USP8 Promotes Smoothened Signaling by Preventing Its Ubiquitination and Changing Its Subcellular Localization

Hedgehog regulates the activity of its signal transducer Smoothened by enhancing its interaction with the deubiquitinase USP8, thereby promoting Smoothened translocation to the cell surface and so enhancing Hh signaling.


Introduction
Hedgehog (Hh) proteins function as morphogens and play critical roles in pattern formation and cell growth control. Hh signaling has also been implicated in tissue repair and stem cell maintenance [1]. Malfunction of Hh signaling causes birth defects as well as several types of cancer [2,3]. The Hh signal is transduced through a receptor complex consisting of Patched (Ptc) and Ihog [4]. The seven transmembrane protein Smoothened (Smo) acts as a signal transducer, and the activity of Smo is inhibited by Ptc in the absence of Hh [5][6][7]. Binding of Hh to Ptc-Ihog relieves the inhibition of Smo by Ptc, which allows Smo to activate the cubitus interuptus (Ci)/Gli family of Zn-finger transcription factors and thereby induce the expression of Hh target genes, such as decapentaplegic (dpp), ptc, and en [5,7]. Posttranslational regulation of the components in the Hh signaling pathway has been shown to be critical for their accumulation and activation. For example, phosphorylation of Smo and Ci/Gli has been extensively studied [8]. In Drosophila, it has been shown that the presence of Hh promotes the hyperphosphorylation and cell surface accumulation of Smo, whereas the absence of Hh allows for the hyperphosphorylation and proteolytic processing of Ci [5]. Although it has been shown that ubiquitination is important for Ci processing and degradation [9], it is not known whether ubiquitination is involved in the posttranslational regulation of other pathway components.
Hh induces the cell surface accumulation and phosphorylation of Smo [10] by multiple kinases, including protein kinase A (PKA) and casein kinase 1 (CK1) [11][12][13], which activate Smo by inducing a conformational change in the protein [14]. It has been shown that forced localization of Smo to the cell surface increases signaling activity, whereas endoplasmic retention of an activated form of Smo blocks activity [15]. The inhibition of endocytosis through the use of a dominant negative form of Rab5 promotes Smo cell surface accumulation [15]. In addition, an antibody uptake experiment using cultured Drosophila S2 cells suggested that Hh regulates Smo cell surface accumulation by blocking endocytosis and/or promoting the recycling of Smo [12]. Notably, Smo is primarily localized to the lysosomes of A-compartment cells in Drosophila imaginal discs, where Hh is not present, and is enriched on the plasma membrane of P-compartment cells where Hh stimulation occurs [16]. Smo has been established as a G protein-coupled receptor-like protein after the recent identification of an association with Gai [17], as well as the finding that G protein-coupled receptor kinase 2 (Gprk2) phosphorylates and regulates Smo accumulation [18,19]. Similar mechanisms have been proposed for the regulation of mammalian Smo, whereby both Smo and Ptc co-localize and internalize in endosomal compartments, and Hh induces the segregation of Smo away from Hh-Ptc complexes that are destined for lysosome degradation [20]. In addition, the association of Smo with b-arrestin 2 appears to promote Smo endocytosis through clathrin-coated pits [21]. Taken together, the controlled accumulation and localization of Smo in the Hh signaling pathway is thought to play a central role in maintaining signaling homeostasis. However, it is yet unclear how Hh controls the intracellular trafficking of Smo [3].
Ubiquitination is the enzymatic process by which proteins are covalently modified with the 76 amino acid protein ubiquitin (Ub). Ubiquitination has been shown to be important not only in the degradation of proteins, but also in the regulation of protein functions, such as protein activity, protein-protein interactions, and protein trafficking [22][23][24]. For example, cell surface receptor tyrosine kinases (RTKs) are monoubiquitinated at multiple sites, which mediates the internalization and subsequent transport of the proteins to the lysosome for destruction [25,26]. The process of ubiquitination was shown to be a reversible modification after the identification of a large family of de-ubiquitinating enzymes (DUBs), which function by removing Ub conjugates from target proteins in order to regulate their biological activity and modulate intracellular trafficking [27]. It has recently been shown that deubiquitinating enzyme UBPY/ubiquitin-specific protease 8 (USP8) regulates the ubiquitination of Frizzled, which is the receptor for Wnt/Wingless [28]. In addition, deletion of UBPY/ UBP8 in mice causes embryonic lethality [29]. However, in Hh signaling, it is currently not known whether the transmembrane protein Smo is regulated by ubiquitination and whether deubiquitination enzymes are involved in the sorting of Smo.
In this study, we found that Smo is monoubiquitinated, and this process is reduced by Hh or by phosphorylation. By using an RNAi screen that targets Drosophila DUBs in both the wings and cultured S2 cells, we identified USP8 as a DUB that prevents Smo ubiquitination and enhances the signaling activity of the protein.
We also show that inactivation of USP8 by RNAi or by overexpressing a dominant negative form of USP8 increases Smo ubiquitination in S2 cells and prevents Smo accumulation in wing discs. The overexpression of USP8 down-regulated Smo ubiquitination and increased Smo accumulation. In addition, we found that USP8 is required for Hh-induced cell surface accumulation of Smo, and that Hh promotes the interaction between Smo and USP8 (see Figure S3). Moreover, we found that USP8 prevents Smo localization to early endosomes that are labeled with Rab5. Taken together, ubiquitination is likely a negative regulatory mechanism in Smo activation and USP8 plays an opposing role in this regulation.

