo reporters; ex-lacZ47 (Fig. 5a-f) and HRE-diap1::GFP48 (Fig. 5g-l). Transgenes were specifically expressed in the posterior compartment of the wing using en-Gal4 and marked by the expression of GFP or RFP, respectively in ex-lacZ or DIAP1::GFP experiments. As a control, we used hpoRNAi, which is known to lead to increased Yki-mediated gene expression7,49 (Fig. 5b, f, h and l). In agreement with our observations regarding the effect of Faf on tissue growth, fafRNAi caused an increase in Yki-mediated transcription, consistent with decreased Hippo signalling activity (Fig. 5c, f, i and l). Accordingly, faf over-expression resulted in decreased Yki activity (Fig. 6a, b and e). We also used the Yki-mediated transcriptional readout to determine whether Faf is acting through Ft to produce these effects. Over-expression of Ft in the posterior compartment of the wing resulted in a significant decrease in Yki-mediated transcription, consistent with its role in activating Hippo signalling (Fig. 5d, 5f, j and l). Interestingly, depletion of faf completely abrogated the effect of Ft over-expression on Yki target gene expression (Fig. 5e, f, k and l). Indeed, the expression levels of the ex-lacZ and DIAP1::GFP reporters were significantly different in the UAS-ft, UAS-fafRNAi condition compared to when Ft was overexpressed in isolation, and were closer to levels seen in controls. Together, this data suggests that Faf promotes tissue growth by positively regulating Hippo signalling and that it does so by modulating the function of Ft.
a–e Regulation of ex-lacZ expression. XY confocal sections of third instar wing imaginal discs containing ex-lacZ (a-e; red in a’-e’ merged images), in which en-Gal4 was used to drive expression of UAS-GFP (a), UAS-hpoRNAi (b), UAS-fafRNAi (c), UAS-ft (d) or UAS-ft and UAS-fafRNAi (e). GFP (green in a’-e’ merged images) indicates posterior compartment where transgenes are expressed. DAPI (blue) stains nuclei. Dashed lines indicate anterior-posterior compartment boundary. f Quantification of ex-lacZ expression levels. Shown are the posterior/anterior (P/A) ex-lacZ ratios for the different genotypes analysed. Data are shown as average ± standard deviation, with all data points represented. (n = 11, 13, 12, 12 and 22). g–k Regulation of DIAP1::GFP expression. XY confocal sections of third instar wing imaginal discs carrying DIAP1::GFP (g-k, shown in green in g’-k’ merged images), in which en-Gal4 was used to drive expression of UAS-lacZRNAi (g), UAS-hpoRNAi (h), UAS-fafRNAi (i), UAS-ft (j) or UAS-ft and UAS-fafRNAi (k). RFP (red in g’-k’ merged images) indicates posterior compartment where transgenes are expressed. DAPI (blue) stains nuclei. Dashed lines indicate boundary between anterior and posterior compartments. l Quantification of DIAP1::GFP expression levels. Shown are posterior/anterior (P/A) DIAP1::GFP ratios for the indicated genotypes. Data are shown as average ± standard deviation, with all data points represented. (n = 12, 12, 15, 12 and 13). Significance was assessed using a one-way ANOVA comparing all genotypes to controls (UAS-GFP in (f) and UAS-lacZRNAi in (l)), with Dunnett’s multiple comparisons test. Pairwise comparisons between UAS-ft and UAS-ft; UAS-fafRNAi (red) were performed using unpaired two-tailed t-test with Welch’s correction. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ns, non-significant. Scale bar: 50 µm.
a–d Faf regulates Yki target gene expression in a DUB-dependent manner. XY confocal sections of third instar wing imaginal discs carrying ex-lacZ (a–d, shown in red in a’–d’ merged images), in which en-Gal4 was used to drive expression of GFP (a), fafisoC (b), fafWT (c) or fafCD (d). GFP (green in a’-d’ merged images) indicates posterior compartment where transgenes are expressed. DAPI (blue) stains nuclei. Dashed lines indicate anterior-posterior compartment boundary. e Quantification of ex-lacZ expression levels. Shown are the posterior/anterior (P/A) ex-lacZ ratios for the different genotypes analysed. Data are shown as average ± standard deviation, with all data points represented. (n = 9, 13, 20 and 18). Significance was assessed using a one-way ANOVA comparing all genotypes to UAS-GFP, with Dunnett’s multiple comparisons test. *, p < 0.05; ***, p < 0.001; ns, non-significant. Scale bar: 50 µm. f–t Effect of Faf on Ft-mediated regulation of growth is DUB-dependent. Shown are adult wings from flies raised at 25°C expressing transgenes under nub-Gal4 control. (j), (o), (t) Quantification of effect of Faf on tissue growth. Shown are relative adult wing sizes from flies expressing the indicated transgenes under nub-Gal4 control. Data are represented as % of the average wing area of controls (nub > GFP, set to 100%). Data are shown as average ± standard deviation, with all data points depicted. (n = 22, 18, 30, 26 for j; n = 22, 21, 22, 29, 32 for o; and n = 22, 19, 19, 29 and 25 for t). Significance was assessed using a one-way ANOVA analysis comparing all genotypes to nub > GFP (black asterisks), to nub > UAS-ft (green asterisks), or to nub > UAS-ftRNAi (blue asterisks) with Dunnett’s multiple comparisons test. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ns, non-significant. Scale bar: 500 µm.
The effects of Faf are dependent on its catalytic activity
Faf is part of the DUB family of proteins, enzymes that regulate ubiquitylation levels and counteract the action of E3 ubiquitin ligases in cells50. Given that Faf is predicted to be an active DUB, we next tested whether the role of Faf in tissue growth and modulation of Ft and Hippo signalling events was dependent on its catalytic activity. To this end, we used previously generated UAS-regulated transgenes encoding either WT or a catalytically inactive form of Faf (respectively, FafWT and FafCD)36 inserted at the same genomic location. First, we assessed the role of the catalytic activity of Faf in the regulation of Yki-dependent transcription using ex-lacZ (Fig. 6a–e) and HRE-DIAP1::GFP (Figure S9a–e). Over-expression of Faf resulted in a reduction in ex-lacZ levels when compared to the UAS-GFP control (Fig. 6a–c and e). Similar results were obtained when the DIAP1::GFP reporter was analysed (Figures S9a–c and S9e). These results confirm the effect of Faf on tissue growth and support the hypothesis that, at least in part, the role of Faf involves the regulation of Hippo signalling activity. Interestingly, the catalytic mutant version of Faf (fafCD) had no effect on either the ex-lacZ (Fig. 6d and e) or the DIAP1::GFP reporters (Figure S9d and S9e). This strongly suggests that the role of Faf in the regulation of tissue growth and Hippo signalling is dependent on its catalytic activity and DUB function, as opposed to a potential role as a scaffolding protein bridging protein-protein interactions.
To further validate these observations, we assessed other Ft-related phenotypes, such as the regulation of D subcellular localisation. As shown in Figures S8 and S9, Faf over-expression affected D membrane localisation and increased D cytoplasmic levels. These results were recapitulated when we overexpressed fafWT in the developing wing disc (Figure S9g). Compared to the respective anterior compartment and controls (lacZRNAi), tissues where fafWT was overexpressed had D subcellular localisation defects. In contrast, when we overexpressed the fafCD mutant, we observed no overt effects on the levels or subcellular localisation of D (Figure S9h). As before, no changes in Arm localisation were observed (Figure S9i–k). Again, this suggests that the effect of Faf on Ft signalling events is dependent on its catalytic activity.
We also directly assessed the role of Faf DUB activity in the regulation of tissue growth and on the genetic interactions with Ft. For this, we expressed FafisoC, FafWT and FafCD in the wing pouch using nub-Gal4 and measured wing size as a proxy for the effects of Faf on tissue growth. When tested in isolation, as expected, we observed phenotypes consistent with the proposed role of Faf in the regulation of tissue growth (Fig. 6f-j). Adult wings from animals expressing UAS-fafisoC (Fig. 6g) or UAS-fafWT (Fig. 6h) were significantly smaller than those of the controls (nub > GFP, Fig. 6f and j). Interestingly, the phenotype elicited by UAS-fafWT was more severe than UAS-faf, which can be explained by the fact that the transgenes have different genomic locations and, therefore, different expression levels. Indeed, RT-PCR experiments revealed increased expression of faf in larvae expressing fafisoC and this was enhanced in fafWT-expressing flies (Figure S9l). Importantly, when UAS-fafCD was expressed, we did not observe tissue growth restriction but, instead, the adult wings were slightly larger than controls (Fig. 6i). This suggests that the catalytic activity of Faf is required for its effect on tissue growth. Moreover, the mild overgrowth of FafCD wings raises the possibility that this catalytically inactive allele may be acting as a mild dominant negative version of Faf.
