Journ

alof

Cell

Scie

nce

Drosophila Vps36 regulates Smo trafficking inHedgehog signaling

Xiaofeng Yang*, Feifei Mao*, Xiangdong Lv, Zhao Zhang, Lin Fu, Yi Lu, Wenqing Wu, Zhaocai Zhou, Lei Zhang`

and Yun Zhao`

State Key Laboratory of Cell Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy ofSciences, 320 Yue-Yang Road, Shanghai 200031, P. R. China

*These authors contributed equally to this study`Authors for correspondence (yunzhao@sibcb.ac.cn; rayzhang@sibcb.ac.cn)

Accepted 17 June 2013Journal of Cell Science 126, 4230–4238� 2013. Published by The Company of Biologists Ltddoi: 10.1242/jcs.128603

SummaryThe hedgehog (Hh) signaling pathway plays a very important role in metazoan development by controlling pattern formation.

Malfunction of the Hh signaling pathway leads to numerous serious human diseases, including congenital disorders and cancers. Theseven-transmembrane domain protein Smoothened (Smo) is a key transducer of the Hh signaling pathway, and mediates the graded Hhsignal across the cell plasma membrane, thereby inducing the proper expression of downstream genes. Smo accumulation on the cell

plasma membrane is regulated by its C-tail phosphorylation and the graded Hh signal. The inhibitory mechanism for Smo membraneaccumulation in the absence of Hh, however, is still largely unknown. Here, we report that Vps36 of the ESCRT-II complex regulatesSmo trafficking between the cytosol and plasma membrane by specifically recognizing the ubiquitin signal on Smo in the absence of Hh.

Furthermore, in the absence of Hh, Smo is ubiquitylated on its cytoplasmic part, including its internal loops and C-tail. Taken together,our data suggest that the ESCRT-II complex, especially Vps36, has a special role in controlling Hh signaling by targeting the membraneprotein Smo for its trafficking in the absence of Hh, thereby regulating Hh signaling activity.

Key words: Drosophila, Hedgehog, Smoothened, Trafficking, Vps36

IntroductionHedgehog (Hh) signaling is essential for organ patterning of

metazoans (Ingham et al., 2011; Jia and Jiang, 2006; Jiang and Hui,

2008). Malfunction of the Hh signaling pathway leads to human

congenital disorders and cancers (Jiang and Hui, 2008). The core

components and regulatory mechanisms of Hh signaling are

conserved between invertebrates and vertebrates, with few

exceptions (Jiang and Hui, 2008; Wilson and Chuang, 2010). In

Drosophila wing imaginal discs, Hh proteins secreted by posterior

(P) compartment cells move into the anterior (A) compartment to

form a local concentration gradient. Genetic and biochemical

studies revealed that several transmembrane proteins, including

Patched (Ptc, 12-transmembrane domain protein), Ihog

(Interference hedgehog, 1-transmembrance domain protein), Boi

(Brother of Ihog, 1-trasmembrance domain protein) and

Smoothened (Smo, 7-transmembrane domain protein), function

as a reception system for signal transduction in Hh signal receiving

cells (Hooper and Scott, 1989; Hooper and Scott, 2005; Yan et al.,

2010; Yao et al., 2006; Zheng et al., 2010).

Important insights into the regulation of Smo and Hh signaling

activity come from studies in Drosophila (Ingham et al., 2011;

Jiang and Hui, 2008). In the absence of Hh, Ptc inhibits Smo

activity and its accumulation on the plasma membrane. Under

these conditions, Smo C-terminal cytoplasmic tail (C-tail)

assumes a ‘closed’ inactive conformation (Zhao et al., 2007).

Cytoplasmic components of the Hh signaling pathway, including

Costal-2 (Cos2), Fused (Fu) and full-length Cubitus interruptus

(Ci), form a large complex. Ci, a zinc-finger transcription factor

of the Hh signaling pathway, is sequentially phosphorylated by

protein kinase A (PKA), glycogen synthase kinase 3 (GSK3) and

casein kinase I (CKI), and then processed to generate a truncated

form (Ci75), which acts as transcriptional repressor to block

downstream gene expression (Smelkinson and Kalderon, 2006;

Zhang et al., 2005). In the presence of Hh, both Ptc and iHog

interact with Hh (Yao et al., 2006), resulting in Smo accumulation

on the plasma membrane and its C-tail phosphorylation by PKA

and CKI, followed by conformational switch to an ‘open’ and

activated state. Activated Smo regulates the expression of distinct

downstream target genes decapentaplegic (dpp), patched (ptc) and

engrailed (en) in a Hh-concentration-dependent manner by

controlling Ci nuclear translocation (Ranieri et al., 2012; Yao

et al., 2006; Zhang et al., 2011; Zhao et al., 2007; Zhou and

Kalderon, 2011). As a membrane protein of Hh pathway central

components, Smo plays a crucial role in transducing the Hh signal

across the cell plasma membrane. Accumulation of Smo on the

plasma membrane is tightly controlled by the Hh signal gradient.

However, how Smo translocation is regulated, especially the

inhibitory mechanism of its plasma membrane accumulation in the

absence of Hh signal, remains largely unknown.

Endosomal sorting complex required for transport (ESCRT)

complexes were initially defined for their involvement in

ubiquitylated membrane protein endosome sorting (Babst et al.,

2002a; Babst et al., 2002b; Katzmann et al., 2001). The ESCRT

machinery consists of four protein complexes, including ESCRT-

0, ESCRT-I, ESCRT-II and ESCRT-III, according to the time

point of their involvement in the sorting process of the

4230 Research Article

Journ

alof

Cell

Scie

nce

ubiquitylated cargo into multivesicular bodies (MVBs) (Wollertand Hurley, 2010). Several mutations are reported in the ESCRT

complex; these disable its ability to sort and degrade componentsof multiple conserved signaling pathways, including receptortyrosine kinase (RTK), EGF receptor, Notch and its ligand Delta,

Wnt receptor and others (Bache et al., 2006; Jekely and Rørth,2003; Lloyd et al., 2002; Piddini et al., 2005; Rusten et al., 2012).Ubiquitylation is an important signal for the endosometrafficking of some membrane proteins mediated by ESCRT

machinery, such as RTK internalization and Notch trafficking(Lloyd et al., 2002; Vaccari and Bilder, 2005). In the Hhsignaling pathway, however, it is largely unknown whether the

ESCRT machinery regulates Smo trafficking and internalization,and if so, how ESCRT is involved in this process.

The Hh signal controls Smo C-tail phosphorylation, and thus

Hh signaling activity. In this study, we identified Vps36(Vacuolar protein sorting 36), a component of the ESCRT-IIcomplex, as an important regulator for Smo endocytic trafficking.

In the absence of Hh, Vps36 interacts with ubiquitylated Smo,thereby negatively regulating the accumulation of Smo on plasmamembrane. Collectively, our data suggest that the ESCRT

pathway plays a special role in controlling Hh signaltransduction by targeting the membrane protein Smo for itstrafficking, thereby regulating proper Hh signaling activity inmetazoan development and pattern formation.

