Hedgehog-Regulated Ubiquitination Controls Smoothened Trafficking and Cell Surface Expression in Drosophila

Hedgehog transduces signal by promoting cell surface expression of the seven-transmembrane protein Smoothened (Smo) in Drosophila, but the underlying mechanism remains unknown. Here we demonstrate that Smo is downregulated by ubiquitin-mediated endocytosis and degradation, and that Hh increases Smo cell surface expression by inhibiting its ubiquitination. We find that Smo is ubiquitinated at multiple Lysine residues including those in its autoinhibitory domain (SAID), leading to endocytosis and degradation of Smo by both lysosome- and proteasome-dependent mechanisms. Hh inhibits Smo ubiquitination via PKA/CK1-mediated phosphorylation of SAID, leading to Smo cell surface accumulation. Inactivation of the ubiquitin activating enzyme Uba1 or perturbation of multiple components of the endocytic machinery leads to Smo accumulation and Hh pathway activation. In addition, we find that the non-visual β-arrestin Kurtz (Krz) interacts with Smo and acts in parallel with ubiquitination to downregulate Smo. Finally, we show that Smo ubiquitination is counteracted by the deubiquitinating enzyme UBPY/USP8. Gain and loss of UBPY lead to reciprocal changes in Smo cell surface expression. Taken together, our results suggest that ubiquitination plays a key role in the downregulation of Smo to keep Hh pathway activity off in the absence of the ligand, and that Hh-induced phosphorylation promotes Smo cell surface accumulation by inhibiting its ubiquitination, which contributes to Hh pathway activation.


Introduction
Hedgehog (Hh) signaling governs cell growth and patterning in species ranging from insects to human [1,2]. Because of its pivotal role in embryonic development and adult tissue homeostasis, misregulation of Hh signaling activity has been linked to many human disorders including birth defects and cancers [1,3,4]. Hh exerts its biological influence through a largely conserved signaling cascade that culminates at the activation of latent transcription factors Cubitus interruptus (Ci)/Gli [1].
The core Hh reception system consists of a 12-transmembrane protein Patched (Ptc) that acts as the Hh receptor and a seventransmembrane protein Smo that acts as the Hh signal transducer [5,6]. Hh and Ptc reciprocally regulate the subcellular localization and active state of Smo [7][8][9][10]. In Drosophila, Hh stimulation or loss of Ptc leads to cell surface accumulation of Smo [7,11]. Increased cell surface expression and activation of Smo are regulated by Hh-induced and PKA/CK1-mediated phosphorylation of Smo carboxyl intracellular tail (C-tail) [12][13][14].
Several observations suggest that Smo cell surface expression is controlled by endocytic trafficking. A transmission electron microscopic study of Drosophila imaginal discs indicated that Smo is localized primarily in the lysosome of anterior compartment cells but is enriched on the plasma membrane of posterior compartment cells [15]. In Drosophila salivary gland cells, blocking endocytosis promotes Smo cell surface accumulation [11]. Using antibody uptake assay in S2 cells, we have shown that Smo reaches the cell surface but quickly internalizes in the absence of Hh and that Hh stimulation diminishes internalized Smo with a concomitant increase in cell surface Smo [13]. Taken together, these observations suggest that Hh signaling may regulate Smo cell surface expression by blocking its endocytosis and/or promoting its recycling back to the cell surface after internalization.
The mechanisms by which Smo endocytic trafficking and cell surface expression are regulated have remained unknown. Smo intracellular regions lack recognizable endosomal-lysosomal sorting signals such as the NPXY and dileucine-based motifs [16]. However, many membrane receptors are internalized after covalently modified by ubiquitination, as has been demonstrated for receptor tyrosine kinases (RTKs) and G protein coupled receptors (GPCRs) [17,18]

Inactivation of the Ubiquitin-Activating Enzyme Uba1 Leads to Smo Accumulation
In Drosophila wing discs, Smo cell surface level is low in anterior (A) compartment cells away from the A/P boundary but is elevated in response to Hh in A-compartment cells near the A/P boundary or in posterior (P) compartment cells ( Figure 1A) [7]. To determine whether Smo is downregulated by the ubiquitin pathway, we generated mutant clones for Uba1, which encodes the only ubiquitin-activating enzyme (E1) in Drosophila [19,20]. We employed a temperature-sensitive allele of Uba1, Uba1 H33 , which behaves like a null allele at the restrictive temperature [19]. Uba1 H33 clones were induced at second instar larval stage (48-72 h AEL) by FRT/FLP mediated mitotic recombination. Larva carrying Uba1 H33 clones were grown at permissive temperature (18uC) for 3 d and then shifted to non-permissive temperature (30uC) for 24 h before dissection for immunostaining. We found that anteriorly situated Uba1 H33 clones accumulated high levels of Smo compared with neighboring wild type cells ( Figure 1A Figure S1B-B'), likely due to the perdurance of Uba1 activity. In general, Smo elevation coincided well with Uba1 mutant clones. Intriguingly, Uba1 H33 mutant cells situated in the posterior compartment also exhibited slightly higher levels of Smo than neighboring wild type cells (arrowhead in Figure 1B), suggesting that a fraction of Smo still undergoes ubiquitin-mediated degradation in the presence of Hh.