Smo Undergoes Monoubiquitination That Is Reduced by Hh or Phosphorylation
In an attempt to explore whether Smo is regulated by ubiquitination, we examined Smo ubiquitination in S2 cells using an immunoprecipitation assay. We found that ubiquitinated Smo was readily detected by a commercial anti-Ub (P4D1) antibody, which detected the immunoprecipitated, endogenous Ub ( Figure 1A Figure 1C, UbK0, a ubiquitin variant where all of the lysine (K) residues were changed to arginine (R), gave rise to an identically smeared banding pattern as wild-type Ub, K48-Ub, and K63-Ub ( Figure 1C, lane 9, compare lanes 1, 7, and 8). This migration pattern was similar to the pattern of Ub variants containing individual lysine mutations (UbKR) ( Figure 1C, lanes 3-6). These data suggest that Smo undergoes multi-monoubiquitination. We

Author Summary
The Hedgehog (Hh) signaling pathway is well known for its role in directing processes such as cell growth, proliferation, and differentiation during embryogenesis. The Figure 2B, middle panel) are consistent with our previous findings [12,30,31].
To determine if mammalian Smo is regulated by ubiquitination, we transfected NIH3T3 cells with Myc-tagged human Smo (Myc-hSmo) and performed a similar immunoprecipitation assay as used for Drosophila S2 cells. We found that hSmo underwent ubiquitination, which was detected by the P4D1 antibody ( Figure 3C, lane 2, top panel). In addition, Shh treatment induced a substantial decrease in hSmo ubiquitination ( Figure 3C, lane 3, top panel). To examine whether hSmo is also multi-monoubiquitinated, we performed an immunoprecipitation assay in HEK293T cells transfected with Myc-hSmo and HA-Ub constructs. We found that Myc-hSmo pulled bound both HA-Ub and HA-UbK0 in a similar pattern ( Figure 3D, top panel). These data suggest a conserved mechanism that regulates mammalian Smo.
It has previously been shown that Hh-induced phosphorylation by PKA and CK1 promotes Drosophila Smo cell-surface accumulation and signaling activity [11][12][13]. Our data suggested that Hh prevents Smo ubiquitination and that ubiquitination may downregulate Smo activity, so we therefore investigated whether there was any relationship between Smo phosphorylation and ubiquitination. As shown in Figure 2E

Identification of a Deubiquitinase for Smo by an RNAi Screen
Ubiquitin ligases and DUBs play opposing roles in the regulation of protein ubiquitination. To explore whether DUBs were involved in the regulation of Smo, we obtained 45 RNAi lines from the Vienna Drosophila RNAi Center (VDRC), which target 33 potential DUBs in the Drosophila genome (Table S1) [32,33], and performed an in vivo screen by overexpressing individual RNAi lines through the wing-specific MS1096 Gal4 to assess for the induction of an adult wing phenotype. The screen showed that RNAi-mediated knockdown of 11 genes affected wing development, as indicated by wing phenotypes or wing blisters (Table S1; unpublished data). Since it was possible that a specific DUB may reduce Smo ubiquitination and cause changes in Smo accumulation, we immunostained the wing imaginal discs using an anti-SmoN antibody when the expression of each of the 11 DUB genes was knocked down by RNAi. Expression of UAS-RNAi lines that targeted USP8 caused a reduction in Hh-induced Smo accumulation in P-compartment cells. Moreover, the expression of Hh target genes, such as ptc, was attenuated in the dorsal compartment cells where the ap-Gal4 was expressed. We also examined the levels of Smo ubiquitination in S2 cells when the expression of each of the 11 DUBs was knocked down by RNAi. As shown in Figure

