Cullin Mediates Degradation of RhoA through Evolutionarily Conserved BTB Adaptors to Control Actin Cytoskeleton Structure and Cell Movement

Molecular Cell

Article Cullin Mediates Degradation of RhoA through Evolutionarily Conserved BTB Adaptors to Control Actin Cytoskeleton Structure and Cell Movement Yuezhou Chen,1,2,5 Zhenxiao Yang,2,3,5 Min Meng,4 Yue Zhao,2 Na Dong,2 Hongming Yan,2 Liping Liu,2 Mingxiao Ding,1 H. Benjamin Peng,4 and Feng Shao2,* 1College

of Life Science, Peking University, Beijing 100871, China Institute of Biological Sciences, Beijing 102206, China 3Graduate School of Chinese Academy of Medical Sciences and Beijing Union Medical College, Beijing 100730, China 4Department of Biology, The Hong Kong University of Science and Technology, Hong Kong, China 5These authors contributed equally to this work *Correspondence: [email protected] DOI 10.1016/j.molcel.2009.09.004 2National

SUMMARY

Cul3, a Cullin family scaffold protein, is thought to mediate the assembly of a large number of SCF (Skp1-Cullin1-F-box protein)-like ubiquitin ligase complexes through BTB domain substrate-recruiting adaptors. Cul3 controls early embryonic development in several genetic models through mechanisms not understood. Very few functional substrate/ adaptor pairs for Cul3 ubiquitin ligases have been identified. Here, we show that Cul3 knockdown in human cells results in abnormal actin stress fibers and distorted cell morphology, owing to impaired ubiquitination and degradation of small GTPase RhoA. We identify a family of RhoA-binding BTB domain adaptors conserved from insects to mammals, designated BACURDs. BACURDs form ubiquitin ligase complexes, which selectively ubiquitinate RhoA, with Cul3. Dysfunction of the Cul3/BACURD complex decreases cell migration potential and impairs RhoA-mediated convergent extension movements during Xenopus gastrulation. Our studies reveal a previously unknown mechanism for controlling RhoA degradation and regulating RhoA function in various biological contexts, which involves a Cul3/BACURD ubiquitin ligase complex. INTRODUCTION Ubiquitination-dependent protein degradation in the proteasome plays key roles in almost every important biological process (Hershko and Ciechanover, 1998; Kerscher et al., 2006). In the final step of polyubiquitin conjugation, ubiquitin ligase recruits the substrate protein and catalyzes formation of polyubiquitin chains on substrate lysine residues. One major class of ubiquitin ligases is a large multisubunit complex mediated by the Cullin family scaffold protein (Petroski and Deshaies, 2005). Cullins share a homologous C terminus that binds the

RING-finger protein Rbx1, but diverge at the N-terminal substrate-recruiting domain. The best-characterized Cullinbased ubiquitin ligase is the SCF complex (for Skp1-Cullin1F-box protein complex) that recognizes and ubiquitinates phosphorylated substrates through a large family of F-box adaptor proteins (Cardozo and Pagano, 2004; Deshaies, 1999). Most eukaryotic genomes encode at least six Cullins (Cul1, 2, 3, 4A, 4B, and 5), and their ubiquitin ligase activities are crucial for a variety of important biological processes. It is not known whether any of the Cullin-based ubiquitin ligases are involved in regulating the actin cytoskeleton dynamics. Similar to Cul1 in the SCF complex, the C-terminal region of Cul3 associates with Rbx1 and its N-terminal domain binds the BTB domain in BTB domain-containing substrate-recruiting adaptors (Furukawa et al., 2003; Geyer et al., 2003; Pintard et al., 2003; Xu et al., 2003). Only a few ubiquitination substrates and corresponding BTB proteins have been identified. The Cul3/ Keap1 complex promotes ubiquitination of the transcription factor Nrf2 to control expression of antioxidant genes upon oxidative stress (Cullinan et al., 2004; Furukawa and Xiong, 2005; Kobayashi et al., 2004), the Cul3/KLHL12 complex regulates the Wnt signaling in development via ubiquitination and degradation of Disheveled (Angers et al., 2006), and the recently identified Cul3/KLHL9/KLHL13 complex is essential for a normal mitosis by targeting Aurora B from mitotic chromosomes for degradation (Sumara et al., 2007). Model eukaryotic organisms (except for yeast) contain dozens or hundreds of BTB proteins that often harbor an additional protein-protein interaction module(s) (Stogios et al., 2005). Thus, Cul3-based ubiquitin ligase complex likely controls a large number of additional conserved pathways via proteasome-dependent substrate degradation. Cul3 is essential for early embryonic development in all tested genetic models, including C. elegans, Drosophila melanogaster, mouse, and even Arabidopsis thaliana (Figueroa et al., 2005; Kurz et al., 2002; Mistry et al., 2004; Singer et al., 1999). However, biochemical understandings of ubiquitination substrates and related BTB proteins for Cul3 complex involved in early embryonic development are lacking. Rho GTPases are molecular switches to control a variety of signaling pathways in eukaryotes (Jaffe and Hall, 2005; Ridley, 2006). RhoA, Rac, and Cdc42, the three best-characterized

Molecular Cell 35, 841–855, September 25, 2009 ª2009 Elsevier Inc. 841

Molecular Cell Ubiquitination of RhoA by the Cul3/BACURD Complex

Rho family members, regulate three separate signaling cascades converged to organization of distinct actin cytoskeleton structures. Specifically, RhoA activation leads to assembly of the contractile actin and myosin filaments structure known as stress fiber; Rac1 promotes formation of actin-rich surface protrusions called lamellipodia; activation of Cdc42 induces filopodia, actin-rich and finger-like structures protruding from the cell membrane (Hall, 1998). Rho GTPases cycle between GTPbound ‘‘on’’ state and GDP-bound ‘‘off’’ state, which is catalyzed by guanine-nucleotide exchange factors and GTPase-activating proteins. Rho GTPases might also be subjected to regulation by ubiquitin-mediated proteasomal degradation (Doye et al., 2002). However, information about regulation of Rho GTPases by the ubiquitin-proteasome pathway is scarce. Smurf1, a HECT domain ubiquitin ligase, is the only known ubiquitin ligase that targets RhoA at the specific cellular protrusion, but cells lacking Smurf1 do not show impaired degradation and an elevated level of total RhoA (Wang et al., 2003). Rho GTPasecontrolled actin cytoskeleton dynamics are a basic determinant of cell morphology and establishment of cell polarity and also provide driving forces for various dynamic cellular processes, including cell migration (Jaffe and Hall, 2005; Ridley, 2006). In early development, plasticity of the actin cytoskeleton is also significant due to the required massive and dynamic cell shape changes and movements of embryonic cells, especially in gastrulation and dorsal closure (Hall and Nobes, 2000). In these processes, rapid actin reorganization in response to developmental cues is mediated by the highly conserved Rho GTPases (Choi and Han, 2002; Kolsch et al., 2007; Settleman, 2001; Tahinci and Symes, 2003; Wunnenberg-Stapleton et al., 1999; Ybot-Gonzalez et al., 2007). In this study, we demonstrate that the Rho family small GTPase RhoA is a direct target of Cul3-based ubiquitin ligase complexes, and cells lacking Cul3 show impaired ubiquitination and degradation of RhoA and exhibit remarkable abnormal actin stress fibers. Using RNAi screen in Drosophila Schneider 2 (S2) cells, we identify a family of BTB proteins conserved from insects to higher eukaryotes, designated BACURDs. Knockdown of BACURDs results in similar abnormal actin stress fibers and markedly increased RhoA protein level. BACURDs form ubiquitin ligase complexes with Cul3, which specifically recruit and ubiquitinate RhoA, in vitro and in vivo. Loss of a functional Cul3/BACURD complex inhibits migration potential of cultured mouse and human cells and leads to defective RhoA-mediated convergent extension movements and multiple RhoA-associated gastrulation defects in developing Xenopus embryos. RESULTS RNAi Knockdown of Cul3 Leads to Remarkable and Abnormal Actin Stress Fibers in Cultured Human Cells We started with RNAi-mediated loss-of-function studies of Cul3 in the cell culture system. Among thirteen human Cul3 (hCul3)-targeting shRNAs tested, two of them (hCul3_shRNA 1# and hCul3_shRNA 2#) markedly reduced the exogenous Flag-Cul3 in 293T cells and endogenous Cul3 in HeLa cells (Figure 1A, left panels). When hCul3_shRNA 1# was either transiently or stably expressed in HeLa cells by retrovirus infection

