Research ArticleCell Biology

Hedgehog reciprocally controls trafficking of Smo and Ptc through the Smurf family of E3 ubiquitin ligases

See allHide authors and affiliations

Sci. Signal.  06 Feb 2018:
Vol. 11, Issue 516, eaan8660
DOI: 10.1126/scisignal.aan8660

Smurf controls reciprocal trafficking of Smo and Ptc

Hedgehog (Hh) stimulates intracellular signaling by binding to Patched (Ptc) at the cell surface, relieving Ptc-mediated repression of the transmembrane protein Smoothened (Smo). Hh also stimulates the ubiquitylation and internalization of Ptc. In the absence of Hh, Smo is inactivated and targeted for ubiquitylation and removal from the membrane. Li et al. identified Smurf family ubiquitin E3 ligases as required for this reciprocal regulation of Ptc and Smo in Drosophila melanogaster. The ability of Smurf to bind to and ubiquitylate Smo depended on phosphorylation of Smurf by G protein–coupled receptor kinase 2 (Gprk2) and was inhibited by Hh-induced phosphorylation of Smo. Hh-induced dissociation of Smurf from Smo freed Smurf to interact with Ptc, thus promoting the ubiquitylation of Ptc. These findings identify Smurf family E3 ligases as key players in the reciprocal accumulation of Smo and Ptc at the cell surface (see Focus by Sharpe and de Sauvage).

Abstract

Hedgehog (Hh) induces signaling by promoting the reciprocal trafficking of its receptor Patched (Ptc) and the signal transducer Smoothened (Smo), which is inhibited by Ptc, at the cell surface. We identified Smurf family E3 ubiquitin ligases as essential for Smo ubiquitylation and cell surface clearance and demonstrated that Smurf family members mediate the reciprocal trafficking of Ptc and Smo in Drosophila melanogaster. G protein–coupled receptor kinase 2 (Gprk2)–mediated phosphorylation of Smurf promoted Smo ubiquitylation by increasing the recruitment of Smurf to Smo, whereas protein kinase A (PKA)–mediated phosphorylation of Smo caused Smurf to dissociate from Smo, thereby inhibiting Smo ubiquitylation. Smo and Ptc competed for the same pool of Smurf family E3 ubiquitin ligases, and Hh promoted Ptc ubiquitylation and degradation by disrupting the association of Smurf family E3 ubiquitin ligases with Smo and stimulating their binding to Ptc. Our study identifies the E3 ubiquitin ligases that target Smo and provides insight into how Hh regulates the reciprocal trafficking of its receptor and signal transducer.

INTRODUCTION

Hedgehog (Hh) signaling plays critical roles in embryonic development and adult tissue homeostasis in species ranging from insects to mammals (13). Aberrant Hh pathway activity contributes to a wide range of human disorders including birth defects and cancer (1, 4, 5). Hh controls cell growth and differentiation through a conserved signal transduction pathway that culminates in the activation of latent transcription factors Cubitus interruptus (Ci) and Gli proteins (1, 2, 6, 7). The core reception system for Hh signaling comprises two multispan transmembrane proteins: the 12-transmembrane protein Patched (Ptc) and the 7-transmembrane protein Smoothened (Smo), which is a member of the G protein (heterotrimeric guanine nucleotide–binding protein)–coupled receptor (GPCR) family. Ptc inhibits Smo in the absence of Hh. Hh binds to Ptc to release its inhibition of Smo, leading to Smo phosphorylation and activation in the presence of Hh (8).

The activity of Smo is controlled through changes in its conformation and subcellular localization. In vertebrates, cilium-localized Ptc restricts Smo ciliary accumulation and keeps Smo in an inactive conformation, whereas Hh binding to Ptc triggers its ciliary exit, leading to ciliary accumulation of phosphorylated Smo that adopts an active conformation (911). In Drosophila melanogaster, most cells lack primary cilia, but Ptc inhibits Smo phosphorylation and cell surface accumulation, whereas Hh promotes Smo phosphorylation, cell surface accumulation, and active conformation (1214). In addition, Hh binding to Ptc facilitates internalization of the Hh-Ptc protein complex, a process required for restricting the range of Hh signaling (12, 1518). Although the Smurf family of E3 ubiquitin ligases has been implicated in the regulation of Ptc ubiquitylation and trafficking in both Drosophila and mammals (1922), the mechanism underlying the regulation of Smo trafficking and cell surface accumulation is still poorly understood. In addition, how Hh signaling coordinates the reciprocal trafficking of Ptc and Smo remains unknown.

Previous studies revealed that the ubiquitylation and subsequent degradation of Smo through both proteasome- and lysosome-dependent mechanisms are responsible for preventing its cell surface accumulation in the absence of Hh (2325). Upon stimulation, Hh induces phosphorylation of the intracellular C-terminal tail of Smo (SmoCT) by protein kinase A (PKA) and casein kinase 1 (CK1), which inhibits Smo ubiquitylation, thereby promoting its cell surface accumulation (23, 24). Furthermore, Hh induces sumoylation of SmoCT at Lys851, which facilitates the recruitment of the deubiquitinase USP8 to antagonize Smo ubiquitylation independently of PKA- and CK1-mediated phosphorylation (26). In addition to ubiquitylation, the Smo-interacting proteins Kurtz (Krz), the Drosophila homolog of β-arrestin 2, and G protein–coupled receptor kinase 2 (Gprk2), the Drosophila homolog of GRK2, promote Smo internalization through unknown mechanisms (23, 2730).

How phosphorylation of Smo inhibits its ubiquitylation has remained a mystery. It has been speculated that phosphorylation of Smo may preclude the binding of an E3 ubiquitin ligase(s) (23); however, previous genetic and RNA interference (RNAi) screens have not identified any E3 ubiquitin ligase that regulates Smo activity or trafficking (3134). One possibility is that multiple E3 ligases are involved in the regulation of Smo ubiquitylation so that perturbation of individual E3s may not result in an obvious change in Smo abundance and Hh pathway activity. Furthermore, Smo ubiquitylation could be catalyzed by E3 ligases that are not dedicated for Smo such that their inactivation might cause pleiotropic phenotypes. Therefore, we decided to carry out an in vitro RNAi screen using a cell-based Smo ubiquitylation assay (23). From this screen, we identified the Smurf family of HECT domain–containing E3s as Smo ubiquitin ligases. We found that Smurf bound to the Smo autoinhibitory domain (SAID) through its HECT domain to promote Smo ubiquitylation. Hh-induced and PKA-mediated phosphorylation of SAID dissociated Smurf from Smo, thereby inhibiting Smo ubiquitylation. We found that the N-terminal region of Smurf bound to its C-terminally localized HECT domain to prevent Smurf from binding to Smo. Gprk2-mediated phosphorylation of the N-terminal region of Smurf alleviated this autoinhibition and freed the HECT domain for binding to Smo. Smo and Ptc competed for the same pool of Smurf family E3s, and Hh promoted Ptc ubiquitylation by releasing Smurf family members from Smo and further stimulating their binding to Ptc.

RESULTS

Cell-based RNAi screen identifies Smurf family members as Smo ubiquitin ligases

To identify E3 ligase(s) that promotes Smo ubiquitylation, we carried out an RNAi screen using a cell-based ubiquitylation assay (23). We first generated a stable Drosophila S2 cell line expressing an inducible Myc-tagged Smo transgene under the control of the metallothionein promoter (pMT-Myc-Smo). Myc-Smo–expressing cells were transfected with Myc-Smo–expressing cells were transfected with double-stranded RNAs (dsRNAs) targeting Drosophila E3 ligases, and cell lysates were subjected to ubiquitylation assay as previously described (23, 35). We initially focused on the HECT family of E3 ubiquitin ligases because E3s in this family have been implicated in the regulation of GPCR endocytosis (36). We targeted 11 HECT domain E3s from Drosophila including Smurf, Nedd4, and Suppressor of deltex [Su(dx)], which together constitute the Smurf subfamily (Fig. 1, A and B). Among the HECT domain E3s tested, we found that knockdown of Smurf (CG4943) reduced Smo ubiquitylation (Fig. 1C). Although knockdown of either Nedd4 (CG7555) or Su(dx) (CG4244) alone did not noticeably change Smo ubiquitylation, knocking them down in combination with Smurf further diminished Smo ubiquitylation compared to Smurf knockdown alone (Fig. 1, C and D), suggesting that the Smurf subfamily of E3s acts in a partially redundant fashion to promote Smo ubiquitylation. Overexpression of Smurf, Nedd4, or Su(dx), but not two other HECT domain E3s (CG3356 and CG6190), increased Smo ubiquitylation (Fig. 1E). Furthermore, mutating a critical residue in the Smurf catalytic domain (C1029A) (37) abolished its ability to promote Smo ubiquitylation (Fig. 1F).