Uba1 Regulates Smo Ubiquitination and Cell Surface Expression
To examine whether Smo is directly ubiquitinated and whether Uba1 is responsible for this activity, we carried out a cell-based ubiquitination assay (see Materials and Methods) [21]. We employed RNAi and/or pharmacological inhibitor to inactivate Uba1. S2 cells stably expressing a Myc-tagged Smo (Myc-Smo) were treated with Uba1 or control doublestranded RNA (dsRNA) in the absence or presence of PYR-41, a cell permeable E1 inhibitor [22]. The efficiency of Uba1 RNAi was confirmed by Western blot analysis of an exogenously expressed tagged Uba1 ( Figure 1C). Myc-Smo was ubiquitinated efficiently in the absence of Uba1 inhibition ( Figure 1D); however, ubiquitination of Smo was attenuated by Uba1 RNAi and more significantly inhibited by PYR-41 ( Figure 1D). The incomplete blockage of Smo ubiquitination by Uba1 RNAi is likely due to partial inactivation of Uba1 by the RNAi approach. Indeed, a combined treatment with Uba1 RNAi and PYR-41 resulted in a more complete inhibition of Smo ubiquitination ( Figure 1D).
We next applied a cell-based immunostaining assay to determine whether Uba1 regulates Smo cell surface expression [13]. Myc-Smo expressing cells were treated with control or Uba1 dsRNA in the absence or presence of PYR-41. Cell surface and total Smo were visualized by immunostaining with an anti-SmoN antibody prior to and after cell membrane permeabilization, respectively. As shown in Figure 1E

Perturbation of Endocytic Machinery Leads to Smo Accumulation and Hh Pathway Activation
Ubiquitinated membrane proteins are internalized through the endocytic pathway and targeted to lysosome for degradation [17]. We therefore examined the effect of inactivation of endocytic components on Smo accumulation in wing imaginal discs. We found that Smo was accumulated in intracellular puncta in mutant clones lacking the Drosophila homolog of HGF-regulated tyrosine kinase substrate (Hrs) (Figure 2A-A'), a protein involved in sorting ubiquitinated membrane proteins into multivesicular bodies (MVBs) [23]. Of note, not all hrs mutant cells exhibited Smo puncta. This could be due to perdurance of Hrs activity and/or disc folding so that Smo puncta are present at different focal planes. RNAi of other endocytic components, including Tsg101 [24], Avalanche (Avl), a Drosophila syntaxin located in early endosomes [25], and Rab5, resulted in Smo accumulation in anterior compartment cells distant from the A/P boundary (arrows in Figure 2B-E), as well as Hh pathway activation as indicated by Ci accumulation and ectopic expression of a Hh target gene decapentaplegic (dpp) ( Figure 2B-E). Taken together, these observations suggest that Smo is downregulated via the endocytic pathway in the absence of Hh.

Author Summary
The Hedgehog (Hh) family of secreted proteins governs cell growth and patterning in diverse species ranging from Drosophila to human. Hh signals across the cell surface membrane by regulating the subcellular location and conformation of a membrane protein called Smoothened (Smo). In Drosophila, Smo accumulates on the cell surface in response to Hh, whereas in the absence of Hh it is internalized and degraded. The molecular mechanisms that control this intracellular trafficking and degradation of Smo were unknown, but here we show that Smo is modified by attachment of several molecules of a small protein called ubiquitin, which tags it for internalization and degradation within the cell. Hh inhibits this ubiquitination of Smo by inducing another modification, phosphorylation, of its intracellular tail by two types of protein kinase enzymes. This loss of ubiquitination and gain of phosphorylation causes the accumulation of Smo at the cell surface. What's more, we find that another protein called Kurtz interacts with Smo and acts in parallel with the ubiquitination process to promote internalization of Smo, and that the deubiquitinating enzyme UBPY/USP8 counteracts ubiquitination of Smo to promote its cell surface accumulation. Our study demonstrates that reversible ubiquitination plays a key role in regulating Smo trafficking to and from the cell surface and thus it provides novel insights into the mechanism of Hh signaling from the outside to the inside of the cell.   [13], treated without or with Hh conditioned medium, and followed by the ubiquitination assay described above. As shown in Figure 4E  To determine whether the SAID domain suffices to promote ubiquitination and internalization of a heterologous membrane protein, we fused it to the C-terminus of the Wingless (Wg) receptor Frizzle 2 (Fz2) to construct Fz2-SAID chimeric protein (FS). When expressed in S2 cells, CFP-tagged Fz2 (CFP-Fz2) was largely accumulated on the cell surface with a small fraction internalized and colocalized with the endosomal marker Rab5 ). In addition, we found that adding the phosphorylation-deficient form but not the phospho-mimetic form of SAID to Fz2 promotes the ubiquitination of the corresponding chimeric protein ( Figure 5E). Taken together, these observations suggest that the SAID domain suffices to promote ubiquitination and internalization of a membrane protein in a manner inhibited by phosphorylation. Combined with our earlier work [8], it seems that the SAID domain autonomously regulates ubiquitination independent of the C-terminal negatively charged region.