USP8 Regulates Smo Ubiquitination
To determine whether USP8 has a role in the Hh-mediated reduction of Smo ubiquitination, we examined Smo ubiquitination when USP8 was inactivated. As shown in Figure 5A Figure 5C) [30], for their ability to bind USP8. We found that amino acids 625-753 of Smo were responsible for the interaction with USP8 ( Figure 5C, unpublished data).
Hh might prevent Smo ubiquitination by regulating the DUB. However, it is unlikely that USP8 activity per se is regulated by Hh, as USP8 has been shown to be involved in multiple signaling pathways. The finding that USP8 interacted with Smo led to the hypothesis that Hh might control the accessibility of Smo to the DUB. To test this, we used a GST pull-down assay previously described [30]. We found that GST-Smo625-753 pulled down more Flag-USP8 in the presence of Hh ( Figure 5D, lane 3, top panel) than that in the absence of Hh, suggesting that Hh regulates the accessibility of USP8 to Smo. We also found that phosphormimetic mutation in GST-Smo enhanced its interaction with USP8 ( Figure 5E

USP8 Promotes Smo Accumulation in Wing Imaginal Discs
We next performed both loss-and gain-of-function studies in wing discs to examine the regulation of Smo accumulation by USP8. We found that the dominant negative USP8C.S blocked Smo accumulation in P-compartment cells ( Figure 6C) and attenuated ptc-lacZ expression ( Figure 6D), which is a phenotype similar to that caused by RNAi of USP8 ( Figure 6B-B0). To further examine the physiological function of USP8 in regulating Smo, we also used an usp8 hypomorphic allele, usp8KO, that has been previously described in the literature [28]. Smo accumulation was severely reduced in usp8KO cells ( Figure 6E), and pupa bearing usp8KO clones were lethal (unpublished data). These loss-of- function studies indicate that USP8 is required for Smo accumulation. In addition, a gain-of-function experiment showed that the overexpression of USP8 caused an accumulation of Smo in both A-and P-compartment cells ( Figure 6F) and caused anterior expansion of ptc-lacZ in cells that received Hh ( Figure 6G). Moreover, inactivation of USP8 by either USP8RNAi or USP8C.S in Drosophila embryo down-regulated En expression ( Figure 6I-J, compare to wild-type En staining in H) and caused lethality of the embryo or pupa (unpublished data). We also examined the expression pattern of endogenous USP8 transcription by in situ hybridization and found that USP8 was highly expressed in the wing pouch that is more sensitive to Hh signaling ( Figure 6K-K0). Our observations strongly suggested that USP8 has a positive role in Hh-induced Smo accumulation, which is likely due to the role in regulating Smo ubiquitination.
Our previous study showed that overexpression of UAS-Smo by MS1096 Gal4 caused albeit low levels of ectopic Hh pathway activation in the anterior compartment of the wing discs [35], which induced overgrowth of the structure between Vein 2 and Vein 3 ( Figure 6M, compare to a wild-type wing in 6L) or the thickness of Vein 3 ( Figure 6N). Here, we found that overexpression of USP8 caused similar phenotypes ( Figure 4P-Q). RNAimediated knockdown of USP8 severely disrupted the wing morphology ( Figure 6O) and USP8C.S expression caused a more severe malformation of the wing as well as lethality of the pupa (unpublished data). The wing phenotypes correlated with the levels of Smo accumulation in wing discs where USP8RNAi, USP8C.S, or USP8 was expressed ( Figure 6B, C, and F), suggesting that the adult wing phenotypes were caused, at least in part, by USP8-mediated regulation of Smo. Overexpression of USP8 elevated the level of Smo but did not induce ectopic Hh signaling activity in A compartment cells located away from the A/P boundary. This could be due to the inhibition of Smo by Ptc, since it has been shown that ptc mutant clones situated in the Acompartment of wing discs accumulate high levels of Smo [16]. To test this possibility, we expressed Flag-USP8 in the ptc mutant background and found that USP8 induced severe wing growth in the ptc heterozygote ( Figure 6S, compared to 6R) and caused sick wings in the ptc homozygous background ( Figure 6T). These results suggest that the USP8-mediated elevation of Smo is still inhibited by Ptc.