Fig 1 A cell-based RNAi screen identifies Smurf family members as Smo E3 ubiquitin ligases.

(A) Family tree of the HECT domain E3 ubiquitin ligases that we targeted by RNAi. (B) Schematic drawings of Smurf, Nedd4, and Su(dx), with the C2, WW, and HECT domains indicated. (C) S2 cells stably expressing Myc-Smo were treated with control dsRNA or dsRNA targeting the indicated HECT domain E3 ligases. After treatment with the proteasome inhibitor MG132, Myc-Smo was immunoprecipitated (IP) with an anti-Myc antibody and immunoblotted (IB) with antibodies recognizing ubiquitin or Myc. Loading was normalized by the amount of Myc-Smo. (D) Cell-based ubiquitylation assay. Immunoblot (IB) showing ubiquitylation of Myc-Smo immunoprecipitated (IP) from extracts from S2 cells treated with the indicated dsRNAs. Luc dsRNA is a negative control. (E) Smo ubiquitylation in S2 cells overexpressing Flag (Fg)–tagged Smurf family members Smurf, Nedd4, Su(dx), CG3356, or CG6190. (F) Smo ubiquitylation in S2 cells overexpressing tagged wild type (WT) (Fg-Smurf) and catalytically inactive (Fg-SmurfC1029A). Immunoblots are representative of three independent experiments.

Smurf family E3 ligases reduce Smo cell surface abundance in cultured cells

Consistent with the effect on Smo ubiquitylation, transfection of S2 cells with dsRNA targeting Smurf, but not of dsRNAs targeting Nedd4 or Su(dx), increased cell surface abundance of Myc-Smo, and combined knockdown of Nedd4 and Su(dx) with Smurf further increased cell surface accumulation of Myc-Smo (Fig. 2, A and B). Conversely, overexpression of Smurf, Nedd4, or Su(dx) decreased the cell surface accumulation as well as the overall protein abundance of a phosphomimetic form of Smo (Myc-SmoSD) that contains Ser-to-Asp substitutions at three PKA and CK1 phosphorylation clusters (Fig. 2, C to F) (13).

Fig 2 The Smurf family of E3s regulates Smo cell surface accumulation in cultured cells.

Immunostaining (A) and quantification (B) of cell surface Smo in S2 cells stably expressing Myc-Smo and treated with control (Luc) or the indicated dsRNA. Data are means ± SD from three independent experiments. n = 10 cells for each experimental condition. **P < 0.01, ***P < 0.001 (Student’s t test). (C) Immunostaining showing a mutant form of Smo that mimics phosphorylation by PKA and CK1 (Myc-SmoSD) and Fg-Smurf, Fg-Nedd4, or Fg-Su(dx) (Flag) in S2 cells expressing the indicated tagged proteins. (D) Quantification of cell surface Myc-SmoSD signals in (C). Data are means ± SD from three independent experiments. n = 10 cells for each experimental condition. ***P < 0.001 (Student’s t test). (E) Immunoblot showing Myc-SmoSD and Fg-Smurf, Fg-Nedd4, or Fg-Su(dx) in S2 cells expressing the indicated combinations of tagged proteins. Myc-CFP (cyan fluorescent protein) is a loading control. The antibody that recognizes green fluorescent protein (GFP) also recognizes CFP. (F) Quantification of Myc-SmoSD abundance in (E). Data are means ± SD from three independent experiments. ***P < 0.001 (Student’s t test). (G) Immunoblot showing Smo protein in extracts from cl-8 cells treated with either Hh-conditioned medium (Hh) or dsRNAs targeting Smurf, Nedd4, Su(dx), or all three (3E3s). (H) Quantification of Smo abundance in (G). Data are means ± SD from three independent experiments. **P < 0.01, ***P < 0.001 (Student’s t test). (I and J) Quantification of smo (I) and ptc (J) mRNA by quantitative reverse transcription PCR (qRT-PCR) in cl-8 cells treated with either Hh-conditioned medium or the indicated dsRNAs. (K) Luciferase activity in cl-8 cells expressing the ptc-luc reporter gene and treated with either Hh-conditioned medium or the indicated dsRNAs. Data are means ± SD from three independent experiments (I to K). Immunofluorescence images are representative of 10 cells for each condition. Immunoblots are representative of three independent experiments. Scale bars, 5 μm.

To examine how the Smurf family E3 ligases regulate Hh signaling responses downstream of Smo, we turned to cl-8 cells, a Drosophila cell line that contains all the Hh pathway components and can recapitulate Hh signaling responses as indicated by the stabilization of Smo and increased expression of ptc, a universal readout of Hh pathway activation (33). Consistent with the results obtained using S2 cells, knockdown of Smurf but not Nedd4 and Su(dx) increased the abundance of endogenous Smo, whereas combined knockdown of all three Smurf family members further increased Smo abundance without altering smo mRNA expression (Fig. 2, G to I). Despite the accumulation of Smo in Smurf family knockdown cells, the expression of endogenous ptc mRNA and of a ptc luciferase (ptc-luc) reporter that reflects Hh pathway activation (33) was not increased (Fig. 2, J and K). By contrast, cl-8 treated with Hh-conditioned medium not only accumulated Smo but also exhibited increased expression of endogenous ptc and the ptc-luc reporter gene (Fig. 2, G, H, J, and K). It has been shown that Hh activates signaling events downstream of Smo not only by increasing Smo cell surface abundance but also by inducing Smo to adopt an active conformation (14). The lack of pathway activation downstream of Smo in Smurf family knockdown cells implies that Smo accumulated in these cells most likely adopted an inactive conformation.

We noticed that knockdown of Smurf family members had a less marked effect on Smo accumulation compared to Hh stimulation (Fig. 2, G and H). Therefore, it is possible that Smurf family members are not the only E3 ligases that target Smo. We found that knockdown of Cul4, a member of the Cullin family of E3 ubiquitin ligases (38), enhanced the effect of Smurf knockdown on Smo ubiquitylation and cell surface accumulation (fig. S1, A and B), implying that Cul4 may act in parallel with the Smurf family to control Smo cell surface abundance. Nevertheless, we focused our effort on the Smurf family for the remainder of this study.

Smurf reduces Smo accumulation and Hh signaling in developing wings

To assess the role of Smurf in vivo, we expressed UAS-Smurf-RNAi [Vienna Drosophila Resource Center (VDRC), 24681] or UAS-Flag (Fg)-Smurf (37) using the wing-specific Gal4 driver MS1096 (MS>Smurf-RNAi or MS>Fg-Smurf). To maximize Gal4 activity and UAS-RNAi expression, larvae were grown at 29°C (39). In addition, we examined wing discs only from male larvae because MS1096 is located on the X chromosome, and males express Gal4 more abundantly than do females. We coexpressed UAS-Dicer2 with the RNAi transgene to improve knockdown efficiency (40). Under these optimized conditions, we found that Smurf knockdown caused increased Smo accumulation in anterior compartment cells away from the anteroposterior compartment boundary (Fig. 3, A and B, and fig. S2, A to E). Notably, the Gal4 driver MS1096 is expressed more strongly in the dorsal compartment cells (arrows in Fig. 3, A and B) than in ventral compartment cells (arrowheads in Fig. 3, A and B) (41). As a consequence, only anterodorsal cells in MS>Smurf-RNAi wing discs ectopically accumulated Smo due to more complete knockdown of Smurf in these cells (arrows in Fig. 3B). Similarly, MS>Smurf-RNAi wing discs exhibited increased Ptc abundance in anterodorsal cells near the anteroposterior boundary (Fig. 3, C and D). Consistent with the experiments in cl-8 cells, knockdown of Smurf in wing discs did not induce ectopic expression of a Hh-responsive reporter gene ptc-lacZ (fig. S2, H and I), suggesting that Smo accumulated in an inactive conformation. The accumulation of Ptc protein in anterodorsal cells near the anteroposterior boundary in Smurf knockdown wing discs is most likely due to the effect of Smurf depletion on Ptc turnover (20).

Fig 3 Smurf reduces Smo cell surface accumulation and Hh signaling in developing wings.

(A to D) Late third-instar control (Ctrl) wing discs (A and C) or wing discs expressing UAS-Smurf-RNAi under control of the MS1096 Gal4 driver (B and D) were immunostained for Ci (red), Smo (green in A and B), or Ptc (green in C and D). Wing discs in these and the following panels are oriented with anterior to the left and ventral up. Arrows and arrowheads indicate the dorsal and ventral compartments of the wing pouch region, respectively. Ci staining marks the anterior compartment cells. (E to I) Late third-instar control wing disc (E) or wing discs expressing UAS-Fg-Smurf under control of the MS1096 Gal4 driver either alone (F to H) or together with UAS-Smurf-RNAi (I) using MS1096 were immunostained for Ci (red), Ptc (green in F), En (green in E and G), Smo (green in H and I), or Flag (blue). Dashed lines demarcate the anteroposterior boundary determined by the Ci expression domain (E and G). Arrows indicate the anterodorsal cells near the anteroposterior boundary (E to G) or posterodorsal compartment cells (H to I) of the wing pouch region. Images are representative of five wing discs per genotype. Scale bars, 50 μm.