ResultsPerturbation of Vps36 leads to Smo accumulation in the Acompartment of wing imaginal discs

As one of the membrane proteins involved in Hh signaling, Smocell plasma membrane accumulation is regulated by the Hh signal,which is critical for its function (Jia and Jiang, 2006; Jiang andHui, 2008). In the absence of Hh, Smo membrane accumulation is

blocked. To figure out how Smo function is regulated, we set outto investigate the inhibitory mechanism of Smo membraneaccumulation. ESCRT complexes, especially the components of

ESCRT-II complex, play an important role in membrane proteintrafficking, thereby regulating multiple conserved signalingpathways (Henne et al., 2011; Marchese et al., 2008; Raiborg

and Stenmark, 2009). Is it possible that the components of ESCRT-II complex are also involved in the regulation of Smo trafficking?To address this possibility, we performed a genetic modifier screenby overexpressing RNAi lines of ESCRT-II components with

C765-Gal4;Smo-PKA12 in Drosophila wing imaginal discs.Smo-PKA12 is partial dominant-negative form of Smo and willlead to a ‘fused wing’ phenotype when it is overexpressed. In the

screen, we identified Vps36, which partially rescued the ‘fusedwing’ phenotype with its RNAi overexpression in Drosophila

wing driven by C765-Gal4; Smo-PKA12, as a player of Hh signaling

(supplementary material Fig. S1A–A0).

Vps36 is a key component of ESCRT-II complex and isinvolved in membrane protein sorting and trafficking (Babst et al.,

2002b; Hierro et al., 2004; Katzmann et al., 2002; Raiborg andStenmark, 2009; Teo et al., 2004). To figure out how Vps36is involved in Hh signaling, we performed in vivo staining

for Hh signaling pathway components. Our results show thatoverexpression of Vps36 RNAi with apterous(AP)-Gal4–GFP inthe dorsal compartment of late third instar larvae wing imaginal

discs led to a specific accumulation of Smo in the Acompartment, but not in the P compartment (Fig. 1A–B0). Toconfirm this finding, we generated another transgenetic fly for

Vps36 RNAi based on the method reported by Wang andcolleagues (Wang et al., 2011). Similar results were obtained

(data not shown). However, overexpression of RNAi mightsometimes cause off-target effects. To rule out this possibility,we used FLP/FRT-mediated recombination to induce mitoticclones of Vps36L5212, in which endogenous Vps36 expression is

absent. Similar to the observation in the experiment where Vps36was knocked down, endogenous Smo specifically accumulated inVps36 mutant clones in the A compartment (Fig. 1C–C0). This

result suggests that Vps36 affects the level of Smo in the absenceof Hh. However, another possibility is that Vps36 affects Hh signaldiffusion, and thereby the Smo protein level, which is also

regulated by the Hh signal. To rule out this possibility, we nextused Act5c.CD2.Gal4(AG4)-Dicer2-GFP to drive Vps36 RNAioverexpression. In Drosophila wing imaginal discs, we found thatSmo was still upregulated in the clones of the A compartment far

from the A–P boundary, where Hh signal is absent (Fig. 1D–E0).Thus our data argue that the regulation of Vps36 on Smo is not dueto Hh signal distribution.

ESCRT complexes play a critical role in the regulation ofmembrane protein trafficking (Henne et al., 2011; Marchese et al.,2008; Raiborg and Stenmark, 2009). Vps36 belongs to the ESCRT-

II complex, a sub-complex of ESCRT machinery. To investigatewhether Vps36 regulates Smo trafficking, we performed anantibody-uptake assay in S2 cells. Cells transfected with Myc–

Smo were incubated with anti-Myc antibody at 4 C, a restrictivetemperature for endocytosis, for 30 minutes to label the cell surfaceMyc–Smo. They were then shifted to 25 C for endocytosis. Wefound that at the beginning (0 minutes), most labeled Smo stayed on

the cell plasma membrane (Fig. 1F,G). After 10 minutes, surface-labeled Smo underwent endocytosis and was mainly located in thecytosol (Fig. 1F9). However, this process was delayed in cells

treated with Vps36 dsRNA (Fig. 1G9), implying that knockdown ofVps36 results in a delay of Smo endocytosis. We also found thatmost of the labeled Smo disappeared 1 hour later, but knockdown of

Vps36 could block this process (Fig. 1F0,G0). The efficiency ofVps36 RNAi in S2 cells was examined by real-time PCR(supplementary material Fig. S1F). These results suggested thatVps36 affects Smo trafficking, endocytosis and eventually its

degradation. To investigate whether other components of theESCRT machinery are also involved in the regulation of Smo, wefurther tested the loss-of-function effects of other ESCRT key

components on Smo. We found that knockdown of Vps25 withAG4-Dicer2–GFP, another component of the ESCRT-II complex,resulted in the accumulation of Smo in only the A compartment

(supplementary material Fig. S1B–B0). Meanwhile, Smo alsoaccumulated in the mitotic clones of Vps25K08904 in the Acompartment (supplementary material Fig. S1C–C0). Similarly,

RNAi overexpression of HRS (a component of ESCRT-0) or Vps28(a component of ESCRT-I) with APGal4–GFP or AG4-Dicer2–GFPalso resulted in specific accumulation of Smo in the A compartment,although not all combinations resulted in membrane accumulation

(supplementary material Fig. S1D–E0, and data not shown). Takentogether, these results suggested that the perturbation of ESCRTcomplexes leads to Smo accumulation in the A compartment of

wing imaginal discs.

Vps36 regulates Smo trafficking and then influences Hhsignaling activity

Smo is the key transducer for the Hh gradient signal across thecell plasma membrane, thereby contributing to the expression of

Vps36 regulates Smo trafficking 4231

Journ

alof

Cell

Scie

nce

downstream genes (Jiang and Hui, 2008; Zhao et al., 2007).

However, it remains largely unknown how Smo trafficking is

regulated, although it is thought that Smo C-tail phosphorylation

plays a critical role in this event (Jia et al., 2004). To further

investigate how the ESCRT-II complex regulates Smo

accumulation, we performed both gain- and loss-of-function

studies in wing imaginal discs. Consistent with results mentioned

above, when Vps36 RNAi was overexpressed driven by MS1096-

Gal4 (drive genes expressed mainly in the wing pouch), the

endogenous Smo accumulated dramatically only in the A

compartment (Fig. 2A9,B9). As shown in the magnified images,

accumulated Smo in A compartment caused by loss-of-Vps36

accumulated in intracellular endosomes and in some cells also

accumulated on the cell plasma membrane (Fig. 2A-,B-). This

implied that knockdown of Vps36 led to a Smo endosome-

trafficking defect and suggests that Vps36 is involved in Smo

trafficking in the absence of Hh. Furthermore, we found that

accumulated Smo upregulated the expression of the downstream

gene dpp (dpp–lacZ) (Fig. 2A,B,A0,B0, arrows in 2B), which is

switched on by low-level Hh signal (Hooper and Scott, 2005; Jia

and Jiang, 2006; Jiang and Hui, 2008). The ptc expression is a

marker of Hh signaling with higher threshold (Hooper and Scott,

1989; Hooper and Scott, 2005; Jia and Jiang, 2006; Jiang and

Hui, 2008). In this experiment, we also found that the ptc

expression pattern (ptc–lacZ) was extended from the A–P

boundary to the A compartment when Vps36 was knocked

down (Fig. 2C–C0,D–D0).

Overexpression of wild-type Smo activates downstream dpp

gene expression (Jia et al., 2004; Ranieri et al., 2012; Zhao et al.,

2007). To confirm that the upregulation of dpp expression is due

to Smo accumulation induced by knockdown of Vps36, we co-

expressed FLAG-tagged wild-type Smo (FLAG–Smo) and Vps36

RNAi in wing imaginal discs driven by MS1096-Gal4. Consistent

with previous results (Jia et al., 2004), overexpression of wild-

type Smo alone weakly upregulated dpp expression (Fig. 2E–E0).

However, knockdown of Vps36 further increased dpp expression

level (Fig. 2F–F0, compared with 2E–E0). Next, we sought to

investigate the gain-of-function effect of the ESCRT-II complex.