Smo Is Ubiquitinated at Multiple Lysine Residues
If Smo ubiquitination is responsible for its internalization and degradation, one would expect that ubiquitination-deficient Smo variants should be stabilized and accumulated on the cell surface. We therefore attempted to identify Lys residues responsible for Smo ubiquitination. In general, ubiquitin acceptor sites lack a strict consensus and target proteins can be ubiquitinated at multiple Lys residues. Smo C-tail and intracellular loops contain a total of 49 Lys residues, many of which may serve as ubiquitin acceptor sites, making it difficult to generate Smo variants devoid of ubiquitination. As deleting the SAID domain diminished Smo ubiquitination ( Figure 4G), we speculated that this region might contain Lys residues critical for Smo ubiquitination. There are a total of 13 Lys residues between aa 661 and aa 818. We therefore constructed Smo Figure 6D). Thus, although Smo K13R exhibits increased stability and cell surface expression, it is still internalized and degraded by lysosome and can be further stabilized by Hh.

Smo K13R Exhibits Increased Stability and Cell Surface Expression
To determine whether the K13R mutation affects Smo stability in vivo, we generated transgenic flies expressing either UAS-Myc-Smo or UAS-Myc-Smo K13R from the same genetic locus using the phiC31 integration system to ensure similar expression level from different constructs [26]. We used the wing specific Gal4 driver MS1096 coupled with tub-Gal80 ts to drive a pulse of UAS-Myc-Smo or UAS-Myc-Smo K13R expression by shifting late third instar larvae to the non-permissive temperature for 12 h. After chasing for

Krz Promotes Smo Internalization by Binding to Its C-Tail
Internalization of Smo K13R is likely due to its residual ubiquitination at a Lys residue(s) outside the SAID domain. In addition, Smo K13R could also be internalized by Smo interacting proteins, as have been shown for other receptors [27,28]. It has been shown that the non-visual arrestin, b-arrestin 2, can bind and internalize mammalian Smo [29]. The Drosophila non-visual arrestin is encoded by krz [30]. We therefore carried out both gain-and lossof-function studies to determine whether Krz regulates Smo cell surface expression. We found that overexpression of Krz in wing imaginal discs using a dorsal compartment specific Gal4 driver, ap-Gal4, blocked Smo accumulation in posterior-dorsal compartment cells (compare Figure 7B with Figure 7A). However, we found that Smo was not accumulated in krz mutant clones located in the anterior compartment of wing discs ( Figure 7C). Similar observations were obtained by a recent study [31].

Smo Ubiquitination Is Counteracted by the Deubiquitinating Enzyme UBPY
Ubiquitination is a reversible process and ubiquitin attached to target proteins can be removed by deubiquitinating enzymes/ DUBs [32]. Compared with the large number of E3 ubiquitin ligases that catalyze ubiquitination of targeted proteins, each genome encodes a much smaller number of DUBs. For example, the Drosophila genome encodes over 200 annotated E3s but less than 30 annotated DUBs (Flybase ; Table S1).  Figure 8B).
We then carried out coimmunoprecipitation assays to determine whether UBPY physically interacts with Smo. As shown in Figure 8C

Smo Is Regulated by Both Mono-and Polyubiquitination
It is generally thought that monoubiquitination or multiubiquitination (monoubiquitination at multiple sites) is responsible for receptor internalization and degradation by lysosome, whereas