Next, we assessed genetic interactions between Faf and Ft by expressing the Faf transgenes in the presence of Ft over-expression (Fig. 6k–o) or Ft RNAi-mediated depletion (Fig. 6p–t). As previously shown, depletion of ft resulted in overgrowth phenotypes (Fig. 6p). In agreement with our previous results, expression of UAS-fafisoC in these conditions abrogated this overgrowth phenotype (Fig. 6q). Similarly, expression of UAS-fafWT blocked the overgrowth caused by depletion of ft and, in fact, caused a significant undergrowth phenotype (Fig. 6r). Again, in agreement with our hypothesis, expression of the inactive form of Faf (UAS-fafCD) did not modify the phenotype of ftRNAi flies and the adult wing sizes were indistinguishable from those of controls (nub > ftRNAi, GFP) (Fig. 6s and t). We validated these observations by assessing the effect of co-expression of Ft and Faf (Fig. 6k–o). Ft over-expression caused a significant undergrowth phenotype when compared to controls (Fig. 6k and f), which was enhanced by co-expression of either UAS-fafisoC (Fig. 6l) or UAS-fafWT (Fig. 6m). In contrast, expression of the mutant UAS-fafCD resulted in a mild rescue of the UAS-ft adult wing phenotype (Fig. 6n and o). Together, our in vivo data support our conclusion that the catalytic activity of Faf is essential for its function in the regulation of Hippo signalling and tissue growth.
Faf-mediated regulation of Ft protein is dependent on its DUB activity
Our in vivo results strongly suggest that Faf DUB activity is required for its function in tissue growth. To test if this effect was directly connected to Ft regulation, we assessed the effect of WT and mutant Faf on the protein levels of Ft, in vivo and in vitro. Firstly, we used hh-Gal4 to express the faf transgenes in the posterior compartment of the developing wing imaginal disc and monitored Ft protein levels using a Ft-specific antibody (Fig. 7a–d). We observed that, in agreement with our previous observations, expression of fafWT resulted in an increase in Ft protein levels (expressed as the ratio between the levels of Ft in the posterior and anterior compartments or normalised to the corresponding Arm levels) (Fig. 7b, d and f), whereas the expression of the fafCD catalytic mutant had no obvious effect on Ft antibody staining (Fig. 7c, d and f). Importantly, despite the fact that fafWT caused significant changes in wing disc morphology, none of the Faf transgenes affected the protein levels of Arm, which was used as a control for a global effect on apical membrane proteins (Fig. 7e).
a–c Faf regulates Ft protein levels in vivo in a DUB-dependent manner. XY confocal micrographs of third instar wing imaginal discs expressing the indicated constructs under the control of hh-Gal4, showing Ft (a–c; red in a”-c”) or Arm antibody staining (a’-c’; cyan in a”-c”), and direct GFP fluorescence (green). Ventral is up in XY sections, whilst GFP marks the hh-Gal4 compartment (right). Dashed white line depicts anterior-posterior compartment boundary. Scale bar: 50 µm. d–f Quantification of in vivo Arm and Ft protein levels. Shown are posterior/anterior (P/A) ratios for Fat (d), Arm protein levels (e), and normalised Ft protein levels ((f); normalised to Arm). Data are shown as average ± standard deviation, with all data points represented. (n = 7, 12, 23 and 18). Significance was assessed using a one-way ANOVA comparing all genotypes to the control (hh>lacZRNAi), with Dunnett’s multiple comparisons test. **, p < 0.01; ***, p < 0.001. g–j Regulation of Ft protein levels in S2 cells requires the DUB activity of Faf. g HA-tagged FtICD was co-expressed with V5-tagged Faf (WT or catalytically-inactive version, CD) in Drosophila S2 cells. Lysates were analysed by immunoblot using the indicated antibodies. Tub (Tub) was used as loading control. h Quantification of effect of Faf catalytic activity on Ft levels. Shown are relative Ft protein levels (fold change relative to controls, set to 1 in cells transfected with empty plasmid (ø); Ft levels were normalised to Tubulin) quantified from Western blot experiments. Data are represented as average ± standard deviation, with all data points represented. (n = 5 independent experiments). Significance was assessed by one-way ANOVA comparing all samples to controls (ø), with Tukey’s multiple comparisons test or an unpaired two-tailed t-test. *, p < 0.05 **, p < 0.01. i DUB inhibition affects Ft protein stability. S2 cells were treated with 5 µM of WP1130 for 6 h before lysis and immunoblot analysis with the indicated antibodies. FLAG-tagged GFP and Tub were used as transfection and loading controls, respectively. j Quantification of effect of DUB inhibition on Ft protein levels. (n = 3 independent experiments). Significance was assessed by an unpaired two-tailed t-test. *, p < 0.05.
Next, we sought to confirm these observations in vitro in Drosophila S2 cells. For that, we generated a catalytically inactive version of Faf (FafCD) in our FafLD cDNA clone and tested whether its expression modulated Ft protein levels. In contrast to WT Faf, which promoted the stabilisation of Ft and resulted in higher levels of Ft protein in cell lysates, expression of FafCD in S2 cells had a minimal effect on Ft levels (Fig. 7g and h). Importantly, as in this situation we assessed the levels of epitope-tagged Ft rather than its endogenous levels, the effects of Faf are likely due to post-translational modifications and not via the regulation of endogenous ft gene transcription and/or translation. Accordingly, we observed that treating S2 cells with the proteasome inhibitor MG132 resulted in higher Ft protein levels (Figure S10a and S10b). Notably, Faf expression stabilised Ft protein levels to such extent that the effect of MG132 was minimal (Figure S10a). As previously observed, depletion of faf resulted in destabilisation of Ft, which was partly rescued by MG132 treatment, suggesting that Ft is degraded by the proteasome when Faf is absent (Figure S10b). Given the prominent effects of modulating Faf expression on Ft protein levels and the limited effect of MG132, it is possible that Ft protein degradation involves other mechanisms beyond proteasome-mediated degradation. Our data is consistent with the notion that Faf regulates Ft in a DUB-dependent manner. We further tested this possibility by treating Drosophila S2 cells with a DUB inhibitor that affects Faf function, WP113051,52 (Fig. 7i and j). WP1130-treated cells exhibited lower levels of Ft protein than vehicle-treated cells (Fig. 7i and j). Although WP1130 is thought to inhibit other DUBs51,52, the effect on Ft levels in S2 cells appears to be due to its effect on Faf since Faf over-expression largely abrogated the effect of WP1130 and faf depletion resulted in reduced Ft levels that were only minimally affected by WP1130 treatment, suggesting that other targets for this inhibitor do not play a major role (Figure S10c). Together, our results reinforce the notion that the catalytic activity of Faf is important for its modulation of Ft protein levels and, subsequently, of Ft-mediated signalling.
Evolutionary conservation of Faf function in tissue growth regulation
Given that Faf is part of an evolutionarily conserved protein family with recognisable orthologues in other species, we next sought to determine whether the function of Faf in the regulation of Ft-mediated signalling is conserved. For this, we initially tested the effect of the mammalian orthologue of Faf, USP9X, in vivo in Drosophila tissues. Using the en-Gal4 driver, we expressed UAS-USP9X in the posterior compartment of the developing wing and assessed whether this impacted on the expression of the reporters of Yki activity, ex-lacZ or HRE-DIAP1::GFP (Fig. 8a–f). Like UAS-fafWT, expression of its mammalian paralog USP9X resulted in a reduction in the posterior/anterior (P/A) ratio of ex-lacZ and DIAP1::GFP expression (Fig. 8a–f). We also determined whether USP9X disrupted D subcellular localisation in the wing disc. Expression of USP9X resulted in a disordered D localisation in the wing epithelial cells (Figure S10d and S10e). Similarly to UAS-faf (Figure S8e’ and S8e”), we observed that cells expressing UAS-USP9X in the posterior compartment, in several instances, appeared to display D discontinuously at the apical membrane or to lack it altogether (Figure S10e’). Importantly, Arm localisation was unaffected (Figure S10f and S10g). These data suggest that, at least when overexpressed, USP9X phenocopies Faf in Drosophila tissues, supporting the hypothesis that the function of Faf in tissue growth is conserved in mammalian tissues.
a–d USP9X regulates Hpo signalling readouts in vivo. XY confocal micrographs of third instar wing imaginal discs carrying ex-lacZ (a, b) or DIAP1::GFP (c, d) and expressing GFP (a), lacZRNAi (c) or UAS-USP9X under the control of en-Gal4 (b,d), showing β-Gal antibody staining (a, b; red in a’, b’) or direct GFP fluorescence (c, d; green in c’, d’). Ventral is up in XY sections. GFP and RFP mark the hh-Gal4-expressing compartment (right) in a, b and c, d, respectively. Dashed lines depict anterior-posterior compartment boundary. Scale bar: 50 µm. e, f Quantification of Yki-mediated transcriptional reporter expression. Shown are posterior/anterior (P/A) ratios for ex-lacZ (e) and DIAP1::GFP levels (f). Data are shown as average ± standard deviation, with all data points represented. (n = 9, 12 in e, and n = 10, 13 in f). Significance was assessed using an unpaired two-tailed t-test, with Welch’s correction. ***, p < 0.001. g–o USP9X regulates tissue growth and modulates Ft-mediated growth phenotypes. Shown are adult wings from flies raised at 25°C expressing the indicated transgenes in the wing pouch under the control of nub-Gal4 (nub > ). Quantification of the effects of USP9X on growth is shown (i, l and o). Data are represented as % of the average wing area of controls (nub > GFP, average set to 100%). Data are shown as average ± standard deviation, with all data points depicted. (n = 22, 24 in I; n = 22, 21, 24 in l; and n = 22, 19, 23 in o). Significance was assessed using one-way ANOVA analysis comparing all genotypes to the nub > GFP control (black asterisks), with Dunnett’s multiple comparisons test. For pairwise comparisons, unpaired two-tailed t-tests with Welch’s correction were used. **, p < 0.01; ***, p < 0.001; ns, non-significant. Scale bar: 500 µm.
