Autophagy Negatively Regulates Wnt Signalling by Promoting Dishevelled Degradation

ARTICLES

Autophagy negatively regulates Wnt signalling by promoting Dishevelled degradation

Chan Gao1,2,6, Weipeng Cao1,2,6, Lan Bao3, Wei Zuo1,2, Guoming Xie1,2, Tiantian Cai2, Wei Fu4, Jian Zhang5, Wei Wu 2, Xu Zhang3 and Ye-Guang Chen1,2,7

In eukaryotic cells, autophagy is a highly conserved self-digestion process to promote cell survival in response to nutrient starvation and other metabolic stresses. Autophagy is regulated by cell signalling such as the mTOR (mammalian target of rapamycin) pathway. However, the significance of autophagy in modulation of signal transduction is unclear. Here we show that autophagy negatively regulates Wnt signalling by promoting Dishevelled (Dvl) degradation. Von Hippel–Lindau protein- mediated ubiquitylation is critical for the binding of Dvl2 to p62, which in turn facilitates the aggregation and the LC3- mediated autophagosome recruitment of Dvl2 under starvation; the ubiquitylated Dvl2 aggregates are ultimately degraded through the autophagy–lysosome pathway. Moreover, a reverse correlation between Dvl expression and autophagy is observed in late stages of colon cancer development, indicating that autophagy may contribute to the aberrant activation of Wnt signalling in tumour formation.

Autophagy is used by eukaryotic cells to self-digest their long-lived Results proteins and dysfunctional organelles and to provide nutrients in Autophagy attenuates Wnt signalling response to cellular metabolic stress1–4. Deregulation of autophagy To investigate whether induction of autophagy has any effect on Wnt has been associated with a variety of human diseases, including signalling, we examined the effect of two autophagy stimuli, the mTOR cancer5–7. During autophagy, double-membrane vesicles, called inhibitor rapamycin and nutrient deprivation medium, on the expres- autophagosomes, are formed to deliver cytoplasmic materials to sion of β-catenin-responsive reporters TopFlash and LEF-luciferase. lysosomes, where the encompassed cargos are ultimately degraded7,8. Both rapamycin and nutrient starvation decreased the Wnt3a-induced Wnt signalling has key functions in development, tissue self-renewal reporter expression (Fig. 1a; Supplementary Information, Fig. S1a, b). and tumorigenesis9–11. Binding of Wnt ligands to their cell-surface Furthermore, inhibition of autophagy with the class III phosphoi- receptors activates several signalling pathways. In the canonical nositide 3-kinase inhibitor 3-methyladenine (3-MA) reversed the pathway, Wnt-initiated binding of Dvl to Frizzled and of axin to inhibitory effect of nutrient starvation (Fig. 1a; Supplementary LRP5/6 results in the disassembly of the β-catenin destruction com- Information, Fig. S1c), suggesting that autophagy negatively regu- plex and consequently leads to the accumulation of β-catenin in the lates Wnt signalling. The inhibitory effect of autophagy on Wnt sig- nucleus. Together with the transcription factors of the lymphoid nalling was confirmed by rapamycin-mediated downregulation of the enhancer-binding factor/T-cell factor family, β-catenin regulates the messenger RNA levels of axin2, c-myc and cyclin D1, well-character- transcription of Wnt target genes. Dvl also has key functions in the ized targets of Wnt/β-catenin signalling17, in the presence of Wnt3a non-canonical Rho/c-Jun N-terminal kinase (JNK) planar cell polar- (Fig. 1b). Rapamycin also attenuated the Wnt5a-stimulated expres- ity and Ca2+-dependent Wnt pathways. However, the mechanism gov- sion of the JNK-responsive reporter AP-1–luciferase (Supplementary erning the activity and stability of Dvl is not fully understood. It has Information, Fig. S1d), suggesting that autophagy can interfere with been suggested that phosphorylation by casein kinases can potenti- non-canonical Wnt signalling. Knockdown of either LC3, LC3-like ate Dvl activity12 and that the stability of Dvl proteins is influenced protein GABARAP (γ-aminobutyric acid receptor-associated protein ) by ubiquitylation-dependent proteasomal degradation13–16. Here we or Beclin 1, essential mediators of autophagy, elevated Wnt3a-induced report that autophagy attenuates Wnt signalling by enhancing Dvl2 TopFlash activity (Fig. 1c; Supplementary Information, Fig. S1e–h). degradation. However, knockdown of another LC3-like protein, GATE-16, had

1The State Key Laboratory of Biomembrane and Membrane Biotechnology, Tsinghua University, Beijing 100084, China. 2School of Life Sciences, Tsinghua University, Beijing 100084, China. 3Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai 200031, China. 4Department of General Surgery, Peking University Third Hospital, Beijing 100191, China. 5Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing 100101, China. 6These authors contributed equally to this work 7Correspondence should be addressed to Y.-G.C. (e-mail: [email protected])

Received 24 May 2010; accepted 5 July 2010; published online 18 July 2010; corrected online 21 July 2010; DOI:10.1038/ncb2082

ARTICLES

a b 50 DMSO ** Control Rap **** Control + Rap Starv. 3.0 Wnt3a CM 40 Starv. + 3-MA Wnt3a CM + Rap 2.5 * ** 30 2.0 **

1.5 20 1.0 Relative mRNA level

TopFlash activity (fold) TopFlash 10 0.5

0 0 Vector Wnt3a axin2 c-myc cyclin D1

cd30 Control Wnt3a CM 12 Atg5+/+ 25 * Atg5–/– * * 10 20 8 15 6

10 4 TopFlash activity (fold) TopFlash TopFlash activity (fold) TopFlash 5 2

0 0 shRNA Control Beclin 1 LC3 Control Wnt3a CM

Figure 1 Autophagy attenuates Wnt signalling. (a) Rapamycin (Rap) treatment asterisks, P < 0.01). Wnt3a CM, Wnt3a conditioned medium. (c) Knockdown of or nutrient deprivation attenuated the expression of the Wnt reporter LC3 or Beclin 1 potentiated Wnt signalling. (n = 3; asterisk, P < 0.05). (d) Atg5 TopFlash in HEK-293T cells. (n = 3; two asterisks, P < 0.01). DMSO, knockout enhanced Wnt signalling. TopFlash reporter activity induced by Wnt3a dimethylsulphoxide. (b) Expression analysis of Wnt target genes by quantitative CM in Atg5–/– or wild-type MEFs. Results are means ± s.d. after normalization to RT–PCR after 4 h of treatment with rapamycin. (n = 3; asterisk, P < 0.05; two Renilla luciferase activity. (n = 3; asterisk, P < 0.05). no effect. In addition, Wnt3a-stimulated TopFlash activity was sig- (Fig. 2b). This was confirmed by reintroduction of Atg5 into Atg5–/– nificantly increased in Atg5–/– cells (Fig. 1d), which are defective in MEFs, which restored the autophagy process, speeded up Dvl2 degra- autophagosome formation18. dation and attenuated the induction of the Wnt target genes c-myc and axin2 (Fig. 2c; Supplementary Information, Fig. S3a, b). Similarly, the Autophagy induces Dvl degradation extended half-life of Dvl2 under starvation was found in Atg7–/– MEFs We then explored which mediator(s) of Wnt signalling might undergo (Fig. 2d; Supplementary Information, Fig. S3c). Atg7 has been shown downregulation on autophagy. One candidate was Dvl, which acts to be essential for autophagy20. Consistent with the previous reports at the node of divergence between the canonical and non-canonical that Dvl can be degraded through the ubiquitylation–proteasome path- Wnt pathways12,19. Indeed, Dvl2 protein levels decreased markedly way13–16, we found that the continued degradation of Dvl2 in autophagy- on autophagy induced by rapamycin or starvation (demonstrated defective Atg5–/– cells was sensitive to the proteasome inhibitor MG132 by an increase of LC3 II), which was accompanied by a concomitant (Fig. 2e). In wild-type cells the Dvl2 degradation induced by starvation decrease in active β-catenin and nuclear accumulation of β-catenin or rapamycin was sensitive to 3-MA and BFA1, but only partly sensitive (Fig. 2a; Supplementary Information, Fig. S2a). All three members of or insensitive to MG132 (Fig. 2e; Supplementary Information, Fig. S3d). the mammalian Dvl family (Dvl1, Dvl2 and Dvl3) were downregulated These data together suggested that autophagy negatively regulates Wnt on rapamycin treatment or starvation (Supplementary Information, signalling by accelerating Dvl2 turnover. However, in cells in which Fig. S2b, c). In contrast, induction of autophagy had no effect on the autophagy is impaired, Dvl2 may also undergo proteasome-mediated protein levels of LRP6 and E-cadherin. Consistent with the role of degradation. autophagy in promoting Dvl2 protein degradation, rapamycin had little effect on Dvl2 mRNA expression (Supplementary Information, Starvation induces localization of Dvl2 to autophagosomes Fig. S2d). Moreover, the decrease in Dvl proteins that was induced by To assess whether Dvl2 is engulfed in autophagosomes, HeLa cells starvation or rapamycin was rescued on inhibition of autophagy by transfected with Flag–Dvl2 were maintained in normal growth medium 3-MA or blockage of lysosomal function by bafilomycin A1 (BFA1) or under starvation for 4 h and followed by immuno-transmission elec- (Fig. 2a; Supplementary Information, Fig. S2c). The half-life of Dvl2 tron microscopy (TEM) analysis. Under normal growth conditions, protein under starvation was prolonged in Atg5–/– murine embry- Dvl2 was localized in the non-membrane electron-dense puncta or scat- onic fibroblasts (MEFs)18 in comparison with that in wild-type cells tered in the cytoplasm, whereas it accumulated in vesicle-like structures