We found that overexpression of V5-tagged Vps36 (Vps36–V5)

alone driven by AG4, or overexpression of the entire ESCRT-II

complex, including Vps22, Vps25 and Vps36, decreased the Smo

level specifically in the A compartment (supplementary material

Fig. S2). These results suggest that Vps36 is involved in the Hh

signaling pathway in the absence of Hh signal.

Vps36 interacts with Smo in the absence of Hh signal

We next sought to identify how the ESCRT-II complex regulates

Smo trafficking in the Hh signaling pathway. To figure out which

Fig. 1. Perturbation of Vps36 leads to Smo accumulation in the A

compartment of the wing imaginal disc. (A–A0) Wing imaginal discs

expressing UAS–GFP alone (A) or with UAS–Vps36 RNAi (B) driven by the

dorsal compartment-specific driver AP-Gal4 were stained with anti-Smo

antibody to show the endogenous Smo (red). GFP signal marks the gene

expression regions. White lines mark the boundary of dorsal and ventral

compartments of the wing imaginal disc. Arrow indicates Smo in the dorsal

compartment and arrowhead indicates Smo in the ventral compartment. Smo

accumulates in the A compartment when Vps36 is knocked down (B).

(C–C0) A wing imaginal disc carrying Vps36 mutant clones stained with anti-

Smo antibody (red). Vps36 mutant clones are marked by the lack of GFP

signal. Arrow indicates a clone in the A compartment in which Smo is

accumulated. Arrowhead indicates a clone in the P compartment in which

Smo staining is no different from that outside the clone. (D,E) Wing imaginal

discs expressing UAS–GFP alone (D) or together with UAS–Vps36 RNAi

(E) driven by AG4-Dicer2 stained with anti-Smo antibody (red). GFP signals

label the target gene expression regions. Arrows indicate clones in the A

compartment in which Smo is upregulated (E). Smo accumulates only in the

clones in the A compartment expressing Vps36 RNAi (E9). (F–G0) S2 cells

were transfected with Myc-tagged wild-type Smo (Myc–SmoWT) and treated

with control dsRNA (Renilla) or Vps36 dsRNA. Antibody-uptake assay with

mouse anti-Myc to label Smo at the surface of live S2 cells. Cells were

subjected to staining at indicated time. Vps36 RNAi delays Smo endocytosis

after 10 minutes (F9,G9) and shows further degradation after 1 hour (F0,G0).

Journal of Cell Science 126 (18)4232

Journ

alof

Cell

Scie

nce

component of the ESCRT-II complex contributes to its

interaction with Smo, we co-expressed Myc-tagged wild-type

Smo (Myc–SmoWT) with V5-tagged Vps22 (Vps22–V5), Vps25

(Vps25–V5) or Vps36 (Vps36–V5) in S2 cells. We then carried

out co-immunoprecipitation (Co-IP) experiments. The results

suggested that Vps36 contributes to the interaction between the

ESCRT-II complex and Smo (Fig. 3A). To further investigate

how Smo interacts with Vps36 and to figure out whether this

interaction is direct or indirect, we performed a yeast two-hybrid

assay using the Smo C-tail as bait and Vps36 as prey. However,

we did not find any clues that Vps36 interacts with the Smo C-tail

directly (data not shown).

Smo trafficking and cell plasma membrane accumulation is

regulated by the Hh signal. We next wanted to ask whether the

interaction between Smo and Vps36 is regulated by the Hh

signal. We co-expressed Myc–Smo and Vps36–V5 in S2 cells

with or without Hh treatment. We found that Vps36 interacted

with Smo in the absence of Hh, and this interaction was blocked

by the Hh signal (Fig. 3B), which was consistent with our

observation that Smo was affected by Vps36 only in the A

compartment of wing imaginal discs (Fig. 1B–E), where Hh

signal was absent. Consistent with this finding, an S2 cell-

staining assay revealed that Vps36 colocalized with Smo only in

the absence of Hh (supplementary material Fig. S3).

Phosphorylation of the Smo C-tail is tightly controlled by the

Hh signal gradient, and is very important for its activity and cell

plasma membrane localization (Jia et al., 2004; Zhao et al.,

2007). The Smo C-tail phosphorylation level and its activation

can be mimicked by SmoSD mutants, in which three PKA sites

(Ser667, Ser687 and Ser740) and adjacent CKI sites are mutated to

aspartic acid. Another Smo mutant, SmoSA, which has three

PKA sites mutated to alanine, can mimic the situation in which

the Smo C-tail fails to be phosphorylated and no longer responds

to the Hh signal (Jia et al., 2004; Zhao et al., 2007). To

Fig. 2. Vps36 regulates the accumulation of Smo and Hh signaling

activity. (A–F) Wing imaginal discs expressing the indicated genes in wing

pouch driven by MS1096-Gal4 were stained with anti-lacZ (green) and anti-

Smo (red) or anti-FLAG (red) antibodies to show the expression of dpp–lacZ

(A,B,E,F) or ptc–lacZ (C,D), endogenous Smo (A–D) or FLAG-tagged Smo

(FLAG–Smo) (E,F). White arrows indicate the ectopic expression of either

dpp–lacZ or ptc–lacZ. Overexpression of Vps36 RNAi caused accumulation

of Smo in the A compartment (compare Smo staining in A9 and B9, C9 and

D9) and upregulates dpp expression (compare dpp–lacZ staining in A and B)

and ptc expression (compare ptc–lacZ staining in C and D). Overexpression of

FLAG–Smo driven by MS1096-Gal4 resulted in the upregulation of dpp

expression (E). Knockdown of Vps36 further increased dpp expression

(compare dpp–lacZ staining in E and F). In addition, the magnified images

showed that knockdown of Vps36 results in Smo accumulation on the cell

plasma membrane in the A compartment (compare Smo staining in A- and B-).

The white squares in A0 and B0 mark the magnified regions for A- and B-,

respectively. Arrows indicate the expression of dpp–lacZ (B,E,F) or ptc–lacZ

(C,D), respectively. Scale bars: 5 mm (A- and B-).

Fig. 3. Vps36 interacts with Smo in the absence of Hh. (A–E) S2 cells

were transfected with the indicated constructs, treated with or without Hh, and

followed by Co-IP with anti-Myc antibody and western blot for anti-Myc to

detect wild-type Smo or Smo mutation expression, or for anti-V5 to detect

wild-type Vps22, Vps25, Vps36, Vps36GLUE expression in the Co-IP

products or cell lysates, respectively. Arrows indicate the bands where

Vps36–V5 or Vps36GLUE–V5 interacted with Myc–SmoWT or Myc-tagged

Smo mutations. (A) In ESCRT-II complex, only Vps36 interacted with wild-

type Smo. Vps36 only interacts with Smo in the absence of Hh (B). This

interaction is regulated by Smo C-tail phosphorylation (C) and the

phosphorylation clusters (D). (E) GLUE domain of Vps36 (Vps36GLUE–V5)

interacts with Smo in the absence of Hh.

Vps36 regulates Smo trafficking 4233

Journ

alof

Cell

Scie

nce

investigate whether the interaction between Smo and Vps36 is

regulated by Smo C-tail phosphorylation, we co-expressedVps36–V5 with Myc–SmoWT, Myc-tagged SmoSA (Myc–SmoSA) or Myc-tagged SmoSD (Myc–SmoSD) in S2 cells,

respectively. In Co-IP experiments, we found that both SmoWTand SmoSA interacted with Vps36; however, SmoSD could notdo so (Fig. 3C). These results imply that the interaction betweenSmo and the ESCRT-II complex mediated by Vps36 is regulated

by the Hh signal and Smo C-tail phosphorylation.