To test this idea further, we analysed tissue growth parameters in Drosophila adult wings using nub-Gal4 to express UAS-USP9X. In agreement with the effect of USP9X on Yki-mediated gene expression, expression of UAS-USP9X in the wing pouch resulted in an undergrowth phenotype (Fig. 8h), when compared with controls (Fig. 8g). Tissue growth was significantly reduced, as evidenced in Fig. 8i. We also tested whether USP9X was able to genetically interact with Ft by combining UAS-USP9X with Ft over-expression or RNAi-mediated depletion. Co-expression of ft and USP9X enhanced the undergrowth phenotype of UAS-ft wings (Fig. 8j–l), indicating that Ft downstream signalling is potentially more active in the presence of ectopic USP9X. In contrast to ft over-expression, depletion of ft (UAS-ftRNAi) resulted in enhanced tissue growth in the wing (Fig. 8m). Expression of USP9X in these conditions resulted in a suppression of the UAS-ftRNAi phenotype (Fig. 8n), and a return to WT wing tissue size (Fig. 8o). This is consistent with the proposed effect of Faf on the regulation of Ft protein levels and suggests that USP9X retains at least some of the functions of Faf in this context.
USP9X-mediated regulation of Ft is conserved
Next, we assessed whether the effects seen with USP9X over-expression in Drosophila tissues are related to its potential regulation of Ft protein levels. To test this, we first monitored Ft protein levels in the Drosophila wing imaginal disc using hh-Gal4 to control USP9X expression (Fig. 9a–e). Compared with the respective control (hh > lacZRNAi; Fig. 9a), Ft protein levels were increased in the posterior compartment of hh > USP9X wing imaginal discs (Fig. 9b, c and e), while Arm levels were unaffected (Fig. 9a’, b’ and d).
a, b USP9X regulates Ft protein levels in vivo. XY confocal micrographs of third instar wing imaginal discs expressing the indicated constructs using hh-Gal4, showing Ft (a, b; red in a”, b”), or Arm antibody staining (a’, b’; cyan in a”, b”), and direct GFP fluorescence (green in a”, b”). Note that panels a-a” are identical to Figs. 7a–7a”. Ventral is up in XY sections, whilst GFP marks the hh-Gal4-expressing compartment (right). Dashed white line depicts anterior-posterior compartment boundary. Scale bar: 50 µm. c–e Quantification of in vivo Arm and Ft protein levels. Shown are posterior/anterior (P/A) ratios for Fat (c), or Arm protein levels (d), as well as the normalised Ft protein levels (e, normalised to Arm). Data are shown as average ± standard deviation, with all data points represented. (n = 7, 13). Significance was assessed using an unpaired two-tailed t-test, with Welch’s correction. ns, non-significant; ***, p < 0.001. f, g USP9X regulates Fat4 protein levels. FLAG-tagged Fat4ΔECD was co-expressed with FLAG-tagged GFP and either empty vector (ø) or HA-tagged USP9X in HEK293 cells before immunoblot analysis using the indicated antibodies. FLAG-GFP and Tubulin were used as transfection and loading controls, respectively. g Quantification of effect of USP9X on Fat4ΔECD protein levels. Shown are relative Fat4 protein levels (fold change relative to empty plasmid (ø; set to 1); Fat4 protein levels were normalised to Tubulin) quantified from Western blot experiments. Data are represented as average ± standard deviation, with all data points represented. (n = 3 independent experiments). Significance was assessed by an unpaired two-tailed t-test. *, p < 0.05. h–j Analysis of USP9X-dependent regulation of Fat4 protein levels via immunostaining. Shown are XY confocal images (h) and insets (i) of HEK293 cells transfected with HA-tagged USP9X and GFP depicting Fat4 antibody staining (h and i; red in h” and i”) or direct fluorescence from GFP (h’ and i’; green in h” and i”). DAPI (blue in h” and i”) stains nuclei. White dashed boxes represent inset images i, i’ and i”. Yellow and white arrowheads in inset images indicate GFP-positive and GFP-negative cells, respectively. Scale bar: 50 µm (h) and 10 µm (i). j Quantification of Fat4 protein levels. Data are represented as average ± standard deviation, with all data points represented (n = 42, 54). Significance was assessed by an unpaired two-tailed t-test with Welch’s correction. ***, p < 0.001.
We next tested whether USP9X could control Ft protein levels in mammalian cells. In mammals, there are multiple genes encoding Ft cadherins, Fat1-412. However, taking into account protein homology and function, Fat4 is the closest mammalian orthologue of Ft12. Therefore, we assessed if the stabilisation of Ft by Faf could be recapitulated in a mammalian setting using the corresponding mammalian proteins, USP9X and Fat4. To this end, we used plasmids encoding different Fat4 truncations (Fat4ICD or Fat4ΔECD), alongside a plasmid encoding USP9X and expressed them in HEK293 cells. Cells were transfected with Fat4 alone or in combination with USP9X and Western blot analysis of cell lysates revealed that Fat4 levels were strongly increased in the presence of USP9X (Figs. 9f, g and S10h). Importantly, this effect was observed for both Fat4ICD and Fat4ΔECD (Fig. 9f and S10h). Given that both Fat4ICD and Fat4ΔECD largely lack the extracellular domain of Fat4, our data indicates that the effect of USP9X on the levels of Fat4 is likely to be independent of any interactions with its cognate partners that associate with the Fat4 extracellular domain. Moreover, since we assessed a FLAG-tagged Fat4 version rather than the endogenous Fat4, the USP9X-mediated regulation of Fat4 levels is predicted to be primarily the result of a post-translational mechanism, rather than an indirect effect on Fat4 gene expression.
We sought to validate our observations by directly assessing Fat4 levels in immunofluorescence experiments (Fig. 9h–j). For this, HEK293 cells were transfected with GFP and either Fat4 (Figure S10i) or USP9X (Fig. 9h and i) and stained with an anti-Fat4 antibody. Subsequently, Fat4 protein levels were analysed and compared between control GFP-negative cells and USP9X-expressing, GFP-positive cells. Analysis of our immunofluorescence experiments revealed that cells expressing GFP and USP9X exhibited higher levels of Fat4 than cells that did not express GFP (Fig. 9h–j). Taken together, our data indicates that, similarly to the role of Faf in Drosophila, USP9X stabilises Fat4 protein in mammalian cells.
Discussion
Ft is an atypical cadherin with essential functions in tissue growth and cell polarity12,53. Despite intense study into its cellular role, the mechanisms regulating Ft function and its downstream effects remain relatively elusive. Ft interacts, both genetically and physically, with many proteins involved in tissue growth and planar cell polarity regulation11,53,54, but it is still unclear how these two functions and the multiple interactions are controlled and coordinated. Interestingly, several Ft-associated proteins and processes are thought to be linked to protein ubiquitylation (e.g., Dlish26, D17,20, Fbxl730,31 and Elgi32). Surprisingly, despite these previous reports, the action of DUBs has not been associated with the regulation of Drosophila Ft. Here, we identified Fat facets (faf) as a regulator of Ft and delineated its role in the regulation of Ft protein stability and function.
Our observations strongly suggest that Faf modulates the role of Ft in the regulation of tissue growth and Hippo signalling, in a manner consistent with a mechanism involving the regulation of Ft protein levels. Tissue growth phenotypes associated with increased or decreased levels of Ft in developing tissues were abrogated or enhanced when Faf levels were modified by over-expression or RNAi-mediated depletion, in agreement with a role for Faf in promoting Ft protein stability. In addition, Faf affected Hippo signalling readouts in vivo. In fact, the effect of Ft on Hippo signalling appears to be partly dependent on Faf function. We also observed a genetic interaction between faf and ft, which further suggests that Faf function is important for the role of Ft in the regulation of tissue growth. However, we cannot fully exclude the hypothesis that Faf also has a parallel function that impinges on imaginal disc growth, particularly as Faf has been previously associated with the Notch, RTK/Ras and JNK pathways33,34,36,55,56,57.
Our results suggest that Faf mainly regulates the tissue growth function of Ft, rather than its PCP functions, since Faf affected Ft-induced tissue size, but not tissue shape phenotypes. This observation was somewhat unexpected, given that Faf has been previously shown to regulate the protein levels of core PCP proteins in the wing via the regulation of Flamingo (Fmi)58 and that its mammalian orthologue USP9X has been associated with PCP regulation via the deubiquitylation of DVL259. However, we have not directly addressed whether Faf regulates Ft-mediated PCP downstream readouts and, therefore, cannot rule out a function for Faf in modulating Ft-related PCP functions. Nevertheless, our results suggest that a putative function of Faf in regulating PCP in the context of Ft signalling is not as relevant as its regulation of tissue growth. It is also possible that ubiquitylation may be a signal or a switch controlling the role of Ft and directing it toward tissue growth regulation or modulation of PCP, by altering Ft protein-protein interactions. Further investigation is required to reconcile these observations. Alternatively, or in addition, Ft ubiquitylation may associated with establishing a specific threshold of Ft signalling at particular developmental stages or in specific tissues, which then influence the response to specific signals, such as mechanical forces.