ARTICLES

a Control Wnt3a CM Vector Wnt3a b Atg5+/+ Atg5–/– 100 +/+ Starv. (h) 0246802468 Atg5 Dvl2 80 Atg5–/– p62 60 Atg5–Atg12

DMSO DMEM DMEM Treatment ––Rap DMSO Rap Starv. + 3-MA Starv. Starv. + 3-MA Starv. 40 LC3 I Dvl2 * LC3 II (percentage) 20 Dvl2 protein level Dvl2 protein LRP6 Tubulin 0 Starv. (h) 02 46 8 E-cadherin Active β-catenin Tubulin LC3 I LC3 II

c Atg5–/– d Atg7+/+ Atg7–/– –/– Ad-GFP Ad-Atg5 Starv. (h) 024681010246 8 0 100 Atg7 Atg7+/+ Starv. (h) 02468024 68 Dvl2 80 Dvl2 Atg5–Atg12 60 p62 LC3 I 40

LC3 II (percentage) 20 LC3 I Tubulin level Dvl2 protein LC3 II 0 Starv. (h) 0 2 4 6 8 10 Tubulin

e Nutrient-rich Starv. (2 h) Nutrient-rich Starv. (4 h) Nutrient-rich Starv. (8 h)

Treatment ––MG132 BFA1 ––MG132 BFA1 –– MG132 BFA1

+/+ Dvl2

Atg5 Tubulin

–/– Dvl2

Atg5 Tubulin

Figure 2 Autophagy induces Dvl degradation. (a) Dvl2 underwent autophagy- GFP or Atg5 were starved for the indicated durations, and the cell lysates mediated degradation. E-cadherin was used as a loading control for were subjected to immunoblotting with the indicated antibodies. (d) Dvl2 membrane fraction, and tubulin as a loading control for cytosolic proteins. degradation was slowed down in Atg7–/– MEFs. MEFs were subjected to The asterisk indicates the phosphorylated form of Dvl2. (b) The half-life starvation for the indicated durations and harvested for immunoblotting of Dvl2 was prolonged in Atg5–/– cells. MEFs were subjected to starvation with the indicated antibodies. Dvl2 levels were quantified (right panel). (starv.) for the indicated durations and harvested for immunoblotting with the (e) The continued degradation of Dvl2 in Atg5–/– cells is proteasome- indicated antibodies. Dvl2 levels were quantified (right panel). Autophagy dependent. MEFs were starved for 2, 4 or 8 h with MG132 (1 μM) or BFA1 induction is indicated by the levels of LC3 II and p62 degradation. (c) (0.1 μM) for different durations, followed by anti-Dvl2 immunoblotting to Reintroduction of Atg5 in Atg5–/– MEFs restored starvation-induced Dvl2 detect endogenous Dvl2 protein levels. Tubulin served as a loading control. degradation. MEFs infected with recombinant adenovirus expressing Uncropped images of blots are shown in Supplementary Information, Fig. S8. during starvation (Fig. 3a). Consistent with this, fluorescence micro- protein GABARAP, could interact directly with Dvl2 in vitro (Fig. 3e, f), scopy examination revealed that clusters of green fluorescent protein implying that LC3 may serve as a receptor to recruit Dvl2 protein into (GFP)–LC3 dots mostly co-localized with endogenous Dvl2 puncta in autophagosomes21. The DEP (for Disheveled, EGL-10, Pleckstrin) domain starved HeLa cells, whereas GFP–LC3 was diffused over the normal of Dvl2, containing a LC3-binding motif (Trp 444-Leu-Lys-Ile 447)22,23, cells (Fig. 3b). The co-localization was confirmed by TEM immuno- was necessary and sufficient for an effective interaction with LC3 gold double labelling: Flag–Dvl2 (arrow) was sequestered in double- (Supplementary Information, Fig. S4), and mutations (W444A, I447A) membrane structures enriched in GFP–LC3 (arrowhead) (Fig. 3c). in this motif decreased the Dvl2–LC3 interaction (Fig. 3g). These data suggest that Dvl2 is engulfed into autophagosomes under starvation conditions. Aggregation of Dvl2 is critical for its ubiquitylation and binding The co-localization of Dvl2 with LC3 suggests that these two proteins to LC3 may interact physically with each other. Indeed, Dvl2 interacted with LC3 Although the isolated DIX domain did not bind to LC3 (Supplementary at the endogenous level in HeLa cells, and this interaction was signifi- Information, Fig. S4c), we found that a deletion mutation (ΔDIX) or point cantly increased under starvation conditions (Fig. 3d). Furthermore, puri- mutations (M1, F43S; M2, V67A/K68A) in the DIX domain abolished fied glutathione S-transferase (GST)–LC3 protein, as well as the LC3-like its LC3-binding activity (Fig. 4a; Supplementary Information, Fig. S4d). nature cell biology advance online publication 3 © 2010 Macmillan Publishers Limited. All rights reserved.

ARTICLES

a b GFP–LC3 Dvl Merge 100 ** 80

N 60 Nutrient-rich Nutrient-rich 40

Co-localization (percentage) 0 N + BFA1 Starvation Starv.

Nutrient-rich Starv. + BFA1 c d IP: IgG Anti-LC3 f Pulldown g GST– GST LC3 Input

+ BFA1 +

HA–Dvl2 WT MUT WT MUT WT MUT GST– – LC3 GABARAP GATE-16 p62 Flag–Dvl2 +++++ Anti-HA

Nutrient-rich Medium: Nutrient-rich Starv. Anti-Flag M (K) GST–LC3 IB: Dvl2 r Coomassie 83 stain GST 62 Dvl2 Coomassie WCL 48 IB: LC3 I stain LC3 II 33 e Input Pulldown 25 GST GST–LC3 Dvl2 GST–LC3 GST

Figure 3 Starvation induces the autophagosome translocation of Dvl2 and interaction in vivo. IB, immunoblotting; IP, immunoprecipitation; S, starvation; promotes the interaction of Dvl2 with LC3. (a) Immunoelectron microscopy WCL, whole-cell lysate; N, nutrient-rich. (e, f) Dvl2 interacts directly with analysis of HeLa cells transfected with Flag–Dvl2 under nutrient-rich or LC3 (e) and GABARAP (f) but not with GATE-16 in vitro. GST pulldown assay starvation conditions. The arrowheads indicate Flag–Dvl2. Scale bars, 1 μm using in vitro-translated Flag–Dvl2 and immobilized GST or the indicated GST (left); 250 nm (right). (b) Localization of endogenous Dvl proteins (red) and fusions. Precipitated proteins were detected with anti-Flag antibody. Coomassie GFP–LC3 (green) in GFP–LC3 stable HeLa cells in nutrient-rich medium or staining of GST and GST–LC3 protein is shown. (g) The LC-3-binding motif under starvation (starv.) conditions with BFA1 for 4 h. The right panel shows is important for the Dvl2–LC3 interaction. GST–LC3-immobilized beads were the percentage co-localization between Dvls and GFP–LC3-positive dots. Scale incubated with the lysates from the cells expressing wild-type (WT) Dvl2 or bar, 5 μm. (c) Immunoelectron microscopy analysis of HeLa cells transfected the W444A, I447A mutant (MUT); the pulldowns were then subjected to anti- with Flag–Dvl2 and GFP–LC3 under starvation conditions. Arrows indicate haemagglutinin (anti-HA) immunoblotting. Uncropped images of blots are Flag–Dvl2; arrowheads indicate GFP–LC3. Scale bar, 250 nm. (d) Dvl2–LC3 shown in Supplementary Information, Fig. S8.