(Figure 1A, right panels), densely packed actin stress fibers appeared in greater than 90% of knockdown cells (for stable knockdown cells, see Figure 1B and light blue columns in Figure 1C; for transient knockdown, see Figure 1F and dark blue columns in Figure 1C). In contrast, control cells harboring an empty vector displayed a normal punctuate distribution of filamentous actin. Similar phenomena were observed with hCul3_shRNA 2# (data not shown). Two other shRNAs that gave little knockdown did not stimulate evident stress fiber formation (Figures S1A–S1C). Consistently, nearly all HeLa cells treated with either of two independent hCul3-specific siRNAs (hCul3_siRNA A and hCul3_siRNA B) (Figure 1D) developed similar abnormal stress-fiber structures (Figure 1E and median blue columns in Figure 1C). The stress fiber phenotype in GFPmarked RNAi cells (hCul3_shRNA 1#) was efficiently rescued by expression of the RNAi-resistant version of hCul3 (Figure 1F and dark blue columns in Figure 1C). Interestingly, Drosophila Cul3 (dCul3) could also efficiently block the stress fiber phenotype in Cul3 knockdown cells (Figure 1F and dark blue columns in Figure 1C). These data firmly establish that reduction of the cellular Cul3 level induces formation of remarkable and abnormal actin stress fibers. Formation of Actin Stress Fibers in Cul3-Depleted Cells Is RhoA Dependent Activation of the RhoA pathway is responsible for stress fiber formation, which requires cooperative activation of two downstream effectors, ROCK and mDia (Leung et al., 1996; Watanabe et al., 1997, 1999) (Figure S2). Y-27632, a potent and specific ROCK inhibitor (Uehata et al., 1997), efficiently blocked actin stress fibers in 70% of hCul3 RNAi HeLa cells, and expression of a dominant-negative RhoA mutant (RhoA N19) also gave a similar effect (Figures 2A and 2B); YopT, a potent bacterial toxin that inactivates RhoA through a proteolytic removal of the lipid membrane anchor of the GTPase (Shao et al., 2002), completely abolished actin stress fibers in Cul3 RNAi cells in its protease activity-dependent manner (Figures 2A and 2B). Furthermore, reduction of RhoA, but not RhoB and RhoC, by siRNA (Figure S3) blocked the stress fiber formation in Cul3 RNAi cells (Figures 2C and 2D). These data together suggest the requirement of RhoA activity for the development of remarkable and abnormal actin stress fibers in Cul3-depleted cells. The RBD pull-down assay (Ren and Schwartz, 2000) was then employed to measure the amount of activated RhoA (RhoAGTP). As shown in Figure 3A, the level of RhoA-GTP is markedly elevated in Cul3 RNAi cells compared to that in control cells. Rho activation turns on transcription of the SRE-driven luciferase through the transcription factor SRF (Hill et al., 1995). Consistently, about 4.5-fold elevation of the SRE-luciferase activity was observed in hCul3 shRNA HeLa cells (data not shown). Moreover, overexpression of RhoA resulted in expected similar stress fibers that largely mimicked the effect of Cul3 knockdown (Figure S4). These data suggest that RNAi depletion of hCul3 in human cells leads to RhoA activation, which in turn induces excessive actin stress fiber formation. We also noticed that Cul3 knockdown HeLa cells became larger, and a significant percentage of them (16%) contain multiple nuclei (Figure 1B), a phenomenon frequently seen in cells with excessive RhoA activation.

842 Molecular Cell 35, 841–855, September 25, 2009 ª2009 Elsevier Inc.

Molecular Cell Ubiquitination of RhoA by the Cul3/BACURD Complex

Figure 1. RNAi Knockdown of hCul3 Leads to Remarkable Actin Stress Fibers in Human Cells (A) Knockdown of hCul3 by shRNAs. Left: intact HeLa cells (upper) or 293T cells expressing Flag-hCul3 (lower) were transiently transfected with indicated hCul3_shRNA plasmid or an empty vector ( ). Right: stable hCul3_shRNA 1# knockdown HeLa cells generated by retroviral infection. Total cell lysates were immunoblotted with antibodies against Flag, Cul3, or tubulin as indicated. (B) Actin stress fibers in stable hCul3_shRNA 1# knockdown HeLa cells or the control cells (Vector). Shown are representative cell images where DAPI staining marks the nucleus and F-actin denotes the filamentous actin stained by rhodamine-phalloidin. Images were taken with the same exposure time. (C) Statistics of cells showing strong abnormal actin stress fibers. Light blue columns, Cul3 stable knockdown cells shown in (B); medium blue columns, Cul3 siRNA transfected cells shown in (E); dark blue columns, GFP-positive Cul3 RNAi cells transfected with rescue plasmid shown in (F). Data are presented as mean ± SD (error bar) of at least two independent determinations. Two hundred and fifty cells were counted for each experiment. (D) Knockdown of endogenous Cul3 by two independent siRNAs. Scramble siRNA or hCul3_siRNA (A or B) was transfected into HeLa cells. Left: qRT-PCR analysis. mRNA level of Cul3 is normalized by that of GAPDH. All the reactions were performed in triplicates, and shown are mean values ± SD (error bar) from three independent experiments. Right: western blots of siRNA-transfected cells using indicated antibodies. (E) Actin stress fibers in Cul3 siRNA HeLa cells. Scramble siRNA or hCul3_siRNA (A or B) was transfected into HeLa cells. Denotations are similar to those shown in (B). (F) Rescue of the actin stress fiber phenotype in Cul3 RNAi cells. Shown are images of HeLa cells cotransfected with GFP-labeled hCul3_shRNA 1# plasmid and an empty vector, RNAi-resistant human Cul3 (hCul3), or Drosophila Cul3 (dCul3).

Impaired Ubiquitination and Degradation of RhoA in Cul3-Depleted HeLa and Drosophila S2 Cells While measuring amounts of RhoA-GTP in Cul3 RNAi cells, to our surprise, we found that the steady level of endogenous RhoA increased significantly in hCul3 RNAi HeLa cells compared with that in control cells (Figures 3A and 3B). In contrast, levels of other closely related small GTPases, including RhoB, RhoC, Rac, and Cdc42, remained unchanged (Figure 3B),

agreeing with the absence of Rac/Cdc42-associated actin cytoskeleton phenotypes in Cul3 RNAi cells. Specific recognition of the endogenous Rho was validated by RhoB- and RhoC-targeting siRNAs (Figure S3). Accumulation of RhoA in Cul3 knockdown cells was also evident from the brighter immunofluorescence signal of RhoA staining (Figure S5). Higher protein level may result from transcriptional upregulation or increased protein stability. In contrast to over 10-fold decrease

Molecular Cell 35, 841–855, September 25, 2009 ª2009 Elsevier Inc. 843

Molecular Cell Ubiquitination of RhoA by the Cul3/BACURD Complex

Figure 2. Stress Fiber Formation in Cul3 RNAi Cells Is RhoA Dependent (A and B) Effects of Y27632, RhoA N19, and YopT (WT or the catalytic inactive C/A mutant) on stress fiber formation in Cul3 RNAi cells. Experiments were performed and data are presented similarly to those shown in Figure 1F (for A) and Figure 1C (for B), except that Y27632 was added directly into the culture medium. Error bar indicates SD. (C and D) Rescue of Cul3 RNAi-induced stress fiber formation by co-knockdown of the RhoA family of small GTPases. siRNAs targeting hRhoA, hRhoB, or hRhoC were delivered into HeLa cells together with hCul3_siRNA A. Data are presented similarly to those shown in Figure 1E (for C) and Figure 1C (for D). Error bar indicates SD.

of hCul3 transcripts, the mRNA level of RhoA in hCul3depleted HeLa cells stayed at the level similar to that in the control cells (Figure 3C). We then measured RhoA protein stability in Cul3 RNAi cells. Following cycloheximide (CHX) treatment to block new protein synthesis, the level of RhoA decreased to half within 4–6 hr in control cells, while no significant reductions were observed until 12 hr in Cul3 RNAi cells (Figure 3D). This suggests that RhoA degradation was severely impaired in Cul3 knockdown cells. Reintroduction of RNAi-resistant Cul3 mutants deficient in either BTB adaptor binding (DN41) or Rbx1 association (DRbx1) (Furukawa et al., 2003) into Cul3 RNAi cells failed to reverse the elevated RhoA protein level (Figure S6A) and to

block development of the strong actin stress fibers (Figures S6B and S6C). Block of the proteasome activity by MG132 led to an increased level of endogenous RhoA (Hoffmann et al., 2007; Lanning et al., 2004) (Figure 3E, left panels), but had little effect in cells lacking Cul3 (Figure 3E, right panels). Indeed, polyubiquitinated species of RhoA were readily detected when endogenous RhoA was immunoprecipitated under denaturing conditions from MG132-treated cells (Figure 3F; compare lane 1 and 3). However, levels of polyubiquitin-modified RhoA were greatly attenuated by hCul3 siRNA treatment (Figure 3F; compare lane 1 and 2). As knockdown of the only reported RhoA ubiquitin ligase, Smurf1, has no effect on the total cellular level of endogenous RhoA (Wang et al., 2003), our data suggest