Overexpression of Smurf in MS>Fg-Smurf wing discs blocked Hh-induced Smo accumulation and diminished the abundance of Ptc and En in anterior compartment cells near the anteroposterior boundary (Fig. 3, E to H), mostly in the dorsal regions of wing discs (arrows). Notably, En expression in anterior compartment cells near the anteroposterior boundary is induced by high levels of Hh (13). Smurf overexpression also reduced En expression in posterior compartment cells (Fig. 3G), which is not under the control of Hh. This is likely due to a pleiotropic effect of excessive Smurf activity. Coexpression of UAS-Smurf-RNAi with UAS-Fg-Smurf using the MS1096 driver effectively eliminated Fg-Smurf expression and restored Smo accumulation in Hh-responding cells in the dorsal region of the wing discs (Fig. 3I).

Consistent with the mild effect of knockdown of Nedd4 or Su(dx) on Smo ubiquitylation in S2 cells, expression of UAS-Nedd4-RNAi or UAS-Su(dx)-RNAi in wing discs under the control of MS1096 did not cause ectopic Smo accumulation (fig. S2, F and G). Wing discs with combined knockdown of Smurf, Nedd4, and Su(dx) were malformed, precluding proper analysis of Smo trafficking in wing discs that lack all the Smurf family members. In contrast to overexpression of Smurf, which inhibited Hh-induced Smo accumulation and ptc-lacZ expression (fig. S3, A and B), overexpression of UAS-Nedd4 or UAS-Su(dx) with the MS1096 driver did not block Hh-induced Smo accumulation and ptc-lacZ expression (fig. S3, C and D). Together, the results from both in vitro and in vivo experiments suggest that Smurf may play a more prominent role than Nedd4 or Su(dx) in the regulation of Smo ubiquitylation and cell surface expression. For this reason, we focused on Smurf but extended some of the key findings to Nedd4 and Su(dx).

Hh inhibits Smo ubiquitylation by dissociating Smurf from Smo

Binding of E3 ubiquitin ligases to their substrates often serves as a regulatory step in the control of protein ubiquitylation (42). Smurf contains a C2 domain in its N-terminal region, a HECT domain in its C-terminal region, and three WW domains in the middle part of the protein (Fig. 4A). As a first step to determine how Hh inhibits Smo ubiquitylation, we asked whether Smurf interacted with Smo in a Hh-dependent manner using coimmunoprecipitation (CoIP) assays. When coexpressed in S2 cells, Flag-tagged Smurf (Fg-Smurf) coimmunoprecipitated with Myc-Smo, and treating cells with Hh-conditioned medium diminished the amount of Fg-Smurf that coimmunoprecipitated with Myc-Smo (Fig. 4B). Deletion of the Smo C-tail (Myc-SmoΔCT) abolished the association between Smo and Smurf (Fig. 4B), suggesting that Smurf binds Smo through the C-tail. Substitution of the three Ser residues in the Smo C-tail that are phosphorylated by PKA (Ser667, Ser687, and Ser740) with Ala (Myc-SmoSA) inhibited Hh-induced dissociation of Smurf from Smo (Fig. 4C). On the other hand, substitution of the three PKA phosphorylation sites and the six adjacent CK1 sites with Asp to mimic phosphorylation (Myc-SmoSD) inhibited the association between Smo and Smurf regardless of the presence of Hh (Fig. 4C). These results demonstrate that Hh-induced phosphorylation of SmoCT by PKA and CK1 is both necessary and sufficient to inhibit Smurf recruitment. Similarly, both Hh and PKA-mediated phosphorylation inhibited the recruitment of Su(dx) and Nedd4 to Smo (fig. S4, A and B). Taken into consideration with our previous finding that ubiquitylation of Myc-SmoSA was resistant to Hh inhibition, whereas Myc-SmoSD exhibited low basal ubiquitylation (23), these results suggest that Hh inhibits Smo ubiquitylation by inhibiting the recruitment of the Smurf family of E3 ligases through PKA-mediated phosphorylation of SmoCT.

Fig 4 Hh-induced phosphorylation of Smo inhibits Smurf recruitment to Smo.

(A) Schematic drawings of full-length Smurf and deletion mutants. (B and C) S2 cells were cotransfected with constructs encoding Fg-Smurf and WT Smo (Myc-SmoWT), a deletion mutant lacking the C-tail (Myc-SmoΔCT), a mutant form of Smo that cannot be phosphorylated by PKA (Myc-SmoSA), or a phosphomimetic form of Smo (Myc-SmoSD) as indicated and treated with control or Hh-conditioned medium (Hh), followed by immunoprecipitation (IP) and immunoblotting (IB) with the indicated antibodies. Cells were treated with the proteasome inhibitor MG132 for 4 hours before lysis. Asterisks indicate the monomeric form of full-length (Myc-SmoWT) and truncated (Myc-SmoΔCT) Smo. (D) S2 cells were cotransfected with Myc-Smo and the indicated deletion forms of Fg-Smurf and subjected to immunoprecipitation (IP) and immunoblotting (IB) with the indicated antibodies. Asterisks indicate the full-length and truncated forms of Smurf (D). (E) Cell-based ubiquitin assay showing ubiquitylation of Myc-Smo immunoprecipitated from S2 cells coexpressing Myc-Smo and a truncated form of Smurf consisting of only the C-terminal HECT domain (Fg-Smurf-HECT). (F) S2 cells were cotransfected with constructs encoding Fg-Smurf-HECT and various forms of Myc-tagged Fz: full-length Fz (Myc-Fz), Fz fused to SAID, the autoinhibitory domain of Smo (Myc-FS), and Fz fused to phosphorylation-deficient (Myc-FSSA) or phosphomimetic (Myc-FSSD) forms of the SAID. The Fz proteins were immunoprecipitated (IP) and then immunoblotted (IB) with antibodies recognizing Myc or Fg. (G) S2 cells were cotransfected with constructs encoding full-length Fg-Smurf and Myc-FS, Myc-FSSA, or Myc-FSSD. The Myc-tagged proteins were immunoprecipitated (IP) and immunoblotted (IB) with antibodies recognizing Myc or Fg. (H) S2 cells with endogenous Smo depleted by dsRNA targeting smo 5′ untranslated region (5′UTR) were cotransfected with constructs encoding Myc-Smo, Myc-SmoΔSAID, and Fg-Smurf. The Myc-tagged proteins were immunoprecipitated (IP) and immunoblotted (IB) with antibodies recognizing Myc or Fg. Blots are representative of three independent experiments.

To identify the Smurf domain responsible for Smo binding, we generated several truncated forms of Smurf (Fig. 4A) and used these in CoIP experiments. Deletion of the C2 domain (SmurfΔC2), WW domains (SmurfΔWW), or both the C2 and WW domains (Smurf-HECT) did not affect Smo binding; however, deletion of the HECT domain (SmurfΔHECT) abolished Smo binding, suggesting that the HECT domain of Smurf is both necessary and sufficient to mediate binding to Smo (Fig. 4D). Consistent with this, coexpression of Smurf-HECT with Myc-Smo increased Smo ubiquitylation (Fig. 4E).

The middle part of SmoCT (amino acids 661 to 818) contains an autoinhibitory domain (SAID) that includes the three PKA and CK1 phosphorylation clusters mutated in Myc-SmoSA and Myc-SmoSD (14). When fused to a heterologous protein such as the Wnt receptor Fz2 to generate Fz2-SAID (FS), the SAID domain confers robust ubiquitylation of FS (23), suggesting that the SAID domain might recruit ubiquitin ligases to Smo. We found that both full-length Smurf and its HECT domain coimmunoprecipitated with Myc-FS (Fig. 4, F and G), whereas deletion of the SAID domain from Smo (SmoΔSAID) abolished Smurf recruitment (Fig. 4H). In addition, the phospho-deficient variant of FS (FSSA) exhibited increased binding to Smurf-HECT, whereas the phospho-mimicking variant (FSSD) exhibited decreased binding to both full-length Smurf and its HECT domain (Fig. 4, F and G). Therefore, the Smo SAID domain recruits the HECT domain of Smurf, and Hh inhibits Smurf recruitment to Smo by inducing phosphorylation of SAID by PKA and CK1.