The DIX domain has been shown to mediate Dvl self-oligomerization24; membrane to target aggregates to autophagosomes for degradation26–28. all of the DIX mutants showed a diffused localization (Supplementary Supporting the possible role of p62 in autophagy-mediated degrada- Information, Fig. S5a), as shown previously24,25. These data implied that tion of Dvl2, starvation promoted Dvl2 to form big puncta, which Dvl2 aggregation is required for the interaction with LC3. A growing were co-localized with dots of p62 that otherwise were more diffuse body of evidence revealed that protein aggregates tagged with ubiquitin in the cytoplasm in nutrient-rich condition (Fig. 4d). Ubiquitin was are selectively targeted for autophagic degradation7,26. We then assessed the also detected in the double-stained dots of Dvl2 and p62, implicating possible involvement of ubiquitylation in autophagy-mediated Dvl2 deg- that starvation promotes p62 binding to ubiquitylated Dvl2 aggre- radation. Ubiquitylation was detected with full-length Dvl2 as well as in gates. Consistently, sucrose gradient ultracentrifugation also revealed ΔDEP and ΔPDZ mutants but was greatly attenuated by mutation or dele- that p62 and Dvl2 were distributed in the same fractions, and starvation of the DIX domain (Fig. 4b; Supplementary Information, Fig. S5b). In tion induced Dvl2 and p62 to form high-molecular-mass aggregates addition, these aggregation-deficient mutants showed increased turnover (Fig. 4e). In the p62 knockdown cells, Dvl2 became more diffuse or kinetics in comparison with wild-type Dvl2 under autophagy induction localized in small puncta (Fig. 4f, cells outlined with a dotted line). (Fig. 4c; Supplementary Information, Fig. S5c, d). Taken together, these Expression of the p62 I431A mutant, which has folding defects in the data indicate that ubiquitylation-associated aggregation of Dvl2 facilitates UBA (ubiquitin-associated) domain27, caused a change in distribu- its binding to LC3 and its autophagy-mediated degradation. tion of Dvl2 from puncta to diffuse (Supplementary Information, Fig. S6a). p62 also promoted Dvl2 ubiquitylation, and p62 knockdown p62 promotes Dvl2–LC3 interaction seemed to have little effect on levels of ubiquitylated Dvl2 protein but Recent studies revealed that p62/SQSTM1 recognizes polyubiquit- decreased the amount of ubiquitylated Dvl2 at high molecular masses ylated protein aggregates and binds to Atg8/LC3 on the autophagosome (Supplementary Information, Fig. S6b). This may be because ubiquit-

ARTICLES

abHA–Dvl2 – WT M1 M2 c HA–Dvl2 WT M1 M2 – WT M1 M2 His–Ub + +++ Starv. (h) 1240 01124 0 24 HA–Dvl2 Mr(K) +++ + Myc–LC3 Anti-HA IP: Anti-Myc IB: Anti-HA Anti-tubulin - 175 Anti-HA 100 WCL Ni-NTA WT IB IB: Anti-HA 80 M1 Anti-Myc 60 M2 40 Protein level Protein - 83 (percentage) 20 0 Starv. (h) 0 1 23 4

e Bottom Top d Ubiquitin p62 Dvl Merge Fraction: 16234 5178 0119

Flag–Dvl2

p62 Nutrient-rich Nutrient-rich

Flag–Dvl2

p62 Starv. + BFA1 Starv. Starv. + BFA1 Starv.

f Control shRNA p62 shRNA Nutrient-rich Starv. + BFA1 Nutrient-rich Starv. + BFA1 p62 Dvl Merge

Figure 4 Aggregation of Dvl2 is critical for its LC3 binding, ubiquitylation kinetics of wild-type Dvl2 and its variants under starvation conditions. Lower and concomitant autophagic degradation. (a) Interaction between LC3 and panel: quantification of the immunoblotting results. (d) Starvation promotes wild-type Dvl2 or its point mutants. (b) Ubiquitylation of wild-type Dvl2 and co-localization of Dvl, ubiquitin and p62. Scale bar, 10 μm. (e) Starvation its variants. Transfected HEK-293T cells were treated with BFA1 (0.2 μM) for induces the formation of high-molecular-mass Dvl2 aggregates. (f) p62 6 h. Ubiquitylated proteins were precipitated with Ni-nitrilotriacetate (Ni-NTA) knockdown alters Dvl localization. Scale bar, 10 μm. Uncropped images of beads, followed by anti-HA immunoblotting. Ub, ubiquitin. (c) Degradation blots are shown in Supplementary Information, Fig. S8. ylation and aggregation facilitate each other. These findings indicated Von Hippel–Lindau protein (pVHL) promotes Dvl2 ubiquitylation that p62 is required for the aggregation of Dvl2 from small to large Several ubiquitin ligases have been suggested to promote Dvl ubiquityla- punctae, consistent with the previously reported role of p62 in promot- tion13–15. pVHL, a component of an SCF (Skp1–Cdc53–F-box)-like ubiq- ing protein aggregation29. uitin E3 ligase complex30,31, has been shown to downregulate β-catenin Although Dvl2 can interact directly with LC3 in vitro, endog- through Jade-1 and to inhibit Wnt signalling32. We found that expression enous interaction between LC3 and Dvl2 might be facilitated by of pVHL efficiently enhanced the level of Dvl2 ubiquitylation (Fig. 6a). p62, because p62 knockdown apparently decreased this interaction Consistently, the pVHL-containing ubiquitin E3 ligase complex could (Fig. 5a). Co-localization of p62 and Dvl2 with GFP–LC3 in the cells ubiquitylate Dvl2 in vitro (Fig. 6b), suggesting that Dvl2 is a direct sub- stably expressing GFP–LC3 was observed under starvation conditions strate of this E3 ligase complex. Starvation enhanced Dvl2 ubiquityla- (Fig. 5b). Furthermore, p62 knockdown slowed starvation-induced Dvl2 tion, and knockdown of pVHL expression resulted in a decrease in Dvl2 degradation (Fig. 5c). These data suggested that p62 functions as an ubiquitylation (Fig. 6c; Supplementary Information, Fig. S7d). It is worth autophagic receptor in cooperation with LC3 to mediate recruitment to noting that starvation enhanced the binding of Dvl2 to Cullin 2 (Cul2), autophagosomes and degradation of Dvl2. which is an essential component of the pVHL–ubiquitin–E3 complex33 nature cell biology advance online publication 5 © 2010 Macmillan Publishers Limited. All rights reserved.

ARTICLES

a Nutrient-rich Starv. + BFA1 IgGIP: Anti-LC3 IP: IgG Anti-LC3

shRNA Control Control p62 shRNA Control Control p62

IB: Dvl2 IB: Dvl2

LC3 LC3

Dvl2 Dvl2

WCL p62 WCL p62 IB: IB:

LC3 I LC3 I LC3 II LC3 II

b GFP–LC3 p62 Dvl Merge c siRNA Control p62 Starv. (h) 0240 24

Dvl2

Nutrient-rich p62

tubulin Starvation

Figure 5 p62 promotes Dvl2–LC3 interaction and Dvl2 degradation under (red) antibodies. Scale bar, 10 μm. (c) p62 knockdown slows the degradation starvation conditions. (a) Knockdown of p62 expression attenuates the LC3– of Dvl2 on starvation. HEK-293T cells were transfected with p62 siRNA. At Dvl2 interaction. shRNA, short hairpin RNA. (b) Co-localization of GFP–LC3, 48 h after transfection, cells were subjected to starvation for various times. The p62 and Dvl in HeLa GFP–LC3 stable cells on starvation. Proteins visualized cell lysates were subjected to immunoblotting with the antibodies indicated. by immunofluorescence with anti-GFP (green), anti-p62 (blue) and anti-Dvl Uncropped images of blots are shown in Supplementary Information, Fig. S8.

(Fig. 6d), although the interaction between Dvl2 and pVHL was not the ubiquitylation and autophagy-mediated degradation of Dvl2, con- affected (Supplementary Information, Fig. S7c). In agreement with tributing to the negative regulation of Wnt signalling. this, Dvl2 could interact with pVHL through its DEP or PDZ domains (Supplementary Information, Fig. S7a, b), but the DEP domain seems to Dvl protein levels are negatively correlated with autophagy in mediate direct interaction (Supplementary Information, Fig. S4b). human colon cancer Consistent with the co-localization of p62 with Dvl2 (Fig. 4d), starva- Although the precise mechanism is still unclear and its role may be tion elevated the interaction between p62 and Dvl2, and this interaction context-dependent, autophagy malfunction has been associated with was eliminated by pVHL knockdown or by the dominant-negative ubiq- tumorigenesis5,34,35. Genetic alterations of autophagy-related genes have uitin mutant UbKOGA (all seven lysines mutated to arginines, plus G76A) been found in colon cancer36,37. To investigate further the autophagic (Fig. 6e). An in vitro binding assay confirmed that the direct interac- regulation of Wnt signalling, we generated a stable Beclin 1-knockdown tion of p62 with Dvl2 required Dvl2 ubiquitylation (Fig. 6f). pVHL SW480 colon cancer cell line. The number of colonies of these cells knockdown and UbKOGA expression also led to a decreased interaction formed in soft agar was significantly greater than that formed by control between Dvl2 and LC3 (Supplementary Information, Fig. S7e). These cells (Fig. 7a, b). Further, the Wnt inhibitor Dkk38,39 decreased the colony data suggested that pVHL-mediated ubiquitylation is critical for Dvl2 formation potential in Beclin 1-knockdown cells. Examination of 126 to interact with p62 and LC3 under starvation conditions. In agree- colon cancer samples revealed a positive correlation between tumour ment with the role of pVHL in starvation-promoted Dvl2 ubiquityla- progression and p62 expression (P = 0.0001; correlation coefficient 0.51) tion and interaction with p62, pVHL knockdown led to an increase in and a negative correlation between tumour progression and Beclin 1 Dvl2 protein level under starvation conditions (Fig. 6g). Furthermore, expression (P = 0.00029; correlation coefficient –0.32). These data are in knockdown of LC3 or p62 rescued pVHL-mediated Dvl2 downregula- accordance with previous observations that the accumulation of p62 is a tion (Fig. 6h), supporting the involvement of autophagy in the negative significant contributor to tumorigenesis34 and that heterozygous disrup- regulation of Dvl2 stability by pVHL. Consistently, pVHL knockdown tion of Beclin 1 promotes tumour formation40,41. In late-stage tumours increased the Wnt3a-promoted expression of TopFlash (Supplementary there was a positive correlation between Dvl and p62 (P = 0.039; correla- Information, Fig. S7f). These results demonstrated that pVHL facilitates tion coefficient 0.60) and a negative correlation between Dvl and Beclin 1