844 Molecular Cell 35, 841–855, September 25, 2009 ª2009 Elsevier Inc.

Molecular Cell Ubiquitination of RhoA by the Cul3/BACURD Complex

Figure 3. Impaired Ubiquitination and Degradation of RhoA in Cul3 RNAi Cells (A) RBD pull-down assays of RhoA-GTP. HeLa cells transfected with scramble siRNA or hCul3_siRNA A were lysed, and RhoA-GTP was precipitated by GSTRBD and analyzed by RhoA immunoblotting (RBD pull-down, upper panel). Immunoblotting of indicated total protein levels is shown in the lower panels (input). (B) Effects of Cul3 knockdown on protein levels of endogenous Rho family GTPases. HeLa cells were transfected with indicated Cul3_shRNA plasmid or a control empty vector, and the total cell lysates were immunoblotted using antibodies as indicated. (C) qRT-PCR measurements of the mRNA level of Cul3 and RhoA in Cul3 RNAi cells. Indicated hCul3_shRNA plasmid or a control empty vector was transfected into HeLa cells. The left three columns are relative mRNA levels of Cul3, and the right three columns correspond to those of RhoA. The mRNA level of GAPDH was used for normalization. All the reactions were performed in triplicates, and shown are mean values ± SD (error bar) from three independent experiments. (D) Stability of endogenous RhoA in hCul3 and hBACURD knockdown cells. HeLa cells were transfected with indicated siRNAs (hCul3_siRNA A, hBACURD1 siRNA 1#, and hBACURD2 siRNA 1#) (see Table S2). Cells were lysed at the indicated time after cycloheximide (CHX) treatment, and shown are RhoA and tubulin immunoblots of total cell lysates. (E) Effects of MG132 treatment on RhoA protein level in hCul3 knockdown cells. Intact HeLa cells (left) or HeLa cells transfected with scramble siRNA or hCul3_siRNA A (right) were treated with DMSO (0) or MG132 at the indicated concentration for 4 hr prior to cell lysis. Immunoblots of total cell lysates are shown. (F) Immunoblotting assays of effects of Cul3 knockdown (hCul3_siRNA A) on in vivo ubiquitination of RhoA. Endogenous RhoA was immunoprecipitated under denaturing conditions from MG132-treated HeLa cells using RhoA or a control antibody. The immunoprecipitates were analyzed by anti-ubiquitin (Ub) and antiRhoA (RhoA) immunoblotting. RhoA-Ub(n) denotes polyubiquitinated forms of RhoA. The lower two panels show knockdown of Cul3 and a tubulin loading control.

that Cul3-based ubiquitin ligase complex plays a dominant role in controlling proteasome-dependent degradation of RhoA in cells. Drosophila Cul3 could complement hCul3 in rescuing Cul3 RNAi-induced actin stress fiber phenotype (Figures 1C and 1F). Indeed, double-strand RNA (dsRNA)-mediated knockdown of dCul3 in S2 cells also led to an increased dRho1 protein level

(RhoA ortholog in Drosophila), but did not affect dRho1 mRNA level (Figures S7A and S7C). This effect was not seen with Drosophila Cdc42 (dCdc42) (Figure S7A) and was further confirmed by the much stronger immunofluorescence staining signal of dRho1 in dCul3 knockdown cells (Figure S7B) (the dRho1 staining signal was validated by dRho1 dsRNA treatment). These results indicate that Drosophila S2 cells employ

Molecular Cell 35, 841–855, September 25, 2009 ª2009 Elsevier Inc. 845

Molecular Cell Ubiquitination of RhoA by the Cul3/BACURD Complex

a similar mechanism involving a dCul3-based ubiquitin ligase to control proteasome-dependent degradation of dRho1. Identification of the Conserved BACURD Family of BTB Proteins Involved in RhoA Accumulation and Stress Fiber Formation Induced by Cul3 Knockdown Cul3DN41 failed to complement the function of Cul3 in regulating RhoA and RhoA function (Figure S6), suggesting the requirement of certain BTB domain adaptors for Cul3-dependent RhoA degradation. In contrast to nearly 200 BTB proteins in humans, D. melanogaster has less than 80 BTB proteins (http://btb. uhnres.utoronto.ca) (Stogios et al., 2005). Out of dsRNAs for a total of 69 Drosophila BTB proteins obtained and assayed, administration of one dsRNA, corresponding to the Drosophila gene CG10465, gave a significant and repeatable increase in the steady protein level of dRho1, but not dCdc42 (Figure S8A). Quantitative real-time PCR (qRT-PCR) measurements not only verified the knockdown efficiency of CG10465 dsRNA, but also indicated an unaffected dRho1 mRNA level (Figure S8B). These data suggest that the Drosophila BTB protein encoded by CG10465 may mediate dCul3-dependent dRho1 degradation. Hence, we name the protein product of CG10465 Drosophila BACURD (dBACURD), for BTB-containing adaptor for Cul3mediated RhoA degradation, hereafter. Sequence analysis suggests that dBACURD belongs to a family of BTB proteins highly conserved from insects to mammals with one member in D. melanogaster (dBACURD), two in Xenopus laevis (named xBACURDa and xBACURDb), and three each in mouse and human (named BACURD1, 2, and 3) (Figure S9). BACURDs contain an N-terminal BTB domain followed by a C-terminal region of about 180 residue sequence with no recognizable motifs. The sequence similarity between dBACURD and hBACURDs is about 60%, and the three hBACURDs are about 70% homologous to each other, with sequence divergences mainly located at the two extreme termini. No BACURDs have a defined biochemical or biological function. Transcripts of only hBACURD1 and hBACURD2 could be detected in HeLa cells (data not shown). HeLa cells transfected with either hBACURD1 or hBACURD2 siRNA (Figure 4B) contained a higher level of RhoA (but not Cdc42) (Figure 4A) and an unchanged RhoA mRNA level (Figure 4B). Also, similarly, to Cul3 knockdown, siRNA knockdown of either hBACURD led to remarkable actin stress fibers (Figures 4C and 4D) as well as increased protein stability of endogenous RhoA (Figure 3D). Meanwhile, RNAi of other BTB proteins, such as Aurora B-targeting KLHL9 and KLHL13, did not affect RhoA level and the actin stress fiber structure (data not shown). Double knockdown of hBACURD1 and hBACURD2 appeared to have some additive effects on both the total RhoA level (Figure 4A) and RhoA-GTP level (Figure S10). Consistently, a further increase of RhoA stability was observed in hBACURD double-knockdown cells (Figure 3D). These data strongly suggest that the BACURD family of BTB proteins shares a similar function as Cul3 in controlling RhoA degradation and in regulating RhoA-associated actin cytoskeleton structure. We then investigated whether hBACURD functions together with Cul3 in regulating RhoA and RhoA function. EGFP-tagged hBACURD1 and hBACURD2, but not EGFP alone, were readily

coimmunoprecipitated with Flag-hCul3 in 293T cells (Figure 4E). The Cul3/BACURD interaction was independent of the presence of RhoA (Figure S11). Mutations in the BTB domain of BACURDs modeled and predicted from the SCF complex structure (Xu et al., 2003) (V84A/L85A/I86A in hBACURD1 denoted as hBACURD1m; I71A/L72A/I73A in hBACURD2 denoted as hBACURD2m) largely abolished the hCul3/hBACURDs interaction (Figures 4E and S11). Furthermore, ectopic expression of EGFP-hBACURD1m or EGFP-hBACURD2m induced striking actin stress fibers in about 80% of transfected HeLa cells (Figure 4F), reminiscent of those observed in BACURD RNAi cells (Figure 4C). These results suggest that hBACURDms act in a dominant-negative fashion to uncouple RhoA from Cul3dependent degradation and strongly support the hypothesis that BACURDs are adaptors for Cul3 to mediate RhoA ubiquitination and degradation. Targeting RhoA for Polyubiquitination by the Cul3/BACURD Ubiquitin Ligase Complex We further examined whether hBACURDs could form complexes with RhoA. As neither RhoA nor BACURD antibodies (see below) are good enough for immunoprecipitation of endogenous proteins, we turned to generating stable HeLa cell lines expressing low level of Flag-hBACURD (Figures S12A and S12B). Endogenous RhoA was apparently precipitated together with either Flag-hBACURDs from these cell lines (Figure S12C). The transiently expressed Flag-hBACURDs and HA-RhoA WT or N19 were more easily coimmunoprecipitated (Figure 5A). The interaction is specific to RhoA, as BACURDs did not coimmunoprecipitate with Rac1 or Cdc42 (data not shown). Recombinant GST-RhoA was capable of binding bacterially purified MBP-hBACURD1 or hBACURD2 in vitro (Figure 5B), indicating a Cul3-independent physical association between RhoA and hBACURDs. The RhoA-binding region was mapped to the C-terminal domain of hBACURDs (Figure S13). Notably, hBACURDs prefer to coimmunoprecipitate with RhoA N19 (Figure 5A), which likely mimics the conformation of RhoAGDP. We then assayed in vitro binding between MBP-BACURDs and purified WT RhoA loaded with GTP or GDP or treated it with EDTA to obtain the Mg2+-free form that adopts a conformation similar to the nucleotide-free state (Shimizu et al., 2000; Soundararajan et al., 2008). As shown in Figure 5B, MBP-BACURD1 or -BACURD2 appeared to selectively precipitate RhoA-GDP versus RhoA-GTP and Mg2+-free RhoA. This result clearly suggests that BACURDs use the C-terminal domain to directly bind RhoA, which has a preference for RhoA-GDP. Furthermore, Flag-hCul3 readily coimmunoprecipitated HARhoA N19 from 293T cells in the presence of EGFP-hBACURDs, but not the Cul3-binding-deficient mutant EGFP-hBACURDms (Figure 5C). Despite numerous antibody generation tries, we were only able to obtain an hBAUCRD2 peptide antibody that is limited to detect the total endogenous hBACURD2, as validated by two independent hBACURD2 and hBACURD1 siRNAs (Figure S14). Using this antibody, both hBACURD2 and RhoA were found to be specifically present in the endogenous Cul3 immunoprecipitates from HeLa cells (Figure 5D). These data strongly suggest that Cul3, BACURD, and RhoA could form a trimeric complex in cells.