Gprk2 regulates Smo ubiquitylation and cell surface expression by promoting Smurf recruitment

Gprk2 plays a dual role in Hh signaling (28, 30). In the presence of Hh, Gprk2 phosphorylates Smo and facilitates dimerization or oligomerization of SmoCT to promote high levels of Smo activity (28, 43). However, in the absence of Hh, Gprk2 reduces Smo cell surface abundance through an unknown mechanism (28, 30). We found that knockdown of Gprk2 reduced ubiquitylation of Myc-Smo in S2 cells and that combined knockdown of Gprk2 and Smurf resulted in a more marked reduction of Myc-Smo ubiquitylation than single knockdown (Fig. 5A). On the other hand, overexpression of Gprk2, but not of the kinase-dead mutant Gprk2KM, enhanced Myc-Smo ubiquitylation (Fig. 5B). These results suggest that Gprk2 reduces cell surface abundance of Smo, at least in part, by promoting Smo ubiquitylation in a manner that depends on Gprk2 kinase activity. This is consistent with our previous finding that Gprk2KM failed to prevent Smo accumulation in gprk2 mutant wing discs (28).

Fig 5 Gprk2 promotes Smurf recruitment to Smo.

(A and B) Cell-based ubiquitylation assay of Myc-Smo in S2 cells treated with Smurf and Gprk2 dsRNAs either alone or in combination (A), or coexpressed with Gprk2 (Fg-Gprk2) or the kinase-dead form Fg-Gprk2KM (B). (C) Immunostaining showing Smo and GFP in late third-instar wing discs expressing UAS-GFP under the control of the brk Gal4 driver. (D to F) Immunostaining showing Smo in late third-instar wing discs expressing UAS-Smurf-RNAi (D), UAS-Gprk2-RNAi (E), or UAS-Smurf-RNAi + UAS-Gprk2-RNAi (F) under control of the brk gal4 driver. Arrows indicate the anterior brk expression domain. (G to I) S2 cells were transfected with the indicated constructs and treated with either control (Luc) dsRNA or dsRNAs targeting the coding sequence (G and I) or 5′UTR (H) of Gprk2, followed by immunoprecipitation (IP) and immunoblotting (IB) with the indicated antibodies. Cells were treated with the proteasome inhibitor MG132 for 4 hours before lysis. Images are representative of five wing discs per genotype. Blots are representative of three independent experiments.

To further determine the functional relationship between Gprk2 and Smurf in the regulation of Smo cell surface accumulation, we carried out genetic interaction experiments. We used brinker-Gal4 (brk-Gal4), which is expressed in the peripheral regions of wing discs (Fig. 5C) (44), to drive expression of UAS-Gprk2-RNAi and UAS-Smurf-RNAi alone and in combination. Knockdown of Smurf (brk>Smurf-RNAi) had little, if any, effect on the abundance of Smo, whereas knockdown of Gprk2 (brk>Gprk2-RNAi) only slightly increased Smo abundance in the anterior brk expression domain, where there was no Hh (Fig. 5, D and E, compared with Fig. 5C). However, combined knockdown of Smurf and Gprk2 resulted in a greater increase in Smo in the anterior brk expression domain (Fig. 5F), indicating that Gprk2 and Smurf cooperatively regulate Smo cell surface abundance.

We next determined whether Gprk2 promotes Smo ubiquitylation by regulating the binding of Smurf to Smo. Knockdown of Gprk2 reduced the amount of Myc-Smo that was associated with Fg-Smurf in S2 cells (Fig. 5G). Expression of Gprk2, but not Gprk2KM, restored the interaction between Smurf and Smo in S2 cells in which the endogenous Gprk2 was depleted by dsRNA targeting the 5′UTR of gprk2 (Fig. 5H), indicating that Gprk2 promotes Smurf recruitment depending on its kinase activity. The regulation of the association between Smo and Smurf by Gprk2 requires the N-terminal region of Smurf because Gprk2 RNAi did not reduce the binding of Smurf-HECT to Smo (Fig. 5I), implying that Gprk2 may promote Smurf binding to Smo by releasing an inhibitory effect imposed by the Smurf N-terminal region.

Gprk2-mediated phosphorylation of Smurf is promoted by Smo but inhibited by Hh

We hypothesized that the N-terminal region of Smurf prevents the HECT domain from binding to Smo and that Gprk2-mediated phosphorylation of Smurf relieves this autoinhibition. Coexpression of Fg-Smurf with Myc-Gprk2, but not Myc-Gprk2KM, in S2 cells resulted in a mobility shift of Fg-Smurf on SDS–polyacrylamide gel electrophoresis, which was abolished by phosphatase treatment (Fig. 6, A and B), suggesting that Gprk2 promotes Smurf phosphorylation. The Gprk2-induced mobility shift of Smurf was inhibited by treating cells with Hh-conditioned medium or by knockdown of Smo (Fig. 6, A to C), suggesting that Smo promotes, whereas Hh inhibits, Smurf phosphorylation by Gprk2. By CoIP assays, we found that Myc-Gprk2 formed a complex with Fg-Smurf and that this association was inhibited by Hh stimulation, Ptc RNAi, or Smo RNAi (Fig. 6D). Because Smo interacts with both Smurf and Gprk2 (Fig. 4) (28), we propose that Smo functions as a scaffold to bring Gprk2 and Smurf into close proximity with one another in the absence of Hh. This facilitates Smurf phosphorylation by Gprk2, which increases the binding affinity of Smurf for Smo and allows Smurf to ubiquitylate Smo. Hh stimulates the phosphorylation of Smo, which causes Smurf to dissociate from Smo and Gprk2, thereby diminishing Gprk2-mediated phosphorylation and activation of Smurf.

Fig 6 Gprk2 promotes Smo-Smurf association by phosphorylating Smurf.

(A and B) Immunoblots (IB) of lysates from S2 cells expressing Fg-Smurf and Myc-Gprk2 or Myc-Grpk2KM as indicated and treated with control or Hh-conditioned medium (Hh) in the absence or presence of λ phosphatase (λ-PP). (C) Immunoblots of lysates from S2 cells expressing Fg-Smurf and Myc-Gprk2 as indicated in the presence or absence of control (Luc) or Smo dsRNA. (D) Immunoblot (IB) showing Fg-Smurf and Myc-Grpk2 in Myc immunoprecipitates (IP) from S2 cells expressing the Myc-Gprk2 and Fg-Smurf and subjected to treatment with Hh-conditioned medium or dsRNAs as indicated. (E and F) Western blot of immunoprecipitates from S2 cells expressing the indicated combinations of the N terminus of Smurf (Fg-SmurfΔHECT), a form of SmurfΔHECT in which the Gprk2 phosphorylation site was mutated (Fg-SmurfΔHECT-SA), or the HECT domain of Smurf (Fg-Smurf-HECT) with WT (Myc-Gprk2) and kinase-dead (Myc-Gprk2KM) forms of Gprk2. (G) Western blot of immunoprecipitates from S2 cells expressing full-length Smurf (Fg-Smurf) or a mutant form of Smurf in which the Gprk2 phosphorylation site was mutated (Fg-SmurfSA) with wild type (WT) Myc-Gprk2. WT sequence of the N-terminal Ser cluster and the SA or SD substitutions are shown. (H) Immunoblot showing Smurf and Smo in immunoprecipitates from cells expressing Myc-Smo and the indicated forms of Smurf. (I) Immunoblot showing ubiquitylation of Smo in S2 cells expressing Myc-Smo and the indicated forms of Smurf. (J) Immunoblot showing Smurf and Smo in immunoprecipitates from cells expressing Myc-Smo and the indicated forms of Smurf and treated with control (Luc) or Gprk2 dsRNAs as indicated. (K and L) Immunoblot analysis of immunoprecipitates (IP) from cells expressing the indicated proteins. Note that Fg-Gprk2 and Fg-SmurfΔHECT or Fg-SmurfΔHECT-SA overlapped on Western blot. Blots are representative of three independent experiments.

Gprk2-mediated phosphorylation of Smurf blocks an intramolecular interaction that prevents Smurf from binding to Smo

To determine the effect of Gprk2-mediated phosphorylation on the ability of Smurf to bind Smo, we first identified the relevant phosphorylation sites in Smurf. By coexpressing Myc-Gprk2 with truncated Fg-Smurf variants in S2 cells, we found that Gprk2 induced a mobility shift of the SmurfΔHECT but not of Smurf-HECT (Fig. 6, E and F), suggesting that Gprk2-induced phosphorylation of Smurf occurred in its N-terminal region.