ARTICLES

abE1 ++++ + cd Myc–VHL ––+ E2 –++ ++ Flag–Dvl2 –++ Cul2 + Rbx1 + Elongin B +–+ ++ Nutrient -rich Nutrient -rich Starv. + BFA1 Starv. + BFA1 His-Ub +++ Mr(K) Elongin C +–+– + 175 pVHL +–++ – Myc-Cul2 –++ Ubiquitin ++++ + Flag-Dvl2 +++ Flag–Dvl2 ++++ + shRNA Control Control VHL Control Control VHL Anti-Flag Mr(K) IP: Anti-Myc Ni-NTA Flag–Dvl2 –++–++ IB: IB: Anti-Flag Anti-Myc His–Ub ++++++ M (K) 83 r IP: Anti-Flag IB: Anti-Myc IB: Anti-Ub 175 WCL 175 Anti-Flag Ni-NTA IB: Anti-Flag 83 62 83

IP: Anti-Flag IB: Anti-Flag

e Nutrient-rich Starv. + BFA1 f g h Ub + E1 + E2 ++ IP: IgG Anti-Dvl2 IgG Anti-Dvl2 VHL E3 complex –+ p62 p62

siRNA Control LC3 Control LC3 Flag–Dvl2 ++ Nutrient -rich Medium: Starv. –––+++ Mr(K) Myc–VHL VHL shRNA –+–+ GA GA Dvl2 Dvl2 p62

VHL shRNA VHL shRNA

UbKO UbKO Control Control Control Control -175 IP: Anti-Flag pVHL p62 IB: Anti-Ub LC3 I IB: Tubulin LC3 II Dvl2 - 83 Myc–VHL Dvl2 IP: Anti-Flag IB: Anti-p62 p62 WCL p62 Tubulin IB: * IgG pVHL IP: Anti-Flag IB: Anti-Flag

Figure 6 pVHL promotes Dvl2 ubiquitylation and degradation by means of or starved with BFA1 (0.1 μM) for 4 h. Cells were harvested for anti-Myc autophagy. (a) Ubiquitylation of Dvl2 enhanced by ectopically expressed pVHL. immunoprecipitation followed by anti-Flag immunoblotting. Total protein (b) In vitro ubiquitylation of Dvl2 by the pVHL E3 ligase complex including expression was confirmed by immunoblotting with whole-cell lysates. (e) pVHL, Elongin B, Elongin C, Rbx1 and Cul2. In vitro-translated Flag–Dvl2 was pVHL knockdown or UbKOGA blocks Dvl2 binding with p62 under starvation incubated with or without purified E1, E2 and E3 ligase complex, Elongin C conditions. (f) Dvl2 ubiquitylation is required for its direct interaction with p62. and pVHL. The reaction was subjected to anti-Flag immunoprecipitation. Dvl2 was ubiquitylated by the pVHL E3 ligase complex in vitro, followed by Dvl2 ubiquitylation was examined by using anti-ubiquitin immunoblotting. in vitro binding assay with recombinant p62. (g) pVHL knockdown decreases (c) Ubiquitylation of Dvl2 is decreased by pVHL knockdown. (d) Starvation Dvl2 degradation under starvation conditions. (h) Knockdown of p62 or LC3 elevates Dvl2–Cul2 interaction. After co-transfection with Myc–Cul2 and attenuates the pVHL-promoted Dvl2 degradation under starvation conditions. Flag–Dvl2 for 36 h, HEK-293T cells were cultured in nutrient-rich medium Uncropped images of blots are shown in Supplementary Information, Fig. S8.

(P = 0.045; correlation coefficient –0.33) (Fig. 7c). Immunoblotting reported to induce Dvl ubiquitylation, thereby promoting Dvl degrada- of fresh colorectal tumours revealed a significant positive correlation tion through the proteasome pathway13–16. Our data strongly suggest between p62 and Dvl2 expression (four of six pairs of samples) (Fig. 7d). that autophagy negatively regulates Wnt signalling by accelerating Dvl A higher expression of the Wnt target cyclin D1 was also observed in turnover under stress conditions such as nutrient starvation. However, the tumours. These data together indicate that impairment of autophagy in the cells in which autophagy was impaired, we found that Dvl also may contribute to aberrant Wnt activation. underwent proteasome-mediated degradation. LC3 not only serves as a marker of autophagosomes but is also impor- Discussion tant in autophagosome formation44,45. We found that LC3 can interact Autophagy has an essential function in the bulk clearance of long-lived directly with Dvl2. A tetrapeptide sequence, WXXL/I, known as an LC3- proteins and cytoplasmic organelles1,3,42. Although signalling cascades binding motif, has been found in several LC3-interacting proteins such as controlling autophagy have begun to emerge43, it is largely unclear p62 and is involved in their direct association with LC3 (ref. 23). Our data whether autophagy can selectively regulate the turnover of signalling also revealed that the sequence WLKI in the DEP domain of Dvl2 is critical molecules. Here we provide evidence that autophagy negatively regu- for interaction with LC3. It is therefore possible that LC3 functions as a lates Wnt signalling by promoting Dvl degradation during metabolic cargo receptor to recruit Dvl2 proteins selectively into autophagosomes. stress. We found that autophagy specifically promotes the degradation Furthermore, we showed that the LC3 homologue GABARAP, but not of all three Dvl isoforms (Dvl1, Dvl2 and Dvl3) but not the Wnt co- GATE-16, binds Dvl2 in vitro, implying distinct roles of Atg8 orthologues receptor LRP6. In support of this, the Dvl turnover rate is much slower in in selective autophagy. As discussed below, Dvl2 can also associate indi- autophagy-deficient cells, Atg5–/– and Atg7–/– MEFs. Proteins have been rectly with LC3 through p62. nature cell biology advance online publication 7 © 2010 Macmillan Publishers Limited. All rights reserved.

ARTICLES

b a Treatment – Dkk CM 1,600 ** shRNA

1354 1,200 1080 800 851 823

400 Number of clones Control Beclin I Control DKK 0 shRNA Control Beclin I

c Beclin I Dvl p62 Dvl d 1 23456 NTNTNTNTNTNT

Dvl2

p62

Cyclin D1

GAPDH Late stages Early stages

e LRP6/ Wnt Nutrient deprivation LRP5 Fz

Phagophore

Autophagosome Dvl

p62

LC3 Ubiquitin Lysosome Autolysosome pVHL

Lysosomal hydrolase

Figure 7 Negative correlation between Dvl protein levels and autophagy from colon cancer tissues by immunohistochemistry staining. Negative in human colon cancer. (a, b) Depletion of Beclin 1 by shRNA promotes correlation between Beclin1 and Dvl was observed in late stages of growth in soft agar. SW480 cells stably expressing Beclin 1 shRNA or tumours, whereas positive correlation between p62 and Dvl in late stages control shRNA were seeded in soft agar and treated with or without Dkk of tumours. Scale bar, 25 μm. (d) Immunoblotting of human colon (a); the colonies were then scored after 10 days (b). Numbers in bars tumour samples versus normal tissues using indicated antibodies. N, indicate the numbers of colonies. Results are means ± s.d. (n = 4). normal tissue; T, tumour. (e) Working model of autophagic degradation of (Asterisk, P < 0.05). Scale bar, 200 μm. (c) Representative images Dvl2.

Dvl undergoes dynamic aggregation in the cytoplasm25. We found We identified the pVHL complex as an E3 ubiquitin ligase to ubiq- that Dvl2 aggregation affects its LC3 binding and ubiquitylation, and uitylate Dvl2 both in vivo and in vitro and to promote the autophagic aggregation-deficient Dvl2 mutants cannot be ubiquitylated and are more degradation of Dvl2 under starvation conditions. This finding extends stable than the wild type under starvation conditions. Recent studies the role of the pVHL complex as an E3 ubiquitin ligase for Dvl2 protein. demonstrated that polymerization contributes to the signalling activity So far, the proteins identified as pVHL targets include hypoxia-inducible of Dvl24,46. It is possible that un-ubiquitylated Dvl2 polymers participate transcription factor 1α48,49, a subfamily of deubiquitylating enzymes50,51, in Wnt signalling, whereas the ubiquitylated Dvl2 aggregates were pref- activated atypical protein kinase C52, and the large subunit (Rpb1) of erentially targeted to the autophagy pathway under conditions of nutrient RNA polymerase II53. The most common mechanism for ubiquitylation starvation. These results are consistent with our early report that Dvl can mediated by the pVHL complex relies on the modulation of the inter- be degraded by means of a lysosome inhibitor-sensitive pathway47. action between pVHL and its targets54. Although we found that pVHL