846 Molecular Cell 35, 841–855, September 25, 2009 ª2009 Elsevier Inc.

Molecular Cell Ubiquitination of RhoA by the Cul3/BACURD Complex

Figure 4. The BACURD Family BTB Proteins Are Involved in RhoA Accumulation and Stress Fiber Formation Induced by Cul3 Knockdown (A) Immunoblotting assays of effects of hBACURD1 and hBACURD2 knockdown on protein levels of endogenous Rho GTPases. HeLa cells were transfected with hBACURD1 siRNA 1#, hBACURD2 siRNA 1#, or scramble siRNA, and total cell lysates were immunoblotted using indicated antibodies. (B) qRT-PCR measurements of mRNA expression of hBACURDs and RhoA in HeLa cells treated with scramble siRNA, hBACURD1 siRNA 1#, or hBACURD2 siRNA 1#. Experiments were performed and data are presented similarly to those in Figure 3C. Error bar indicates SD. (C and D) Actin stress fibers in hBACURD knockdown HeLa cells. hCul3_siRNA A, hBACURD1 siRNA 1#, and hBACURD2 siRNA 1 were used here. Shown in (C) are representative cell images where DAPI staining marks the nucleus and F-actin denotes the filamentous actin stained by rhodamine-phalloidin. Statistics of cells showing strong actin stress fibers are shown in (D) and presented as mean ± SD (error bar) of at least two independent determinations. Two hundred and fifty cells were counted for each experiment. (E) Coimmunoprecipitation assays of binding between Cul3 and hBACURDs. Flag-hCul3 and indicated EGFP-tagged hBACURD1 or hBACURD2 constructs were cotransfected into 293T cells. The total cell lysates (Input) and Flag immunoprecipitates (Flag-IP) were blotted using indicated antibodies. hBACURD1m and hBACURD2m denote the hBACURD1 V84A/L85A/I86A and hBACURD2 I71A/L72A/I73A triple mutant, respectively. (F) Actin stress fibers induced by dominant-negative mutants of hBACURDs. HeLa cells were transfected with indicated EGFP-tagged Cul3-binding-deficient hBACURD mutants identified in (E). Upper panels, rhodamine-phalloidin staining of the filamentous actin; lower panels, transfected cells marked by GFP staining; numbers in the parenthesis, statistics of GFP-positive cells showing actin stress fibers.

We then carried out in vitro ubiquitination reactions using His33Flag-RhoA N19, GST-hCul3/Rbx1, His-33HA-hBACURD1/2, and the hCul3/Rbx1/hBACURD complexes purified from insect

cells. Flag-RhoA N19, immunopurified from the reaction mixtures under denaturing conditions, was found to be readily and heavily polyubiquitinated following the ubiquitination

Molecular Cell 35, 841–855, September 25, 2009 ª2009 Elsevier Inc. 847

Molecular Cell Ubiquitination of RhoA by the Cul3/BACURD Complex

Figure 5. RhoA Is a Direct Target and Substrate of the Cul3/BACURD Ubiquitin Ligase Complex (A) Coimmunoprecipitation assays of RhoA and hBACURDs in transfected cells. Flag-hBACURD1 (left) or Flag-hBACURD2 (right) was cotransfected with HA-RhoA (WT or N19) into 293T cells. Total cell lysates were subjected to Flag immunoprecipitation followed by immunoblotting using indicated antibodies. (B) In vitro interaction between purified hBACURDs and RhoA loaded with different nucleotides. MBP-hBACURD1 (left) or MBP-hBACURD2 (right) were incubated with GST or GST-RhoA immobilized on glutathione beads and preloaded with indicated nucleotides. MBP proteins bound on glutathione beads were recovered and blotted with MBP antibody (upper panel). The Coomassie blue staining (lower panels) shows inputs of GST, GST-RhoA, and MBP-hBACURD proteins. *, EDTA-treated RhoA that is Mg2+-free and adopts a conformation similar to the nucleotide-free state. (C) Trimeric complex formation of exogenous hCul3, hBACURDs, and RhoA in cells. HA-RhoA N19, Flag-hCul3, and indicated GFP-tagged hBACURDs or their BTB domain mutants were cotransfected into 293T cells. Flag immunoprecipitates (Flag-IP, upper three panels) and total cell lysates (Input, lower two panels) were blotted using indicated antibodies. hBACURD1m and hBACURD2m are Cul3-binding deficient mutants identified in Figure 4E. (D) Endogenous RhoA, hBACURD2, and hCul3 form a complex. Lysates of intact HeLa cells were subjected to immunoprecipitation using the Cul3 antibody or the control IgG. Cell lysates prior to immunoprecipitation (Input) and the immunoprecipitates (IP) were blotted using the indicated antibodies. (E and F) In vitro ubiquitination of RhoA by the Cul3/hBACURD complex. Recombinant complexes of GST-hCul3/Rbx1, GST-hCul3/Rbx1/hBACURD1, or GST-hCul3/Rbx1/hBACURD2 purified from insect cells were used in ubiquitination assays supplemented with indicated components. Ubiquitinated forms of Flag-RhoA were immunoprecipitated by Flag antibody under denaturing conditions and blotted with ubiquitin antibody (top panel). RhoA-Ub(n) and RhoA-Ub

848 Molecular Cell 35, 841–855, September 25, 2009 ª2009 Elsevier Inc.

Molecular Cell Ubiquitination of RhoA by the Cul3/BACURD Complex

reaction (Figure 5E). Omission of either hCul3 or hBACURDs from the reaction abolished RhoA ubiquitination (Figures 5E and 5F). We also performed ubiquitination assays using WT RhoA loaded with GTP or GDP or the Mg2+-free form. Consistent with the binding results (Figure 4B), robust ubiquitination activities were observed with RhoA-GDP, and the other two forms of RhoA were much poorer substrates in this assay (Figures 5F and S15A). The dCul3/Rbx1/dBACURD complex purified from Sf9 cells also exhibited robust ubiquitin ligase activities toward the dRho1 protein (Figure S15B). These results provide direct evidence that RhoA is a direct ubiquitination target of the Cul3/ BACURD complex, and the preference for RhoA-GDP further suggests that activity of the Cul3/BACURD complex likely limits the amount of RhoA-GDP or a specific pool of RhoA-GDP in cells that otherwise could be activated by RhoGEFs. Dysfunction of the Cul3/BACURD Complex Attenuates the Migration Potential of MEF and HeLa Cells RhoA-controlled actin cytoskeleton dynamics determine cell migration potential and provide driving forces for cell movement, including that in cancer progression (Ridley, 2006). Our above data indicate a housekeeping-type role of the Cul3/BACURD/ RhoA axis in maintaining an appropriate level of RhoA in cells. Thus, we hypothesized that the ability of cells to migrate should be attenuated in the absence of a functional Cul3/BACURD complex. To this end, the Boyden chamber assay was employed to examine cell migration properties. In MEF cells, siRNA knockdown of mCul3 (Figure S16A) led to severe migration defects that could be rescued by simultaneous knockdown of mRhoA (Figures S16B, S17A, and S17B). Transcripts of all three mouse BACURDs were detected in MEF cells (data not shown). Striking migration defects were observed in MEF cells treated with mBACURD1 or mBACURD3 siRNA, and cells transfected with mBACURD2 siRNA showed a negligible weak defect (Figures S17A and S17B). Variations of migration defects among the three mBACURD knockdown cells appeared to correlate well with differential silencing efficiencies of the corresponding siRNAs (Figure S16C). Treatment of MEF cells with all three mBACURDs at the same time gave slightly more severe migration defects than the individual mBACURD1 or mBACURD3 knockdown. These results suggest an important role of the Cul3/BACURD complex in controlling cell migration potential and also highlight its evolutionary conservation in mouse cells. Cell migration is key to many biological processes, including cancer cell metastasis, would healing, and embryonic development. In the Boyden chamber assay, severely impaired migrations were also observed in the tumorigenic HeLa cells treated with siRNAs targeting hCul3, hBACURD1, or hBACURD2 (Figures 6A and 6B). Overexpression of RhoA resulted in an expected similar migration phenotype (Figures S17C and S17D). Double knockdown of hBACURD1 and hBACURD2 gave a slightly more severe defect than individual knockdown. Similar to stress fiber formation, migration defects caused by