The Smo auto-inhibitory domain (SAID domain, Smo C-tailamino acids 661–818) contains several arginine motifs, which

can block the phosphorylation of their adjacent target sites (Jiangand Hui, 2008; Zhao et al., 2007). To further test howphosphorylation of Smo regulates its trafficking, as well as itsinteraction with Vps36, we performed a Co-IP assay. Similar to

SmoWT, Myc-tagged SmoD818 (Myc–SmoD818), in whichamino acids 819 to the end of Smo were deleted and thephosphorylation clusters were kept, still interacted with Vps36–

V5. However, both Myc-tagged SmoD661 (Myc–SmoD661), inwhich amino acids 662 to the end of Smo, including thephosphorylation clusters, were deleted and Myc-tagged

SmoD661–818 (Myc–SmoD661–818), in which the SAIDdomain was deleted, did not interact with Vps36–V5 (Fig. 3D).These data indicated that the Smo SAID domain and its

phosphorylation clusters play an important role in theregulation of Smo function by the ESCRT-II complex. Takentogether, ESCRT-II complex interacts with Smo via Vps36, andthis interaction is regulated by the Hh signal and Smo C-tail

phosphorylation.

Yeast Vps36 contains one GLUE (GRAM-like ubiquitin-binding in Eap45) domain and two NZF (Npl4-type zinc finger)

domains. The GLUE domain interacts with 3-phosphorylatedphosphoinositides and one of the NZF domains is responsible forrecognizing ubiquitin (Teo et al., 2006). However, the GLUEdomain rather than the NZF domain, has been found in the

mammalian and Drosophila homologs of Vps36 (Alam et al.,2006; Hirano et al., 2006; Irion and St Johnston, 2007; Slagsvoldet al., 2005). To test whether the Drosophila Vps36 GLUE

domain plays a role in Smo interaction with Vps36, weperformed additional Co-IP experiments. As shown in Fig. 3E,the V5-tagged Vps36 GLUE domain (Vps36GLUE–V5, amino

acids 1–150) interacted with Smo, and this interaction wasregulated by the Hh signal. Collectively, these data suggest thatthe interaction between Smo and Vps36 is mediated by the Vps36

GLUE domain, and regulated by the Hh signal and Smo C-tailphosphorylation.

The ubiquitin signal mediates the interaction between Smoand Vps36

Ubiquitin (Ub) is a 76 amino acid protein that is highly conservedfrom yeast to human. During protein ubiquitylation, Ub iscovalently attached to the lysine residues via its C-terminus.

Ubiquitylation is an important protein post-translationalmodification, which was primarily found as a signal for proteindegradation by the proteasome, and was later identified as an

important regulatory mechanism for a variety of cellularprocesses including signal transduction, receptor modificationand DNA repair (Haglund and Dikic, 2005; Haglund and Dikic,

2012; Hayden and Ghosh, 2008). Previous studies suggested thatubiquitylation of certain membrane proteins is a very criticalsignal for its sorting into intraluminal vesicles (ILVs) of

endosomes, then MVBs (Henne et al., 2011; Marchese et al.,2008; Raiborg and Stenmark, 2009). In addition, in yeast and

mammalian cells, Vps36 can recognize the ubiquitylated proteins(Alam et al., 2006; Hirano et al., 2006; Teo et al., 2006). In theexperiment, we found that Vps36 recognizes Smo in the absenceof Hh, but not in yeast two-hybrid assay. On the basis of this, we

hypothesized that ubiquitylation of Smo mediates the interactionbetween Smo and Vps36. To test this possibility, we firstperformed a Co-IP assay to check the ubiquitylation state of Smo.

Consistent with recent findings from other groups (Li et al., 2012;Xia et al., 2012), when Myc–SmoWT was co-expressed with HA-tagged ubiquitin (HA–Ub) in S2 cells, Smo was ubiquitylated in

the absence of Hh, yet Hh signal stimulation dramaticallydecreased the Smo ubiquitylation level (Fig. 4A). Meanwhile, wefound that Myc–SmoD661–818, which did not interact withVps36 (Fig. 3D), had a lower ubiquitylation level compared with

Myc–SmoWT (supplementary material Fig. S4A). Furthermore,knockdown of Vps36 increased the in vivo ubiquitylated substratelevel stained by FK2 antibody, which recognizes the ubiquitin

conjoining substrates (supplementary material Fig. S4B–C0).Together with the result that Smo only interacts with Vps36 inthe absence of Hh, these observations suggest that Smo is

ubiquitylated in the absence of Hh, and then recognized by Vps36for its trafficking and endocytosis. In retrospect, it also explainswhy we did not observe interaction between the Smo C-tail and

Vps36 in the yeast two-hybrid experiment because the Smo C-tailmight not be modified by ubiquitin in yeast. To confirm thisconclusion, we generated a fusion protein, Myc–Smo–Ub, inwhich a ubiquitin is fused to the end of Myc-tagged wild-type

Smo. Co-IP suggested that, compared with Myc–SmoWT, Myc–Smo–Ub had a stronger interaction with Vps36, even in thepresence of Hh (Fig. 4B). This result supports our notion that

ubiquitylation of Smo mediates the interaction between Smo andVps36.

USP8 is a Smo deubiquintylase that can decrease the Smo

ubiquitylation level (Li et al., 2012; Xia et al., 2012). To furtherdissect how ubiquitylation regulates the interaction between Smoand Vps36, we used USP8. When we co-expressed Myc–SmoWTwith FLAG-tagged USP8 (FLAG–USP8), we found that, indeed,

overexpression of USP8 decreased not only the Smo ubiquitylationlevel, but also the interaction between Vps36 and Smo (Fig. 4C).These results further support our conclusion that Vps36 interacts

with Smo through ubiquitin in the absence of Hh.

Considering that ubiquitin might function as a signal forendocytosis and trafficking of many membrane proteins (Henne

et al., 2011; Marchese et al., 2008; Raiborg and Stenmark, 2009)and that ubiquitin mediates the binding between Smo and Vps36in the absence of Hh, we hypothesized that ubiquitylation is also

involved in the regulation of Smo endocytosis and trafficking.Cell staining suggested that, unlike Myc–SmoWT accumulationon the cell plasma membrane in the presence of Hh, Myc–Smo–Ub was detained in cytosol under the same conditions (Fig. 4D).

This result suggests that ubiquitylation is involved in regulatingSmo endocytosis and trafficking.

UBP2 is a deubiquintylase in yeast and was also reported to

have a deubiquitylation function in Drosophila (DiAntonio et al.,2001). We next used the UBP2 transgenic fly to further studyhow ubiquitylation mediates Smo endocytosis and trafficking.

We found that when UBP2 was overexpressed with AG4–GFPand labeled with GFP in wing imaginal discs, Smo onlyaccumulated in GFP-positive clones of the A compartment

Journal of Cell Science 126 (18)4234

Journ

alof

Cell

Scie

nce

(Fig. 4E–E0). We did not observe obvious Smo accumulation in

the GFP-positive clones of the P compartment. The magnified

images showed us in detail that, in the GFP-positive clones of A

compartment, the accumulated Smo localized to the cell plasma

membrane (Fig. 4F–F0). However, Smo staining was no different

between inside or outside the GFP-positive clones of the P

compartment (Fig. 4G–G0). Similar results were also observed in

the A compartment when the FLAG–Usp8 transgenic fly was

overexpressed with AG4 (supplementary material Fig. S4D–F0).

Therefore, our results here suggest that the Hh signal regulates

Smo trafficking by controlling its ubiquitylation. As a result, the

ubiquitylated Smo is recognized by Vps36 through the ubiquitin

signal for its trafficking.