Discussion
Regulation of Smo cell surface expression is a key step in Hh signal transduction [7,11,13], but the underlying mechanism has  Early studies with yeast membrane receptors provided evidence that monoubiquitination of GPCRs mediates their agonistinduced internalization [33,34]. Later studies with mammalian GPCRs and other receptors suggested that both mono-and polyubiquitination could be involved in receptor endocytosis and degradation [18]. However, it has been shown that ''polyubiquitination'' of some receptors is due to monoubiquitination at multiple sites (multiubiquitination) instead of forming a polyubiquitination chain at a single site [35,36]. Here we provide evidence that Smo is both mono-and polyubiquitinated. It is possible that mono-or multiubiquitination may lead to Smo internalization and that internalized Smo could be further ubiquitinated in the endocytic pathway, leading to the formation of Lys 48-linked polyubiquitin chain that targets Smo for proteasome-mediated degradation ( Figure 10). Thus, multiple ubiquitination events provide a robust mechanism for Smo downregulation to prevent aberrant Smo activity in the absence of Hh.
Regulation of Smo trafficking and cell surface expression provides a new paradigm for how the ubiquitin pathway controls the activity of a membrane receptor. Unlike all the other cases whereby receptor ubiquitination is triggered by ligand or agonist stimulation and serves as a mechanism to control the duration of cell signaling, Smo ubiquitination occurs in the absence of ligand stimulation and serves as a mechanism to keep the basal pathway The mechanisms that regulate Smo trafficking and cell surface expression exhibit interesting similarities to as well as important differences from those regulating GPCRs. For example, it has been shown that agonist-induced downregulation of b2-Adrenergic Receptor (b2AR) is mediated by both b-arrestin and receptor ubiquitination [27]. In addition, b2AR internalization and degradation is regulated by both proteasome-and lysosomedependent mechanisms [27,37]. However, b2AR ubiquitination is induced by agonist and serves as a mechanism for desensitization [27,37], whereas Smo ubiquitination is inhibited by Hh and serves as a mechanism for keeping pathway activity off in the absence of the ligand. b-arrestin binding to b2AR is induced by agonists and requires GRK2-mediated phosphorylation of the activated receptor [27], whereas Krz binding to Smo is attenuated by Hh and Smo phosphorylation (Figure 7). Although GPRK2/GRK2 also regulates Smo in Drosophila, its function appears to be uncoupled from that of Krz because loss of GPRK2 exhibits a phenotype distinct from that exhibited by loss of Krz [31,[38][39][40]. Furthermore, Krz can internalize Smo in the absence of GPRK2 [31]. b-arrestin is required for b2AR ubiquitination [27,37], whereas Krz inactivation does not significantly affect Smo ubiquitination (unpublished observations). Finally, while the proteasome inhibitor MG132 blocks agonist-induced b2AR internalization [27], it does not prevent Smo internalization but instead inhibits Smo degradation after internalization (Figure 3).
It is also interesting to note that b-arrestin has been implicated in the regulation of Smo trafficking and Shh signaling in vertebrates [29,41,42]. Furthermore, b-arrestin binds to mammalian Smo (mSmo) in a manner promoted by Shh and GRK2mediated phosphorylation of mSmo C-tail [42,43], which is analogous to agonist-induced b-arrestin binding to GPCRs. However, instead of internalizing mSmo for degradation, barrestin appears to promote mSmo ciliary accumulation [42], which correlates with its positive role in Shh signaling. Both Drosophila and vertebrate Smo proteins can activate trimeric Gproteins [44][45][46], suggesting that they are not only structurally but also functionally related to GPCRs. It is conceivable that Smo proteins may employ multiple mechanisms utilized by GPCRs to control their intracellular trafficking and activity. Thus, it will be interesting to determine whether vertebrate Smo is also regulated by the ubiquitin pathway.

Ubiquitination Assay
Ubiquitination assays were carried out based on the protocol described previously [21]. Briefly, Myc-Smo stably expressing S2 cells or S2 cells transfected with Smo variants with or without HA-Ub (wild type or mutants) were treated with MG132 or NH 4 Cl before harvesting. Cells were lysed in 100 ml of denaturing buffer (1% SDS/50 mM Tris, pH 7.5/0.5 mM EDTA/1 mM DTT). After incubation for 5 min at 100uC, the lysates were diluted 10fold with lysis buffer and then subjected to immunoprecipitation and Western blot analysis.  (Table S1) were generated by PCR and used for generating dsRNA. Ptc RNAi was carried out as previously described [51]. dsRNA targeting the Fire Fly Luciferase coding sequence was used as a control. For RNAi knockdown experiments, S2 cells were cultured in serum free medium containing indicated dsRNA at 23uC for 8 h. After adding fetal bovine serum to a final concentration of 10%, dsRNA treated cells were cultured overnight before transfection. 48 h after transfection, cells were harvested for further analysis.