The effects of Faf on tissue growth and Hippo signalling regulation in Drosophila were recapitulated with the mammalian orthologue USP9X, suggesting that the role of Faf in regulating Ft function may be evolutionarily conserved. Indeed, USP9X has been previously associated with Hippo signalling regulation, albeit not with the regulation of the Ft orthologue Fat4. USP9X is thought to regulate Hippo signalling via the modulation of YAP160, LATS61,62 and Angiomotin proteins63,64. In addition, USP9X has been shown to interact with and to deubiquitylate YAP160, LATS61,62, WW4561, KIBRA61, AMOT61,64 and AMOTL263. It is still unclear how USP9X affects Hippo signalling, as some of its reported substrates have opposing effects on the pathway. Whilst some of the differences can be potentially explained by cell-type-specific functions of USP9X, these are not sufficient to reconcile all previous observations. Moreover, none of these reports has assessed the role of USP9X in regulating Fat4 or other mammalian Ft proteins. Therefore, it remains a possibility that some of the previously reported effects of USP9X on Hippo signalling could be related to Fat4 regulation, particularly for those proteins known to be in the vicinity of Fat4 at the membrane. Interestingly, in the mouse heart, Fat4 regulates Hippo signalling and YAP1 activity via the Angiomotin protein Amotl165, raising the possibility that USP9X may be required for this function. USP9X has a complex role in cancer, with both pro- and anti-tumourigenic functions and this may be related to the fact that it not only regulates Hippo signalling, but also other pathways such as TGF-β, Wnt and JAK-STAT, both in mammals59,66,67,68,69,70, as well as in Drosophila35.
Our data firmly establishes Faf as a regulator of Ft function and tissue growth. However, questions remain regarding the precise molecular mechanisms involved. Our data is consistent with a direct effect of Faf on Ft, given that the two proteins interact in co-IP experiments. Faf controls Ft protein levels post-translationally in a catalytically-dependent manner, which is supported by the effect of the DUB inhibitor WP1130 on Ft protein levels. Although WP1130 is thought to inhibit other DUBs (USP5, USP14, UCH37 and UCH-L1)52, the orthologous Drosophila genes were either not identified as hits in our in vivo RNAi screen or had the opposite effect to Faf, suggesting that Faf is the relevant target for WP1130 in the context of Ft regulation.
However, whilst the DUB activity of Faf seems to be required for the regulation of Ft protein levels and function, it is still unclear whether Faf directly deubiquitylates Ft or if its effect is indirect. Analysis of a direct effect of Faf in deubiquitylating Ft is complicated by the lack of evidence regarding direct Ft ubiquitylation and the fact that no E3 ligase has been identified that targets Ft. Previous efforts have failed to establish Elgi as a Ft E3 ligase32 and, therefore, we have not tested whether Faf-mediated regulation of Ft is related to the E3 ligase activity of Elgi. Without a clear E3 ligase candidate to promote ectopic Ft ubiquitylation in Drosophila S2 cells, it would be challenging to firmly establish Faf as a direct Ft DUB. Alternative models that are not dependent on Faf directly acting on Ft are also plausible. For instance, although the precise mechanistic details remain unclear, in their role as modulators of Ft-mediated signalling, both Fbxl7 and Elgi have been proposed to regulate protein trafficking30,32. Therefore, it is possible that the effect of Faf is also related to protein trafficking. Indeed, it has been previously reported that the mechanism by which Faf regulates PCP in Drosophila is potentially related to protein trafficking, as loss of faf results in increased lysosomal degradation of Fmi, a possible consequence of improper vesicle accumulation of internalised Fmi58. Moreover, Faf and its putative substrate Lqf (Liquid facets, a Drosophila Epsin) have been proposed to control endocytosis of several cargoes, including the Notch ligand, Delta55,57. Whether Faf-mediated regulation of Ft is dependent on Ft endocytosis and trafficking is still unknown.
Finally, while the level at which Faf acts within the Ft signalling pathway remains to be precisely determined, some hints come from our in vivo genetic interaction experiments, the analysis of D subcellular localisation, and the effects on Ft protein stability. Our data suggest that Faf likely acts at a regulatory node including Ft, Fbxl7 and App and, given the phenotypes observed, it is possible that it antagonises the action of Dlish in regulating D function. This is supported by the fact that Faf appears to act upstream of the Hpo kinases Hpo and Wts. Further biochemical experiments are required to determine if the interactions between these Ft signalling components are modulated by Faf, if App-mediated palmitoylation is affected by Faf activity, or if any of these proteins is a bona fide Faf substrate. Another question to address in future experiments is whether Faf function is regulated by specific cues, for instance mechanical forces, nutrient supply and hormonal signals, all inputs that control Hpo signalling and, by extension, tissue growth.
Methods
Drosophila cell culture, expression constructs and chemical treatments
Work involved the use of the Drosophila cell line Schneider S2 (RRID:CVCL_Z232), obtained from the ATCC and screened for mycoplasma presence, showing no contamination. Drosophila S2 cells were grown in Drosophila Schneider’s medium (Thermo Fisher Scientific) supplemented with 10% (v/v) FBS, 50 µg/mL penicillin and 50 µg/mL streptomycin. Expression plasmids were transfected using Effectene transfection reagent (QIAGEN). Plasmids were generated via Gateway® technology (Thermo Fisher Scientific). Open reading frames (ORFs) were PCR amplified from cDNA clones obtained from the Drosophila Genomics Resource Centre (DGRC, https://dgrc.cgb.indiana.edu/vectors/Overview) and cloned into the pDONR207 or pDONR-Zeo Entry vectors. Destination vectors used were obtained from the Drosophila Gateway Vector Collection or generated in-house as previously described28. All Entry vectors were verified by sequencing. Point mutations were generated using the Quikchange Site-Directed Mutagenesis kit (Agilent) according to the manufacturer’s instructions. The FtICD plasmid has been previously described43. Where indicated, inhibition of Faf was achieved by treating cells with 5 µM of the DUB inhibitor WP1130, also known as Degrasyn (Cambridge Bioscience) for 6 h before cell lysis. Proteasome inhibition was achieved by treating S2 cells with 50 µM of MG132 (Cambridge Bioscience) for 4 h before cell lysis.
Mammalian cell culture and expression constructs
Mammalian in vitro work involved the use of HEK293 cells (RRID:CVCL_0045), obtained from the ATCC and screened for mycoplasma presence, showing no contamination. HEK293 cells were grown in Dulbecco’s Modified Eagle Medium (DMEM, Thermo Fisher Scientific) supplemented with 10% (v/v) FBS, 50 µg/mL penicillin and 50 µg/mL streptomycin. Expression plasmids were transfected using Lipofectamine LTX Transfection Reagent (Thermo Fisher Scientific). USP9X plasmid (pCMV-HA-USP9X (DU10171)) was obtained through the MRC PPU Reagents and Services facility (MRC PPU, College of Life Sciences, University of Dundee, Scotland, mrcppureagents.dundee.ac.uk). Fat4 plasmids (pcDNA Fat4[ICD]::FLAG and pCMV5 Fat4[ΔECD]::FLAG) were a kind gift from Helen McNeill (Washington University School of Medicine, St. Louis). The pCMV-FLAG-EGFP plasmid was a kind gift from Nic Tapon (Francis Crick Institute, London).
RNAi production and treatment
dsRNAs were synthesised using the Megascript T7 kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. DNA templates for dsRNA synthesis were PCR amplified from genomic DNA or plasmids encoding the respective genes using primers containing the 5’ T7 RNA polymerase-binding site sequence. dsRNA primers were designed using the DKFZ E-RNAi design tool (https://www.dkfz.de/signaling/e620rnai3/). The following primers were used: lacZ (Fwd –TTGCCGGGAAGCTAGAGTAA and Rev – GCCTTCCTGTTTTTGCTCAC); faf (Fwd – CATCGCGTTTAGGCGAGTA and Rev – CGCACCACGCTGATGAGTA). After cell seeding, S2 cells were incubated with 20 μg dsRNA for 1 h in serum-free medium, before complete medium was added. 72 h after dsRNA treatment, cells were lysed and processed as detailed below.
RNA isolation and RT-PCR analysis
For faf expression analysis in UAS-faf transgenes, hsFLP; UAS-lacZRNAi; Act > CD2 > RFP females were crossed with control flies (UAS-mCherry), UAS-faf isoC or UAS-fafWT. 48 h after egg laying (AEL) developing larvae were heat-shocked at 37°C for 1 h. Late L3 larvae were collected, homogenised using disposable pellet pestles and total RNA was extracted using the QIAshredder and RNeasy kits (QIAGEN), according to the manufacturer’s instructions. 1 µg of total RNA was used for cDNA production using the QuantiTect Reverse Transcription kit (QIAGEN), following the manufacturer’s instructions. cDNA was used to conduct RT-PCR analysis using the following primers: rp49 (Fwd – GACGCTTCAAGGGACAGTATCTG and Rev – GCAGTAAACGCGGTTCTGCATGAG); faf (Fwd – TAATACGACTCACTATAGGGACGCCAGAGCAAATGTTTTT and Rev – TAATACGACTCACTATAGGGACATGCTAAAGTCTTGCCCG). RT-PCR reactions were analysed in a 2% agarose gel and imaged in a Chemidoc MP Imaging System (Bio-Rad).