Ubiquitylation sites in Smo

Because our results suggested that ubiquitylation of Smo plays a very

important role in Smo trafficking, we next tried to figure out how

ubiquitylation affects Smo trafficking and which lysine residue(s) in

the cytoplasmic part of Smo, including its C-tail and internal loops,

is/are ubiquitylated and contribute(s) to this event. As we have shown

that the Smo SAID domain is very important for Smo trafficking and

recognition by Vps36, it is likely that lysine residues within this

region are targeted for ubiquitylation. We then first mutated 13 lysine

residues to arginine in amino acids 661–818 of Smo (Myc-tagged

SmoK13R, Myc–SmoK13R). Our results showed that the Myc–

SmoK13R ubiquitylation level was comparable with that of wild-

type Smo (Fig. 5A), suggesting that these lysine residues might not

be key sites for Smo ubiquitylation, although the SAID domain is

very important for regulating Smo ubiquitylation. To figure out

which lysine residue(s) is/are important, we then mutated all 42

lysine residues to arginine in the Smo C-tail to generate Myc-tagged

SmoK42R (Myc–SmoK42R) or mutated six lysine residues in the

internal loops to arginine to generate Myc-tagged SmoK6R (Myc–

SmoK6R). Unexpectedly, we found that Myc–SmoK42R or Myc–

SmoK6R was still ubiquitylated although to a dramatically decreased

extent, especially for Myc–SmoK42R (Fig. 5A). We then generated

Myc-tagged SmoKallR (Myc–SmoKallR), in which 48 lysine

residues in both the Smo C-tail and Smo internal loops were

mutated to arginine. The ubiquitylation level of Myc–SmoKallR was

further decreased (Fig. 5A), suggesting that multiple lysine residues

in the Smo C-tail and internal loops contribute to its ubiquitylation in

the absence of Hh. We next examined the cellular localization of

SmoKallR. As expected, unlike SmoWT, SmoK13R or SmoK6R,

SmoK42R and SmoKallR largely accumulated on the cell plasma

membrane in the absence of Hh (Fig. 5B, supplementary material

Fig. S5A–D9). Western blotting showed that the binding between

Vps36 and Smo was weakened when all lysine residues in the Smo

C-tail and internal loops were mutated to arginines (Fig. 5C). These

results further suggest that Vps36 regulates Smo trafficking by

recognizing the ubiquitin signals on Smo.

Because overexpressed SmoK42R and SmoKallR accumulate on

the cell surface in the absence of Hh, we next investigated their

functions in vivo. We generated transgenic flies of attB-UAS-Smo

variants at the 25C attP locus, which ensures equal expression of

different forms of Smo mutants, including wild-type Smo,

SmoK42R and SmoKallR. We then overexpressed wild-type Smo

and these two Smo mutants using MS1096-Gal4. We found that,

compared with the wild-type, Smo overexpression of SmoK42R–

FLAG or SmoKallR–FLAG increased dpp expression (Fig. 5D–F0).

DiscussionAs one of essential membrane proteins of the Hh signaling

pathway, Smo contributes to transducing the graded Hh signal

across the cell plasma membrane. Accumulation of Smo on the

Fig. 4. Ubiquitin mediates the interaction between

Smo and Vps36. (A–C) S2 cells were transfected

with the indicated constructs, treated with or without

Hh, followed by Co-IP with anti-Myc antibody and

western blot for anti-Myc to detect wild-type Smo

(A,C) or Smo–Ub (B) expression, for anti-HA to

detect the overexpressed Ub (A,C), for anti-V5 to

detect expression of Vps36 (B,C) or for anti-FLAG to

detect expression of USP8 (C) in the Co-IP product or

cell lysates. Arrows indicate the bands where Vps36–

V5 interacted with Myc–SmoWT or Myc–SmoWT–

Ub. (A) In the absence of Hh, the ubiquitylation level

of Smo was higher than that in the presence of Hh.

(B) Smo–Ub interacts with Vps36 regardless of Hh.

(C) USP8 decreases the ubiquitylation level of Smo

and its interaction with Vps36. (D) S2 cells were

transfected with the indicated constructs and treated

with or without Hh, followed by immunostaining with

anti-Myc antibody. Ub blocks Smo cell plasma

membrane accumulation induced by Hh signal.

(E–E0) A wing imaginal disc expressing UBP2 with

AG4–GFP stained with anti-Smo antibody. White

squares indicate the magnified regions in the A and P

compartment in E0, respectively. GFP signal indicates

target gene UBP2 expression regions (E–G, E0–G0). In

the magnified images, expression of UBP2

accumulated Smo on the cell plasma membrane only

in the A compartment (compare Smo staining in F9

and G9). Scale bars: 5 mm (F0,G0).

Vps36 regulates Smo trafficking 4235

Journ

alof

Cell

Scie

nce

cell plasma membrane and its C-tail phosphorylation are finely

controlled by the Hh signal. In the absence of Hh, Smo fails to

accumulate on the cell plasma membrane. In the presence of Hh,

Ptc relieves its inhibition of Smo activity by binding Hh, resulting

in Smo cell plasma membrane accumulation. However, it

remains largely unknown how Smo trafficking is regulated

during Hh signaling. In this study, we first found that the ESCRT-

II complex regulated Smo trafficking in the absence of Hh,

through the interaction between Vps36 and the ubiquitylated

Smo. Our in vitro and in vivo observations show that Vps36

specifically recognizes Smo and then regulates its endosome

trafficking. We also found that the recognition of Smo by Vps36

is dependent on multiple ubiquitylation sites of its C-tail and

internal loops. Meanwhile, Smo association with Vps36 is

regulated by its C-tail phosphorylation and Hh signal.

Moreover, the GLUE domain of Drosophila Vps36 is involved

in recognizing ubiquitin, as does its mammalian homolog.

However, in this experiment, we also observed that loss of Hrs,

a component of the ESCRT-0 complex, resulted in Smo

intracellular endosome accumulation. This also implied that,

unlike Hrs, Vps36 might play a specific role in Smo regulation.

Ubiquitin can function as a signal for the endosome sorting and

trafficking of many membrane proteins (Henne et al., 2011;

Marchese et al., 2008; Raiborg and Stenmark, 2009). Here, we

report that the Hh signal regulates Smo trafficking by controlling

its ubiquitylation. In the absence of Hh, Smo was ubiquitylated

and recognized by ESCRT complexes for endosome trafficking

and for further sorting, therefore inhibiting its accumulation on

the cell plasma membrane and blocking its activity. However, in

the presence of Hh, the ubiquitylation of Smo was inhibited, and

the cytosol to plasma membrane trafficking of Smo was blocked,

resulting in its accumulation on the cell membrane and activation

of the Hh pathway.

In our experiment, we found that the SAID domain of Smo is

important to its recognition by Vps36, which is required for Smo

trafficking. The Smo SAID domain has three PKA and CKI

phosphorylation clusters and four arginine motifs, which are very

important for Smo activity regulation and its conformational

change in response to Hh (Zhao et al., 2007). A recent study

suggested that Smo is ubiquitylated at multiple lysine residues in

the SAID domain (Li et al., 2012). To characterize Smo

ubiquitylation, we first mutated all 13 lysine residues located in

the SAID domain (SmoK13R). Unexpectedly, we found that

SmoK13R, in which all lysine residues in the SAID domain were

mutated to arginine, showed equivalent ubiquitylation levels

compared with SmoWT. To further address this question, we then

mutated more lysine residues in Smo cytoplasmic regions,

including its C-tail and internal loops. Finally, we found that

only overexpression of SmoKallR, in which all the lysine

residues in the cytoplasmic part were mutated to arginine, leads

to a dramatic decrease of the Smo ubiquitylation level, and

thereby its recognition by Vps36. One explanation for this

phenomenon is that Smo ubiquitylation sites are not limited to

only a few lysine residues. It is likely that the Hh signal regulates

Smo trafficking by controlling Smo ubiquitylation on multiple

lysine sites. Meanwhile, even SmoKallR still has basal level

ubiquitylation. At this stage, we cannot rule out the possibility

that, in the absence of Hh, other amino acid residues, such as

cysteine and serine residues, can be ubiquitylated, as previously

reported (Cadwell and Coscoy, 2005; Wang et al., 2007).