ARTICLES can interact directly with Dvl2 through the DEP domain, pVHL–Dvl2 6. Mathew, R., Karantza-Wadsworth, V. & White, E. Role of autophagy in cancer. Nat. Rev. Cancer 7, 961–967 (2007). binding is not affected by starvation. Instead, starvation promotes the 7. Mizushima, N., Levine, B., Cuervo, A. M. & Klionsky, D. J. Autophagy fights disease interaction of Dvl2 with the Cul2 subunit of the pVHL complex. The through cellular self-digestion. Nature 451, 1069–1075 (2008). 8. Xie, Z. & Klionsky, D. J. Autophagosome formation: core machinery and adaptations. pVHL-mediated ubiquitylation of Dvl2 is essential for the interaction of Nat. Cell Biol. 9, 1102–1109 (2007). Dvl2 with p62, a protein that can bind to ubiquitin and target polyubiq- 9. Moon, R. T., Kohn, A. D., Ferrari, G. V. D. & Kaykas, A. WNT and β-catenin signalling: diseases and therapies. Nat. Rev. Genet. 5, 691–701 (2004). uitylated protein aggregates to autophagosomes by means of its interac- 10. Clevers, H. Wnt/β-catenin signaling in development and disease. Cell 127, 469–480 tion with LC3 on the autophagosome membrane26–28. (2006). 11. Logan, C. Y. & Nusse, R. The Wnt signaling pathway in development and disease. Annu. Autophagy has been linked to various diseases such as cancer, neuro- Rev. Cell Dev. Biol. 20, 781–810 (2004). degeneration, cardiomyopathies and pathogen infection5,7. An essential 12. Wharton, K. A. Runnin’ with the Dvl: proteins that associate with Dsh/Dvl and their significance to Wnt signal transduction. Dev. Biol. 253, 1–17 (2003). autophagy regulator, Beclin 1, is monoallelically deleted in a high per- 13. Angers, S. et al. The KLHL12–Cullin-3 ubiquitin ligase negatively regulates the Wnt–β- centage of human breast, ovarian and prostate cancers, and Beclin 1+/– catenin pathway by targeting Dishevelled for degradation. Nat. Cell Biol. 8, 348–357 40,41 (2006). mice are more prone to the development of spontaneous tumours . We 14. Chan, D. W., Chan, C. Y., Yam, J. W., Ching, Y. P. & Ng, I. O. Prickle-1 negatively regu- demonstrated that the level of autophagy was inversely correlated with lates Wnt/β-catenin pathway by promoting Dishevelled ubiquitination/degradation in liver cancer. Gastroenterology 131, 1218–1227 (2006). Dvl expression and activation of Wnt signalling in colorectal tumours, 15. Miyazaki, K. et al. NEDL1, a novel ubiquitin-protein isopeptide ligase for dishevelled-1, in which autophagy and Wnt signalling might cooperate with each other. targets mutant superoxide dismutase-1. J. Biol. Chem. 279, 11327–11335 (2004). 16. Simons, M. et al. Inversin, the gene product mutated in nephronophthisis type II, func- Because loss-of-function mutations of the VHL gene are associated with tions as a molecular switch between Wnt signaling pathways. Nat. Genet. 37, 537–543 kidney cancers30, upregulated Wnt/Dvl/β-catenin signalling might also (2005). 17. Reya, T. & Clevers, H. Wnt signalling in stem cells and cancer. Nature 434, 843–850 contribute to renal carcinogenesis related to VHL mutation. These find- (2005). ings link pVHL to oncogenic Wnt signalling. 18. Kuma, A. et al. The role of autophagy during the early neonatal starvation period. Nature 432, 1032–1036 (2004). Our findings shed light on a new function of autophagy in the regula- 19. Gao, C. & Chen, Y. G. Dishevelled: the hub of Wnt signaling. Cell Signal. 22, 717–727 tion of signal transduction. Our data suggest the working model shown (2010). 20. Komatsu, M. et al. Impairment of starvation-induced and constitutive autophagy in in Fig. 7e, in which Wnt ligand triggers Dvl recruitment to the plasma Atg7-deficient mice. J. Cell Biol. 169, 425–434 (2005). membrane by Frizzled receptor, which functions with the LRP5/6 co- 21. Yu, L., Strandberg, L. & Lenardo, M. J. The selectivity of autophagy and its role in cell death and survival. Autophagy 4, 567–573 (2008). receptor to transduce signalling. On starvation, Dvl is ubiquitylated by 22. Ichimura, Y. et al. Structural basis for sorting mechanism of p62 in selective autophagy. pVHL. The ubiquitylation provides a signature for recognition by p62, J. Biol. Chem. 283, 22847–22857 (2008). 23. Noda, N. N. et al. Structural basis of target recognition by Atg8/LC3 during selective and p62 association in turn promotes Dvl aggregation and binding to autophagy. Genes Cells 13, 1211–1218 (2008). LC3. Dvl aggregates are then selectively fetched by LC3 into autophago- 24. Schwarz-Romond, T. et al. The DIX domain of Dishevelled confers Wnt signaling by dynamic polymerization. Nat. Struct. Mol. Biol. 14, 484–492 (2007). somes through its direct interaction with Dvl or by means of p62 as a 25. Schwarz-Romond, T., Merrifield, C., Nichols, B. J. & Bienz, M. The Wnt signalling bridge, followed by their degradation in lysosomes. effector Dishevelled forms dynamic protein assemblies rather than stable associations with cytoplasmic vesicles. J. Cell Sci. 118, 5269–5277 (2005). 26. Kirkin, V., McEwan, D. G., Novak, I. & Dikic, I. A role for ubiquitin in selective Methods autophagy. Mol. Cell 34, 259–269 (2009). 27. Bjorkoy, G. et al. p62/SQSTM1 forms protein aggregates degraded by autophagy and Methods and any associated references are available in the online version has a protective effect on huntingtin-induced cell death. J. Cell Biol. 171, 603–614 of the paper at http://www.nature.com/naturecellbiology/ (2005). 28. Pankiv, S. et al. p62/SQSTM1 binds directly to Atg8/LC3 to facilitate degradation of ubiq- Note: Supplementary Information is available on the Nature Cell Biology website. uitinated protein aggregates by autophagy. J. Biol. Chem. 282, 24131–24145 (2007). 29. Komatsu, M. et al. Homeostatic levels of p62 control cytoplasmic inclusion body Acknowledgements formation in autophagy-deficient mice. Cell 131, 1149–1163 (2007). 30. Kaelin, W. G. Jr. The von Hippel–Lindau tumour suppressor protein: O sensing and We thank Terje Johansen for p62 constructs, Long Yu for human LC3 construct, 2 –/– –/– cancer. Nat. Rev. Cancer 8, 865–873 (2008). Noboru Mizushima for Atg5 MEFs, Masaaki Komatsu for Atg7 MEFs, and Li 31. Lisztwan, J., Imbert, G., Wirbelauer, C., Gstaiger, M. & Krek, W. The von Hippel–Lindau Yu, Roel Nusse, Daniel J. Klionsky, Juan Liang and Zhao Chen for suggestions. tumor suppressor protein is a component of an E3 ubiquitin-protein ligase activity. This work was supported by grants from the National Natural Science Foundation Genes Dev. 13, 1822–1833 (1999). of China (30930050 and 30921004) and the 973 Program (2006CB943401 and 32. Chitalia, V. C. et al. Jade-1 inhibits Wnt signalling by ubiquitylating β-catenin and 2010CB833706) to Y.-G.C. mediates Wnt pathway inhibition by pVHL. Nat.Cell Biol. 10, 1208–1216 (2008). 33. Pause, A. et al. The von Hippel–Lindau tumor-suppressor gene product forms a stable Author contributions complex with human CUL-2, a member of the Cdc53 family of proteins. Proc. Natl C.G.., W.C., L.B., J.Z., W.W., X.Z. and Y.-G.C. designed the experiments and Acad. Sci. USA 94, 2156–2161 (1997). 34. Mathew, R. et al. Autophagy suppresses tumorigenesis through elimination of p62. Cell analysed data. C.G., W.C., L.B., W.Z., G.X., T.C. and W.F. performed the 137, 1062–1075 (2009). experiments. C.G. and Y.-G.C. wrote the manuscript. 35. Liang, X. H. et al. Induction of autophagy and inhibition of tumorigenesis by beclin 1. Nature 402, 672–676 (1999). Competing financial interests 36. Kim, M. S. et al. Frameshift mutation of UVRAG, an autophagy-related gene, in gastric The authors declare no competing financial interests. carcinomas with microsatellite instability. Hum. Pathol. 39, 1059–1063 (2008). 37. Coppola, D. et al. Down-regulation of Bax-interacting factor-1 in colorectal adenocar- Published online at http://www.nature.com/naturecellbiology cinoma. Cancer 113, 2665–2670 (2008). Reprints and permissions information is available online at http://npg.nature.com/ 38. Glinka, A. et al. Dickkopf-1 is a member of a new family of secreted proteins and func- reprintsandpermissions/ tions in head induction. Nature 391, 357–362 (1998). 39. Fedi, P. et al. Isolation and biochemical characterization of the human Dkk-1 homo- 1. Mizushima, N. Autophagy: process and function. Genes Dev. 21, 2861–2873 (2007). logue, a novel inhibitor of mammalian Wnt signaling. J. Biol. Chem. 274, 19465– 2. Kim, J. & Klionsky, D. J. Autophagy, cytoplasm-to-vacuole targeting pathway, and pex- 19472 (1999). ophagy in yeast and mammalian cells. Annu. Rev. Biochem. 69, 303–342 (2000). 40. Qu, X. et al. Promotion of tumorigenesis by heterozygous disruption of the beclin 1 3. Baehrecke, E. H. Autophagy: dual roles in life and death? Nat. Rev. Mol. Cell Biol. 6, autophagy gene. J. Clin. Invest. 112, 1809–1820 (2003). 505–510 (2005). 41. Yue, Z., Jin, S., Yang, C., Levine, A. J. & Heintz, N. Beclin 1, an autophagy gene 4. Mehrpour, M., Esclatine, A., Beau, I. & Codogno, P. Overview of macroautophagy essential for early embryonic development, is a haploinsufficient tumor suppressor. regulation in mammalian cells. Cell Res. 20, 748–762 (2010). Proc. Natl Acad. Sci. USA 100, 15077–15082 (2003). 5. Levine, B. & Kroemer, G. Autophagy in the pathogenesis of disease. Cell 132, 27–42 42. He, C. & Klionsky, D. J. Regulation mechanisms and signaling pathways of autophagy. (2008). Annu. Rev. Genet. 43, 67–93 (2009).