Inspection of the Smurf sequence identified a cluster of Ser residues (SEDDSSEDS; Fig. 6G) that conforms to consensus sites for Gprk2, which, like other members of the GRK family of kinases, tends to phosphorylate Ser and Thr residues in an acidic environment (28, 45). We mutated the Ser residues in this cluster to either Ala (SA; Fig. 6G) or Asp (SD; Fig. 6G) in both full-length Smurf (SmurfSA or SmurfSD) and SmurfΔHECT (SmurfΔHECT-SA). We found that Gprk2 did not induce mobility shift of either SmurfΔHECT-SA or SmurfSA (Fig. 6, E and G), indicating that Gprk2-mediated phosphorylation of Smurf occurred at this Ser cluster. Consistent with Gprk2-mediated phosphorylation of Smurf promoting its association between Smurf and Smo, SmurfSD exhibited increased binding and ubiquitylation activity toward Smo compared with wild-type Smurf, whereas SmurfSA showed decreased binding to and ubiquitylation of Smo (Fig. 6, H and I). Furthermore, the binding of SmurfSD or SmurfSA to Smo no longer depended on Gprk2 (Fig. 6J). Together, these results demonstrate that Gprk2 promotes Smurf recruitment to and ubiquitylation of Smo by phosphorylating the Smurf N-terminal region.

In considering the mechanism by which Gprk2-mediated phosphorylation of Smurf promotes its binding to Smo, we speculated that the N-terminal region of Smurf may interact with its HECT domain to interfere with the binding of the HECT domain to Smo and that this autoinhibition was prevented by Gprk2-mediated phosphorylation of Smurf. We found that hemagglutinin (HA)–Smurf–HECT coimmunoprecipitated with Fg-SmurfΔHECT when cotransfected into S2 cells, and this association was diminished by cotransfection with Myc-Gprk2 (Fig. 6K). By contrast, the association between HA-Smurf-HECT and Fg-SmurfΔHECT-SA was not inhibited by Gprk2 (Fig. 6K), suggesting that Gprk2-mediated phosphorylation inhibits an intramolecular interaction between the Smurf N terminus and its HECT domain. Furthermore, coexpression of Fg-SmurfΔHECT interfered with the binding of HA-Smurf-HECT to Myc-Smo, but this inhibition was released by Gprk2 (Fig. 6L). By contrast, Gprk2 failed to release the inhibition of Smo binding to Smurf-HECT by Fg-SmurfΔHECT-SA (Fig. 6L). Hence, the intramolecular interaction between the N- and C-terminal regions, which may cause Smurf to adopt a closed conformation, inhibits Smurf recruitment to Smo, and Gprk2-mediated phosphorylation of Smurf alleviates this autoinhibition.

Hh promotes Ptc-Smurf complex formation and Ptc ubiquitylation by releasing the Smurf family of E3s from Smo

Hh controls reciprocal trafficking of Ptc and Smo (12). In the absence of Hh, Ptc is internalized and degraded at modest levels, and binding of Hh to Ptc further stimulates Ptc internalization and degradation (12, 18, 19, 21, 46). Whereas ligand-independent Ptc internalization modulates cell sensitivity to the Hh morphogen (18), internalization and degradation of Hh-bound Ptc play a critical role in limiting the spread of the Hh ligand (15, 16). Previous studies suggested that the Smurf family of E3 ubiquitin ligases is responsible for Ptc ubiquitylation and internalization (20, 21). Consistent with these findings, Hh stimulation increased ubiquitylation of Myc-tagged Ptc (Myc-Ptc) in S2 cells, and Smurf RNAi inhibited both basal and Hh-stimulated Myc-Ptc ubiquitylation (Fig. 7A). Nedd4 or Su(dx) RNAi also reduced Myc-Ptc ubiquitylation, and combined RNAi of all three Smurf family members resulted in a greater reduction of Myc-Ptc ubiquitylation than did RNAi of any one Smurf family member alone (fig. S5). Furthermore, Hh increased the binding of Myc-Ptc to Fg-Smurf (Fig. 7B) as well as to Flag-tagged Nedd4 and Su(dx) (fig. S6, A and B), suggesting that Hh promotes Ptc internalization and degradation by recruiting Smurf family members to ubiquitylate Ptc.

Fig 7 Hh promotes Ptc ubiquitylation by dissociating Smurf from Smo and facilitating its binding to Ptc.

(A) Ptc ubiquitylation assay in S2 cells treated with control or Hh-conditioned medium (Hh) in the presence of Luc, Smo, Smurf, or both Smo and Smurf dsRNA. (B) Immunoblotting (IB) or immunoprecipitates (IP) from S2 cells coexpressing Myc-Ptc and Fg-Smurf and treated with Hh-conditioned medium in the presence or absence of the indicated dsRNAs. (C and D) L>Ptc-GFP expression in control wing discs (C) or wing discs containing clones expressing UAS-Smo-RNAi under the control of act-Gal4 (D). Smo depletion decreased Ptc-GFP in anterior compartment cells away from the anteroposterior boundary (arrows in D) but not in posterior compartment cells (arrowheads in D). Images are representative of five wing discs per genotype. (E and F) Ptc-Smurf association (E) or Ptc ubiquitylation (F) in S2 cells expressing the indicated constructs in the presence of Luc or Smo dsRNA. (G and H) Ubiquitylation of Myc-Ptc or Myc-PtcΔL2 in S2 cells treated with control or Hh-conditioned medium in the presence of Luc or Ptc dsRNA (G) or in the presence of both Smo and Ptc dsRNA (H). (I) Smurf-Ptc (or PtcΔL2) association in S2 cells transfected with the indicated dsRNA and treated with control or Hh-conditioned medium. Blots are representative of three independent experiments.

Because Hh stimulates the dissociation of Smurf family members from Smo (Fig. 4, B and C, and fig. S4, A and B), which should increase the pool of free Smurf family E3s that are available to bind to Ptc, we wondered whether the increase in Ptc-Smurf association and Ptc ubiquitylation induced by Hh were, at least in part, mediated by Hh-induced dissociation of Smurf, Nedd4, and Su(dx) from Smo. We found that Smo RNAi phenocopied Hh stimulation because it enhanced both the interaction between Myc-Ptc and Fg-Smurf, Fg-Nedd4, or Fg-Su(dx) and the ubiquitylation of Myc-Ptc (Fig. 7, A and B, and fig. S6, A and B). Furthermore, Smo knockdown did not block Hh-stimulated ubiquitylation of Myc-Ptc (Fig. 7A) or Hh-induced binding of Fg-Smurf to Myc-Ptc (Fig. 7B), suggesting that Smo activity and signaling downstream of Smo are not required for the regulation of Ptc ubiquitylation by Hh.

To determine the effect of Smo depletion on Ptc turnover in vivo, we took advantage of a Ptc reporter, L>Ptc-GFP, which expresses GFP-tagged Ptc under the control of a ribosomal promoter (18). Although the expression of L>Ptc-GFP is ubiquitous, the amount of Ptc-GFP protein is lower in posterior compartment cells than in anterior compartment cells away from the anteroposterior boundary due to the presence of Hh in the posterior compartment cells (Fig. 7C) (18). The extremely low abundance of Ptc-GFP in anterior compartment cells near the anteroposterior boundary is due to the presence of high levels of endogenous Ptc that can oligomerize with Ptc-GFP and promote its internalization and degradation (18). Smo RNAi clones in wing discs reduced the abundance of Ptc-GFP when the clones were situated in the anterior compartment away from the anteroposterior boundary (arrows in Fig. 7D). By contrast, Smo RNAi clones did not change Ptc-GFP abundance in the posterior compartment where the Ptc-GFP level was already decreased in response to Hh (arrowheads in Fig. 7D).

These results suggest that Smo and Ptc may compete for binding to Smurf family members and that Smo depletion or Hh-induced dissociation of the Smurf family members from Smo increases their availability for binding to Ptc, leading to increased Ptc ubiquitylation and turnover. In further support of this notion, overexpression of Smo-CFP or SmoSA-CFP reduced the association between Fg-Smurf and Myc-Ptc (Fig. 7E) and diminished Myc-Ptc ubiquitylation (Fig. 7F).

Because Smurf dissociated from Smo is not phosphorylated by Gprk2 and thus may adopt a closed conformation (Fig. 6), we asked whether the phosphorylation state of Smurf affects the interaction between Ptc and Smurf. CoIP experiments indicated that Myc-Ptc interacted equally well with Fg-SmurfWT, Fg-SmurfSA, and Fg-SmurfSD (fig. S7), suggesting that Ptc binds Smurf regardless of whether or not Smurf is phosphorylated by Gprk2. This is in contrast to Smo, which preferentially binds Smurf that has been phosphorylated by Gprk2 (Fig. 6H).

