Hrs Promotes Ubiquitination and Mediates Endosomal Trafficking of Smoothened in Drosophila Hedgehog Signaling

In Hedgehog (Hh) signaling, the seven-transmembrane protein Smoothened (Smo) acts as a signal transducer that is regulated by phosphorylation, ubiquitination, and cell surface accumulation. However, it is not clear how Smo cell surface accumulation and intracellular trafficking are regulated. Here, we demonstrate that inactivation of Hrs by deletion or RNAi accumulates Smo in the late endosome that is marked by late endosome markers. Inactivation of Hrs enhances the wing defects caused by dominant-negative Smo. We show that Hrs promotes Smo ubiquitination, deleting the ubiquitin-interacting-motif (UIM) in Hrs abolishes the ability of Hrs to regulate Smo ubiquitination. However, the UIM domain neither recognizes the ubiquitinated Smo nor directly interacts with Smo. Hrs lacking UIM domain still downregulates Smo activity even though to a less extent. We have characterized that the N-terminus of Hrs directly interacts with the PKA/CK1 phosphorylation clusters to prevent Smo phosphorylation and activation, indicating an ubiquitin-independent regulation of Smo by Hrs. Finally, we found that knockdown of Tsg101 accumulates Smo that is co-localized with Hrs and other late endosome markers. Taken together, our data indicate that Hrs mediates Smo trafficking in the late endosome by not only promoting Smo ubiquitination but also blocking Smo phosphorylation.


Introduction
The Hedgehog (Hh) morphogen controls such development processes as cell proliferation, embryonic patterning, and cell growth [1][2][3]. Dysregulation of Hh signaling has been implicated in many human disorders, including several cancer types [4][5][6]. Smoothened (Smo), an atypical G protein-coupled receptor (GPCR), is essential in both insects and mammals for transduction of the Hh signal [2,3,7]. Abnormal Smo activation results in basal cell carcinoma (BCC) and medulloblastoma, so it remains an attractive therapeutic target.
In Drosophila, the Hh signal is transduced through a reception system at the plasma membrane, which includes the receptor complexes Ptc-Ihog and the signal transducer Smo [7][8][9]. Binding of Hh to Ptc-Ihog relieves the inhibition of Smo by Ptc, allowing Smo to ultimately 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 [7,10]. Intracellular Smo is unstable and very low levels of Smo has been observed in the absence of Hh. Upon different levels of Hh stimulation, Smo is differentially phosphorylated [11,12] and phosphorylation increases the level of cell surface Smo [2,7], although the mechanism is not completely understood. Using an antibody uptake assay, we found that Hh treatment inhibited Smo endocytosis and reduced the ratio of early endosome-localized Smo [13]. In addition, ubiquitination promotes endocytosis of Smo, whereas deubiquitination prevents the process [13,14]. A transmission electron microscopic study of Drosophila imaginal discs indicated that Smo is directed primarily to the lysosomes of A compartment cells, but is enriched at the plasma membrane of P compartment cells [15]. It is clearly important to understand how Hh regulates Smo trafficking and how ubiquitination promotes Smo endocytosis. Among the proteins that regulate receptor intracellular trafficking, the components from the Endosomal Sorting Complex Required for Transport (ESCRT) are critical. In Drosophila, the endosomal sorting of Smo is likely regulated by the Drosophila homolog of HGF-regulated tyrosine kinase substrate (Hrs), as Smo accumulation has been observed in cells mutating hrs [14,16]. In mammals, Hh signal transduction depends on the primary cilium, and ciliary accumulation is required for Smo activation [17][18][19], thus the cilium represents a signaling center for Hh pathway in mammals [20]. Similarly, phosphorylation by multiple kinases promotes the ciliary localization of mammalian Smo [21]. However, the mechanisms by which the ciliary localization of vertebrate Smo is controlled remain unclear.
In this study, we provide genetic and biochemical evidence that the ubiquitination and intracellular trafficking of Smo are regulated by Hrs through direct interaction that inhibits the phosphorylation of Smo at the carboxyl-terminal domain. We further provide evidence that both Hrs and tumor susceptibility gene 101 (Tsg101) mediate Smo trafficking in the late endosomes, likely downstream of Shibire (Shi, the Drosophila homolog of the dynamin GTPase).