ARTICLES

43. Yang, Z. & Klionsky, D. J. Mammalian autophagy: core molecular machinery and signal- 49. Kamura, T. et al. Activation of HIF1α ubiquitination by a reconstituted von Hippel–Lindau ing regulation. Curr. Opin. Cell Biol. 22, 124–131 (2010). (VHL) tumor suppressor complex. Proc. Natl Acad. Sci. USA 97, 10430–10435 (2000). 44. Kirisako, T. et al. Formation process of autophagosome is traced with Apg8/Aut7p in 50. Li, Z. et al. Ubiquitination of a novel deubiquitinating enzyme requires direct binding to yeast. J. Cell Biol. 147, 435–446 (1999). von Hippel–Lindau tumor suppressor protein. J. Biol. Chem. 277, 4656–4662 (2002). 45. Tanida, I., Ueno, T. & Kominami, E. LC3 conjugation system in mammalian autophagy. 51. Li, Z. et al. Identification of a deubiquitinating enzyme subfamily as substrates of the Int. J. Biochem. Cell Biol. 36, 2503–2518 (2004). von Hippel–Lindau tumor suppressor. Biochem. Biophys. Res. Commun. 294, 700–709 46. Bilic, J. et al. Wnt induces LRP6 signalosomes and promotes dishevelled-dependent (2002). LRP6 phosphorylation. Science 316, 1619–1622 (2007). 52. Okuda, H. et al. The von Hippel–Lindau tumor suppressor protein mediates ubiquitination 47. Zhang, L., Gao, X., Wen, J., Ning, Y. & Chen, Y. G. Dapper 1 antagonizes Wnt of activated atypical protein kinase C. J. Biol. Chem. 276, 43611–43617 (2001). signaling by promoting dishevelled degradation. J. Biol. Chem. 281, 8607–8612 53. Kuznetsova, A. V. et al. von Hippel–Lindau protein binds hyperphosphorylated large (2006). subunit of RNA polymerase II through a proline hydroxylation motif and targets it for 48. Ohh, M. et al. Ubiquitination of hypoxia-inducible factor requires direct binding ubiquitination. Proc. Natl Acad. Sci. USA 100, 2706–2711 (2003). to the β-domain of the von Hippel–Lindau protein. Nat. Cell Biol. 2, 423–427 54. Frew, I. J. & Krek, W. pVHL: a multipurpose adaptor protein. Sci. Signal. 1, pe30 (2000). (2008).

DOI: 10.1038/ncb2082 METHODS

Methods Adenoviral expression system. Recombinant adenovirus containing mouse Atg5 Plasmids and RNA interference. LC3B was cloned into the XhoI and or EGFP complementary DNA was prepared with the AdEasy Adenoviral Vector BamHI sites of pEGFP-C3 (Invitrogen) or into the BamHI sites of pGEX- System (Stratagene). A blunted fragment of mouse Atg5 or EGFP cDNA was cloned 4T1 (Amersham Pharmacia Biotech). Dvl2 was cloned into the BamHI site of into the EcoRV site of the pShuttle–CMV vector, and the resultant plasmids were pDsRed–N1 (Invitrogen) and pCS2(+)–Flag. VHL was cloned into the EcoRI linearized by digestion with PmeI. The linearized plasmids were co-transformed and XhoI sites of pCMV–Myc (Clontech). pDestHA-p62 I431A was constructed into Escherichia coli BJ5183 cells with an adenoviral backbone plasmid, pAdEasy-1. by the mutagenesis of pDestHA–p62 (a gift from Terje Johansen) using the Recombinants were selected, the linearized recombinant plasmids were transfected QuikChange Site-Directed Mutagenesis Kit (Stratagene). To generate Beclin 1 into 293A cells, and high-titre viral stocks were prepared. and LC3B shRNA, the double-stranded oligonucleotides targeting specific genes were cloned into BglII-digested and HindIII-digested pSUPER (Oligoengine). Transfection, reporter assay, immunoprecipitation and immunoblotting, and The sequences of the oligonucleotides are as follows: Beclin 1-1, 5ʹ-gatcc quantitative real time RT–PCR (qRT–PCR) analysis. Transfection, reporter cctgaggatgacagtgaacagttcaagagactgttcactgtcatcctcatttttggaaa-3ʹ (forward) assay, immunoprecipitation and immunoblotting were performed as described and 5ʹ-agcttttccaaaaatgaggatgacagtgaacagtctcttgaactgttcactgtcatcctcaggg-3ʹ previously47,56. (reverse); Beclin 1-2, 5ʹ-gatcccctttcagagataccgacttgttcaagagacaagtcggtatctct- Total RNA was prepared with TRIzol (Invitrogen) and cDNA was synthe- gaaatttttggaaa-3ʹ (forward) and 5ʹ-agcttttccaaaaatttcagagataccgacttgtctcttg sized from 1 μg of RNA with Revertra Ace (Toyobo). qRT–PCR was per- aacaagtcggtatctctgaaaggg-3ʹ (reverse); LC3, 5ʹ-gatcccccaaagttccttgtacctgattca formed with the SYBR Green detection method with a Mx3000p Quantitative agagatcaggtacaaggaactttgtttttggaaa-3ʹ (forward) and 5ʹ-agcttttccaaaaacaaagtt PCR system (Stratagene). The primers used were as follows: human β-actin ccttgtacctgatctcttgaatcaggtacaaggaactttgggg-3ʹ (reverse); (5ʹ-gtaccactggcatcgtgatggact-3ʹ and 5ʹ-ccgctcattgccaatggtgat-3ʹ), human axin2 p62-1, 5ʹ-gatccccacacggacacttcgggtggttcaagagaccacccgaagtgtccgtgttttttggaaa-3ʹ (5ʹ-agtgtgaggtccacggaaac-3ʹ and 5ʹ-cttcacactgcgatgcattt-3ʹ), human c-myc (forward) and 5ʹ-agcttttccaaaaaacacggacacttcgggtggtctcttgaaccacccgaagt- (5ʹ-tctccttgcagctgcttag-3ʹ and 5ʹ-gtcgtagtcgaggtcatag-3ʹ), human cyclin D1 gtccgtgtggg-3ʹ (reverse); p62-2 5ʹ-gatccccggtgaaacacggacacttcttcaag (5ʹ-ctggccatgaactacctgga -3ʹ and 5ʹ-ctccgcctctggcattttgg-3ʹ), human Dvl2 agagaagtgtccgtgtttcacctttttggaaa-3ʹ (forward) and 5ʹ-agcttttccaaaaacggacacttcg (5ʹ-gcttccacatggccatgggc-3ʹ and 5ʹ-tggcactgctggtgagagtcacag-3ʹ), human LC3 gtgaaacatctcttgaagaagtgtccgtgtttcaccggg-3ʹ (reverse). (5ʹ-ccacacccaaagtcctcact-3ʹ and 5ʹ-cactgctgctttccgtaaca-3ʹ), human GABARAP The control shRNA sequence was 5ʹ-gatccccaattctccgaacgtgtcacttcaagagagt- (5ʹ-atgtcattccacccaccagt-3ʹ and 5ʹ-gaagtgccggtcctgaataa-3ʹ), human GATE-16 gacacgttcggagaatttttttggaaa-3ʹ (forward) and 5ʹ-agcttttccaaaaaaattctccgaacgtgt- (5ʹ-aaatatcccgacagggttcc-3ʹ and 5ʹ-gtgttctctccgctgtaggc-3ʹ), mouse GAPDH cactctcttgaagtgacacgttcggagaattggg-3ʹ (reverse). (5ʹ-aagaaggtggtgaagcag-3ʹ and 5ʹ-tcataccaggaaatgagc-3ʹ.), mouse axin2 VHL shRNAs were purchased from Open Biosystems. (5ʹ-tgagtagcgccgtgttagtg-3ʹ and 5ʹ-ctcggtctgtgcctggtcaa-3ʹ), mouse c-myc Transfection of HEK-293T cells with siRNA was performed with Lipofectamine (5ʹ-tgcgacgaggaagagaattt-3ʹ and 5ʹ-atcgcagatgaagctctggt-3ʹ), mouse cyclin D1 2000 (Invitrogen). The sequences of siRNA were as follows: NC (negative con- (5ʹ-gccatccaaactgaggaaaa-3ʹ and 5ʹ-gatcctgggagtcatcggta-3ʹ) and mouse atg5 trol), 5ʹ-ttctccgaacgtgtcacgt-3ʹ; LC3, 5ʹ-gaaggcgcttacagctcaa-3ʹ; GABARAP A, (5ʹ-atgacagatgacaaagat-3ʹ and 5ʹ-tcaatctgttggctgtgg-3ʹ). 5ʹ-cgagaagatccgaaagaaa-3ʹ; GABARAP B, 5ʹ-gcatcttgtaactctcctt-3ʹ; GATE-16 A, 5ʹ-gtgcggtattaaagtgaaa-3ʹ; GATE-16 B, 5ʹ-cctcttcacatagacctat-3ʹ. Autophagy analysis. For starvation, cells were washed three times with PBS and incubated for 4–6 h in Hanks balanced salt solution (HBSS; Invitrogen) or com- Reagents and antibodies. Rapamycin, 3-MA and BFA1 were obtained from plete medium containing 2 μM rapamycin at 37 °C. Autophagy was assessed by Sigma and MG132 was from Calbiochem. Antibodies were from various sources, GFP–LC3 redistribution, LC3 cleavage, p62 degradation and electron microscopy. including Abcam (anti-tubulin), BD Biosciences (anti-Beclin 1, anti-p62 and anti- To inhibit starvation-mediated autophagy, cells were treated with HBSS contain- pVHL), Biomol (anti-ubiquitin), Cell Signaling (anti-Dvl2, anti-E-cadherin, anti- ing 10 mM 3-MA (Sigma). Axin2, anti-c-Myc, anti-cyclin D1, anti-pS6K, anti-S6K and anti-GAPDH), MBL (anti-LC3), Millipore (anti-active β-catenin), Santa Cruz (anti-GFP, anti-tubulin, In vitro pulldown assay. GST or MBP fusion proteins were purified from E. coli anti-Myc, anti-HA, anti-Dvl1, anti-Dvl2 and anti-Dvl3), Sigma (anti-LC3, anti- and immobilized for 30 min on glutathione-Sepharose (Amersham Pharmacia Atg5 and anti-Flag M2) and Jackson ImmunoResearch (AMCA-conjugated Biotech) or amylose resin (New England Biolabs) at 4 °C. The beads were washed anti-rabbit, Aminomethylcoumarin acetate (AMCA)-conjugated anti-mouse, extensively with binding buffer (50 mM Tris-HCl pH 8.0, 250 mM NaCl) and fluorescein isothiocyanate-conjugated anti-mouse, tetramethylrhodamine were subsequently incubated for 90 min with HA-tagged or Flag-tagged Dvl2 β-isothiocyanate (TRITC)-conjugated anti-goat, TRITC-conjugated anti-rabbit). in vitro translated by a reticulocyte lysate system (Promega). Bound proteins The anti-LRP6 antibody T1479 was described previously55. were extracted with loading buffer and analysed by immunoblotting with the indicated antibodies. Cell lines and tumour samples. HeLa, HeLa GFP–LC3 stable cell line (provided by Li Yu, Tsinghua University), HEK-293T, SW480, immortalized Atg5–/– Ubiquitylation analysis. In vivo ubiquitylation was performed as follows. HEK- MEFs (provided by Noboru Mizushima, Tokyo Medical and Dental University) 293 cells were transfected with pHis–ubiquitin along with other expression and Atg7–/– MEFs (provided by Masaaki Komatsu, Tokyo Metropolitan Institute vectors as described. At 36 h after transfection, HEK-293T cells were treated Medical Science) were maintained in DMEM medium supplemented with 10% with 200 nM BFA1 for 6 h. The cells were lysed with buffer A (6 M guanidine,