It has been suggested that activation of Smurf by Smo mediates the effect of Hh on Ptc ubiquitylation and degradation because overexpression of SmoSD increases Ptc ubiquitylation (20). Therefore, we were surprised to see that depletion of Smo also increased Ptc ubiquitylation. Consistent with the observation that SmoSD increases Ptc ubiquitylation (20), overexpression of SmoSD-CFP increased the amount of Fg-Smurf that coimmunoprecipitated with Myc-Ptc from S2 cell extracts (Fig. 7E) and enhanced Myc-Ptc ubiquitylation (Fig. 7F). Our previous study revealed that SmoSD can recruit PKAc to transphosphorylate wild-type Smo that dimerizes with it (47). Therefore, we hypothesized that SmoSD may stimulate phosphorylation of endogenous Smo to release Smurf, thereby mimicking the effect of Hh stimulation. SmoSD-CFP did not stimulate the binding of Fg-Smurf to Myc-Ptc, nor did it increase Myc-Ptc ubiquitylation, in S2 cells in which endogenous Smo was depleted by dsRNA targeting the 5′UTR of smo (Fig. 7, E and F). Furthermore, the increased association between Myc-Ptc and Fg-Smurf and the ubiquitylation of Myc-Ptc in Smo-depleted cells were suppressed by the expression of Smo-CFP or SmoSA-CFP (Fig. 7, E and F), suggesting that unphosphorylated Smo competes with Ptc for the Smurf family of E3s.

Binding of Hh to Ptc further stimulates the recruitment of Smurf to Ptc independent of Smo

We noticed that, in contrast to SmoSD, Hh could further stimulate Ptc-Smurf association and Ptc ubiquitylation in Smo-depleted cells (Fig. 7, A and B), suggesting that Hh may promote Ptc ubiquitylation through a mechanism independent of Smo. One possibility is that binding of Hh to Ptc increases its binding affinity to Smurf. To test this, we took advantage of a mutant form of Ptc, PtcΔL2, that no longer binds Hh but retains its ability to inhibit Smo (17, 48, 49). Unlike Myc-Ptc ubiquitylation, which was markedly stimulated by Hh, Myc-PtcΔL2 ubiquitylation was only marginally induced by Hh (Fig. 7G). The residual regulation of Myc-PtcΔL2 by Hh is likely due to endogenous wild-type Ptc that can oligomerize with Myc-PtcΔL2 (18, 19). Knockdown of endogenous Ptc by targeting its 5′UTR and 3′UTR completely abolished the regulation of Myc-PtcΔL2 ubiquitylation by Hh but did not affect Hh-stimulated ubiquitylation of Myc-Ptc (Fig. 7G). These results demonstrate that binding of Hh to Ptc is required to stimulate Ptc ubiquitylation by Smurf.

Despite Myc-PtcΔL2 being unable to respond to Hh, Smo RNAi increased the ubiquitylation of Myc-PtcΔL2 similarly to that of Myc-Ptc in cells with endogenous Ptc depleted (Fig. 7H). Consistent with this, the binding of Fg-Smurf to both Myc-Ptc and Myc-PtcΔL2 was increased in Smo-depleted cells (Fig. 7I), confirming that loss of Smo increases the pool of Smurf that is available for binding to Ptc. In Smo- and Ptc-depleted cells, Hh further increased the binding of Fg-Smurf to Myc-Ptc (Fig. 7I) as well as the ubiquitylation of Myc-Ptc (Fig. 7H). In contrast, Hh did not increase the binding of Smurf to and ubiquitylation of Myc-PtcΔL2 in these cells (Fig. 7, H and I). These results suggest that binding of Hh to Ptc could further increase the recruitment of Smurf to Ptc even after Smurf was released from Smo.

DISCUSSION

It has been well documented that Hh stimulates reciprocal trafficking of Ptc and Smo and that Hh-induced phosphorylation of Smo promotes Smo cell surface accumulation; however, the underlying molecular mechanisms have remained obscure. Here, we identified the Smurf family of HECT domain–containing ubiquitin E3 ligases as essential for Smo ubiquitylation and cell surface clearance. We demonstrated that Hh inhibits Smo ubiquitylation by preventing the recruitment of these E3 ligases through PKA-mediated phosphorylation of Smo. In addition, we provided evidence that Smo and Ptc compete for a common pool of E3s and that Hh stimulates Ptc ubiquitylation by both liberating Smurf family members from Smo and further stimulating their binding to Ptc (Fig. 8).

Fig. 8 Model for how Hh regulates the reciprocal trafficking of Smo and Ptc through Smurf.

In the absence of Hh (left), Gprk2-mediated phosphorylation of Smurf promotes Smurf association with Smo, leading to Smo ubiquitylation, internalization, and degradation, and Ptc is modestly ubiquitylated by Smurf. In the presence of Hh (right), Smo recruits PKA, which phosphorylates the Smo C-tail, thus inhibiting Smurf binding to and ubiquitinating Smo. Hh promotes the binding of released Smurf to Ptc, leading to increased Ptc ubiquitylation and degradation. The different shapes of unphosphorylated and phosphorylated Smurf represent open (ovals) and closed (circles) conformations.

Regulation of Smo ubiquitylation and cell surface expression by multiple E3 ubiquitin ligases, Gprk2, and PKA

It has been shown that ubiquitylation of Smo controls its cell surface abundance and that the Hh-induced phosphorylation of Smo by PKA inhibits Smo ubiquitylation to promote its cell surface accumulation (23, 24). Hh also stimulates sumoylation of Smo, which, in turn, recruits USP8 to deubiquitylate Smo in a manner that does not depend on phosphorylation of Smo by PKA (26), leaving open the question of how PKA-mediated phosphorylation of Smo inhibits its ubiquitylation and degradation. Using a cell-based Smo ubiquitylation assay that is sensitive to perturbation, we identified the Smurf family of HECT domain–containing E3 ubiquitin ligases, including Smurf, Nedd4, and Su(dx), as Smo E3s. Functional studies using both gain- and loss-of-function approaches revealed that Smurf plays a more prominent role than Nedd4 or Su(dx) in the regulation of Smo ubiquitylation and cell surface accumulation.

The Smurf family E3s contain multiple WW domains that normally bind to substrates containing a PPXY motif, as is the case for Ptc (1921). However, we found that the region containing all three WW domains of Smurf was dispensable for its binding to Smo. Instead, Smurf interacted with Smo through its HECT domain. A truncated form of Smo consisting of only the HECT domain (Smurf-HECT) can bind and ubiquitylate Smo. The interaction between Smurf-HECT and Smo is inhibited by the N-terminal region of Smurf, and this autoinhibition is alleviated by Gprk2-mediated phosphorylation of a Ser cluster near the N-terminal region of Smurf. Depletion of Gprk2 compromised Smurf recruitment to Smo and the subsequent ubiquitylation of Smo in S2 cells, which explains, at least in part, why Smo abundance is increased in gprk2 mutant or Gprk2 RNAi wing discs (2830). Hence, Gprk2 regulates Smo trafficking by increasing a local pool of phosphorylated Smurf that is competent to bind to and ubiquitylate Smo (Fig. 8). We propose that Smurf may adopt two different conformations depending on its phosphorylation state. Unphosphorylated Smurf adopts a closed conformation in which the N-terminal region interacts with the C-terminal region to prevent its interaction with Smo. Gprk2-mediated phosphorylation of the N-terminal region disrupts its interaction with the C-terminal region, allowing Smurf to adopt an open conformation capable of binding to Smo. In contrast, Gprk2-mediated phosphorylation of Smurf did not seem to affect its interaction with Ptc, because Ptc bound to wild-type, phosphorylation-deficient, and phosphomimetic forms of Smurf equally well (fig. S7). Because Ptc interacts with the WW domains of Smurf (20), this observation suggests that the WW domains of Smurf are likely to be exposed in both the closed and open conformations. An analogous conformational change has been attributed to the regulation of the E3 ubiquitin ligase Itch by Jun N-terminal kinase 1 (JNK1), in which phosphorylation of Itch disrupts an inhibitory interaction between its WW and HECT domains (50). However, in the case of Itch, the phosphorylation-induced conformational change appears to enhance its catalytic activity rather than increase substrate binding. It would be interesting to determine whether other HECT domain–containing E3s also recognize substrates through their HECT domains in a manner that is regulated by autoinhibition and phosphorylation.