Cul3 knockdown could be rescued either by expression of the RNAi-resistant hCul3 or by simultaneous knockdown of hRhoA. hBACURD knockdown-triggered migration defects were also reversed by reintroducing the RNAi-resistant version of the corresponding BACURD gene (Figures 6A and 6B). Moreover, the ‘‘wound healing’’ properties of hCul3 or hBACURD RNAi HeLa cells were severely affected, and reduction of the RhoA level by siRNA also blocked the ‘‘would healing’’ defects in hCul3 RNAi cells (Figure 6C and 6D). Taken together, these results suggest that the Cul3/BACURD complex plays an important role in controlling cell migration potential by maintaining an appropriate cellular level of RhoA. The Cul3/BACURD Complex Is Critical for RhoA-Mediated Convergent Extension during Gastrulation in Xenopus Embryos In early Xenopus embryos, overexpression of either RhoA L63 or N19 leads to defective convergent extension cell movement and multiple anterior and axial gastrulation phenotypes (Tahinci and Symes, 2003). As both Xenopus BACURDs interacted with Cul3 and RhoA N19 in heterologous 293T cells in a manner similar to hBACURDs (Figure S18A), we asked whether loss of xBACURDs could affect RhoA-mediated convergent extension cell movements during Xenopus gastrulation. Antisense Morpholino oligonucleotide (MO) capable of blocking in vitro translation of the xBACURD mRNA (Figure 7A) was used to silence expression of endogenous xBACURDs in Xenopus embryos. The classical Keller explant assay (Keller, 1991) was carried out to directly examine the axial mesoderm cell movement. Whereas open-faced Keller explants from control MO-injected embryos continued to elongate, explants from xBACURD MO-injected embryos stopped elongating and behaved similarly to those from RhoA L63-overexpressing embryos reported previously (Tahinci and Symes, 2003) or obtained in our hands (Figures 7B and 7C). Almost all the reported convergent extension phenotypes in Xenopus are caused by dysfunction of signaling pathways upstream of Rho GTPases (Kuhl, 2002) or direct manipulations of Rho GTPase activities (Tahinci and Symes, 2003; Wallingford et al., 2002). Our results clearly demonstrate that depletion of xBACURDs inhibits RhoA-dependent convergent extension cell movement of the axial mesoderm in Xenopus. A significant portion of xBACURD MO-treated embryos or embryos injected with dominant-negative mutants of BACURD mRNA exhibited multiple anterior and axial gastrulation defects despite the normal development during cleavage (Figures 7D and S18B; Table S1). MO-induced phenotypes were blocked by coinjection of MO-resistant EGFP-xBACURD mRNA (Figure 7D). Anterior defects were mainly associated with reduced or absent head structure, consistent with a previously established function of Rho GTPase in controlling the head mesoderm cell migration (Ren et al., 2006; Tahinci and Symes, 2003; Wunnenberg-Stapleton et al., 1999). Axial gastrulation defects

denote polyubiquitinated and monoubiquitinated forms of RhoA, respectively. Unmodified RhoA on the beads and Cul3/hBACURD proteins added into the reaction are shown in the lower three panels. Flag-RhoA N19 was used in (E); WT Flag-RhoA loaded with indicated different nucleotides or the Mg2+-free form (Flag-RhoA-EDTA) was used in (F). HC, antibody heavy chain.

Molecular Cell 35, 841–855, September 25, 2009 ª2009 Elsevier Inc. 849

Molecular Cell Ubiquitination of RhoA by the Cul3/BACURD Complex

Figure 6. The Cul3/BACURD Complex Is Essential for Migration of Cultured Cells (A and B) Boyden chamber cell-migration assays. HeLa cells were treated with hCul3_siRNA A or hBACURD siRNAs (1#, see Table S2). Co-knockdown of hRhoA or expression of the RNAi-resistant hCul3 or hBACURDs was carried out to test their rescue effects. Assays were performed as described under Experimental Procedures. Migrated cells on the lower side of the filter were stained with crystal violet and shown in (A). Relative cell migration shown in (B) was determined by counting nine random fields of the filter, and the value from the control RNAi cells was arbitrarily set at 100%. Error bar indicates SD. (C and D) Wound-healing assays of hCul3 or hBACURD knockdown HeLa cells. The siRNA oligos used are the same as those in (A). Confluent HeLa cells treated with indicated siRNAs were scratched by a plastic pipette. Cells were photographed at 0 or 24 hr after wounding (C). The width of the wound was measured to determine the relative wound closure (D), with that of the control siRNA-treated cells arbitrarily set as 100%. Error bar indicates SD.

in xBACURD MO-injected embryos was characterized by a failure of the proctodeum to close and spinal bifida, which resulted from inability of the blastopore to close, as well as a shortening of the anterior-posterior axis (Djiane et al., 2000; Sokol, 1996; Wallingford and Harland, 2001). These anterior and axial gastrulation phenotypes also appeared in embryos

overexpressing RhoA L63, as reported previously (Tahinci and Symes, 2003) and confirmed in our hands (Figure S18B). Taken together, our biochemical and functional analysis with Xenopus BACURDs reveals a critical role of the Cul3/BACURD ubiquitin ligase complexes in regulating RhoA-mediated convergent extension cell movements during Xenopus early gastrulation.

850 Molecular Cell 35, 841–855, September 25, 2009 ª2009 Elsevier Inc.

Molecular Cell Ubiquitination of RhoA by the Cul3/BACURD Complex

Figure 7. The Cul3/BACURD Complex Is Essential for RhoA-Mediated Convergent Extension Cell Movement during Xenopus Gastrulation (A) Suppression of in vitro translation of xBACURD mRNA by Morpholino. Indicated xBACURDs were in vitro translated in the presence of specific or control Morpholino oligos (MOs). Shown is the autoradiography of in vitro translation products. (B and C) The Keller explant assay of convergent extension. Indicated xBACURD MOs or RhoA L63 mRNA was injected into 2- to 4-cell-stage Xenopus embryos, and the Keller explants were removed when the injected embryos reached the early gastrula stage (stage 10–10+). Shown in (B) are representative images of in vitro cultured explants at stage 20. The length of the longest aspect and the width of constriction point of each explant were measured, and statistics of the length-to-width ratio (LWR) are shown in (C). Significant differences among explants were calculated by using the Student’s t test (p < 0.05). Error bar indicates SD. (D) Effects of xBACURD Morpholino on Xenopus embryo development and rescue analysis. Xenopus embryos at stage 26 were examined after injection of 2- to 4-cell-stage embryos with a control MO (left panels), indicated xBACURD MO (middle panels), or the indicated combination of xBACURD MO and excessive MO-resistant xBACURD mRNAs (right panels) (4 ng and 1 ng mRNA for xBACURDa and xBACURDb, respectively). The amount of MO injected for each embryo is about 4.2 pM. Statistics of embryos showing different gastrulation phenotypes are detailed in Table S1.

DISCUSSION Biochemical Modes of the Cul3/BACURD Ubiquitin Ligase Complex and Its Recognition of RhoA In contrast to the SCF complex, little is known about the biochemical mechanism of Cul3-based complexes. In the best-studied Cul3/Keap1/Nrf2 complex, a Keap1 homodimer recognizes both the DLG and ETGE motif within one Nrf2 molecule (Tong et al., 2006). We have demonstrated a functional RhoA-targeting Cul3/BACURD ubiquitin ligase complex. However, it remains to be determined whether homo- or heterodimerization of BACURDs is required for assembly of a fully functional Cul3/BACURD complex, as dimerization is a common property for many BTB domains (Stogios et al., 2005). Indeed, KLHL9 and KLHL13 that mediate ubiquitination of Aurora B stay in

a single complex (Sumara et al., 2007). One useful experiment is to generate dimerization-deficient mutants of BACURDs and test whether they could rescue loss-of-BACURD-induced actin stress fiber phenotypes in cultured cells or developmental defects in Xenopus embryos. Protein ubiquitination is subjected to fine regulation to achieve timely destruction of the substrate. RhoA degradation by the Cul3/BACURD complex must be regulated, given that temporal and spatial control of RhoA activity is crucial for RhoA-regulated biological processes. One possible strategy is through dimerization of the BACURD protein, if dimerization occurs in a regulated manner. Alternatively, posttranslational modifications of RhoA could offer a potential layer of regulation. Moreover, RhoA might exist in a complex with other proteins, and the complex provides multiple substrate recognition surfaces required for maximal