Fig. 5. Mapping of the ubiquitylation sites in

Smo cytoplasmic region. (A,C) S2 cells were

transfected with the indicated constructs,

followed by Co-IP with anti-Myc antibody and

western blot for anti-Myc to detect the indicated

Smo mutation expression (A,C), for anti-Ub to

detect endogenous Ub level (A) and for anti-V5

to detect Vps36 expression (C). (B) Staining for

anti-Myc was performed to detect change in

Myc–SmoWT or Myc–SmoKallR localization

after treatment with or without Hh. Western blot

results show that, compared with other Smo

mutations, ubiquitylation level of SmoKallR

dramatically decreased (A), and this mutation

did not interact with Vps36 (C). (B) Cell

staining shows that Myc–SmoKallR

accumulates on the cell plasma membrane even

without Hh treatment. (D–F0) Wing imaginal

discs expressing the indicated genes in wing

pouch driven by MS1096-Gal4 stained with

anti-lacZ (green) and anti-FLAG (red)

antibodies to show the expression of dpp–lacZ

and overexpressed genes. Overexpression of

SmoK42R–FLAG and SmoKallR–FLAG further

increased dpp–lacZ expression (compare dpp–

lacZ staining in D–F).

Journal of Cell Science 126 (18)4236

Journ

alof

Cell

Scie

nce

Collectively, our work uncovered a mechanism whereby

the Hh signal controls Smo trafficking by regulating Smo

ubiquitylation. The next question would be how the Hh signal

regulates Smo ubiquitylation. One hypothesis is that the Hh

signal induces Smo phosphorylation and conformational change,

which results in an ‘open’ and activated conformation of Smo

that fails to recruit E3 ubiquitin ligase for ubiquitylation.

Consistent with this notion, the SAID domain of Smo, which is

important for the maintenance of Smo ‘closed’ conformation by

arginine motifs, has been found to be critical for Smo cell plasma

membrane localization (Zhao et al., 2007). We further found that

Smo lacking the SAID domain had a lower ubiquitylation level

independent of the ubiquitylation of lysine residues in this

domain (Fig. 5A, supplementary material Fig. S4A). It is also

possible that the binding of E3 ubiquitin ligase to Smo is

regulated by the Hh signal and Smo C-tail phosphorylation. In

addition, our results suggest that Smo trafficking is regulated by

the ESCRT-II complex only in the absence of Hh. How Smo

internalization is regulated in the presence of Hh, as well as the

specific E3 ligase and components involved in this process

warrants further study.

Materials and MethodsDNA constructs

All Drosophila genes used in this study were constructed in pUAST vectors. Thegeneration of Myc-tagged Smo wild-type, SmoD661, SmoD818, SmoD661–818,SmoSA and SmoSD were described previously (Zhao et al., 2007). Smo variantswith substitutions at lysine residues to arginine were generated using PCR-basedsite-directed mutagenesis at the background of Myc–Smo. The internal loops ofSmo were predicted as reported previously (Carroll et al., 2012). For Myc–SmoK6R, we mutated all the lysine residues (K282, K286, K378, K379, K460 andK467) located in the internal loops to arginine. For Myc–SmoK13R, we mutatedall the 13 lysine residues to arginine in amino acids from 661 to 818 of Smo. Myc–SmoK42R was a Smo mutant with all the 42 lysine residues in Smo C-tail mutatedto arginine. In Myc–SmoKallR, all the lysine residues in Smo cytoplasmic partwere mutated to arginine. Myc–SmoUb and Smo–CFP were constructed by sub-cloning the Ub or CFP sequence in frame with the C-terminus of Myc–Smo orSmo, respectively. Vps36–V5, Vps22–V5 and Vps25–V5 were constructed byinserting the coding sequences into the UAST–V5 vector. For Vps36GLUE–V5,amino acids 1–150 of Vps36 were kept.

Drosophila mutants and transgenes

Drosophila strains used in this study were maintained under standard conditions.The yw strain was used as host for all the P-element mediated transformations.MS1096- Gal4, act .CD2 .Gal4(AG4), AG4-Dicer2, AP-Gal4, Ci-Gal4, C765-

Smo-PKA12, dpp-lacZ, ptc-lacZ, UAS-GFP, UAS-FLAG-SmoWT have beendescribed (Flybase) (Jia et al., 2004; Zhao et al., 2007). Vps36 RNAi (VDRC,#16847), Vps25 RNAi (NIG, 14750 R-1), Smo RNAi (11561R-1), Vps25k08904

(Bloomington, #10839), HRS RNAi (Bloomington, #28026), Vps28 RNAi(VDRC, #31894) and UBP2 (Bloomington, #9907), were obtained from NIG,Bloomington or VDRC. UAS-FLAG-USP8 transgenic fly was a generous gift fromDr Jianhang Jia (Xia et al., 2012). Vps36L5212 is a null allele and a generous giftfrom Dr Daniel St Johnston (Irion and St Johnston, 2007). Another Vps36 RNAitransgenic fly was generated as reported (Wang et al., 2011). The target sequenceswere GAGCAGCTTAGTGGTCCAATA and GCAATTCTGCCTACTTCAGTA.To generate mutant clones, FRT/FLP-mediated mitotic recombination was used forthe following genotypes: Vps36 clone: yw 122, FRT2A Vps36L5212/FRT2A hs-

GFP; Vps25 clone: yw 122; FRT42 Vps25k08904/FRT42 hs-GFP.

Cell culture, transfection, immunoprecipitation, western blot analysisand immunostaining

S2 cells were cultured in Schneider’s Drosophila Medium (Invitrogen) with 10%fetal bovine serum, 100 U/ml penicillin and 100 mg/ml streptomycin. Transfectionwas carried out by using a Calcium Phosphate Transfection Kit (Specialty Media)according to manufacturer’s instructions. An ub-Gal4 plasmid was co-transfectedwith pUAST expression vectors for all the transfection experiments. 5 mg DNA forub-Gal4 and 5 mg DNA for each pUAST expression vector were used in a typicaltransfection experiment for 10 cm dish. Cells were harvested 48 hours aftertransfection with indicated buffers for different assays. For immunoprecipitationassay, cells were lysed in NP-40 buffer (50 mM Tris-HCl pH 8.0, 0.1 M NaCl,10 mM NaF, 1 mM Na3VO4, 1% NP-40, 10% Glycerol, 1.5 mM EDTA, protease

inhibitor cocktail) for 30 minutes at 4 C. After centrifugation, lysates wereincubated with 2 mg indicated antibodies for 2 hours at 4 C. Samples werecombined with 20 ml Protein A/G PLUS agarose (Santa Cruz) and incubated for1 hour on a rotator at 4 C. Beads were washed three times with 1 ml NP-40 buffer.For detecting Smo, the IP products and cell lysates were incubated at 37 C for30 minutes, to avoid boiling, before loading on an SDS gel, and then subjected towestern blot with indicated antibodies.