Smo recruits Smurf through the SAID located in the Smo C-tail, which explains, at least in part, why deletion of SAID resulted in Smo cell surface accumulation, whereas addition of SAID to Fz2 resulted in ubiquitylation and internalization of the fusion protein (14, 23). SAID contains three Ser clusters phosphorylated by PKA and CK1 upon Hh stimulation (14). Phosphomimetic mutations of the three PKA and CK1 phosphorylation clusters (SmoSD) diminished the binding of Smurf, Su(dx), and Nedd4 to Smo, whereas phospho-deficient mutations of the three PKA sites (SmoSA) prevented Hh-induced Smurf dissociation from the mutant Smo. In conjunction with our previous findings that SmoSD exhibited diminished ubiquitylation and accumulated at high levels on the cell surface in the absence of Hh, whereas SmoSA was ubiquitylated and failed to accumulate on the cell surface even after Hh stimulation (13, 23), these results suggest that Hh-stimulated phosphorylation of Smo by PKA and CK1 inhibits Smo ubiquitylation by dissociating the Smurf family of E3s from Smo.

The phenotypes associated with Smurf knockdown were weak in general. To observe ectopic Smo accumulation in wing discs, we had to optimize the RNAi efficiency (Fig. 3), which may explain why a previous study failed to observe this phenotype (20). Aside from the partially redundant functions of Nedd4 and Su(dx), additional E3(s) may also participate in the regulation of Smo cell surface accumulation, which could explain why the effect of combined knockdown of all the Smurf family members on Smo accumulation was less marked compared with Hh stimulation (Fig. 2E). Knockdown of Cul4 enhanced the effect of Smurf depletion on Smo ubiquitylation and cell surface expression (fig. S1), suggesting that Cul4 may act in parallel with the Smurf family to regulate Smo ubiquitylation and trafficking. Further study is needed to determine how Cul4 regulates Smo ubiquitylation and how this regulation is inhibited by Hh.

It has been shown that Hh stimulates sumoylation of Smo in a manner that does not depend on Smo phosphorylation by PKA and that sumoylation recruits USP8 to deubiquitylate Smo (26, 51). Hence, Hh regulates Smo trafficking and cell surface expression through two paralleled mechanisms: PKA-dependent inhibition of ubiquitylation and PKA-independent but sumoylation-dependent deubiquitylation. We speculate that the involvement of multiple parallel regulatory mechanisms that control Smo trafficking and cell surface expression may allow graded Hh signals to fine-tune the subcellular localization and abundance of Smo to generate precise signaling outputs. It would be interesting to determine whether these two mechanisms are differentially regulated by graded Hh signals.

The Smurf family of E3s mediates the reciprocal trafficking of Smo and Ptc

It is interesting that the Smurf family of E3 ligases also regulates the trafficking of Ptc (1921). Like the regulation of Smo, Smurf also appears to play a more prominent role in the regulation of Ptc ubiquitylation and turnover than the other Smurf family members (fig. S5) (20), raising an important question of how Hh signaling coordinately regulates the activity of the Smurf family members to achieve appropriate regulation of reciprocal trafficking of Ptc and Smo. Here, we demonstrate that Hh inhibits the ubiquitylation of Smo while enhancing the ubiquitylation of Ptc by switching Smurf substrates from Smo to Ptc. Mechanistically, we found that Smo and Ptc compete for the same pool of Smurf and that Gprk2-mediated phosphorylation of Smurf increases its binding to Smo, thus limiting the amount of Smurf that is accessible to Ptc. Binding of Hh to Ptc alleviates its inhibition of Smo, allowing PKA to phosphorylate Smo to release Smurf, thus increasing the amount of free Smurf accessible to Ptc (Fig. 8). Depletion of Smo increased the amount of Smurf bound to Ptc, and consequently the ubiquitylation of Ptc in S2 cells, and decreased Ptc abundance in wing discs, which mimics the effect of Hh stimulation. In addition, Hh further increased Ptc-Smurf association and Ptc ubiquitylation in Smo-depleted cells, and this stimulation depended on the binding of Ptc to Hh, suggesting that Hh-bound Ptc may have increased binding affinity for Smurf. It is possible that binding of Hh to Ptc changes the conformation of Ptc or elicits a modification of its intracellular tail to facilitate Smurf binding. Because both Hh stimulation and Smo deletion increased the association of Ptc with Su(dx) and Nedd4 (fig. S6), it is likely that the mechanism underlying Smurf regulation could also be extended to Su(dx) and Nedd4.

A previous study suggested that Smo activates Smurf to regulate Ptc turnover in response to Hh because a constitutively active form of Smo (SmoSD) promotes Ptc ubiquitylation (20). However, we showed that depletion of Smo did not block Hh-stimulated ubiquitylation of Ptc, nor did it inhibit the binding of Smurf to Ptc (Fig. 7, A and B), suggesting that Smo activity per se is not required for Hh-stimulated ubiquitylation of Ptc. Because our previous study demonstrated that Smo can form dimers or oligomers and SmoSD can recruit PKAc to transphosphorylate the wild-type Smo that dimerizes with it (47), it is conceivable that overexpression of SmoSD induces phosphorylation of endogenous Smo by PKA to release Smurf, thereby increasing the accessibility of Smurf to Ptc. In support of this view, depletion of Smo phenocopied overexpression of SmoSD in the regulation of Ptc-Smurf complex formation and Ptc ubiquitylation. Furthermore, overexpression of SmoSD had no effect on Ptc-Smurf interaction and Ptc ubiquitylation in Smo-depleted cells (Fig. 7, E and F). Hence, SmoSD acts through a derepression mechanism, alleviating the inhibition of Ptc ubiquitylation by endogenous Smo. Because other Smurf family members including Su(dx) and Nedd4 regulate both Ptc and Smo ubiquitylation and their interactions with Ptc and Smo are reciprocally regulated by Hh, we propose that Hh signaling may redirect multiple E3s from Smo to Ptc to regulate their reciprocal trafficking. It is interesting to note that Hh also redirects PKA and CK1 to reciprocally regulate the phosphorylation of Smo and Ci (47, 52). Hence, employment of the same set of enzymes to reciprocally regulate the activities of different pathway components in signaling “on” and “off” states is a recurring theme in Hh signal transduction.

Vertebrate Hh signaling also affects reciprocal trafficking of Ptc and Smo, with Ptc exiting from the primary cilium while Smo moves in (9, 10). It has been shown that the mammalian Smurf family members Smurf1 and Smurf2 promote Ptch1 ubiquitylation and that Smurf-mediated endocytosis of Ptch1 is required for its ciliary clearance and proper Hh signaling (22). Although it is not clear whether the ubiquitylation pathway is involved in the regulation of mammalian Smo ciliary trafficking, it has been shown that sumoylation of mammalian Smo facilitates its ciliary localization (26). Therefore, it would be interesting to determine in the future whether Smo trafficking is regulated by ubiquitylation and whether Hh regulates the reciprocal trafficking of Smo and Ptch1 in mammals through common E3 ligases.

MATERIALS AND METHODS

Constructs and transgenes

The following transgenic fly stocks were used: UAS-Smurf-RNAi (VDRC, 24681), UAS-Su(dx)-RNAi (VDRC, 21814 and 103814), UAS-Nedd4-RNAi (VDRC, 13121; BL, 34741), UAS-Gprk2-RNAi (BL, 34843), UAS-Smurf-Flag (37), UAS-Nedd4 and UAS-Su(dx) (53), L>Ptc-GFP (18), ptc-lacZ (second and third chromosomal insertion lines) (13), MS1096-Gal4 (54), and brk-Gal4 (BL, 53707). The UAS-Smo constructs Myc-Smo, Myc-SmoSA (SmoSA123), Myc-SmoSD (SmoSD123), Myc-SmoΔCT, Myc-SmoΔSAID, Smo-CFP, SmoSA-CFP, and SmoSD-CFP were previously described (13, 14, 55). The Fz2 and Fz2-Smo fusion constructs Myc-Fz, Myc-FS, Myc-FSSA, and Myc-FSSD were previously described (23). Fg-Gprk2, Fg-Gprk2KM, Myc-Gprk2, and Myc-Gprk2 KM were previously described (28). DNA constructs for Fg-Smurf and its deletion mutants WW, HECT, ΔC2, ΔWW, ΔHECT, Fg-Nedd4, and Fg-Su(dx) were generated by subcloning the corresponding complementary DNA (cDNA) coding sequence into the pUAST vector containing a Flag tag (56). To construct HA-Smurf-HECT, the corresponding cDNA fragment was amplified by polymerase chain reaction (PCR) and subcloned in frame with three copies of HA tags in the pUAST-3HA vector (56). PCR-based site-directed mutagenesis was used to generate SmurfSA, SmurfSD, and SmurfΔHECT-SA. Myc-Ptc has six copies of Myc tags fused to the C terminus of full-length Ptc. For making UAS-Myc-PtcΔL2, PtcΔL2 coding sequence was amplified from PtcΔL2-GFP (provided by A. Casali) to replace Ptc coding sequence in UAS-Myc-Ptc.