Molecular Cell 35, 841–855, September 25, 2009 ª2009 Elsevier Inc. 851

Molecular Cell Ubiquitination of RhoA by the Cul3/BACURD Complex

recognition and ubiquitination of RhoA by the Cul3/BACURD complex. This hypothesis, if true, could also help to explain why other closely related Rho GTPases, such as RhoB and RhoC, are not targeted by the Cul3/BACURD complex. Differential Functions of Different BACURD Family Members in Vertebrates Insects have a single BACURD protein, while vertebrates have evolved multiple highly homologous BACURDs. Statistics analysis indicates that loss of xBACURDb gives more severe gastrulation phenotypes than that of xBACURDa in Xenopus embryos (Table S1). This suggests that xBACURDb plays a predominant role in regulating RhoA function during gastrulation, and xBACURDa might regulate other aspects of RhoA function in Xenopus. hBACURD1 and hBACURD2 have comparable binding affinities for Cul3 and RhoA, and both form active ubiquitin ligase complexes toward RhoA in vitro. RNAi knockdown of each of them in HeLa cells causes elevated levels of RhoA and actin stress fiber formation, suggesting that their functions are not completely redundant. Thus, each BACURD in the same cell might control levels of RhoA in a specific population and/or regulate a distinct RhoA-controlled signaling pathway. hBACURD2 was originally identified as one of the tumor necrosis factor (TNF)-induced genes in human endothelial cells (Wolf et al., 1992). Microarray data mined from the Genomics Institute of the Novartis Research Foundation (http://biogps.gnf.org) suggest that BACURD2 are ubiquitously expressed in various human and mouse tissues, while transcripts of BACURD1 are more enriched in oocytes, germ cells, and some brain tissues. hBACURD3 has higher expressions in lymphoid tissues, lung, smooth muscle, and thyroid, while mouse BACURD3 transcripts are relatively concentrated in ovary, placenta, umbilical cord, and lung. The different expression patterns of BACURDs also indicate their possible tissue-specific functions of regulating RhoA activity under different biological contexts. The Role of Cul3-Based Ubiquitin Ligase Complexes in Mitosis and Early Embryonic Development In addition to controlling RhoA protein level, Cul3 is also essential for a normal mitotic progression (Sumara et al., 2007). A subpopulation of RhoA is localized at the equatorial cell cortex at the site of the nascent cleavage furrow, and its activation plays a key role in contractile-ring and cleavage-furrow formation during cytokinesis (Piekny et al., 2005). Thus, the Cul3-BACURD pathway might be involved in regulation of RhoA during cytokinesis. The Cul3 RNAi cells proliferate much more slowly than the control cells (data not shown). Deficiencies in the RhoAtargeting Cul3/BACURD complex and the Aurora B-targeting Cul3/KLHL9/13 complex could both contribute to cell growth defects, particularly if BACURDs are involved in regulating RhoA during cytokinesis. Cul3 is essential for embryonic development in all the tested genetic models. We demonstrate here that the Cul3-BACURDRhoA pathway plays a critical role in Xenopus embryo gastrulation. Knockout of Cul3 in mice results in embryonic lethality prior to 7.5 days of gestation, and the embryos show multiple defects in both embryonic and extraembryonic compartments, including completely disorganized extraembryonic tissues and abnormal

trophectoderm (Singer et al., 1999). Increased levels of free cyclin E were observed in the ectoplacental cone and extraembryonic ectoderm of Cul3 / mouse embryos, which leads to a proposal that Cul3 is involved in controlling cyclin E level during G1/S transition, but the required BTB adaptor remains elusive. In Cul3 RNAi HeLa cells, we were unable to detect any changes of endogenous cyclin E level, and flow cytometry analysis revealed no defects in G1/S transition (data not shown). It is unlikely that the early embryonic lethality of Cul3-deficient mice is solely attributed to defective degradations of a subpopulation of cyclin E. Instead, recent studies in Cul3 conditional knockout mice suggest that turnover of cyclin E by Cul3 is involved in maintaining cell quiescence (McEvoy et al., 2007). The embryonic lethality caused by Cul3 knockout likely results from deficiencies of multiple Cul3-mediated degradation pathways involving different BTB adaptors. As gastrulation is also abnormal in Cul3-deficient embryos (Singer et al., 1999), malfunction of RhoA could be a major cause of the abnormal development observed with Cul3deficient mouse embryos. EXPERIMENTAL PROCEDURES Plasmids, Antibodies, and Reagents Plasmids, antibodies, and reagents are described in the Supplemental Data. Cell Culture, Transfection, and Generation of Stable Cell Lines 293T and HeLa cells (American Type Culture Collection [ATCC]; Manassas, VA) were grown in Dulbecco’s modified Eagle’s medium (Hyclone; Logan, UT) containing 10% FBS and 2 mM L-glutamine at 37 C in a 5% CO2 incubator. D. melanogaster S2 cells were maintained in Schneider’s Drosophila medium supplemented with 10% heat-inactivated FBS and penicillin/streptomycin at 26 C. Lipofectamine 2000 (Invitrogen) was used to transfect plasmids and siRNAs into 293T and HeLa cells. Detailed protocols for RNAi in S2 cells are described in the Supplemental Data. To generate stable cell lines, retroviruses expressing hCul3 shRNA 1# in the pMKO.1 vector and Flag-hBACURD1 (-hBACURD2) in the pBabe-puro vector were packaged from Phoenix GP cells (ATCC 3514) by cotransfection of VSV-G plasmid (Clontech; Palo Alto, CA). The retrovirus in the supernatant collected 48 hr after transfection was used to infect HeLa cells. Selection with 0.8 mg/ml puromycin was started 48 hr after infection. Ten days later, individual colonies were lifted and tested by immunoblotting analysis with appropriate antibodies. For Flag-hBACURD1 and -hBACURD2 stable cell lines, clones with an expression level comparable with that of the endogenous protein were selected. Immunoprecipitation, GST Pull-Down Assay, and Western Blotting Detailed procedures are described in the Supplemental Data. Fluorescence Microscopy and qRT-PCR HeLa cells cultured on coverslips were fixed with 4% paraformaldehyde (Electron Microscopy Sciences; Hatfield, PA) for 10 min at room temperature (RT), washed with PBS, and then permeabilized for 10 min in PBS containing 0.5% Triton. Drosophila S2 cells were transferred to the poly-L-lysine-treated coverslips before fixation with 4% paraformaldehyde and permeabilization with 0.5% Triton. Cells were blocked in PBS containing 1% BSA and stained with rhodamine-phalloidin (Invitrogen) for 2 hr at RT to visualize the filamentous actin or with RhoA or Rho1 antibody diluted in PBS containing 1% BSA (1:100). Alexa Fluor 488- or 546-labeled secondary antibodies (Invitrogen) were used. Images were acquired using a Zeiss SM 510 META laser scanning microscope. For real-time PCR analysis, total RNA were extracted by TRIzol (Invitrogen) and digested with DNase I (Invitrogen). One microgram of total RNA was reverse-transcribed into cDNA in the presence or absence of the transcriptase SuperScript II (Invitrogen). qRT-PCR analysis was performed using the Power SYBR Green PCR Master Mix (Applied Biosystems; Foster City, CA) on