Immunoprecipitation and immunoblot analysis
For purification of FLAG-tagged proteins, cells were lysed in lysis buffer (50 mM Tris pH 7.5, 150 mM NaCl, 1% Triton X-100, 10% (v/v) glycerol, and 1 mM EDTA), to which 0.1 M NaF, phosphatase inhibitors 2 and 3 (Sigma) and protease inhibitor cocktail (Complete, Roche) were added. Cell extracts were spun at 17,000 g for 10 min at 4 °C. FLAG-tagged proteins were purified using anti-FLAG M2 Affinity agarose gel (Sigma) for > 1 h at 4 °C. FLAG immunoprecipitates were then washed four times with lysis buffer before elution using 150 ng/µl 3x FLAG peptide for 15–30 minutes at 4 °C. Detection of purified proteins and associated complexes was performed by immunoblot analysis using chemiluminescence (Thermo Fisher Scientific). Western blots were probed with mouse anti-FLAG (M2; Sigma; RRID:AB_262044), rat anti-HA (3F10; Roche Applied Science; RRID:AB_2314622), mouse anti-V5 (Thermo Fisher Scientific; RRID:AB_2556564), or mouse anti-tubulin (E7; DSHB; RRID:AB_528499). Secondary antibodies used included HRP-conjugated sheep anti-mouse (Amersham) and HRP-conjugated goat anti-rat (Thermo Fisher Scientific). For densitometry analysis of immunoblots, X-ray blots were scanned using an Epson Perfection V700 flatbed scanner and further analysed with the Gel Analyser function on ImageJ (RRID:SCR_003070). Alternatively, immunoblots were analysed in an ImageQuant 600 (Amersham).
Immunostaining
Larval tissues were processed as previously described71. Primary antibodies were incubated overnight at 4°C unless otherwise stated. Mouse anti-Armadillo (N2 7A1; DSHB; RRID:AB_528089) was used at 1:50, rat anti-Fat (kind gift from Helen McNeill) was used at 1:500 and mouse anti-β-galactosidase (Z3781, Promega; RRID:AB_430877) was used at 1:500. Anti-mouse Rhodamine Red-X-conjugated (Jackson ImmunoResearch) secondary antibodies were used at 1:500. Anti-mouse or anti-rat Alexa Fluor 647-conjugated (Jackson ImmunoResearch) secondary antibodies were used at 1:500. Secondary antibodies were incubated for at least 2 h at room temperature. After washes, tissues were stained with DAPI (1µg/mL) for 10 minutes before clearing in Vectashield (without DAPI) (H-1200, Vector Labs; RRID:AB_2336790, respectively), and mounting with Mowiol 40-88 (Sigma). Fluorescence images were acquired on Zeiss LSM710 or Zeiss LSM880 confocal laser scanning microscopes (40x or 63x objective lens).
Mammalian cell immunofluorescence
HEK293 cells were seeded onto glass coverslips coated with Poly-L-Lysine (Sigma) and left to adhere overnight prior to transfection with plasmid DNA. 24-48 h after transfection, coverslips were washed in PBS, fixed with 4% Paraformaldehyde and permeabilised in 0.1% Triton-X-100 in PBS before incubating with the primary antibody Rabbit anti-Fat4 (PA5-116735; Thermo Fisher Scientific; RRID: AB_2901366) at a concentration of 1:100. Donkey anti-rabbit Rhodamine Red X-conjugated secondary antibody (Jackson Immunoresearch) was used at a concentration of 1:500. After washes, the cells were stained with DAPI (1µg/mL) for 5 minutes. Coverslips were then rinsed in distilled H2O, dried on tissue and mounted onto microslides containing a droplet of Mowiol 40-88 (Sigma). Slides were left to dry overnight before imaging on a LSM 880 confocal microscope (40x objective).
Drosophila genetics and genotypes
Transgenic RNAi stocks were obtained from the Vienna Drosophila Resource Centre (VDRC; RRID:SCR_013805) and the Kyoto Stock Centre (DGRC; RRID:SCR_008469). Details of RNAi stocks used in the in vivo RNAi screens are detailed in Table S1. ftG-rv and ft8 were obtained from Yanlan Mao (UCL). en-Gal4, UAS-GFP, ex-lacZ (ex697) and mirr-Gal4 were obtained fron Nic Tapon (Francis Crick Institute). en-Gal4, HRE-diap1::GFP was obtained from Jin Jiang (UT Southwestern). Fbxl7 stocks were obtained from Barry Thompson (ANU, Canberra). en-Gal4, UAS-RFP was obtained from Jean-Paul Vincent (Francis Crick Institute). The faf fly stocks fafB3 (BL25100), fafBX3 (BL25101), fafBX4 (BL25107), fafFO8 (BL25108), and UAS-faf (BL25102) were obtained from Bloomington. UAS-fafWT, UAS-fafC1677S (UAS-fafCD) and UAS-USP9X were a kind gift from Bassem A. Hassan (Paris Brain Institute, France) and have been previously described36.
All crosses were raised at 25°C unless otherwise stated. Genotypes were as follows:
Figures 1a, 3a, 6f, 8g, S1d, S2a, S4a, S5e: nub-Gal4, UAS-GFP / UAS-GFP
Figure 1b, S1e, S2g: nub-Gal4, UAS-GFP / +; + / UAS-fafRNAi 55GD (VDRC 2955GD)
Figures 1c, 6g, S1h: UAS-fafisoC / +; nub-Gal4, UAS-GFP
Figures 1d, 6k, 8j: nub-Gal4, UAS-ftHA / UAS-GFP
Figure 1e: nub-Gal4, UAS-ftHA / +; + / UAS-fafRNAi 55GD (VDRC 2955GD)
Figures 1f, 6l: UAS-faf / +; nub-Gal4, UAS-ftHA / +
Figures 1g, 6p, 8m: nub-Gal4, UAS-ftRNAi (VDRC 108863KK) / UAS-GFP
Figure 1h: nub-Gal4, UAS-ftRNAi (VDRC 108863KK) / +; + / UAS-fafRNAi 55GD (VDRC 2955GD)
Figures 1i, 6q: UAS-faf / +; nub-Gal4, UAS-ftRNAi (VDRC 108863KK) / +
Figure 2d: ftG-rv / ft8; + / faf FO8 (BL25108)
Figure 2e: ftG-rv / ft8; + / faf BX4 (BL25107)
Figure 2f, S3c: ftG-rv / ft8; + / faf B3 (BL25100)
Figure 2g: ftG-rv / ft8; + / faf BX3 (BL25101)
Figure 3b: nub-Gal4, UAS-ds / UAS-GFP
Figure 3c: nub-Gal4, UAS-ds / UAS-fafRNAi 79GD (VRDC 30679GD)
Figure 3d: nub-Gal4, UAS-ds / UAS-fafRNAi KK (VRDC 107716KK)
Figure 3e: nub-Gal4, UAS-ds / +; + / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure 3f: UAS-fafisoC / +; nub-Gal4, UAS-ds / +
Figure 3g: nub-Gal4, UAS-DlishRNAi (VDRC 104282KK) / UAS-GFP
Figure 3h: nub-Gal4, UAS-DlishRNAi (VDRC 104282KK) / +; + / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure 3i: UAS-fafisoC / +; nub-Gal4, UAS-DlishRNAi (VDRC 104282KK) / +
Figure 3j: nub-Gal4, UAS-DlishRNAi (VDRC 104282KK) / UAS-ftHA
Figure 3k: nub-Gal4, UAS-DlishRNAi (VDRC 104282KK) / UAS-ftHA; + / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure 3l: nub-Gal4 / UAS-GFP; UAS-Fbxl7GFP / +
Figure 3m: nub-Gal4 / + ; UAS-Fbxl7GFP / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure 3n: nub-Gal4 / UAS-ftHA; UAS-Fbxl7GFP / +
Figure 3o: nub-Gal4 / UAS-ftHA; UAS-Fbxl7GFP / UAS-fafRNAi 55GD (VDRC 2955GD)
Figures 4e, 7a, 9a: + / UAS-lacZRNAi; hh-Gal4, UAS-GFP / +
Figure 4f: hh-Gal4, UAS-GFP / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure 4g: UAS-fafisoC / +;; hh-Gal4, UAS-GFP / +
Figure 4h: UAS-ftRNAi (VDRC 108863KK) / +; hh-Gal4, UAS-GFP / +
Figures 5a, 6a, 8a: en-Gal4, UAS-GFP, ex-lacZ (ex697) / UAS-GFP; MKRS / +
Figure 5b: en-Gal4, UAS-GFP, ex-lacZ (ex697) / UAS-hpoRNAi (VDRC 104169KK); MKRS / +
Figure 5c: en-Gal4, UAS-GFP, ex-lacZ (ex697) / + ; MKRS / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure 5d: en-Gal4, UAS-GFP, ex-lacZ (ex697) / UAS-ftHA; MKRS / +
Figure 5e: en-Gal4, UAS-GFP, ex-lacZ (ex697) / UAS-ftHA; MKRS / UAS-fafRNAi 55GD (VDRC 2955GD)
Figures 5g, 8c, S9a: en-Gal4, UAS-RFP / UAS-lacZRNAi; HRE-diap1::GFP / +
Figure 5h: en-Gal4, UAS-RFP / UAS-hpoRNAi (VDRC 104169KK); HRE-diap1::GFP / +