USP8 Promotes Smo Signaling Activity
Overexpression of USP8 caused an elevation of Smo in both Aand P-compartment cells but did not ectopically activate Hh target genes, such as ptc, in A-compartment cells located away from the A/P boundary ( Figure 6G). This could have been due to insufficient levels of Smo. To test this possibility, we coexpressed Flag-USP8 with GFP-Smo, a construct that had been developed in an earlier study [35]. Co-expression of Flag-USP8 with GFP-Smo by MS1096 Gal4 caused induction of dpp-lacZ and ptc expression in A-compartment cells both close to and away from the A/P boundary ( Figure 7G-H), whereas the expression of Flag-USP8 alone did not induce ectopic dpp-lacZ and ptc expression ( Figure 7E-F). Moreover, the expression of GFP-Smo only induced a low level of ectopic dpp-lacZ ( Figure 7C) and no ectopic ptc expression ( Figure 7D) in cells located away from the A/P boundary. Similar results were obtained when act.CD2.Gal4 was used to co-express Flag-USP8 with GFP-Smo, which suggested a cell-autonomous regulation ( Figure S1A-B9). We also found that USP8 did not change the dominant negative activity of Smo PKA123 in wing discs (unpublished data), suggesting that phosphorylation and dimerization is required for Smo activation even though Smo can be stabilized by deubiquitination.
We had previously shown that Smo SD123 has constitutive cellsurface expression and signaling activity and is still regulated by Hh [12]. To test whether USP8 regulated the activity of Smo SD123 in vivo, Smo SD123 was expressed either alone or together with USP8RNAi and the expression of ptc-lacZ and en were examined. As shown in Figure 7N-O9, Smo SD123 showed constitutive signaling activity and induced the ectopic expression of ptc-lacZ and en. Knockdown of USP8 by RNAi attenuated the ectopic ptc-lacZ expression ( Figure 7O) and greatly reduced the ectopic expression of en ( Figure 7O9), which was likely due to the effect of USP8 regulating Smo SD123 ubiquitination ( Figure 5B). Similar results were achieved by using act.CD2.Gal4 ( Figure S1C-D9). These data suggest that Smo SD123 underwent a low level of ubiquitination, which prevents further activation.  from the A/P boundary ( Figure 7I-J). We also examined the activity of USP8 mutants that contain the Cys572Ser mutations. Both USP8C.S and USP8NT1C.S had dominant negative effects, but USP8NT2C.S and USP8NT3C.S did not ( Figure 8C). In addition, USP8NT3C.S had no effect on Smo accumulation in wing discs ( Figure 8E-E9) or on GFP-Smo activity (unpublished data). These data suggest that the MIT and RHOD domains are required for USP8C.S to act in a dominant negative manner. We further examined the subcellular localization of USP8 variants and found that USP8C.S and USP8NT1C.S, but not the other forms, were primarily localized to Rab5-labeled and enlarged early endosomes, which was caused by the overexpression of these mutants (Figure 9). Rab5 is a small GTPase that is often used as a marker of the early endosome [36]. The results from the localization assay suggest that the N-terminal domains of USP8 are responsible for its accumulation in the early endosome and for the dominant negative activity.

The Function of USP8 Domains in Regulating Smo
We also carried out a series of immunoprecipitation experiments with S2 cells to map the USP8 domain that is responsible for the interaction with Smo. Collectively, the data showed that USP8NT3 was responsible for the interaction with Smo ( Figures 8A and S2). Importantly, these results were in agreement with the finding that the catalytic domain of USP8 was capable of preventing Smo ubiquitination.