Hrs Regulates Smo Activity through Mediating Smo Trafficking in the Late Endosome
Hh induces stabilization and accumulation of Smo at the cell surface [22,23]. In Drosophila wing imaginal disc, the posterior (P) compartment cells express and secrete Hh proteins that act upon neighboring anterior (A) compartment cells located adjacent to the A/P boundary to induce the expression of Hh target genes. Pcompartment cells as well as A-compartment cells near the A/P boundary exhibit high levels of Smo cell surface accumulation (Fig. 1A). In A-compartment cells away from the A/P boundary, Smo levels are extremely low and the intracellular puncta of Smo suggest the trafficking of Smo inside the cell that leads to degradation of the protein (Fig. 1A). We recently showed that ubiquitination promotes Smo intracellular trafficking that is mediated by endosomes [13]. It was also reported that Smo accumulates in cells mutating hrs that encodes a protein involved in sorting ubiquitinated membrane proteins in the endosomes [14,16], raising the possibility that Hrs may facilitate endosomal sorting of Smo. Smo accumulated as puncta in mutant clones lacking hrs [14] (Fig. 1B). Inactivation of Hrs by RNAi also accumulates Smo in puncta (Fig. 1C). The phenotype of HrsRNAi is unlikely due to an off-target effect because expression of different transgenic lines from Bloomington (#28026 and #28964) and VDRC (v20933) targeting different non-overlapping regions produced a similar phonotype that is consistent with the phenotype caused by hrs mutation. We thus used the HrsRNAi lines to examine the localization of Smo when Hrs is inactivated. We found that Smo puncta co-localized neither with the early endosome marker Rab5 (Fig. 1B), nor with the recycling endosome marker Rab11 (Fig. 1D). Instead, Smo puncta colocalized with the late endosome marker Rab7 (Fig. 1E). In addition, Smo puncta co-localized with the overexpressed GFP-Rab7 and GFP-Lamp1 (Fig. 1F, data not shown), which are often expressed in the late endosomes. These data suggest that Hrs facilitates Smo sorting into the late endosome.
Knockdown of Hrs by RNAi using the wing-specific MS1096-Gal4 caused a severe phenotype in adult wings (Fig. 2B, compared to WT wing in Fig. 2A). To better explore the Hh wing phenotype and to examine whether Hrs plays a role in regulating Smo activity, we used a weaker Gal4 line, C765-Gal4. As we previously described, expressing the phospho-deficient Smo mutant, Smo 2PKA12 (a weak dominant-negative form), by C765-Gal4 caused a reproducible wing phenotype with partial fusion between Vein 3 and 4 (arrow in Fig. 2D, compared to WT wing structure in Fig. 2A) [24]. This phenotype provided a sensitized genetic background for screening novel components involved in Hh signaling [24]. We reasoned that if Hrs regulates Smo activity in wing development, manipulating Hrs expression levels may dominantly modify this phenotype. Indeed, knockdown of Hrs by RNAi in Smo 2PKA12 expressing wing caused further fusion and narrower Vein 3 and Vein 4 (Fig. 2E, compared to Fig. 2D), although HrsRNAi alone driven by the C765-Gal4 did not cause any phenotype in the wing (Fig. 2C). We further generated UAS-HA-Hrs transgene and assessed its ability to regulate Smo activity. We found that coexpressing Smo 2PKA12 with HA-Hrs reduced the Smo 2PKA12 phenotype (Fig. 2G), even though expressing HA-Hrs alone produced wild-type wings (Fig. 2F). These data suggest that changing Hrs levels in wing discs leads to changes in the dominant-negative activity of Smo 2PKA12 .  [13,14]. However, the machinery responsible for Smo ubiquitination is not known, although inactivation of the ubiquitin activating enzyme Uba1 downregulates Smo ubiquitination [14]. Since Hrs is a key endocytic regulator recognizing the ubiquitinated receptors and mediates receptor sorting onto multivesicular bodies [25,26], we wondered whether Hrs regulates Smo by conserved mechanisms. Surprisingly, we found that Hrs regulates the levels of Smo ubiquitination even though Hrs is unlikely the ubiquitin-ligase for Smo. We examined Smo ubiquitination in S2 cells using the immunoprecipitation assay we have established [13]. The ubiquitination of Smo was readily detected by the anti-Ub antibody that recognized the immunoprecipitated endogenous Ub (Fig. 3A, top panel). Interestingly, the levels of Hrs normally interacts with ubiquitinated cargoes through its UIM domain and plays an essential role in endosomal sorting. We wondered whether it would be possible for Hrs to recognize the ubiquitinated Smo. Surprisingly, we found that the interaction between Smo and Hrs was not changed (  Taken together, UIM is a highly conserved sequence that can bind ubiquitinated cargos and is found in a number of proteins involved in endocytosis and protein trafficking, and our results further indicate that it also functions to promote the ubiquitination of a target protein.