Figure 5i: en-Gal4, UAS-RFP / + ; HRE-diap1::GFP / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure 5j: en-Gal4, UAS-RFP / UAS-ftHA; HRE-diap1::GFP / +
Figure 5k: en-Gal4, UAS-RFP / UAS-ftHA; HRE-diap1::GFP / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure 6b: UAS-fafisoC / +; en-Gal4, UAS-GFP, ex-lacZ (ex697) / UAS-GFP; MKRS / +
Figure 6c: en-Gal4, UAS-GFP, ex-lacZ (ex697) / UAS-GFP; MKRS / UAS-fafWT
Figure 6d: en-Gal4, UAS-GFP, ex-lacZ (ex697) / UAS-GFP; MKRS / UAS-fafCD (fafC1677S)
Figure 6h: nub-Gal4, UAS-GFP / +; + / UAS-fafWT
Figure 6i: nub-Gal4, UAS-GFP / +; + / UAS-fafCD (fafC1677S)
Figure 6m: nub-Gal4, UAS-ftHA / +; + / UAS-fafWT
Figure 6n: nub-Gal4, UAS-ftHA / +; + / UAS-fafCD (fafC1677S)
Figure 6r: nub-Gal4, UAS-ftRNAi (VDRC 108863KK) / +; + / UAS-fafWT
Figure 6s: nub-Gal4, UAS-ftRNAi (VDRC 108863KK) / +; + / UAS-fafCD (fafC1677S)
Figure 7b: hh-Gal4, UAS-GFP / UAS-fafWT
Figure 7c: hh-Gal4, UAS-GFP / UAS-fafCD (fafC1677S)
Figure 8b: en-Gal4, UAS-GFP, ex-lacZ (ex697) / UAS-GFP; MKRS / UAS-USP9X
Figure 8d: en-Gal4, UAS-RFP / + ; HRE-diap1::GFP / UAS-USP9X
Figure 8h: nub-Gal4, UAS-GFP / +; + / UAS-USP9X
Figure 8k: nub-Gal4, UAS-ftHA / +; + / UAS-USP9X
Figure 8n: nub-Gal4, UAS-ftRNAi (VDRC 108863KK) / +; + / UAS-USP9X
Figure 9b: hh-Gal4, UAS-GFP / UAS-USP9X
Figure S1f: nub-Gal4, UAS-GFP / UAS-fafRNAi KK (VDRC 107716KK)
Figure S1g: nub-Gal4, UAS-GFP / UAS-fafRNAi 79GD (VDRC 30769GD)
Figure S4b: nub-Gal4, UAS-ex2 / UAS-GFP
Figure S4c: nub-Gal4, UAS-ex2 / +; + / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure S4d: nub-Gal4, UAS-ex2 / UAS-ftHA
Figure S4e: nub-Gal4, UAS-ex2 / UAS-ftHA; + / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure S4f: nub-Gal4, UAS-elgiRNAi (VDRC 109617KK) / UAS-GFP
Figure S4g: nub-Gal4, UAS-elgiRNAi (VDRC 109617KK) / +; + / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure S4h: UAS-fafisoC / +; nub-Gal4, UAS-elgiRNAi (VDRC 109617KK) / +
Figure S4i: nub-Gal4, UAS-elgiRNAi (VDRC 109617KK) / UAS-ftHA
Figure S4j: nub-Gal4, UAS-elgiRNAi (VDRC 109617KK) / UAS-ftHA; + / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure S4k: nub-Gal4, UAS-appRNAi (VDRC 32863GD) / UAS-GFP
Figure S4l: nub-Gal4, UAS-appRNAi (VDRC 32863GD) / +; + / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure S4m: UAS-fafisoC / +; nub-Gal4, UAS-appRNAi (VDRC 32863GD) / +
Figure S4n: nub-Gal4, UAS-appRNAi (VDRC 32863GD) / UAS-ftHA
Figure S4o: nub-Gal4, UAS-appRNAi (VDRC 32863GD) / UAS-ftHA; + / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure S5a: nub-Gal4, UAS-hpoRNAi (VDRC 104169KK) / UAS-GFP (Room temperature cross)
Figure S5b: UAS-fafisoC / +; nub-Gal4, UAS-hpoRNAi (VDRC 104169KK) / + (Room temperature cross)
Figure S5c: nub-Gal4, UAS-hpoRNAi (VDRC 104169KK) / +; + / UAS-fafRNAi 55GD (VDRC 2955GD) (Room temperature cross)
Figure S5f: nub-Gal4 / UAS-GFP; + / UAS-wts
Figure S5g: UAS-fafisoC / +; nub-Gal4 / + ; UAS-wts / +
Figure S5h: nub-Gal4 / + ; UAS-wts / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure S5i: nub-Gal4 / UAS-ftHA; UAS-wts / +
Figure S5j: nub-Gal4 / UAS-ftHA; UAS-wts / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure S7a: UAS-ftHA / +; hh-Gal4, UAS-GFP / +
Figure S7b: UAS-ftHA / +; hh-Gal4, UAS-GFP / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure S7g: UAS-GFP / UAS-GFP; mirr-Gal4 / +
Figure S7h: UAS-GFP / +; mirr-Gal4 / UAS-fafWT
Figure S7i: UAS-GFP / +; mirr-Gal4 / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure S8b, S8h, S9f, S9i, S10d, S10f: en-Gal4, UAS-RFP / UAS-lacZRNAi; Dachs::GFP / +
Figure S8c, S8i: en-Gal4, UAS-RFP / UAS-ftHA; Dachs::GFP / +
Figure S8d, S8j: en-Gal4, UAS-RFP / UAS-ftHA; Dachs::GFP / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure S8e, S8k: UAS-fafisoC / +; en-Gal4, UAS-RFP / +; Dachs::GFP / +
Figure S8f, S8l: en-Gal4, UAS-RFP / UAS-ftRNAi (VDRC 108863KK); Dachs::GFP / +
Figure S8g, S8m: en-Gal4, UAS-RFP / +; Dachs::GFP / UAS-fafRNAi 55GD (VDRC 2955GD)
Figure S9b: UAS-fafisoC / +; en-Gal4, UAS-RFP / + ; HRE-diap1::GFP / +
Figure S9c: en-Gal4, UAS-RFP / + ; HRE-diap1::GFP / UAS-fafWT
Figure S9d: en-Gal4, UAS-RFP / + ; HRE-diap1::GFP / UAS-fafCD (fafC1677S)
Figure S9g, S9j: en-Gal4, UAS-RFP / UAS-lacZRNAi; Dachs::GFP / UAS-fafWT
Figure S9h, S9k: en-Gal4, UAS-RFP / UAS-lacZRNAi; Dachs::GFP / UAS-fafCD (fafC1677S)
Figure S9l: hsFLP / + ; UAS-lacZRNAi / UAS-mCherry; act > CD2>Gal4, UAS RFP / + (mCherry sample)
Figure S9l: hsFLP / UAS-fafisoC; UAS-lacZRNAi / +; act > CD2>Gal4, UAS RFP / + (fafisoC sample)
Figure S9l: hsFLP / + ; UAS-lacZRNAi / +; act > CD2>Gal4, UAS RFP / UAS-fafWT (fafWT sample)
Figure S10e, S10g: en-Gal4, UAS-RFP / UAS-lacZRNAi; Dachs::GFP / UAS-USP9X
Immunofluorescence quantification and statistical analyses
For quantification of Ft and Arm protein levels in vivo (Figs. 7d-f, 9c-e, S7c-e), relative ex-lacZ (Figs. 5f, 6e and 8e) or diap1::GFP levels (Figs. 5l, 8f and S9e), ratios in posterior versus anterior compartment (P/A) were calculated by manually drawing around each compartment of the wing disc in maximum intensity projections, then measuring the mean grey pixel value in Fiji (RRID:SCR_002285). For quantification of Ft protein levels in eye imaginal discs (Figure S7f), ratios between GFP-positive and GFP-negative areas were calculated by manually drawing around the relevant sections of the eye disc in maximum intensity projections and measuring mean grey pixel values in Fiji. Significance was calculated using a Brown-Forsythe and Welch ANOVA comparing all genotypes to their respective controls (for Ft and Arm protein levels), or a one-way ANOVA comparing all means to the en-Gal4 control (UAS-GFP for ex-lacZ; UAS-lacZRNAi for diap1::GFP), both with Dunnett’s multiple comparisons test. Pairwise comparisons between other samples were performed using unpaired two-tailed t-tests or unpaired two-tailed t-tests with Welch’s correction.
Analysis of larval wing imaginal disc sizes
For tissue size analysis of ft trans-heterozygous mutants, crosses were setup at 25°C and flies were placed in new vials every 24 h for precise staging of larval development. Larvae were dissected at the late L3 stage, ~5 days AEL.
Analysis of genetic interactions in Drosophila adult wings
For analysis of genetic interactions in the Drosophila wing, flies with the genotypes of interest were collected and preserved in 70% EtOH for at least 24 h. Adult wing samples were prepared as previously described72. Briefly, wings were removed in 100% isopropanol, mounted in microscope slides using Euparal (Anglian Lepidopterist Supplies) as mounting medium and baked at 65°C for at least 5 h. Adult wing images were captured using a Pannoramic 250 Flash High Throughput Scanner (3DHISTECH) and extracted using the Pannoramic Viewer software (3DHISTECH) or captured using an Axioscan 7 (Zeiss). Wing area was quantified using Fiji (the alula and costal cell of the wing were both excluded from the analysis). Images were processed using Adobe Photoshop (RRID:SCR_014199).