For immunostaining of wing imaginal discs, third-instar larvae were cut in halfand fixed in freshly made 4% formaldehyde in PBS buffer at room temperature for20 minutes, then rinsed with buffer PBT (PBS, 0.1% Triton X-100) and washedfour times with buffer PBTA (PBS, 0.1% Triton X-100, 1% BSA). Larvae wereincubated overnight with primary antibody diluted in PBTA at 4 C, then washedwith PBT and incubated with secondary antibody diluted in PBTA for 2 hours atroom temperature. After washing, wing imaginal discs were dissected and mountedin 40% glycerol. A Leica LAS SP5 confocal microscope was used to recordimmunostaining images. Primary antibodies used in this study: mouse anti-Smo(DSHB), rabbit anti-FLAG (Sigma), rat anti-Ci (2A1) (DSHB), rabbit anti-lacZ(MP Biomedicals), mouse anti-HA (Sigma), mouse anti-Myc (Sigma), mouse anti-V5 (Sigma) and mouse anti-FK2 (Millipore). Secondary antibodies used in thisstudy were purchased from Millipore.

Ubiquitylation assay

S2 cells transfected with Smo variants with or without HA-tagged Ub (wild-type ormutant) were lysed in denaturing buffer (1% SDS, 50 mM Tris-HCl, pH 7.5,0.5 mM EDTA, 1 mM DTT) by ultrasound. Then the lysates were diluted tenfoldwith regular lysis buffer and subjected to immunoprecipitation with anti-Mycantibody. The samples were incubated at 37 C for 30 minutes before SDS gelelectrophoresis, and then subjected to western blot analysis with anti-HA or mouseanti-ubiquitin (Santa Cruz) antibody.

Antibody-uptake assay

S2 cells transfected with Myc-tagged wild-type Smo (Myc-SmoWT) were treatedwith control dsRNA (Renilla) or Vps36 dsRNA. After cultured for 36 hours, S2cells were then incubated with mouse anti-Myc (Sigma, 1:20) antibody for30 minutes at 4 C. Then cells were washed twice with PBS at 4 C and furtherincubated in Schneider’s Drosophila Medium (Invitrogen) with 10% fetal bovineserum at 25 C for the indicated time periods. After washing with PBS at 4 C, cellswere fixed, permeabilized and stained with the secondary antibody at roomtemperature. Immunostaining images were recorded with Leica LAS SP5 confocalmicroscope.

Knockdown in Drosophila S2 cells and real-time PCR

Double-strand RNA was synthesized using the in vitro Transcription T7 Kit fromTakaRa. After cells were transfected with the indicated constructs for 12 hours, theculture medium was changed to Serum Free Medium with 20–50 mg dsRNA/106

cells for 1 hour starvation. Then fresh medium with serum was added and cellswere cultured for 36 hours. The dsRNA targeting the Renilla luciferase codingsequence was used as a control. The targeting amino acids and primer sequencesare as follows: Vps36, targeting amino acids 1–700; Vps36-RNAi-F, 59-GATCACTAATACGACTCACTATAGGGAGAAGAGGACAGGAATTGGG-39,Vps36-RNAi-R, 59-GATCACTAATACGACTCACTATAGGGTTAGTCCCGG-CCTAGCAGAA-39; Renilla luciferase, targeting amino acids 1–407; Renillaluciferase RNAi-F, 59-GATCACTAATACGACTCACTATAGGGATGACTTC-GAAAGTTTATGATCCAG-39, Renilla luciferase RNAi-R, 59-GATCACTAAT-ACGACTCACTATAGGGTTATCTTGATGCTCATAGCTATAATG-39. Vps36RNAi efficiency was tested by real-time PCR with RPL32 as an internalcontrol. Primer sequences are as follows: Vps36-realtime-F, 59-TTCTGCCT-ACTTCAGTAGCC-39, Vps36-realtime-R, 59-GTTTACACGACAGTAGACAT-CAG-39; RPL32-realtime-F, 59-CTAAGCTGTCGCACAAATGG-39, RPL32-realtime-R, 59-AGGAACTTCTTGAATCCGGTG-39.

AcknowledgementsWe apologize to colleagues whose work is not cited owing to spacelimitation. We thank Drs Daniel St Johnston, Jianhang Jia, DSHB,VDRC, NIG and Bloomington Stock Center for reagents and flystocks.

Author contributionsX.Y. and F.M. designed and performed all experiments, andinterpreted data; X.L. performed some experiments andparticipated in experiment design and data analysis; Z.Z. and L.F.were involved in experiment design and interpreted data; Y.L. andW.W. contributed to making fly stocks and reagents; Z.Z.contributed to experiment design; Y.Z. and L.Z. directed theexperiment; X.Y. and Y.Z. wrote the manuscript.

Vps36 regulates Smo trafficking 4237

Journ

alof

Cell

Scie

nce

FundingThis work was supported by grants from the National Basic ResearchProgram of China [973 Program: grant number 2011CB943902,2010CB912101, 2012CB945001]; the National Natural ScienceFoundation of China [grant numbers 31171414 and 31171394];and the ‘Strategic Priority Research Program’ of the ChineseAcademy of Sciences [grant numbers XDA01010405 andXDA01010406].

Supplementary material available online at

http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.128603/-/DC1

ReferenceAlam, S. L., Langelier, C., Whitby, F. G., Koirala, S., Robinson, H., Hill, C. P. and

Sundquist, W. I. (2006). Structural basis for ubiquitin recognition by the humanESCRT-II EAP45 GLUE domain. Nat. Struct. Mol. Biol. 13, 1029-1030.

Babst, M., Katzmann, D. J., Estepa-Sabal, E. J., Meerloo, T. and Emr, S. D. (2002a).Escrt-III: an endosome-associated heterooligomeric protein complex required for mvbsorting. Dev. Cell 3, 271-282.

Babst, M., Katzmann, D. J., Snyder, W. B., Wendland, B. and Emr, S. D. (2002b).Endosome-associated complex, ESCRT-II, recruits transport machinery for proteinsorting at the multivesicular body. Dev. Cell 3, 283-289.

Bache, K. G., Stuffers, S., Malerød, L., Slagsvold, T., Raiborg, C., Lechardeur, D.,

Walchli, S., Lukacs, G. L., Brech, A. and Stenmark, H. (2006). The ESCRT-IIIsubunit hVps24 is required for degradation but not silencing of the epidermal growthfactor receptor. Mol. Biol. Cell 17, 2513-2523.

Cadwell, K. and Coscoy, L. (2005). Ubiquitination on nonlysine residues by a viral E3ubiquitin ligase. Science 309, 127-130.

Carroll, C. E., Marada, S., Stewart, D. P., Ouyang, J. X. and Ogden, S. K. (2012).The extracellular loops of Smoothened play a regulatory role in control of Hedgehogpathway activation. Development 139, 612-621.

DiAntonio, A., Haghighi, A. P., Portman, S. L., Lee, J. D., Amaranto, A. M. andGoodman, C. S. (2001). Ubiquitination-dependent mechanisms regulate synapticgrowth and function. Nature 412, 449-452.

Haglund, K. and Dikic, I. (2005). Ubiquitylation and cell signaling. EMBO J. 24, 3353-3359.

Haglund, K. and Dikic, I. (2012). The role of ubiquitylation in receptor endocytosis andendosomal sorting. J. Cell Sci. 125, 265-275.

Hayden, M. S. and Ghosh, S. (2008). Shared principles in NF-kappaB signaling. Cell

132, 344-362.Henne, W. M., Buchkovich, N. J. and Emr, S. D. (2011). The ESCRT pathway. Dev.

Cell 21, 77-91.Hierro, A., Sun, J., Rusnak, A. S., Kim, J., Prag, G., Emr, S. D. and Hurley, J. H.

(2004). Structure of the ESCRT-II endosomal trafficking complex. Nature 431, 221-225.