USP8 Regulates Hh-Induced Cell Surface Accumulation of Smo
In the absence of Hh, Smo remains in an unphosphorylated or hypophosphorylated state and can be removed from the cell surface by endocytosis. To confirm this presumption, we examined Smo levels in wing discs when the endocytosis machinery was inactivated. Shibire (Shi) is the Drosophila homolog of the dynamin GTPase that plays an essential role in regulating the internalization and sorting of membrane proteins [37] and is required for Ptc endocytosis [38]. We found that the accumulation of Smo was elevated in wing discs expressing Shi RNAi ( Figure 10A) or expressing a dominant negative Shi (Shi-DN, Figure 10B), suggesting that Smo may function through Shi-mediated endocytosis. In addition, we found that the inactivation of Shi caused an increase in Ci ( Figure 10A9 and B9), which suggested that Shi has a negative role in Hh signaling.
We next wished to assess whether the accumulation of Smo induced by the overexpression of USP8 or the inactivation of Shi was due to changes in the cell surface accumulation of Smo. CFP-Smo was transfected into S2 cells that were treated with GFP dsRNA (control), USP8 dsRNA, or Shi dsRNA and then treated with Hh-conditioned medium or control medium. Consistent with our previous observations [31], Hh stimulation causes the accumulation of Smo on the cell surface in the presence of control dsRNA ( Figure 10D). Interestingly, the cell surface accumulation of Smo was elevated by the RNAi-mediated reduction of Shi ( Figure 10E We have previously shown that Smo SD123 has constitutive cell surface expression [12]. To examine if USP8 regulates the cell surface accumulation of Smo SD123 , we co-transfected CFP- Figure 9. The localization of USP8 variants in S2 cells. S2 cells were cotransfected with USP8 constructs and RFP-Rab5 followed by immunostaining with an anti-HA antibody to label the expression of USP8 constructs. Rab5 marks the early endosome. Of note, the expression of USP8C.S and USP8NT1C.S causes enlarged early endosomes, and both forms of USP8 localize in these enlarged early endosomes. USP8, USP8NT1, USP8NT2, and USP8NT2C.S are evenly distributed in the cytosol without a specific pattern. USP8NT3 behaves like USP8NT2 and USP8NT3C.S behaves like USP8NT2C.S (unpublished data). doi:10.1371/journal.pbio.1001238.g009 Smo SD123 with USP8 or USP8C.S in S2 cells. The cell surface accumulation of Smo SD123 increased in the presence of USP8 ( Figure 10N) and was reduced in the presence of USP8C.S or USP8 RNAi ( Figure 10O, unpublished data), which suggested that the decrease in Smo SD123 activity (Figure S1D-D9) was due to the attenuation of cell surface accumulation of Smo SD123 . Consistent with these data, the overexpression of USP8 reduced Smo SD123 ubiquitination, whereas the inactivation of USP8 increased Smo ubiquitination ( Figure 5B). The quantification analysis of Smo cell surface accumulation further supported our findings ( Figure 10P).

USP8 Prevents the Localization of Smo to the Early Endosome
To determine whether the cell surface accumulation of Smo was due to a change in subcellular localization, we performed an antibody uptake assay with the anti-SmoN antibody to examine whether internalized Smo co-localized with Rab5. We hypothesized that blocking endocytosis may decrease Smo localization to the early endosomes, while inhibiting Smo recycling may lead to an accumulation of Smo in the early endosomes. As shown in Figure 11B, 99% of the internalized Smo protein co-localized with Rab5 positive particles, which occupied nearly 70% of the endosomes ( Figure 11B). We also found that Hh treatment inhibited Smo endocytosis and reduced the ratio of Rab5-localized Smo ( Figure 11C), which was consistent with our previous observation that Smo internalization is largely inhibited by Hh [12]. In addition, most Smo SD123 was not internalized (unpublished data). Therefore, Hh treatment and phosphorylation of Smo likely blocked Smo endocytosis. Moreover, Shi RNAi attenuated the localization of Smo in the early endosomes ( Figure 11D), whereas USP8 RNAi elevated the localization ( Figure 11E). The overexpression of USP8 decreased the amount of Smo that colocalized with Rab5 ( Figure 11F), which suggested that USP8 may prevent the accumulation of Smo in early endosomes and therefore promote the cell surface accumulation of the protein ( Figure 10I-J) as well as increase the signaling activity ( Figure 7G-H). We also found that inactivation of USP8 by RNAi or by USP8C.S overexpression caused an enlargement of the early endosome ( Figure 11E and Figure 9), which was consistent with a previous finding that USP8 deficient mouse primary cells exhibit enlarged early endosomes [29]. In addition, quantification of the amount of Smo that co-localized with Rab5 supported these conclusions ( Figure 11G). Taken together, our data suggest that USP8 up-regulates the cell surface accumulation and signaling activity of Smo by deubiquitinating the protein ( Figure 11H).