Analysis of planar cell polarity phenotypes
For analysis of tissue roundness (Fig. 1k, S2c, S2d), images of adult wings were analysed in Fiji and the ratio between the proximal-distal (PD) axis (defined by a line that bisects the area between the L3 and L4 veins in half) and the anterior-posterior axis (defined by a line that crosses the L4/L5 crossvein and generally defines the widest part of the adult wing) was calculated (Figure S2b). Additionally, circularity was calculated using the Fiji area selection tool to define the adult wing, or by fitting an ellipse to the wing blade section (Figure S2b’). For analysis of wing hair orientation (Figure S2i), adult wing images were used to define 3 regions of interest (ROI) within the section of the wing between the L3 and L4 veins. ROIs were used to train a model to extract features in Ilastik (RRID:SCR_015246) by defining specific pixels as part of a wing hair or as part of the wing background. Images corresponding to the probability model from Ilastik73 were analysed in Fiji, thresholded to include only the wing hair and wing hair orientation was quantified using OrientationJ (RRID:SCR_014796) in Fiji74. Data was plotted as normalised pixel frequencies (0 to 3.5) for each 0.5° angle and 180° rose plots were generated using R studio.
Analysis of wing vein defects
For analysis of L2 vein defects in adult wings (Figure S2h), images of adult wings were analysed for the presence or absence of vein defects and the phenotype frequency was calculated. Significance was calculated in Prism using Chi-square analysis, adjusted Fisher’s exact test and Bonferroni correction for multiple comparisons.
Statistical analysis
Statistical analyses were performed using GraphPad Prism 8, 9 and 10 (RRID:SCR_002798). Significance values corresponding to comparisons between two groups were calculated using unpaired t-test or unpaired t-test with Welch’s correction. Significance values corresponding to comparisons between multiple groups were calculated using one-way ANOVA, Brown-Forsythe and Welch ANOVA or Kruskal-Wallis ANOVA analyses with Dunnett’s, Tukey’s or Dunn’s multiple comparisons tests. p-values: ns denotes not significant; * denotes I < 0.05; ** denotes p < 0.01; *** denotes p < 0.001. Significance values are provided in Table S2.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
Source data are provided with this paper.
References
Davis, J. R. & Tapon, N. Hippo signalling during development. Development 146, 51–63 (2019).
Fulford, A., Tapon, N. & Ribeiro, P. S. Upstairs, downstairs: spatial regulation of Hippo signalling. Curr. Opin. Cell Biol. 51, 22–32 (2018).
Misra, J. R. & Irvine, K. D. The hippo signaling network and its biological functions. Annu Rev. Genet 52, 65–87 (2018).
Richardson, H. E. & Portela, M. Tissue growth and tumorigenesis in drosophila: cell polarity and the Hippo pathway. Curr. Opin. Cell Biol. 48, 1–9 (2017).
Manning, S. A., Kroeger, B. & Harvey, K. F. The regulation of Yorkie, YAP and TAZ: new insights into the Hippo pathway. Development 147, dev17906 (2020).
Zheng, Y. & Pan, D. The hippo signaling pathway in development and disease. Dev. Cell 50, 264–282 (2019).
Dong, J. et al. Elucidation of a universal size-control mechanism in drosophila and mammals. Cell 130, 1120–1133 (2007).
Chen, Q. et al. Homeostatic control of Hippo signaling activity revealed by an endogenous activating mutation in YAP. Genes Dev. 29, 1285–1297 (2015).
Genevet, A. & Tapon, N. The Hippo pathway and apico-basal cell polarity. Biochem J. 436, 213–224 (2011).
Moroishi, T. et al. A YAP/TAZ-induced feedback mechanism regulates Hippo pathway homeostasis. Genes Dev. 29, 1271–1284 (2015).
Blair, S. & McNeill, H. Big roles for Fat cadherins. Curr. Opin. Cell Biol. 51, 73–80 (2018).
Fulford, A. D. & McNeill, H. Fat/Dachsous family cadherins in cell and tissue organisation. Curr. Opin. Cell Biol. 62, 96–103 (2020).
Ishikawa, H. O., Takeuchi, H., Haltiwanger, R. S. & Irvine, K. D. Four-jointed is a Golgi kinase that phosphorylates a subset of cadherin domains. Science 321, 401–404 (2008).
Brittle, A. L., Repiso, A., Casal, J., Lawrence, P. A. & Strutt, D. Four-jointed modulates growth and planar polarity by reducing the affinity of dachsous for fat. Curr. Biol. 20, 803–810 (2010).
Cho, E. & Irvine, K. D. Action of fat, four-jointed, dachsous and dachs in distal-to-proximal wing signaling. Development 131, 4489–4500 (2004).
Simon, M. A., Xu, A., Ishikawa, H. O. & Irvine, K. D. Modulation of fat:dachsous binding by the cadherin domain kinase four-jointed. Curr. Biol. 20, 811–817 (2010).
Cho, E. et al. Delineation of a Fat tumor suppressor pathway. Nat. Genet 38, 1142–1150 (2006).
Mao, Y. et al. Dachs: an unconventional myosin that functions downstream of Fat to regulate growth, affinity and gene expression in drosophila. Development 133, 2539–2551 (2006).
Silva, E., Tsatskis, Y., Gardano, L., Tapon, N. & McNeill, H. The tumor-suppressor gene fat controls tissue growth upstream of expanded in the hippo signaling pathway. Curr. Biol. 16, 2081–2089 (2006).
Rauskolb, C., Pan, G., Reddy, B. V., Oh, H. & Irvine, K. D. Zyxin links fat signaling to the hippo pathway. PLoS Biol. 9, e1000624 (2011).
Vrabioiu, A. M. & Struhl, G. Fat/dachsous signaling promotes drosophila wing growth by regulating the conformational state of the NDR kinase warts. Dev. Cell 35, 737–749 (2015).
Matakatsu, H. & Blair, S. S. The DHHC palmitoyltransferase approximated regulates Fat signaling and Dachs localization and activity. Curr. Biol. 18, 1390–1395 (2008).
Matakatsu, H., Blair, S. S. & Fehon, R. G. The palmitoyltransferase approximated promotes growth via the Hippo pathway by palmitoylation of Fat. J. Cell Biol. 216, 265–277 (2017).
Misra, J. R. & Irvine, K. D. Vamana couples fat signaling to the hippo pathway. Dev. Cell 39, 254–266 (2016).
Zhang, Y., Wang, X., Matakatsu, H., Fehon, R. & Blair, S. S. The novel SH3 domain protein Dlish/CG10933 mediates fat signaling in Drosophila by binding and regulating Dachs. Elife 5, e22672 (2016).
Wang, X., Zhang, Y. & Blair, S. S. Fat-regulated adaptor protein Dlish binds the growth suppressor expanded and controls its stability and ubiquitination. Proc. Natl Acad. Sci. USA 116, 1319–1324 (2019).
Fulford, A. D. et al. Casein kinase 1 family proteins promote Slimb-dependent expanded degradation. Elife 8, e46592 (2019).
Ribeiro, P., Holder, M., Frith, D., Snijders, A. P. & Tapon, N. Crumbs promotes expanded recognition and degradation by the SCF(Slimb/beta-TrCP) ubiquitin ligase. Proc. Natl Acad. Sci. USA 111, E1980–E1989 (2014).
Zhang, H. et al. SCF(Slmb) E3 ligase-mediated degradation of expanded is inhibited by the hippo pathway in Drosophila. Cell Res 25, 93–109 (2015).
Bosch, J. A. et al. The Drosophila F-box protein Fbxl7 binds to the protocadherin fat and regulates Dachs localization and Hippo signaling. Elife 3, e03383 (2014).
Rodrigues-Campos, M. & Thompson, B. J. The ubiquitin ligase FbxL7 regulates the Dachsous-Fat-Dachs system in Drosophila. Development 141, 4098–4103 (2014).
Misra, J. R. & Irvine, K. D. Early girl is a novel component of the Fat signaling pathway. PLoS Genet 15, e1007955 (2019).
Chen, X., Zhang, B. & Fischer, J. A. A specific protein substrate for a deubiquitinating enzyme: Liquid facets is the substrate of Fat facets. Genes Dev. 16, 289–294 (2002).
Isaksson, A., Peverali, F. A., Kockel, L., Mlodzik, M. & Bohmann, D. The deubiquitination enzyme fat facets negatively regulates RTK/Ras/MAPK signalling during Drosophila eye development. Mech. Dev. 68, 59–67 (1997).
Stinchfield, M. J. et al. Fat facets deubiquitylation of Medea/Smad4 modulates interpretation of a Dpp morphogen gradient. Development 139, 2721–2729 (2012).
Koch, M. et al. A fat-facets-Dscam1-JNK pathway enhances axonal growth in development and after injury. Front Cell Neurosci. 11, 416 (2017).
Huang, Y., Baker, R. T. & Fischer-Vize, J. A. Control of cell fate by a deubiquitinating enzyme encoded by the fat facets gene. Science 270, 1828–1831 (1995).
Garoia, F. et al. Cell behaviour of Drosophila fat cadherin mutations in wing development. Mech. Dev. 94, 95–109 (2000).