Cell culture and transfection

Drosophila S2 cells were cultured in Drosophila serum-free medium (SFM) (Invitrogen) with 10% fetal bovine serum, penicillin (100 U/ml), and streptomycin (100 mg/ml) at 23°C. cl-8 cells were cultured in Shields and Sang M3 Insect Medium (Sigma) with 2.5% fetal bovine serum, 2.5% fly extract, insulin (0.125 IU/ml; 0.5 mg/ml) (Sigma), penicillin (100 U/ml), and streptomycin (100 mg/ml) at 24°C. Transfection was carried out using the Calcium Phosphate Transfection Kit (Specialty Media) according to the manufacturer’s instructions. HhN-conditioned medium treatment was carried out as described (33). Briefly, S2 cells stably expressing the N-terminal fragment of Hh (HhN) that is active for signaling were selected in hygromycin (200 μg/ml). Hh-conditioned medium was prepared by culturing cells without hygromycin but with 0.7 mM CuSO4 for 1 day. The medium was harvested and sterilized by filtration. Unless mentioned otherwise, Hh-conditioned medium was used at a 6:4 dilution ratio by fresh medium.

Immunoprecipitation, Western blotting, and immunostaining

Immunoprecipitation and Western blot analysis were carried out using standard protocols, as previously described (57). Cells were treated with 50 μM MG132 (Calbiochem) for 4 hours or 20 mM NH4Cl (Sigma) for 18 hours to stabilize Ptc before harvesting and lysing the cells. For Smo cell surface staining assay, S2 cells stably expressing Myc-Smo were harvested and washed with phosphate-buffered saline (PBS), fixed with 4% formaldehyde at room temperature for 20 min, and incubated with the mouse anti-Myc antibody in PBS at room temperature for 30 min. Cells were washed three times by PBS followed by secondary antibody staining. Immunostaining of imaginal discs was carried out as described (31). Antibodies used in this study were as follows: mouse anti-EN [Developmental Studies Hybridoma Bank (DSHB)], mouse anti-Ptc and mouse anti-SmoN (DSHB), rabbit and mouse anti-Flag (Sigma), mouse anti-Myc (Santa Cruz Biotechnology), mouse anti-HA (Santa Cruz Biotechnology), mouse anti-Ci 2A1 (58), mouse anti-GFP (Millipore), and rabbit anti-GFP (Santa Cruz Biotechnology).

Quantitative RT-PCR

Total RNA was extracted from 1 × 106 cells using the RNeasy Plus Mini Kit (Qiagen), and cDNA was synthesized using the iScript cDNA Synthesis Kit (Bio-Rad Laboratories). qRT-PCR was performed using iQ SYBR Green System (Bio-Rad Laboratories). qRT-PCR was performed in triplicate on each of three independent biological replicates. Primer sequences used are as follows: Smo, 5′-ATGCAGTACTTAAACTTTCCG-3′ and 5′-GTAGCAACGGGCACGTCGGAC-3′; Ptc, 5′-ATGGACCGCGACAGCCTCCCA-3′ and 5′-CGACGCAGAAGGTGCTCAGCA-3′. Actin was used as a normalization control. Quantification of mRNA levels was calculated using the comparative CT method.

RNAi and dual-luciferase assay

dsRNA was generated using the MEGAscript High Yield Transcription Kit (Ambion, AM1334) according to the manufacturer’s instruction. The following primers were used for generating the dsRNA targeting each gene: Smurf, 5′-GAATTAATACGACTCACTATAGGGAGAAGCGGAGGAGGAAGTAGATCC-3′ and 5′-GAATTAATACGACTCACTATAGGGAGATCGAATCATTTGCAAAATACT-3′; Nedd4, 5′-GAATTAATACGACTCACTATAGGGAGAGTCGCACATTTGTTGCTTCAG-3′ and 5′-GAATTAATACGACTCACTATAGGGAGACCGCTTATAAAAATCATTAAA-3′; Su(dx), 5′-GAATTAATACGACTCACTATAGGGAGAGGGACTGGATTATGGTGGTG-3′ and 5′-GAATTAATACGACTCACTATAGGGAGAACGCTAAACCAGAGCTCCAA-3′; Cul4, 5′-GAATTAATACGACTCACTATAGGGAGACGCGAATTCGCTGCAAAATTC-3′ and 5′-GAATTAATACGACTCACTATAGGGAGAGCGTCTAGAGCTCGTTCTCCT-3′; Smo 5′UTR, 5′-GAATTAATACGACTCACTATAGGGAGAGTCGCACATTTGTTGCTTCAG-3′ and 5′-GAATTAATACGACTCACTATAGGGAGACCGCTTATAAAAATCATTAAA-3′; Gprk2, 5′-GAATTAATACGACTCACTATAGGGAGAGGGGGCGACGCCTTGGACGCC-3′ and 5′-GAATTAATACGACTCACTATAGGGAGAAAAATACATAGAGCTCTCAAA-3′; Gprk2 5′UTR, 5′-GAATTAATACGACTCACTATAGGGAGATTTCACTAACTCAACGACGGT-3′ and 5′-GAATTAATACGACTCACTATAGGGAGAAAGAACAAGAAGAAGCAGCGA-3′; Ptc 5′UTR, 5′-GAATTAATACGACTCACTATAGGGAGACAGACTGCGTGCGATCCTCGA-3′ and 5′-GAATTAATACGACTCACTATAGGGAGAGTTATTGGAATCTCGTATTTT-3′; Ptc 3′UTR, 5′-GAATTAATACGACTCACTATAGGGAGACACTAGCACTAGTTCCTGTAG-3′ and 5′-GAATTAATACGACTCACTATAGGGAGAGTACATATTTAAACTAAATTA-3′.

dsRNA targeting the firefly luciferase coding sequence was used as a control. For knockdown experiments, S2 or cl-8 cells were cultured in SFM containing the indicated dsRNA at 23°C for 8 hours. After adding fetal bovine serum to a final concentration of 10% (S2 cells) or 2.5% (cl-8 cells), dsRNA-treated cells were cultured for 2 days before analysis or overnight before transfection with DNA constructs. The transfected cells were cultured for additional 2 days before analysis. Dual-luciferase activity was measured with Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer’s instructions. Briefly, cl-8 cells were treated with corresponding dsRNA in SFM for 8 hours at 24°C, and then fetal bovine serum was added to a final concentration of 2.5%. Cells were cultured for 24 hours before transfected with the constructs of ptc–firefly luciferase and Pol III–Renilla luciferase. For Hh treatment, two-thirds of the medium was replaced with Hh-conditioned medium, and cells were incubated for 24 hours before harvest. Each sample was measured in triplicate using a FLUOstar OPTIMA plate reader (BMG Labtech).

Ubiquitylation assay

Ubiquitylation assays were carried out based on a protocol described preciously (23). Briefly, cells were treated with 50 μM MG132 (Calbiochem) for 4 hours to inhibit proteasome-mediated degradation or 20 mM NH4Cl (Sigma) for 18 hours to inhibit lysosome-mediated degradation (Ptc only) before harvest. Cells were lysed in 100 μl of denaturing buffer [1% SDS, 50 mM tris, (pH 7.5), 0.5 mM EDTA, 1 mM dithiothreitol]. After incubation for 5 min at 100°C, the lysates were diluted 10-fold with lysis buffer and then subjected to immunoprecipitation and Western blot analysis.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/11/516/eaan8660/DC1

Fig. S1. Cul4 acts in parallel with Smurf to promote Smo ubiquitylation and cell surface clearance.

Fig. S2. Effect of Smurf family knockdown on Smo cell surface accumulation and Hh signaling in wing discs.

Fig. S3. Effect of overexpression of Smurf family members on Smo cell surface accumulation and Hh signaling in wing discs.

Fig. S4. Hh-stimulated and PKA-mediated phosphorylation of Smo inhibit the recruitment of Su(dx) and Nedd4.

Fig. S5. Hh regulates Ptc ubiquitylation through the Smurf family of E3s.

Fig. S6. Regulation of Ptc-E3 interaction by Hh and Smo.

Fig. S7. Ptc interacts with Smurf regardless of Smurf phosphorylation by Gprk2.

REFERENCES AND NOTES

Acknowledgments: We thank A. Casali, M. O’Connor, M. Baron, X. Feng, and D. Chen for providing reagents; Bloomington and VDRC stock centers for fly stocks; and DSHB for antibodies. Funding: This work was supported by grants from the NIH (GM118063) and Welch Foundation (I-1603) to J.J. J.J. is a Eugene McDermott Endowed Scholar in Biomedical Science at the University of Texas Southwestern Medical Center. Author contributions: J.J. and Shuang Li designed the experiments; Shuang Li, Shuangxi Li, and B.W. performed the experiments; Shuang Li, Shuangxi Li, and J.J. analyzed the data; and J.J. wrote the manuscript. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article