Hrs Promotes the Ubiquitination of smo
The finding that Hrs DU was still able to downregulate GFP-Smo activity, even though to a lesser extent compared to full-length Hrs   (Fig. 6F, top panel). OA treatment blocked Smo ubiquitination (not shown), similarly to that caused by phosphomimetic mutation in Smo [13,14]. These data suggest that phosphorylation dissociates Smo from Hrs interaction. Compared to wild-type Smo, Smo SD123 that has phospho-mimetic mutations in the residues of three phosphorylation clusters interacted with Hrs weakly and such interaction was completely blocked by the treatment with OA (Fig. 6F, top panel), suggesting that phosphorylation of Smo in cultured cells blocks its interaction with Hrs. Since Smo SD123 still has very weak interaction with Hrs and OA treatment completely blocks this weak interaction, it is possible

Tsg101 Regulates Smo Trafficking in the Late Endosomes
After endocytosis from the plasma membrane into early endosomes, ubiquitinated receptors are bound to the ESCRT-0 complex that contains Hrs [26,29]. Hrs selects ubiquitinated cargos and recruits components of other ESCRT complexes including ESCRT-I, II and III. Tsg101, a subunit of ESCRT-I, binds the Hrs-receptor complex and recognizes ubiquitinated cargoes. To determine whether Tsg101 is also involved in Smo trafficking, we turned to the Drosophila imaginal disc to examine whether inactivation of Tsg101 affect Smo accumulation. We found that knockdown of Tsg101 by RNAi accumulated Smo in A compartment cells (Fig. 7A) leading to Smo puncta (Fig. 7B). However, Smo puncta did not reside in the early endosomes that were labeled with Rab5 (Fig. 7B). Instead, Smo puncta were colocalized with both the late endosome marker Rab7 (Fig. 7C) and Hrs (Fig. 7D). Consistently, Smo puncta co-localized with Rab7-GFP (not shown). These data suggest that Tsg101 regulates Smo in late endosomes.
To examine whether Tsg101 regulates Smo activity, we expressed Tsg101RNAi with Smo 2PKA12 in the wing and found that the intervein fusion phenotype was enhanced (Fig. 2I). However, Tsg101RNAi alone driven by C765-Gal4 did not induce any obvious phenotype (Fig. 2H), even though Tsg101RNAi driven by MS1096-Gal4 produced an extreme wing phenotype similar to that caused by HrsRNAi (Fig. 2B, data not shown). These data suggest that inactivation of Tsg101 enhances the dominant-negative activity of Smo 2PKA12 .
To further examine the function of the endosomal compartments in regulating Smo activity, we carried out a ptc-luciferase (ptc-luc) reporter assay to monitor the activity of Smo in Hh signling. As shown in Fig. 7E, compared to GFP RNAi, Shi RNAi elevated ptc-luc activity (Fig. 7E) because the knockdown of Shi increases the cell surface accumulation of Smo [13]. RNAi of Hrs or Tsg101 also elevated the ptc-luc activity even though to a lesser extent (Fig. 7E), suggesting that inactivation of Hrs or Tsg101 may not highly accumulate Smo on the cell surface, and that Hrs and Tsg101 act downstream of Shi in regulating Smo intracellular trafficking.