Hirano, S., Suzuki, N., Slagsvold, T., Kawasaki, M., Trambaiolo, D., Kato, R.,

Stenmark, H. and Wakatsuki, S. (2006). Structural basis of ubiquitin recognition bymammalian Eap45 GLUE domain. Nat. Struct. Mol. Biol. 13, 1031-1032.

Hooper, J. E. and Scott, M. P. (1989). The Drosophila patched gene encodes a putativemembrane protein required for segmental patterning. Cell 59, 751-765.

Hooper, J. E. and Scott, M. P. (2005). Communicating with Hedgehogs. Nat. Rev. Mol.

Cell Biol. 6, 306-317.

Ingham, P. W., Nakano, Y. and Seger, C. (2011). Mechanisms and functions ofHedgehog signalling across the metazoa. Nat. Rev. Genet. 12, 393-406.

Irion, U. and St Johnston, D. (2007). bicoid RNA localization requires specific bindingof an endosomal sorting complex. Nature 445, 554-558.

Jekely, G. and Rørth, P. (2003). Hrs mediates downregulation of multiple signallingreceptors in Drosophila. EMBO Rep. 4, 1163-1168.

Jia, J. and Jiang, J. (2006). Decoding the Hedgehog signal in animal development.Cell. Mol. Life Sci. 63, 1249-1265.

Jia, J., Tong, C., Wang, B., Luo, L. and Jiang, J. (2004). Hedgehog signalling activityof Smoothened requires phosphorylation by protein kinase A and casein kinase I.Nature 432, 1045-1050.

Jiang, J. and Hui, C. C. (2008). Hedgehog signaling in development and cancer. Dev.

Cell 15, 801-812.

Katzmann, D. J., Babst, M. and Emr, S. D. (2001). Ubiquitin-dependent sorting intothe multivesicular body pathway requires the function of a conserved endosomalprotein sorting complex, ESCRT-I. Cell 106, 145-155.

Katzmann, D. J., Odorizzi, G. and Emr, S. D. (2002). Receptor downregulation andmultivesicular-body sorting. Nat. Rev. Mol. Cell Biol. 3, 893-905.

Li, S., Chen, Y., Shi, Q., Yue, T., Wang, B. and Jiang, J. (2012). Hedgehog-regulatedubiquitination controls smoothened trafficking and cell surface expression inDrosophila. PLoS Biol. 10, e1001239.

Lloyd, T. E., Atkinson, R., Wu, M. N., Zhou, Y., Pennetta, G. and Bellen, H. J.

(2002). Hrs regulates endosome membrane invagination and tyrosine kinase receptorsignaling in Drosophila. Cell 108, 261-269.

Marchese, A., Paing, M. M., Temple, B. R. and Trejo, J. (2008). G protein-coupledreceptor sorting to endosomes and lysosomes. Annu. Rev. Pharmacol. Toxicol. 48,601-629.

Piddini, E., Marshall, F., Dubois, L., Hirst, E. and Vincent, J. P. (2005). Arrow(LRP6) and Frizzled2 cooperate to degrade Wingless in Drosophila imaginal discs.Development 132, 5479-5489.

Raiborg, C. and Stenmark, H. (2009). The ESCRT machinery in endosomal sorting ofubiquitylated membrane proteins. Nature 458, 445-452.

Ranieri, N., Ruel, L., Gallet, A., Raisin, S. and Therond, P. P. (2012). Distinctphosphorylations on kinesin costal-2 mediate differential hedgehog signalingstrength. Dev. Cell 22, 279-294.

Rusten, T. E., Vaccari, T. and Stenmark, H. (2012). Shaping development withESCRTs. Nat. Cell Biol. 14, 38-45.

Slagsvold, T., Aasland, R., Hirano, S., Bache, K. G., Raiborg, C., Trambaiolo, D.,Wakatsuki, S. and Stenmark, H. (2005). Eap45 in mammalian ESCRT-II bindsubiquitin via a phosphoinositide-interacting GLUE domain. J. Biol. Chem. 280,19600-19606.

Smelkinson, M. G. and Kalderon, D. (2006). Processing of the Drosophila hedgehogsignaling effector Ci-155 to the repressor Ci-75 is mediated by direct binding to theSCF component Slimb. Curr. Biol. 16, 110-116.

Teo, H., Perisic, O., Gonzalez, B. and Williams, R. L. (2004). ESCRT-II, anendosome-associated complex required for protein sorting: crystal structure andinteractions with ESCRT-III and membranes. Dev. Cell 7, 559-569.

Teo, H., Gill, D. J., Sun, J., Perisic, O., Veprintsev, D. B., Vallis, Y., Emr, S. D. andWilliams, R. L. (2006). ESCRT-I core and ESCRT-II GLUE domain structuresreveal role for GLUE in linking to ESCRT-I and membranes. Cell 125, 99-111.

Vaccari, T. and Bilder, D. (2005). The Drosophila tumor suppressor vps25 preventsnonautonomous overproliferation by regulating notch trafficking. Dev. Cell 9, 687-698.

Wang, X., Herr, R. A., Chua, W. J., Lybarger, L., Wiertz, E. J. and Hansen, T. H.

(2007). Ubiquitination of serine, threonine, or lysine residues on the cytoplasmic tailcan induce ERAD of MHC-I by viral E3 ligase mK3. J. Cell Biol. 177, 613-624.

Wang, H., Mu, Y. and Chen, D. (2011). Effective gene silencing in Drosophila ovariangermline by artificial microRNAs. Cell Res. 21, 700-703.

Wilson, C. W. and Chuang, P. T. (2010). Mechanism and evolution of cytosolicHedgehog signal transduction. Development 137, 2079-2094.

Wollert, T. and Hurley, J. H. (2010). Molecular mechanism of multivesicular bodybiogenesis by ESCRT complexes. Nature 464, 864-869.

Xia, R., Jia, H., Fan, J., Liu, Y. and Jia, J. (2012). USP8 promotes smoothenedsignaling by preventing its ubiquitination and changing its subcellular localization.PLoS Biol. 10, e1001238.

Yan, D., Wu, Y., Yang, Y., Belenkaya, T. Y., Tang, X. and Lin, X. (2010). The cell-surface proteins Dally-like and Ihog differentially regulate Hedgehog signalingstrength and range during development. Development 137, 2033-2044.

Yao, S., Lum, L. and Beachy, P. (2006). The ihog cell-surface proteins bind Hedgehogand mediate pathway activation. Cell 125, 343-357.

Zhang, W., Zhao, Y., Tong, C., Wang, G., Wang, B., Jia, J. and Jiang, J. (2005).Hedgehog-regulated Costal2-kinase complexes control phosphorylation and proteolyticprocessing of Cubitus interruptus. Dev. Cell 8, 267-278.

Zhang, Y. Y., Mao, F. F., Lu, Y., Wu, W. Q., Zhang, L. and Zhao, Y. (2011).Transduction of the Hedgehog signal through the dimerization of Fused and thenuclear translocation of Cubitus interruptus. Cell Res. 21, 1436-1451.

Zhao, Y., Tong, C. and Jiang, J. (2007). Hedgehog regulates smoothened activity byinducing a conformational switch. Nature 450, 252-258.

Zheng, X., Mann, R. K., Sever, N. and Beachy, P. A. (2010). Genetic and biochemicaldefinition of the Hedgehog receptor. Genes Dev. 24, 57-71.

Zhou, Q. and Kalderon, D. (2011). Hedgehog activates fused through phosphorylationto elicit a full spectrum of pathway responses. Dev. Cell 20, 802-814.

Journal of Cell Science 126 (18)4238