Mao, Y., Kucuk, B. & Irvine, K. D. Drosophila lowfat, a novel modulator of fat signaling. Development 136, 3223–3233 (2009).
Bryant, P. J., Huettner, B., Held, L. I. Jr., Ryerse, J. & Szidonya, J. Mutations at the fat locus interfere with cell proliferation control and epithelial morphogenesis in Drosophila. Dev. Biol. 129, 541–554 (1988).
Mahoney, P. A. et al. The fat tumor suppressor gene in Drosophila encodes a novel member of the cadherin gene superfamily. Cell 67, 853–868 (1991).
Matakatsu, H. & Blair, S. S. Separating the adhesive and signaling functions of the fat and Dachsous protocadherins. Development 133, 2315–2324 (2006).
Sopko, R. et al. Phosphorylation of the tumor suppressor fat is regulated by its ligand Dachsous and the kinase discs overgrown. Curr. Biol. 19, 1112–1117 (2009).
Sing, A. et al. The atypical cadherin fat directly regulates mitochondrial function and metabolic state. Cell 158, 1293–1308 (2014).
Ambegaonkar, A. A., Pan, G., Mani, M., Feng, Y. & Irvine, K. D. Propagation of Dachsous-fat planar cell polarity. Curr. Biol. 22, 1302–1308 (2012).
Mao, Y. et al. Planar polarization of the atypical myosin Dachs orients cell divisions in Drosophila. Genes Dev. 25, 131–136 (2011).
Boedigheimer, M. & Laughon, A. Expanded: a gene involved in the control of cell proliferation in imaginal discs. Development 118, 1291–1301 (1993).
Wu, S., Liu, Y., Zheng, Y., Dong, J. & Pan, D. The TEAD/TEF family protein Scalloped mediates transcriptional output of the Hippo growth-regulatory pathway. Dev. Cell 14, 388–398 (2008).
Huang, J., Wu, S., Barrera, J., Matthews, K. & Pan, D. The Hippo signaling pathway coordinately regulates cell proliferation and apoptosis by inactivating Yorkie, the Drosophila Homolog of YAP. Cell 122, 421–434 (2005).
Clague, M. J., Urbe, S. & Komander, D. Breaking the chains: deubiquitylating enzyme specificity begets function. Nat. Rev. Mol. Cell Biol. 20, 338–352 (2019).
Bartholomeusz, G. A. et al. Activation of a novel Bcr/Abl destruction pathway by WP1130 induces apoptosis of chronic myelogenous leukemia cells. Blood 109, 3470–3478 (2007).
Kapuria, V. et al. Deubiquitinase inhibition by small-molecule WP1130 triggers aggresome formation and tumor cell apoptosis. Cancer Res 70, 9265–9276 (2010).
Gridnev, A. & Misra, J. R. Emerging mechanisms of growth and patterning regulation by Dachsous and fat protocadherins. Front Cell Dev. Biol. 10, 842593 (2022).
Strutt, H. & Strutt, D. How do the Fat-Dachsous and core planar polarity pathways act together and independently to coordinate polarized cell behaviours? Open Biol. 11, 200356 (2021).
Cadavid, A. L., Ginzel, A. & Fischer, J. A. The function of the Drosophila fat facets deubiquitinating enzyme in limiting photoreceptor cell number is intimately associated with endocytosis. Development 127, 1727–1736 (2000).
Fischer-Vize, J. A., Rubin, G. M. & Lehmann, R. The fat facets gene is required for Drosophila eye and embryo development. Development 116, 985–1000 (1992).
Overstreet, E., Fitch, E. & Fischer, J. A. Fat facets and liquid facets promote delta endocytosis and delta signaling in the signaling cells. Development 131, 5355–5366 (2004).
Strutt, H., Searle, E., Thomas-Macarthur, V., Brookfield, R. & Strutt, D. A Cul-3-BTB ubiquitylation pathway regulates junctional levels and asymmetry of core planar polarity proteins. Development 140, 1693–1702 (2013).
Nielsen, C. P., Jernigan, K. K., Diggins, N. L., Webb, D. J. & MacGurn, J. A. USP9X Deubiquitylates DVL2 to Regulate WNT Pathway Specification. Cell Rep. 28, 1074–1089.e1075 (2019).
Li, L. et al. The deubiquitinase USP9X promotes tumor cell survival and confers chemoresistance through YAP1 stabilization. Oncogene 37, 2422–2431 (2018).
Toloczko, A. et al. Deubiquitinating Enzyme USP9X Suppresses Tumor Growth via LATS Kinase and Core Components of the Hippo Pathway. Cancer Res. 77, 4921–4933 (2017).
Zhu, C. et al. Deubiquitylase USP9X suppresses tumorigenesis by stabilizing large tumor suppressor kinase 2 (LATS2) in the Hippo pathway. J. Biol. Chem. 293, 1178–1191 (2018).
Kim, M., Kim, M., Park, S. J., Lee, C. & Lim, D. S. Role of Angiomotin-like 2 mono-ubiquitination on YAP inhibition. EMBO Rep. 17, 64–78 (2016).
Thanh Nguyen, H. et al. Deubiquitylating enzyme USP9x regulates hippo pathway activity by controlling angiomotin protein turnover. Cell Discov. 2, 16001 (2016).
Ragni, C. V. et al. Amotl1 mediates sequestration of the Hippo effector Yap1 downstream of Fat4 to restrict heart growth. Nat. Commun. 8, 14582 (2017).
Chou, D. H. et al. Kinase-independent small-molecule inhibition of JAK-STAT signaling. J. Am. Chem. Soc. 137, 7929–7934 (2015).
Jie, X. et al. USP9X-mediated KDM4C deubiquitination promotes lung cancer radioresistance by epigenetically inducing TGF-beta2 transcription. Cell Death Differ. 28, 2095–2111 (2021).
Meng, Y. et al. Roles of USP9X in cellular functions and tumorigenesis (Review). Oncol. Lett. 26, 506 (2023).
Murray, R. Z., Jolly, L. A. & Wood, S. A. The FAM deubiquitylating enzyme localizes to multiple points of protein trafficking in epithelia, where it associates with E-cadherin and beta-catenin. Mol. Biol. Cell 15, 1591–1599 (2004).
Wu, Y. et al. Aberrant phosphorylation of SMAD4 Thr277-mediated USP9x-SMAD4 interaction by free fatty acids promotes breast cancer metastasis. Cancer Res. 77, 1383–1394 (2017).
Genevet, A., Wehr, M. C., Brain, R., Thompson, B. J. & Tapon, N. Kibra is a regulator of the Salvador/Warts/Hippo signaling network. Dev. Cell 18, 300–308 (2010).
Fulford, A. D. & Ribeiro, P. S. Drosophila genetics: analysis of tissue growth in adult tissues. Methods Mol. Biol. 1893, 43–51 (2019).
Berg, S. et al. ilastik: interactive machine learning for (bio)image analysis. Nat. Methods 16, 1226–1232 (2019).
Rezakhaniha, R. et al. Experimental investigation of collagen waviness and orientation in the arterial adventitia using confocal laser scanning microscopy. Biomech. Model Mechanobiol. 11, 461–473 (2012).
Acknowledgements
We thank the Vienna Drosophila Resource Centre and the Bloomington Drosophila Stock Centre (NIH P40OD018537) for transgenic fly stocks. The antibodies E7 (DSHB Hybridoma Product E7), 2A1 (DSHB Hybridoma Product 2A1) and N2 7A1 (DSHB Hybridoma Product N2 7A1 Armadillo) were deposited by M. Klymkowsky, R. Holmgren and E. Wieschaus to the Developmental Studies Hybridoma Bank, created by the NICHD of the NIH and maintained at The University of Iowa, Department of Biology. We thank N. Tapon, H. McNeill, B. Thompson, Y. Mao, J. P. Vincent and B. Hassan for providing fly stocks; Sam Wallis and Linda Hammond for assistance with microscopy; and C. Mohanathas for assistance in preliminary experiments. We thank members of the Ribeiro lab for helpful discussions and N. Tapon, M. Holder, S.A. Martin, and H. McNeill for critical reading of the manuscript. This work was supported by funding from Cancer Research UK (C16420/A18066), The Academy of Medical Sciences/Wellcome Trust Springboard Award (SBF001/1018), Barts Charity (G-002400) and from the Biotechnology and Biological Sciences Research Council (BB/T004576/1). L.E.D. was supported by a Biotechnology and Biological Sciences Research Council LIDo PhD studentship (BB/M009513/1). R.I.L. was supported by a Medical Research Council MRC-DTP PhD studentship (MR/W007045/1).
Author information
Authors and Affiliations
Contributions
P.S.R. conceived the project and supervised the study; P.S.R., L.E.D., A.S., A.D.F. designed the experiments; P.S.R., L.E.D., A.S., A.D.F., R.I.L., H.S.B performed the experiments and analysed the data; P.S.R., L.E.D., A.S. wrote the manuscript; P.S.R., L.E.D., A.S. contributed to manuscript editing.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Source data
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Dawson, L.E., Sekar, A., Fulford, A.D. et al. The deubiquitylating enzyme Fat facets promotes Fat signalling and restricts tissue growth. Nat Commun 16, 1938 (2025). https://doi.org/10.1038/s41467-025-57164-3
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467-025-57164-3