852 Molecular Cell 35, 841–855, September 25, 2009 ª2009 Elsevier Inc.

Molecular Cell Ubiquitination of RhoA by the Cul3/BACURD Complex

Applied Biosystems 7500 Real-Time PCR System. The expression level of target genes was normalized to that of GAPDH. Purification of Recombinant Proteins Detailed procedures for purification of recombinant proteins are described in the Supplemental Data. In Vitro Binding and RhoA Nucleotide Loading Two micrograms of MBP-hBACURD1 or MBP-hBACURD2 immobilized on amylose beads were incubated with 10 mg of GST or GST-RhoA (WT or mutants) in the binding buffer (50 mM Tris-HCl [pH 7.6], 150 mM NaCl, 5 mM MgCl2, 0.5% NP-40, and 0.5 mg/ml BSA) at 4 C for 1 hr. The beads were washed five times with the binding buffer, and bound proteins recovered in the SDS loading buffer were subjected to immunoblotting analysis. To prepare nucleotide-free RhoA, purified RhoA (WT) immobilized on beads was incubated in a buffer containing 50 mM Tris-HCl (pH 7.6) and 20 mM EDTA for 30 min at RT and then washed three times with the buffer containing 50 mM Tris-HCl (pH 7.6), 50 mM NaCl, and 1 mM DTT. To obtain GDP- or GTPgSbound RhoA, 10 mg of nucleotide-free RhoA on beads was loaded with 250 mM GDP or 25 mM GTPgS in the buffer containing 50 mM Tris-HCl (pH 7.6), 50 mM NaCl, and 10 mM MgCl2 for 30 min at 30 C. For in vitro binding, 10 mg of GST, nucleotide-free GST-RhoA, or nucleotide-loaded GST-RhoA bound on glutathione beads was incubated with 1 mg of MBPhBACURD1 or MBP-hBACURD2 in the binding buffer (50 mM Tris-HCl [pH 7.6], 50 mM NaCl, 10 mM MgCl2, and 0.5% NP-40) at 4 C for 2 hr. The beads were washed four times with the binding buffer and analyzed by SDS-PAGE and immunoblotting. For in vitro ubiquitination assay, nucleotide-loaded His-33Flag-RhoA or the nucleotide-free form was used. In Vivo and In Vitro Ubiquitination Assays HeLa cells transfected with scramble siRNA or hCul3 siRNAs for 60 hr were treated with 25 mM MG132 for 4 hr. Cells were lysed in 1% SDS buffer (50 mM Tris-HCl [pH 7.6], 150 mM NaCl, 1% SDS, 1% sodium deoxycholate, and 1% NP-40) and boiled for 30 min. The lysates were centrifuged and diluted by 10-fold with a buffer containing 50 mM Tris-HCl (pH 7.6) and 150 mM NaCl. The diluted lysates were precleared by protein G Sepharose for 4 hr at 4 C and subjected to overnight immunoprecipitation using either RhoA antibody (sc-418) or Cdc42 antibody (sc-8401) (as a control). The immunoprecipitates were washed five times with the wash buffer (50 mM Tris-HCl [pH 7.6], 150 mM NaCl, 0.1% SDS, 0.1% sodium deoxycholate, and 0.5% NP-40), eluted with 200 mM DTT-supplemented SDS loading buffer, and analyzed by immunoblotting. For in vitro ubiquitination, 200 ng of GST/Cul3 or GST/Cul3/BTB complexes purified from insect cells were mixed with 60 ng of E1 (BIOMOL International; Plymouth Meeting, PA), 250 ng of UbcH5c (BIOMOL International), 12 mg of Ubiquitin (Boston Biochem; Cambridge, MA), and 300 ng of RhoA (or dRho1) in the reaction buffer containing 50 mM Tris (pH 7.4), 5 mM MgCl2, 0.5 mM DTT, 2 mM NaF, and 4 mM ATP. Following incubation at 30 C for 1 hr, reactions were subjected to Flag immunoprecipitation under denaturing conditions as described (Shiio and Eisenman, 2003). Reactions were terminated by adding equal volume of 2% SDS buffer (50 mM Tris [pH 7.6], 150 mM NaCl, 2% SDS, 1% sodium deoxycholate, and 1% NP-40) and boiled for 30 min. Samples were further diluted by 1-fold with the buffer containing 50 mM Tris-HCl (pH 7.6) and 150 mM NaCl and subjected to Flag immunoprecipitation overnight after preclearing for 2 hr using protein G Sepharose. The beads were washed five times with the wash buffer (50 mM Tris-HCl [pH 7.6], 150 mM NaCl, 0.5% SDS, 0.5% sodium deoxycholate, and 0.5% NP-40) and eluted with 200 mM DTT-supplemented SDS loading buffer. The eluted proteins were analyzed by SDS-PAGE and immunoblotting. Transwell Boyden Chamber Assay and Monolayer Wounding Assay The Transwell migration assays were performed as previously described (Wang et al., 2003). Briefly, cells (2 3 104/well) were plated on the Transwell filters with 8 mm pores (Corning; Lowell, MA) coated with 10 mg/ml fibronectin. Eight hours (for MEF) or twenty-four hours (for HeLa) later, filters were fixed with 4% formaldehyde and stained with 0.4% crystal violet in 10% ethanol. Nonmigrating cells on the upper side of the filters were removed with a cotton

swab, and cells on the underside of the filters were photographed. To quantify cell motility, cells stained positively with crystal violet in nine random fields of each filter were counted, and three independent filters were analyzed. In the wound-healing assay, confluent HeLa cells were scratched by a plastic pipette and washed three times with PBS. Images were taken at 0 or 24 hr after wounding, and the wound closure was calculated 24 hr later. Xenopus Embryo Manipulations and Assays Xenopus assays are described in the Supplemental Data.

SUPPLEMENTAL DATA Supplemental Data include Supplemental Experimental Procedures, Supplemental References, three tables, and 18 figures and can be found online at http://www.cell.com/molecular-cell/supplemental/S1097-2765(09)00632-7. ACKNOWLEDGMENTS We thank Ning Zheng for providing the Cul3 and Rbx1 baculoviruses and Caroline Worby for suggestions on S2 cell RNAi experiments. We are grateful to Xiaojuan Shi and the NIBS antibody facility for assistance in generating antibodies and to Chen Zhan, Liqin Fu, and the NIBS imaging facility for their help on use of confocal microscopes. We thank Fenghe Du and Xiaodong Wang for obtaining reagents and Xiaodong Wang for reading this manuscript. We also thank members of the Shao lab for helpful discussions and technical assistance. This work was supported by the Chinese Ministry of Science and Technology 863 Grant (2005AA210950) and 973 National Basic Research Plan of China (2006CB806502) to F.S. Received: January 26, 2009 Revised: May 3, 2009 Accepted: June 29, 2009 Published: September 24, 2009 REFERENCES Angers, S., Thorpe, C.J., Biechele, T.L., Goldenberg, S.J., Zheng, N., MacCoss, M.J., and Moon, R.T. (2006). The KLHL12-Cullin-3 ubiquitin ligase negatively regulates the Wnt-beta-catenin pathway by targeting Dishevelled for degradation. Nat. Cell Biol. 8, 348–357. Cardozo, T., and Pagano, M. (2004). The SCF ubiquitin ligase: insights into a molecular machine. Nat. Rev. Mol. Cell Biol. 5, 739–751. Choi, S.C., and Han, J.K. (2002). Xenopus Cdc42 regulates convergent extension movements during gastrulation through Wnt/Ca2+ signaling pathway. Dev. Biol. 244, 342–357. Cullinan, S.B., Gordan, J.D., Jin, J., Harper, J.W., and Diehl, J.A. (2004). The Keap1-BTB protein is an adaptor that bridges Nrf2 to a Cul3-based E3 ligase: oxidative stress sensing by a Cul3-Keap1 ligase. Mol. Cell. Biol. 24, 8477–8486. Deshaies, R.J. (1999). SCF and Cullin/Ring H2-based ubiquitin ligases. Annu. Rev. Cell Dev. Biol. 15, 435–467. Djiane, A., Riou, J., Umbhauer, M., Boucaut, J., and Shi, D. (2000). Role of frizzled 7 in the regulation of convergent extension movements during gastrulation in Xenopus laevis. Development 127, 3091–3100. Doye, A., Mettouchi, A., Bossis, G., Clement, R., Buisson-Touati, C., Flatau, G., Gagnoux, L., Piechaczyk, M., Boquet, P., and Lemichez, E. (2002). CNF1 exploits the ubiquitin-proteasome machinery to restrict Rho GTPase activation for bacterial host cell invasion. Cell 111, 553–564. Figueroa, P., Gusmaroli, G., Serino, G., Habashi, J., Ma, L., Shen, Y., Feng, S., Bostick, M., Callis, J., Hellmann, H., and Deng, X.W. (2005). Arabidopsis has two redundant Cullin3 proteins that are essential for embryo development and that interact with RBX1 and BTB proteins to form multisubunit E3 ubiquitin ligase complexes in vivo. Plant Cell 17, 1180–1195.

Molecular Cell 35, 841–855, September 25, 2009 ª2009 Elsevier Inc. 853

Molecular Cell Ubiquitination of RhoA by the Cul3/BACURD Complex

Furukawa, M., and Xiong, Y. (2005). BTB protein Keap1 targets antioxidant transcription factor Nrf2 for ubiquitination by the Cullin 3-Roc1 ligase. Mol. Cell. Biol. 25, 162–171. Furukawa, M., He, Y.J., Borchers, C., and Xiong, Y. (2003). Targeting of protein ubiquitination by BTB-Cullin 3-Roc1 ubiquitin ligases. Nat. Cell Biol. 5, 1001–1007. Geyer, R., Wee, S., Anderson, S., Yates, J., and Wolf, D.A. (2003). BTB/POZ domain proteins are putative substrate adaptors for cullin 3 ubiquitin ligases. Mol. Cell 12, 783–790. Hall, A. (1998). Rho GTPases and the actin cytoskeleton. Science 279, 509–514. Hall, A., and Nobes, C.D. (2000). Rho GTPases: molecular switches that control the organization and dynamics of the actin cytoskeleton. Philos. Trans. R. Soc. Lond. B Biol. Sci. 355, 965–970. Hershko, A., and Ciechanover, A. (1998). The ubiquitin system. Annu. Rev. Biochem. 67, 425–479.

Ren, R., Nagel, M., Tahinci, E., Winklbauer, R., and Symes, K. (2006). Migrating anterior mesoderm cells and intercalating trunk mesoderm cells have distinct responses to Rho and Rac during Xenopus gastrulation. Dev. Dyn. 235, 1090–1099. Ren, X.D., and Schwartz, M.A. (2000). Determination of GTP loading on Rho. Methods Enzymol. 325, 264–272. Ridley, A.J. (2006). Rho GTPases and actin dynamics in membrane protrusions and vesicle trafficking. Trends Cell Biol. 16, 522–529. Settleman, J. (2001). Rac ‘n Rho: the music that shapes a developing embryo. Dev. Cell 1, 321–331. Shao, F., Merritt, P.M., Bao, Z., Innes, R.W., and Dixon, J.E. (2002). A Yersinia effector and a Pseudomonas avirulence protein define a family of cysteine proteases functioning in bacterial pathogenesis. Cell 109, 575–588. Shiio, Y., and Eisenman, R.N. (2003). Histone sumoylation is associated with transcriptional repression. Proc. Natl. Acad. Sci. USA 100, 13225–13230.