FBS (Hyclone) in a 37 °C humidified incubator containing 5% CO2. Transient 50 mM Tris-HCl, 250 mM NaCl, 5 mM imidazole pH 8.0). The resulting lysate transfection was performed with VigoFect (Vigorous) in accordance with was sonicated to decrease viscosity before the additino of 20 μl of Ni2+-NTA beads the manufacturer’s recommendations for the biochemical experiments. To (Qiagen), and the mixture was rotated overnight at 4 °C. Subsequently, the beads establish a Beclin 1 knockdown stable cell line, SW480 cells were transfected were washed twice with buffer B (3 M guanidine, 50 mM Tris-HCl, 250 mM NaCl, separately with control shRNA or Beclin 1 shRNA constructs by Lipofectamine 5 mM imidazole pH 8.0) and twice with buffer C (50 mM Tris-HCl, 250 mM 2000 (Invitrogen); 1 μg ml–1 puromycin was added 30 h after transfection. NaCl, 5 mM imidazole, 0.2% Triton X-100 pH 8.0) and subjected to immunoblot- About 10 days later, stably transfected cells were expended and then stored ting. In vitro ubiquitylation was performed as follows. About 200 ng of Myc–Cul2 for further analysis. produced by in vitro transcription and translation was immobilized on Protein A Fresh tumour and normal sample pairs were obtained immediately after sur- beads (Invitrogen) with anti-Myc antibody. The beads were incubated for 1 h at gery at the Department of General Surgery, Peking University Third Hospital (in 4 °C with recombinant proteins (500 ng of VHL, 500 ng of Elongin C, 500 ng of accordance with hospital regulations) and stored at –80 °C. The frozen tissue was Elongin B, 500 ng of Rbx1) to precipitate the E3 complex. The beads were washed extracted and homogenized in a fivefold volume of lysis buffer (50 mM Tris-HCl, three times with buffer containing 50 mM Tris-HCl pH 7.4, 250 mM NaCl and 150 mM NaCl, 0.1 M 2-mercaptoethanol, 2% SDS, complete protease inhibitor 0.1% Triton-X. In vitro ubiquitylation assays were performed in a total volume cocktail and PhosSTOP phosphatase inhibitor cocktail (Roche)). After sonication of 50 μl with the ubiquitinylation kit from Enzo Life Science, in accordance with for 2 min, samples were centrifuged at 15,000g for 15 min and heated at 95 °C for the protocol provided by the manufacturer. Then 500 ng of recombinant human 5 min. The protein was detected by immunoblotting. UbcH5a (E2; Enzo Life Science), beads with immunocomplex of E3, and 200 ng

METHODS DOI: 10.1038/ncb2082 of in vitro-translated Flag–Dvl2 were added to the reaction system and incubated overnight at 4 °C with the primary antibodies was followed by the secondary at 37 °C for 1 h. The reaction was terminated by the addition of 1% SDS and antibody for 3 h at 4 °C. Endogenous Dvl was detected with antibody (Santa Cruz) boiling for 2 min. The supernatant was subjected to immunoprecipitation with that recognized all three Dvl isoforms. Confocal laser scanning of fixed cells was anti-Flag antibody, followed by immunoblotting with anti-ubiquitin antibody, done on a Zeiss LSM 710 NLO Laser Scanning Microscope. or used for a pulldown assay. Soft-agar colony-formation assay. For soft-agar assay, 3 × 103 cells per 35-mm Membrane fraction and sucrose gradient sedimentation. To isolate the membrane dish were seeded in 1 ml of 0.35% (w/v) agarose (Difco) in complete DMEM fraction, HEK-293T cells were lysed in a low-salt buffer (20 mM HEPES pH 7.9, medium overlaid on 1 ml of a 0.5% bottom agar layer. Colonies were scored and

1.5 mM MgCl2, 20 mM KCl, 25% glycerol, protease inhibitors). The cell lysates were photographed 10 days later. first centrifuged at 1,000g to remove the nuclei, and the supernatant was subjected to centrifugation at 40,000g for 30 min. The pellets were then dissolved in 2 × loading Immunohistochemistry. Immunohistochemistry was performed with human buffer. Sucrose gradient sedimentation was performed as described previously46. colon carcinoma (grades I–III with normal controls) tissue array (CC05-01-004) by Cybrdi Inc. A Spearman’s correlation test was used to analyse the relationship Immunoelectron microscopy and immunofluorescence. For immunoeletron between Dvl and Beclin 1, or p62. microscopy analysis, HeLa cells were transfected with Flag–Dvl2 with or without GFP–LC3 for 36 h and cultured under normal or starvation conditions for 4 h. Statistical analysis. Statistical analyses were performed with a two-tailed unpaired Transfected cells were processed for pre-embedding immunogold-silver labelling t-test. P < 0.05 was considered statistically significant. (Flag–Dvl2) and/or post-embedding labelling (GFP–LC3) as described previ- ously57. Immunofluorescence analysis was performed as previously47; cells were 55. Davidson, G. et al. Casein kinase 1γ couples Wnt receptor activation to cytoplasmic grown in six-well plates on glass coverslips and transfected with the indicated signal transduction. Nature 438, 867–872 (2005). 56. Gao, X. et al. Dapper1 is a nucleocytoplasmic shuttling protein that negatively modu- plasmids for 36 h. Before incubation with antibodies, cells were fixed for 30 min lates Wnt signaling in the nucleus. J. Biol. Chem. 283, 35679–35688 (2008). with 4% paraformaldehyde at room temperature, permeabilized for 30 min with 57. Guan, J. S. et al. Interaction with vesicle luminal protachykinin regulates surface 0.2% Triton X-100 (Sigma) and blocked for 2 h in 10% horse serum. Incubation expression of Δ-opioid receptors and opioid analgesia. Cell 122, 619–631 (2005).