Discussion
The regulation of Smo intracellular trafficking has been a critical step in understanding the molecular mechanisms of cytosolic Hh signal transduction [20,28]. In this study, we have identified and characterized the role of Hrs and Tsg101 in the endosomal sorting of Smo. Similar to some other membrane proteins, Smo shares conserved mechanisms by which the multivesicular body (MVB) controls the sorting of ubiquitinated proteins. Novel mechanisms have also been identified in this study. We show that Hrs prevents Smo phosphorylation by directly binding to the phosphorylation sites, which blocks the cell surface accumulation and prevents the activation of the receptor. In addition, Hrs mediates Smo trafficking in the late endosome rather than in the early endosome.
Ubiquitinated membrane receptors are normally internalized through the endocytic pathway and targeted to MVBs and eventually to lysosome for degradation. The ESCRT machinery comprises four protein complexes (ESCRT-0, I, II, and III) that are required for membrane receptors to be sequentially targeted to the MVBs. Although some studies showed that the ESCRT-0 consisting of Hrs can be used as early endosome marker [29], many studies have shown that Hrs mediates the trafficking of  In this study, we found that the ubiquitination is not required for the interaction between Smo and Hrs. Hrs normally recognizes the ubiquitinated cargos through its UIM domain, but the UIM domain obviously does not specifically recognize the ubiquitinated membrane receptor. It has been shown that Hrs, Tsg101, and other component protein from the ESCRTs undergo ubiquitination [31,32], raising the possibility that Hrs may recognize ubiquitinated proteins from the endosomal sorting machinery. Alternatively, the UIM domain of Hrs might recognize the ubiquitination pathway proteins that carry ubiquitin required for Smo ubiquitination.
We showed that Hrs promotes the ubiquitination of Smo (Fig. 3). It is possible that Hrs competes with USP8 for binding Smo, as Hrs and USP8 directly interact with the same domain of Smo. However, we found that Hrs blocks Smo phosphorylation (Fig. 6C-D), whereas USP8 does not [13], suggesting that Hrs and USP8 bind Smo in different conformation. In support of this hypothesis, manipulating the levels of USP8 does not change the physical interaction between Smo and Hrs (Fig. 4C). In an immunoprecipitation assay, we did not observe a strong physical interaction between Hrs and USP8 (not shown). Hh likely downregulates Smo ubiquitination by promoting Smo-USP8 interaction [13] and by disassociating Hrs (Fig. 3B-C).
It is also possible that Hrs regulates Smo ubiquitination by facilitating the ubiquitin ligase(s). However, the ubiquitin conjugating enzyme(s) and ubiquitin ligase(s) are unknown although it has been shown that mutating the ubiquitin activating enzyme increases the levels of Smo in wing discs [14]. It is also possible that other domains in Hrs help to position the N-terminal direct interacting domain of Hrs to bind Smo. Smo are both monoubiquitinated and poly-ubiquitinated at many lysine residues in Smo C-tail [13,14,33]. We were unable to narrow down the specific residues that are regulated by Hrs because we found that Hrs promoted the ubiquitination of SmoK13R that contains 13 lysine residues mutated in Smo C-tail [14].
Not only the ubiquitination machinery but also the degradation pathways for Smo have not been clearly addressed, although a recent study has demonstrated VPS36-mediated trafficking of Smo [33]. It is likely that Smo utilizes both the proteasome-and lysosome-mediated degradation pathways [13,14,34]    to the GST backbone. His-HrsCT1 and His-HrsUIM were generated by fusing Hrs aa1-240 (corresponding nucleotide 1-720) and aa242-300 (corresponding nucleotide 724-900) to the pET-30a-His backbone, respectively. Fly mutants used: hrs D28 [36]. Hrs RNAi lines were obtained from either Bloomington (#28026 and #28964) or VDRC (v20933), and line #28964 was used for most of the experiments as all those lines gave rise to similar phenotypes. Tsg101 RNAi line (v23944) was obtained from VDRC and was characterized by previous studies [14]. MS1096 Gal4, ap-Gal4, C765-Gal4, and UAS-GFP-Smo have been described [24,35]. HA-Hrs and HA-Hrs DU transgenic lines were generated at the VK5 attP locus to ensure the proteins are expressed at the same levels without positional effects [27].

In vitro Kinase Assay and GST Fusion Protein Pull-down
For the in vitro kinase assay, GST-Smo fusion proteins were expressed in bacteria, which is harvested, washed with PBS, and suspended with lysis buffer (PBS supplied with 1% Triton X100 and protease inhibitor) at the ratio of 50 ml buffer/1 ml culture. After sonication and centrifugation, glutathione beads were added to the supernatant aliquots (15 ml beads/500 ml lysate), followed by incubation for one hour and washing with PBS for three times at 4uC. The GST fusion protein was then subjected to a kinase assay with commercial PKA and CK1 (New England Biolabs) according to the supplier's protocols. Phosphorylation of Smo was detected by western blot with the phospho-SmoP antibody that recognizes the phosphorylated forms of Smo [11]. The assay of GST fusion proteins pull-down with His-tagged proteins has been previously described [27].

Immunostaining of Wing Imaginal Discs
Wing discs from third instar larvae were dissected in PBS then fixed with 4% formaldehyde in PBS for 20 min. After permeabilization with PBT (PBS supplemented with 1% Triton X100), discs were incubated with the indicated primary antibodies for three hours and the corresponding second antibodies for one hour sequentially, and washed with PBT for three times, 20 min per wash, following incubations. Primary antibodies used in this study were as follows: mouse anti-SmoN (DSHB, 1:10); rabbit anti-b-Gal (Cappel, 1:1,500), anti-Rab5 (Abcam, 1:300), anti-Rab7 (gift from Dr. Akira Nakamura, 1:3000), and anti-Rab11 (gift from Dr. Donald Ready, 1:3000); guinea pig anti-Hrs (gift from Dr. Hugo Bellen). Secondary antibodies were from Jackson ImmunoResearch Laboratories Inc., affinity-purified for multiple labeling (1:500). Samples were mounted on slides in 80% glycerol. Fluorescence signals were acquired with the 20 x objective on an Olympus confocal microscope and images were processed with Olympus Fluoview Ver.1.7c. About 15 imaginal discs were screened and 3-5 disc images were taken for each genotype.