Hill, C.S., Wynne, J., and Treisman, R. (1995). The Rho family GTPases RhoA, Rac1, and CDC42Hs regulate transcriptional activation by SRF. Cell 81, 1159–1170.

Shimizu, T., Ihara, K., Maesaki, R., Kuroda, S., Kaibuchi, K., and Hakoshima, T. (2000). An open conformation of switch I revealed by the crystal structure of a Mg2+-free form of RHOA complexed with GDP. Implications for the GDP/ GTP exchange mechanism. J. Biol. Chem. 275, 18311–18317.

Hoffmann, C., Aktories, K., and Schmidt, G. (2007). Change in substrate specificity of cytotoxic necrotizing factor unmasks proteasome-independent down-regulation of constitutively active RhoA. J. Biol. Chem. 282, 10826–10832.

Singer, J.D., Gurian-West, M., Clurman, B., and Roberts, J.M. (1999). Cullin-3 targets cyclin E for ubiquitination and controls S phase in mammalian cells. Genes Dev. 13, 2375–2387.

Jaffe, A.B., and Hall, A. (2005). Rho GTPases: biochemistry and biology. Annu. Rev. Cell Dev. Biol. 21, 247–269. Keller, R. (1991). Early embryonic development of Xenopus laevis. Methods Cell Biol. 36, 61–113. Kerscher, O., Felberbaum, R., and Hochstrasser, M. (2006). Modification of proteins by ubiquitin and ubiquitin-like proteins. Annu. Rev. Cell Dev. Biol. 22, 159–180. Kobayashi, A., Kang, M.I., Okawa, H., Ohtsuji, M., Zenke, Y., Chiba, T., Igarashi, K., and Yamamoto, M. (2004). Oxidative stress sensor Keap1 functions as an adaptor for Cul3-based E3 ligase to regulate proteasomal degradation of Nrf2. Mol. Cell. Biol. 24, 7130–7139. Kolsch, V., Seher, T., Fernandez-Ballester, G.J., Serrano, L., and Leptin, M. (2007). Control of Drosophila gastrulation by apical localization of adherens junctions and RhoGEF2. Science 315, 384–386. Kuhl, M. (2002). Non-canonical Wnt signaling in Xenopus: regulation of axis formation and gastrulation. Semin. Cell Dev. Biol. 13, 243–249. Kurz, T., Pintard, L., Willis, J.H., Hamill, D.R., Gonczy, P., Peter, M., and Bowerman, B. (2002). Cytoskeletal regulation by the Nedd8 ubiquitin-like protein modification pathway. Science 295, 1294–1298. Lanning, C.C., Daddona, J.L., Ruiz-Velasco, R., Shafer, S.H., and Williams, C.L. (2004). The Rac1 C-terminal polybasic region regulates the nuclear localization and protein degradation of Rac1. J. Biol. Chem. 279, 44197–44210. Leung, T., Chen, X.Q., Manser, E., and Lim, L. (1996). The p160 RhoA-binding kinase ROK alpha is a member of a kinase family and is involved in the reorganization of the cytoskeleton. Mol. Cell. Biol. 16, 5313–5327. McEvoy, J.D., Kossatz, U., Malek, N., and Singer, J.D. (2007). Constitutive turnover of cyclin E by Cul3 maintains quiescence. Mol. Cell. Biol. 27, 3651–3666. Mistry, H., Wilson, B.A., Roberts, I.J., O’Kane, C.J., and Skeath, J.B. (2004). Cullin-3 regulates pattern formation, external sensory organ development and cell survival during Drosophila development. Mech. Dev. 121, 1495–1507. Petroski, M.D., and Deshaies, R.J. (2005). Function and regulation of cullinRING ubiquitin ligases. Nat. Rev. Mol. Cell Biol. 6, 9–20. Piekny, A., Werner, M., and Glotzer, M. (2005). Cytokinesis: welcome to the Rho zone. Trends Cell Biol. 15, 651–658. Pintard, L., Willis, J.H., Willems, A., Johnson, J.L., Srayko, M., Kurz, T., Glaser, S., Mains, P.E., Tyers, M., Bowerman, B., and Peter, M. (2003). The BTB protein MEL-26 is a substrate-specific adaptor of the CUL-3 ubiquitin-ligase. Nature 425, 311–316.

Sokol, S.Y. (1996). Analysis of Dishevelled signalling pathways during Xenopus development. Curr. Biol. 6, 1456–1467. Soundararajan, M., Turnbull, A., Fedorov, O., Johansson, C., and Doyle, D.A. (2008). RhoB can adopt a Mg2+ free conformation prior to GEF binding. Proteins 72, 498–505. Stogios, P.J., Downs, G.S., Jauhal, J.J., Nandra, S.K., and Prive, G.G. (2005). Sequence and structural analysis of BTB domain proteins. Genome Biol. 6, R82. Sumara, I., Quadroni, M., Frei, C., Olma, M.H., Sumara, G., Ricci, R., and Peter, M. (2007). A Cul3-based E3 ligase removes Aurora B from mitotic chromosomes, regulating mitotic progression and completion of cytokinesis in human cells. Dev. Cell 12, 887–900. Tahinci, E., and Symes, K. (2003). Distinct functions of Rho and Rac are required for convergent extension during Xenopus gastrulation. Dev. Biol. 259, 318–335. Tong, K.I., Katoh, Y., Kusunoki, H., Itoh, K., Tanaka, T., and Yamamoto, M. (2006). Keap1 recruits Neh2 through binding to ETGE and DLG motifs: characterization of the two-site molecular recognition model. Mol. Cell. Biol. 26, 2887–2900. Uehata, M., Ishizaki, T., Satoh, H., Ono, T., Kawahara, T., Morishita, T., Tamakawa, H., Yamagami, K., Inui, J., Maekawa, M., and Narumiya, S. (1997). Calcium sensitization of smooth muscle mediated by a Rho-associated protein kinase in hypertension. Nature 389, 990–994. Wallingford, J.B., and Harland, R.M. (2001). Xenopus Dishevelled signaling regulates both neural and mesodermal convergent extension: parallel forces elongating the body axis. Development 128, 2581–2592. Wallingford, J.B., Fraser, S.E., and Harland, R.M. (2002). Convergent extension: the molecular control of polarized cell movement during embryonic development. Dev. Cell 2, 695–706. Wang, H.R., Zhang, Y., Ozdamar, B., Ogunjimi, A.A., Alexandrova, E., Thomsen, G.H., and Wrana, J.L. (2003). Regulation of cell polarity and protrusion formation by targeting RhoA for degradation. Science 302, 1775–1779. Watanabe, N., Madaule, P., Reid, T., Ishizaki, T., Watanabe, G., Kakizuka, A., Saito, Y., Nakao, K., Jockusch, B.M., and Narumiya, S. (1997). p140mDia, a mammalian homolog of Drosophila diaphanous, is a target protein for Rho small GTPase and is a ligand for profilin. EMBO J. 16, 3044–3056. Watanabe, N., Kato, T., Fujita, A., Ishizaki, T., and Narumiya, S. (1999). Cooperation between mDia1 and ROCK in Rho-induced actin reorganization. Nat. Cell Biol. 1, 136–143.

854 Molecular Cell 35, 841–855, September 25, 2009 ª2009 Elsevier Inc.

Molecular Cell Ubiquitination of RhoA by the Cul3/BACURD Complex

Wolf, F.W., Marks, R.M., Sarma, V., Byers, M.G., Katz, R.W., Shows, T.B., and Dixit, V.M. (1992). Characterization of a novel tumor necrosis factor-alphainduced endothelial primary response gene. J. Biol. Chem. 267, 1317–1326.

Xu, L., Wei, Y., Reboul, J., Vaglio, P., Shin, T.H., Vidal, M., Elledge, S.J., and Harper, J.W. (2003). BTB proteins are substrate-specific adaptors in an SCF-like modular ubiquitin ligase containing CUL-3. Nature 425, 316–321.

Wunnenberg-Stapleton, K., Blitz, I.L., Hashimoto, C., and Cho, K.W. (1999). Involvement of the small GTPases XRhoA and XRnd1 in cell adhesion and head formation in early Xenopus development. Development 126, 5339–5351.

Ybot-Gonzalez, P., Savery, D., Gerrelli, D., Signore, M., Mitchell, C.E., Faux, C.H., Greene, N.D., and Copp, A.J. (2007). Convergent extension, planarcell-polarity signalling and initiation of mouse neural tube closure. Development 134, 789–799.

Molecular Cell 35, 841–855, September 25, 2009 ª2009 Elsevier Inc. 855