supplementary information

DOI: 10.1038/ncb2082

Figure S1 Autophagy attenuates the expression of TopFlash or AP-1 reporter with LC3 siRNA, GABARAP siRNA (B), or two GATE-16 siRNA (A, B). At induced by Wnt3a or Wnt5a. (a, b) HEK293T cells transfected with 24 h post-transfection, the cells were transfected with TopFlash reporter TopFlash or LEF-1 luciferase reporter were maintained in control medium and then were maintained in control medium or Wnt3a condition medium or Wnt3a condition medium (CM) overnight, and then treated with DMSO or (CM) overnight. Then cells were harvested and subjected to luciferase rapamycin (2μM). After 2h treatment, luciferase activity was measured. The activity assay. (f) Efficiency of shRNA-mediated downregulation of LC3 and pRL-tk Renilla reporter (5ng) was co-transfected to normalize transfection Beclin 1. HEK293T cells were transfected with LC3 shRNA or two Beclin 1 efficiency. (c) HEK293T cells transfected with LEF-1 luciferase reporter shRNAs (B1, B2). The cell lysates were subjected to anti-Beclin or anti-LC3 with or without Wnt3a were in HBSS solution (starvation) in the absence immunoblotting. Tubulin served as a loading control. (g, h) Efficiency of or presence of 10 mM 3-MA for 2 h, and then luciferase activity was siRNA-mediated downregulation of LC3, GABARAP and GATE-16. Reporter measured. (d) HeLa cells transfected with AP-1 luciferase reporter or co- assays were performed in triplicate and the data represent the mean ± S.D. transfected with Wnt5a were treated with DMSO or rapamycin for 2 h, and after normalized to Renilla activity. All the other data are also presented then luciferase activity was measured. (e) Knockdown of LC3 or GABARAP, as the mean ± S.D. (n=3). The asterisks indicate a statistically significant but not GATE-16 potentiated Wnt signaling. HEK293T cells were transfected difference (* p<0.05; ** p<0.01).

supplementary information

Figure S2 Dvl undergoes autophagy-mediated degradation. (a) Effect of served as a loading control. (c) Autophagy induced degradation of ectopically autophagy on Wnt3a-induced nuclear accumulation of β-catenin. HEK293T expressed Dvl1, Dvl2, and Dvl3. HEK293T cells transfected with human cells transfected with Wnt3a (0.5 μg) were treated with 2 μM rapamycin Flag-Dvl1, -Dvl2, or -Dvl3 were treated with DMSO or 2 μM rapamycin, or or starved for 4h with or without 10 mM 3-MA. Then the cell nuclei were placed in DMEM or HBSS solution (starvation, S) for 4h in the absence harvested for immunoblotting analysis of the endogenous β-catenin protein. or presence of 3-MA or BFA1. Then, cells were harvested for anti-Flag Lamin B served as a loading control. (b) Rapamycin treatment promoted immunoblotting. Tubulin served as a loading control. (d) Rapamycin had no turnover of endogenous Dvl3 protein. HEK293T cells maintained in effect on Dvl2 mRNA levels shown by quantitative RT-PCR. Total RNA was complete medium were treated with DMSO or 2 μM rapamycin. Then, cells extracted from HEK293T cells treated with DMSO or 2 μM rapamycin for 4 h were harvested for immunoblotting to detect endogenous Dvl3. Tubulin and subjected to q-RT-PCR. The results are shown as means ± S.D. (n=3).

supplementary information

Figure S3 (a) Re-introduction of Atg5 in atg5-/- cells repressed the induction triplicate and the data represent the mean ± S.D. (n=3). (b) Atg5 expression of Wnt target genes. GFP or Atg5 was re-introduction in atg5-/- MEFs, and was confirmed by PCR. (c) Atg7 expression was examined by PCR. (d) the cells were then treated with control or Wnt3a condition medium. Total HEK293T cells were treated with MG132 (1μM) or 3-MA (10 mM) in the RNA was harvested for q-PCR. The bars indicated the induction folds of absence or presence of rapamycin (2μM) for 4 h, followed by anti-Dvl2 target genes under Wnt stimulation. Each experiment was performed in immunoblotting to detect endogenous Dvl2 protein level.

supplementary information

Figure S4 The DEP domain of Dvl2 directly interacts with LC3. (a) A Interaction between LC3 and wild-type or various mutants Dvl2. After co- schematic representation of wild-type mouse Dvl2 and different domains, transfected with myc-LC3 and Flag-Dvl2 variants as indicated, HEK293T deletions or point mutations used in this study. (b) Pull-down assay using cells were harvested for anti-myc immunoprecipitation followed by anti-Flag various recombinant Dvl2 domains and immobilized GST, GST-LC3, immunoblotting. Total protein expression was confirmed by immunoblotting MBP, or MBP-VHL. Proteins were detected by Coomassie staining. (c, d) with whole cell lysates (WCL).

supplementary information

Figure S5 The DIX and DEP domains of Dvl2 are important for autophagy- Degradation kinetics of Dvl2 full-length and variants under starvation (c) or induced Dvl2 degradation. (a) HeLa cells transfected with Flag-tagged full rapamycin treatment (d). HEK293T cells were transfected with the plasmids length (FL) and various mutant Dvl2 were processed for immunofluorescence. as indicated. At 36 h post-transfection, cells were washed three times with The nuclei were counter-stained with DAPI (blue). Scale bar: 10μm. (b) PBS, and subjected to starvation with HBSS or 2μM rapamycin treatment in Ubiquitination of full-length Dvl2 and its variants. HEK293 cells transfected the presence of 20mM cycloheximide. At a various time points, cells were with His-tagged ubiquitin and Flag-tagged Dvl2 variants were harvested for harvested for anti-Flag immunoblotting. Total protein expression was detected Ni-NTA beads precipitation followed by anti-Flag immunoblotting. (c, d) by immunoblotting. Tubulin served as a loading control.

supplementary information

Figure S6 p62 promotes Dvl2 aggregation. (a) Immunofluorescence of HeLa p62-knockdown cells under normal or starvation condition. HEK293T cells cells transfected with Flag-Dvl2 and wild-type or I431A mutant HA-p62. transfected with His-tagged ubiquitin and Flag-tagged Dvl2 with control At 36 h post-transfection, the cells were starved for 4 h in the presence of or p62 siRNA. After 36 h post-transfection, the cells were cultured under BFA1. Proteins visualized by immunofluorescence with anti-HA (green) and normal or starvation condition with BFA1 for 4 h, and then harvested in GTN anti-Flag (red) antibodies. Scale bar: 10μm. (b) Ubiquitination of Dvl2 in buffer for Ni-NTA beads precipitation, followed by anti-Flag immunoblotting.

supplementary information

Figure S7 (a, b) Interaction between pVHL and Dvl2 mutants. After ubiquitin. After transfection for 36 h, the cells were treated with BFA1 transfected with myc-VHL together with various mutants of Flag-Dvl2 for 36 for 6 h under normal or starvation condition. Cell extracts were subjected h, HEK293T cells were harvested for anti-myc immunoprecipitation followed to anti-Flag immunoprecipitation. The ubiquitinated Dvl2 was revealed by anti-Flag immunoblotting. Total protein expression was confirmed by using anti-Ub antibody. (e) Knockdown of pVHL expression attenuates immunoblotting with whole cell lysates (WCL) (lower panels). (c) Interaction the LC3-Dvl2 interaction. HeLa cells transfected with either control or between Dvl2 and pVHL. After transfected with myc-pVHL and Flag-Dvl2 VHL shRNA were placed in starvation medium with BFA1 for 4 h. The for 36 h, HEK293T cells were cultured in nutrient-rich medium or starved cell lysates were subjected to anti-LC3 immunoprecipitation followed by with BFA1 for 4 h. Cells were harvested for anti-myc immunoprecipitation anti-Dvl2 immunoblotting. The total protein expression was confirmed followed by anti-Flag immunoblotting. Protein expression was confirmed by immunoblotting with whole cell lysates (WCL). (f) pVHL knockdown by immunoblotting with whole cell lysates (WCL). (d) Ubiquitination of promoted TopFlash reporter activity. Each experiment was performed in Dvl2. HEK293T cells were transfected with Flag-Dvl2 or together with triplicate and the data represent the mean ± S.D. (n=3).

supplementary information

Figure S8 Full scans of immunoblots shown in Figs 2-7.