Discussion
Smo is a key Hh signal transduction molecule located on the plasma membrane and there is a good correlation between the cell surface accumulation and signaling activity. However, little is known regarding the dynamic activation of Smo or the factors that regulate Smo trafficking. The present study has established that the multi-monoubiquitination of Smo is a reversible process that is impeded by an upstream signal. Moreover, this work has shown that a specific deubiquitinase inhibits Smo ubiquitination. Using an RNAi screen in both the Drosophila wing and cultured cells, we identified and characterized USP8 as the DUB that is required for Smo deubiquitination, accumulation, and activity.

Smo Ubiquitination
In this study, we have uncovered that Smo is multimonoubiquitinated and that the ubiquitination of Smo is reduced by Hh. However, it is currently unknown how Smo is ubiquitinated. Both Drosophila and vertebrate Smo contain multiple Lys residues that are highly conserved among various species [39], but it is not clear whether these residues can act as linkages for Ub. In an effort to characterize Smo ubiquitination, we mutated many of the conserved Lys residues in Smo C-tail and examined their ubiquitination by an immunoprecipitation assay and tested in vivo activities by expressing them at the equivalent levels in wing discs. However, we did not observe any changes in ubiquitination or activity (unpublished data). It is likely that monoubiquitination occurs at many sites on Smo, which requires the simultaneous mutation of most sites in order to detect an ubiquitination change.
In It has been shown that the Smo C-tail mediates binding with Cos2, Fu, and PP4 [30,35,[40][41][42]. In this study, we add USP8 as another protein that binds to the Smo C-tail region between amino acids 625-753, which includes the three PKA and CK1 phosphorylation clusters. It is likely that phosphorylation at this region promotes the binding of USP8. In support of this notion, we found that Hh promotes the interaction between Smo and USP8 ( Figure 5D). We further found that a phospho-mimetic mutation of Smo promotes its binding with USP8 ( Figure 5E-F

The Role of USP8 in Regulating Smo
USP family members have been shown to be involved in the regulation of different cellular pathways [44]. USP8 was first described as a growth-regulated ubiquitin isopeptidase that plays a possible role in the control of mammalian cell proliferation [45]. There is increasing evidence that USP8 has a role in endosomal sorting [28,29,46]. The identification and characterization of USP8 in this study as a DUB involved in the reduction of Smo ubiquitination has provided the opportunity to further understand the functions of USP8 in regulating membrane protein trafficking. We have characterized the role of USP8 in the regulation of Smo cell surface expression that is critical for Smo activation. We have also provided evidence that USP8 is required for Hh-induced Smo accumulation. However, although the overexpression of USP8 increased the levels of Smo in vivo, it failed to induce Hh target gene expression in A-compartment cells located away from the A/ P boundary. Our explanation is that the increase in Smo levels that was induced by USP8 was still inhibited by Ptc, since we found that USP8 induced more severe overgrowth phenotypes in the ptc mutant background ( Figure 6R-T). It is also possible that USP8 might have additional function(s) in the Hh pathway by targeting other component(s).

Constructs, Mutants, and Transgenes
The Myc-Smo, Myc-SmoCT, and Myc-SmoDCT constructs have been previously described [35]. To construct Flag-tagged DUBs, we obtained full-length cDNA from either DGRC or by RT-PCR from fly embryonic RNA, amplified the cDNA fragments, and sub-cloned them into the 2xFlag-UAST vector.

FISH, Embryo Staining, and Wing Disc Immunostaining
For fluorescence in situ hybridization (FISH), yw wing discs from third instar larvae were collected and USP8 RNA probes were prepared according to the instruction of the DIG RNA labeling kits (Roach). After fixation, hybridization, and posthybridization washes, wing discs were subjected to secondary antibody incubation and then washed with PBT buffer. For embryo immunostaining, stage 11 embryos from specific genotypes were dechorionated, fixed in solution containing 80% Heptane, and immunostained with the indicated antibodies. Act5C-Gal4 (Flybase) was used to drive the expression of UAS-USP8RNAi or UAS-USP8C.S in embryos. A standard protocol was used for the immunostaining of wing discs from late third instar larvae. Antibodies used in this study were as follows: