Developmental Biology 346 (2010) 247–257

Contents lists available at ScienceDirect

Developmental Biology

journal homepage: www.elsevier.com/developmentalbiology

Drosophila Cand1 regulates Cullin3-dependent E3 ligases by affecting the neddylation of Cullin3 and by controlling the stability of Cullin3 and adaptor

Song-Hee Kim, Hyung-Jun Kim 1, Seonmi Kim, Jeongbin Yim ⁎

School of Biological Science, Seoul National University, Seoul 151-742, Republic of Korea article info abstract

Article history: -RING ligases (CRLs), which comprise the largest class of E3 ligases, regulate diverse cellular Received for publication 11 February 2010 processes by targeting numerous . Conjugation of the ubiquitin-like protein Nedd8 with Cullin Revised 22 June 2010 activates CRLs. Cullin-associated and neddylation-dissociated 1 (Cand1) is known to negatively regulate CRL Accepted 27 July 2010 activity by sequestering unneddylated Cullin1 (Cul1) in biochemical studies. However, genetic studies of Available online 4 August 2010 Arabidopsis have shown that Cand1 is required for optimal CRL activity. To elucidate the regulation of CRLs by Cand1, we analyzed a Cand1 mutant in Drosophila. Loss of Cand1 causes accumulation of neddylated Cullin3 Keywords: Cullin-associated and neddylation-dissociated 1 (Cul3) and stabilizes the Cul3 adaptor protein HIB. In addition, the Cand1 mutation stimulates protein (Cand1) degradation of Cubitus interruptus (Ci), suggesting that Cul3-RING ligase activity is enhanced by the loss of Cullin-RING ubiquitin ligases Cand1. However, the loss of Cand1 fails to repress the accumulation of Ci in Nedd8AN015 or CSN5null mutant Cullin3 clones. Although Cand1 is able to bind both Cul1 and Cul3, mutation of Cand1 suppresses only the Drosophila accumulation of Cul3 induced by the dAPP-BP1 mutation defective in the neddylation pathway, and this Nedd8 effect is attenuated by inhibition of function. Furthermore, overexpression of Cand1 stabilizes the Cul3 protein when the neddylation pathway is partially suppressed. These data indicate that Cand1 stabilizes unneddylated Cul3 by preventing proteasomal degradation. Here, we propose that binding of Cand1 to unneddylated Cul3 causes a shift in the equilibrium away from the neddylation of Cul3 that is required for the degradation of substrate by CRLs, and protects unneddylated Cul3 from proteasomal degradation. Cand1 regulates Cul3-mediated E3 ligase activity not only by acting on the neddylation of Cul3, but also by controlling the stability of the adaptor protein and unneddylated Cul3. © 2010 Elsevier Inc. All rights reserved.

Introduction APC11 (Hershko and Ciechanover, 1998; Ohta et al., 1999). The various Cullin proteins function as a rigid scaffold for the assembly of this Cullin-RING ubiquitin ligases (CRLs) are a family of multi-subunit modular class of ligase. All associate with a RING protein through ubiquitin ligases with diverse cellular functions, including regulation of their C-terminal domain, whereas the N-terminal region recruits a wide the cell cycle, the DNA damage response, and various transcription variety of receptor proteins that confer substrate specificity. CRL activity factors (Ciechanover, 1998; Kerscher et al., 2006; Petroski and Deshaies, can be regulated by numerous mechanisms, such as the turnover of 2005; Willems et al., 2004). The mammalian Cullin family contains six substrate receptors and the reversible attachment of the ubiquitin-like closely related proteins (Cul1, Cul2, Cul3, Cul4A, Cul4B and Cul5) and protein Nedd8 to Cullins (neddylation) (Hori et al., 1999; Pan et al., three distantly related proteins (Cul7, PARC and APC2) (Dias et al., 2002; 2004). The neddylation of Cullin proteins involves a ubiquitin-like Hori et al., 1999; Kamura et al., 2004; Osaka et al., 1998; Pintard et al., process utilizing the Nedd8-specific E1, APP-BP1-Uba3 heterodimeric 2004; Shiyanov et al., 1999; Skaar et al., 2007; Wertz et al., 2004; Xu enzyme, and the E2 enzyme, Ubc12 (Gong and Yeh, 1999; Liakopoulos et al., 2003). All Cullins contain a conserved carboxy-terminal domain of et al., 1998; Rabut and Peter, 2008; Yeh et al., 2000). Cullin neddylation approximately 100 amino acids that binds to the small RING finger is essential for the function of Cullin-containing ubiquitin E3 ligases proteins ROC1 (RING of Cullins, also known as Hrt1 and Rbx1), ROC2 or (Bornstein et al., 2006; del Pozo et al., 2002; Furukawa et al., 2000; Lammer et al., 1998; Ohh et al., 2002; Podust et al., 2000; Read et al.,

Abbreviations: Cand1, Cullin-associated and neddylation-dissociated 1; Ci, Cubitus 2000; Wu et al., 2000). It can be reversed by the COP9 signalosome interruptus; CRLs, Cullin-RING ubiquitin ligases; CSN, COP9 signalosome; Cul3, Cullin3; (CSN), a highly conserved protein complex that consists of eight Da-GAL4, daughterless-GAL4; DTS, dominant temperature-sensitive; HIB, hedgehog- subunits (CSN1–8) (Cope and Deshaies, 2003; Doronkin et al., 2003; induced BTB protein; rdx, roadkill; SCF, Skp1/Cullin1/F-box proteins. Stuttmann et al., 2009), and deneddylation of Cullin is mediated by the ⁎ Corresponding author. Fax: +82 2 871 4315. MPN/JAMM motif of the CSN5 subunit (Cope et al., 2002). E-mail address: [email protected] (J. Yim). 1 Present address: Department of Biology, University of Pennsylvania, Howard Cand1 has been reported to bind selectively to unneddylated Hughes Medical Institute, Philadelphia, Pennsylvania, USA. Cullin1 (Goldenberg et al., 2004; Hwang et al., 2003; Liu et al., 2002;

0012-1606/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.ydbio.2010.07.031 248 S.-H. Kim et al. / Developmental Biology 346 (2010) 247–257

Zheng et al., 2002) and sequester Cullins (Goldenberg et al., 2004; Lo at room temperature for 1.5 h. After being washed three times with and Hannink, 2006). These biochemical studies suggest that Cand1 the wash solution, the larval discs were mounted in an antifade acts as a negative regulator of CRLs. In mammalian cell lines, Cand1 solution. The primary antibodies were mouse anti-GFP (1:200; regulates the formation of the SCF (Skp1/ Cullin1/F-box proteins) Sigma), rabbit anti-GFP (1:500; Abcam), rat anti-CiFL (1:2; 2A1, complex, and its dissociation from Cul1 is coupled to the incorpora- Iowa Hybridoma Bank), mouse anti-Myc (1:200; Roche), and rabbit tion of F-box proteins into the SCF complex (Liu et al., 2002; Zheng anti-β-gal (1:1000; Sigma). Secondary antibodies coupled to the et al., 2002). However, the Arabidopsis Cand1 null mutant displays appropriate fluorophore (FITC, Cy2, or Cy3; Jackson ImmunoResearch defects in growth, development and auxin response, indicating that Laboratories) were diluted to 1:200 in the blocking solution. All SCF ligase activity is compromised in the mutant (Chuang et al., 2004; confocal images were taken on a DE/LSM510 NLO (Carl Zeiss). Feng et al., 2004; Stuttmann et al., 2009). These data indicate that Cand1 is required for efficient CRL function in Arabidopsis. Further- more, Cand1 may participate in Cul3-based BCR3 complex (BTB DNA constructs for germline transformation domain-containing protein, Cul3 and a RING protein) cycling, given its competition with the BTB protein Keap1, the substrate receptor of Wild-type Cand1 cDNAs (RE54013 and RE15044 from the Drosophila BCR3, for binding to Cul3 (Lo and Hannink, 2006; Wu et al., 2006). Genome Resource Center) were cloned into the pUAST vector and These results indicate that Cand1 has contradictory effects on the pUAST-HA vector as KpnI–XbaI fragments. P-element-mediated germ- regulation of CRL activity. To explain these paradoxical results, it has line transformation was performed according to the methods described been suggested that CSN and Cand1 play critical roles through previously (Rubin and Spradling, 1982). The parental strain for all deneddylation and neddylation of Cullins, which could induce the germline transformations was w1118. Flies bearing autosomal transgenes cycle of assembly and disassembly of CRL complexes (Cope and were identified and used for all analyses. Deshaies, 2003; Lo and Hannink, 2006; Zhang et al., 2008). It is obvious that Cand1 has an important function in modulating the CRL activity, but despite much progress, the regulation of CRL activation Fly strains still remains unclear. To understand the mechanism of regulation of CRLs by Cand1, we All stocks were maintained and raised under standard conditions generated Cand1 mutant in Drosophila. We found that loss of Cand1 unless otherwise specified. w1118 was used as a control strain. w;; leads to accumulations of neddylated Cul3 and Cul3 adaptor protein CSN5null/TM3 Ser act-GFP was a gift from D. Segal (Oron et al., 2002) HIB/rdx, and stimulates protein degradation of CiFL by upregulating and tub-myc-slimb was from J. Jiang (Ko et al., 2002). The mutants Cul3-based E3 ligase activity. We also present evidence that Cand1 CSN5null FRT82B, Nedd8AN015 FRT40A, UAS-HIB-RNAi, pUAS-Flag-HIB and stabilizes unneddylated Cul3 protein by preventing proteasomal Ci-lacZ have been previously described (Doronkin et al., 2002; Ou degradation. Our results support a role for Cand1 in fine-tuning et al., 2002; Zhang et al., 2006). pUAST-HA-HIB/rdx (CG9924-RB Cul3-mediated E3 ligase activity by controlling the equilibrium state isoform cloned into the pUAST vector as an EcoRI–XhoI fragment) was of Cul3 neddylation and the stability of adaptor protein, and by a kind gift from H. Suh of our lab. The Bloomington Stock Center protecting unneddylated Cul3 from proteasomal degradation. provided other fly stocks.

Material and methods Drosophila genetics and FLP/FRT mosaic clonal analysis Western blotting The Cand116 mutants were created by imprecise excision of P- Larval brain and eye imaginal discs were collected in 2× SDS element G2412 (Genexel Inc.). Breakpoints were mapped by genomic sample buffer and boiled for 7 min. The extracts were then run on a PCR and sequenced. The sequence of the deletion junction of Cand116 7.5% or 8% Tris–glycine gel. Western blotting was performed mutant flies is AACTGGCCGCGGTGAGCGGG/TGTCTGCGGCAG- according to the standard protocol. The primary antibodies used CACCTGTA, with the slash indicating the deletion breakpoint. The were rabbit anti-Cul1 (1:500; Zymed), mouse anti-Cul3 (1:1000; BD sequence of the deletion junction of Cand1100 flies is AATATTTAAATAA- Biosciences), mouse anti-β-tubulin (1:2000; E7, Iowa Hybridoma TATTATT/ATGCTAAACCAATCAATAAA. Clones of mutant cells were Bank), rat anti-HA (1:1000; Roche), and mouse anti-Myc (1:1000; generated by FLP/FRT-mediated mitotic recombination as previously Roche). After incubation with HRP-coupled secondary antibodies described (Jiang and Struhl, 1995). Mitotic recombination clones for the (goat anti-rat diluted 1:2000, Jackson ImmunoResearch Laboratories; wing disc were induced 48±3 h after egg laying (AEL) in staged larvae donkey anti-mouse diluted 1:2000, Jackson ImmunoResearch Labo- by heat shock at 37 °C for 80 min. The larval genotype was yw P{ey-FLP}; ratories; and goat anti-rabbit diluted 1:2000, Pierce), the blots were P{Ubi-GFP} FRT40A/Cand116 FRT40A. The discs were dissected at 110±3 h visualized using a chemiluminescent detection kit (ECL PLUS, AEL, fixed in fixation solution, and stained with anti-GFP antibody to Amersham Biosciences). The images were processed on a LAS-3000 mark the clones. The following flies were used to generate clones in the (Fujifilm) and quantified using Multi Gauge software (V3.0, Fujifilm). eye disc: yw P{ey-FLP};P{ubi-GFP} FRT40A/Cand116 FRT40A and yw P{ey- FLP};P{ubi-GFP}FRT40A/Cand116,Nedd8AN015 FRT40A. To generate Cand116 Immunohistochemistry of larval discs or CSN5null mutant clones expressing myc-slimb in the eye discs, the following flies were used: yw P{ey-FLP};P{ubi-GFP} FRT40A/Cand116 The staged larvae were dissected in 1× PBS. Dissected imaginal FRT40A;tub-myc-slimb/+ and yw P{ey-FLP};tub-myc-slimb/+;P{ubi- 82B null 82B 16 discs were incubated in a fixation solution (0.1 M PIPES pH 6.9, 1 mM GFP} FRT /CSN5 FRT . To generate Cand1 mutant clones in Ci- 40A 16 EGTA, 1.0% Triton X-100, 2 mM MgSO4, 1% EM grade formaldehyde) lacZ expression in the eye discs, yw P{ey-FLP};P{ubi-GFP} FRT /Cand1 for 30 min. After being washed three times with a washing solution FRT40A;;Ci-lacZ was used. To generate CSN5 null mutant clones in the 16 82B (50 mM Tris–Cl pH 6.8, 150 mM NaCl, 0.1% Triton X-100, and 1 mg/ml Cand1 mutant background, yw P{ey-FLP};Cand1 ;P{ubi-GFP} FRT / null 82B 16 BSA), the discs were incubated in a blocking solution (50 mM Tris–Cl CSN5 FRT was used. To express Cand1 in the Cand1 mutant 16 16 pH 6.8, 150 mM NaCl, 0.1% Triton X-100, and 10 mg/ml BSA) at 4 °C for background, w;Cand1 /UAS-HA-Cand1,Cand1 ;hs-GAL4/+ was heat 2 hr. The discs were stained overnight at 4 °C with appropriate shocked at 37 °C for 1 h. To repress proteasome function by expressing antibodies in the blocking solution. After being washed four times in UAS-Prosβ21 (from the Bloomington Stock Center) with the Da-GAL4 the wash solution, the discs were incubated with secondary antibody driver, larvae are raised at the restrictive temperature (29 °C). S.-H. Kim et al. / Developmental Biology 346 (2010) 247–257 249

RT-PCR and quantitative RT-PCR expressing the Cand1 protein (Supplemental Table S1). Because the deleted region of Cand116 mutant flies contains the CG5045 and Cand1 For RT-PCR, total RNA was prepared using TRIzol (Invitrogen) and (CG5366) , we generated Cand1100 flies that have a single reverse transcribed using M-MLV reverse transcriptase (Promega). PCR deletion of CG5045. Cand1100 flies survive to the adult stage and show was performed using Cand1-specific primers with the forward sequence normal mRNA levels of Cand1 (Fig. 1A and Supplemental Fig. S1A). being 5′ GACATGCTGCAGAATGAGCT 3′ and the reverse sequence being The temporal expression pattern of Cand1, examined using quantita- 5′ TGGATGGCACCGAGGGTAG 3′, to obtain a product of 1042 bp. The 5′- tive RT-PCR, showed expression at all developmental stages (Fig. 1C). undeleted region of Cand116 mutant was amplified using primers with the forward sequence being 5′ GGACAAGGACTTCCGTTTC 3′ and the Loss of Cand1 affects neddylation of Cul3 and Ci protein stability reverse sequence being 5′ TGGCCGCATTCGGTTCTAA 3′.TheGAP-DH control was amplified with the forward primer 5′-GTCAACGATCCCTT- Previously, it was reported that Cand1 binds selectively to CATCGA-3′ and the reverse primer 5′-TGTACGATAGTTTTGGCTAG-3′. unneddylated Cullin1 (Cul1) and forms a large compact complex For quantitative RT-PCR, total RNA was isolated from each develop- with the SCF catalytic core, Cul1 and Roc1 (Goldenberg et al., 2004; Liu mental stage of w1118. First-strand cDNA was synthesized from 3 μgof et al., 2002). To investigate how Cand1 affects the Cullin protein in total RNA with a total cycle number of 29. rp49 primers were used for the Drosophila,wefirst examined an immunoblot of Cullin protein in control reaction. Cand116 mutant larvae. Interestingly, the ratio of Nedd8 modified to unmodified Cul3 is markedly increased in the Cand1 mutant (Figs. 2B GST pull-down analysis and D) and the total amount of Cul3 protein is slightly decreased (Fig. 2F), whereas the ratio of Nedd8 modified to unmodified Cul1 [35S] labeled Cul1 and Cul3 were synthesized in vitro using the protein is not affected (Figs. 2A and C) with little increase in the pcDNA 3.1-Cul1, -Cul3 and TNT Quick-Coupled Transcription/Trans- relative Cul1 protein levels by the loss of Cand1 (Fig. 2E). Cand1100 flies lation system (Promega), according to the manufacturer's protocol. did not show this defect of Cul3 protein (Supplemental Fig. S1B), and BL21 cells containing GST and GST–Cand1 expression plasmids (pGEX the expression of Cand1 driven by ubiquitous daughterless GAL4 (Da- 6P-1) were grown in LB medium. The expression of fusion proteins GAL4) can fully rescue the reduced protein level of Cul3 and the ratio was induced by treating the cells with 0.1 mM isopropyl-1-thio-β-D- of neddylated to unneddylated Cul3 induced by the loss of Cand1 (lane galactopyranoside for 16 h at 18 °C. The GST fusion protein was 4inFig. 2B). These results indicate that the increase of neddylated purified by the batch GST Purification Module (Amersham Bios- Cul3 is due to the specific loss of the Cand1 . When we expressed ciences). GST–Cand1 fusion protein and GST were isolated using HA-tagged Cand1 in the Cand116 mutant background, the ratio of glutathione Sepharose 4 fast flow beads. The immobilized GST and neddylated to unneddylated Cul3 was decreased depending on the GST–Cand1 protein beads were washed with phosphate-buffered level of Cand1 expression (Supplemental Figs. S1C and D). The level of saline and incubated with [35S] labeled Cul1 or Cul3 at 4 °C for 12 h in unneddylated Cul3 was maximized at the highest level of Cand1 binding buffer (50 mM HEPES pH 7.8, 5 mM EDTA, 250 mM NaCl, 5% expression (7.5 h after heat shock) and was again lowered with glycerol, 0.1% Triton X-100, 1 mM DTT, and protease inhibitor reduced Cand1 (3.5 h and 12 h after heat shock). The expression of cocktail). The beads were washed three times with the binding buffer Cand1 is weak under the control of heat shock-GAL4 (hs-GAL4) and the and subjected to SDS-polyacrylamide gel electrophoresis and lower transgene expression levels can only partially rescue the defect autoradiography. of the Cand1 mutant. These results show that Cand1 is involved in For competition GST pull-down, His-tagged HIB/rdx protein (pET suppression of Cul3 neddylation. In CSN5null mutants that lack Cullin 15b) was purified with His·Bind Resin (Novagen) according to the deneddylation activity, only the neddylated forms of Cullins are manufacturer's instructions. GST–Cand1 bound to glutathione beads accumulated, while the total amounts of Cullins are significantly was mixed with in vitro translated Cul3 and purified His-HIB protein decreased (lane 3 in Figs. 2 A–F; these data are discussed in Fig. 4). in the binding buffer and incubated overnight at 4 °C. The reaction Previous studies have reported contradictory effects of Cand1 on CRL mixtures were analyzed by SDS-PAGE and western assay with anti- activity. In vitro studies have suggested that Cand1 negatively regulate Cul3 and anti-His (27E8, Cell Signaling) antibodies. CRL activity by binding to Cul1 and dissociating the CRL complex For immunoprecipitation, His-tagged HIB/rdx and in vitro trans- (Hwang et al., 2003; Liu et al., 2002; Zheng et al., 2002), whereas loss-of- lated Cul3 were incubated with anti-His antibody (27E8, Cell function alleles of Arabidopsis Cand1 have shown a reduced auxin Signaling) and protein G Sepharose 4 Fast Flow (Amersham response, indicating that Cand1 is required for optimal CRL activity Biosciences). The immunocomplexes were electrophoresed by SDS- (Chuang et al., 2004; Feng et al., 2004). To determine how loss of Cand1 PAGE and detected using the anti-Cul3 antibody. affects CRL activity in vivo Drosophila model system, we examined the protein stability of a substrate of Cullin-RING ligase in the developing Results eye disc of Drosophila.IntheeyediscofDrosophila, Cubitus interruptus (Ci/Gli) degradation is regulated by two distinct Cullin-RING ligases. CG5366 encodes the Drosophila ortholog of Cand1 The full length Ci protein (CiFL) accumulated in morphogenetic furrow (MF) in response to hedgehog signaling, but not in anterior and The Drosophila melanogaster ortholog of the human Cand1 is posterior cells. In anterior cells, CiFL is degraded by the Cul1-based encoded by CG5366 (FlyBase ID; FBgn0027568) and was named SCFslimb complex, whereas, in posterior cells behind morphogenetic Cand1. The G2412 fly line with an EP-element inserted into a genomic furrows (MF), CiFL is depleted by a Cul3-dependent pathway (Ou et al., region near the Cand1 locus was purchased from Genexel, Inc. 2002; Smelkinson and Kalderon, 2006). To circumvent pupae lethality (Fig. 1A). To investigate the function of Cand1 in vivo, we generated of the Cand1 mutant, we used the directed mosaic FLP/FRT system two deletion mutant lines in the Cand1 gene via imprecise excision of (Theodosiou and Xu, 1998). FLP (Flipase)-mediated recombination can the P-element G2412. We identified a deletion allele, designated be used to generate mitotic clones by creating flies with transgenic FRT Cand116, which lacks exon 6 to the 3′ end of Cand1 (Fig. 1A). The (FLP recombination target) sites at identical positions on homologous Cand116 mutant is a loss-of-function allele of Cand1. No expression of . As a result of recombination, homologous Cand1 mRNA was detected in the Cand116 mutants by RT-PCR with arms harboring allelic FRT sites can be exchanged, generating one the primer set1, although the 5′-undeleted region of Cand1 was still mutant daughter cell and one wild-type daughter cell (“twin spot cell”). transcribed (primer set 2) (Figs. 1A and B). The Cand116 mutant is Mutant clones of Cand1 (−/−) were identifiable by their lack of GFP lethal in the pupal stage, and its lethality is rescued by a transgene expression and proximity to homozygous twin-spots, which had two 250 S.-H. Kim et al. / Developmental Biology 346 (2010) 247–257

Fig. 1. Molecular map and developmental expression of Cand1. (A) The transposon insertion site is indicated by a triangle above the map. The Cand116 and Cand1100 lines were obtained by P-element imprecise excision of G2412, and the excised regions are represented by vertical lines. (B) Cand1 mRNA expression analysis by RT-PCR in third-instar larvae. Primer pairs amplifying either the deleted or undeleted regions were used to amplify wild-type and the Cand116 mutant allele, and amplified regions are indicated in A. PCR amplification with the primer set 1 does not produce mRNA for Cand1 in Cand116 mutant homozygotes, although mRNA was detected with the primer set 2. It confirms that the Cand116 is a loss-of-function allele of Cand1. GAP-DH was used as a control. (C) Quantitative RT-PCR analysis of Cand1 expression during development. As a control, transcripts of rp49 were amplified. Cand1 transcripts can be detected during all developmental stages. copies of the GFP gene (+/+, 2× GFP), and otherwise normal from Cullins, but also by maintaining adaptor stability by recruiting heterozygous cells (+/−, 1× GFP). In and behind the morphogenetic deubiquitination enzymes (Wee et al., 2005). We observed that CSN5 furrows (MF: open arrowheads), CiFL (red) was decreased in Cand1 null mutant clones had decreased the levels of Myc-tagged Slimb mutant clones (marked by the absence of GFP, filled arrowheads) protein (filled arrowheads in Figs. 3D-F), whereas Cand1 mutant compared to wild-type clones (marked by GFP expression) (Figs. 2G–I). clones did not (Figs. 3A–C). To analyze the levels of HIB/rdx protein in The induction of the Cand1 mutant clone in the wing disc leads to a Cand1 mutant cells, Flag tagged HIB protein was expressed ubiqui- reduced protein level of CiFL in the anterior compartment where CiFL is tously using a heat shock-GAL4 driver, because high levels of processed to the repressor form of Ci by the Cul1-mediated E3 ligase expression of the HIB/rdx caused lethality. Flag-HIB was detected (filled arrowheads in Figs. 2J–L). These observations indicate that loss of with anti-Flag antibody, and it shows that the level of HIB/rdx protein Cand1 promotes CRL-dependent substrate degradation. The decrease of is increased in Cand1 mutant cells of the eye imaginal disc (Figs. 3G–I). CiFL in Cand116 mutant clones was not caused by a change in Ci For western blotting, protein was extracted from imaginal discs and transcription level because the expression of Ci-lacZ remained constant brain lobes expressing HA-tagged HIB/rdx using the elav-GAL4 driver. in the mutant cells (Figs. 2M–O). These results suggest that the level of It confirmed that the level of HIB is also increased in Cand1 mutant CiFL is regulated post-transcriptionally by Cand1 and that the reduction flies (Fig. 3K). Cand1 affects only the level of the Cul3 adaptor, HIB/rdx of CiFL in Cand1 mutant clones might be the result of increased levels of protein, but it fails to affect the Cul1 adaptor Slimb that is destabilized neddylated Cul3. by loss of CSN5. These observations suggest that the increased level of HIB/rdx could be a cause of the enhanced degradation of CiFL in the Loss of Cand1 stabilizes the Cul3 adaptor protein HIB/rdx, but not Slimb, Cand1 mutants, and Cand1 could be involved in the suppression of F-box of Cul1 HIB/rdx, leading to the regulation of Cul3-based RING ligase activity.

Proteolytic processing of CiFL is mediated by two types of Cullin- Mutations in Nedd8 or CSN5 induce accumulation of CiFL regardless of RING E3 ligase containing Cul1 and Cul3, respectively. Cul1 forms a Cand1 mutation SCFSlimb complex (Skp1/Cullin1/F-box proteins), while Slimb act as an F-box for the recognition of CiFL (Ou et al., 2002; Smelkinson et al., The loss of Drosophila Cand1 induces accumulation of neddylated 2007; Wu et al., 2005). Cul3 forms a complex with HIB (hedgehog- Cul3 and increases CiFL degradation with enhanced CRL activity induced BTB protein)/rdx (roadkill) that functions as the substrate (Fig. 2). In addition to the important role played by the neddylation of recognition component of a Cul3-based E3 complex Cullins in the regulation of the E3 activity of Cullin-RING ligases and promotes degradation of CiFL (Kent et al., 2006; Zhang et al., (CRLs) (Kawakami et al., 2001; Wu et al., 2002), the crystal structure 2006). Because we observed that mutation of Cand1 increases CiFL of Nedd8 modified and unmodified form of CRLs has shown that the degradation, we investigated the levels of adaptor proteins in the conjugation of Nedd8 to Cullin induces a conformational rearrange- Cand1 mutant. To this end, we expressed Myc-Slimb or Flag-HIB/rdx ment of the CRL complex in open forms (Duda et al., 2008). This result in the eye discs carrying Cand1 mutant clones and examined the level suggests that the neddylated form of Cullin is essential in forming an of adaptor proteins with anti-Myc and anti-Flag antibodies, respec- active CRL complex. To examine whether accumulation of neddylated tively. Myc-tagged Slimb protein shows no remarkable change in the Cul3 in the Cand1 mutant is responsible for increased degradation of eye disc of Cand116 mutant clones (filled arrowheads in Figs. 3A–C); CiFL protein, we generated Nedd8, Cand1 double-mutant clones in the this result was confirmed by western blot (Fig. 3J). There have been Drosophila eye disc and monitored CiFL protein levels. In the reports that CSN regulates CRL activity not only by cleaving Nedd8 Drosophila eye disc, the loss of Nedd8 showed a great accumulation S.-H. Kim et al. / Developmental Biology 346 (2010) 247–257 251

Fig. 2. Loss of Cand1 affects the neddylation of Cullin3 and Ci protein stability. (A–B) Western blot analysis of Cul1 (A) and Cul3 (B). Cell extracts were prepared from third-instar eye discs and brain lobes of wild-type (WT: lane 1) and Cand116 (lane 2), CSN5null mutants (lane 3), and Cand116 mutant expressing UAS-Cand1 driven by Da-GAL4 (lane 4). In the Cand116 mutant, the ratio of neddylated to unneddylated Cul3 is increased and the total amount of Cul3 is decreased slightly. Cand1 expression is able to rescue the reversed ratio of modified to unmodified Cul3 seen in homozygous Cand116 mutants. (C–D) Quantification of the ratio of neddylated to unneddylated Cul1 (C) and Cul3 (D) in WT, Cand116, and CSN5null mutants. (E–F) Quantification of relative total protein levels of Cul1 (E) and Cul3 (F) in WT, Cand116, and CSN5null flies. Error bars indicate the standard deviation (S.D.) of a minimum of three independent experiments. Asterisks indicate significance by t-test (*pb0.01, **pb0.001). (G–O) For imaginal eye discs of third-instar larva, anterior is to the left. For wing discs, dorsal is up and anterior is left. The scale bar represents 20 μm. (G–L) A third-instar eye (G–I) and wing (J–L) disc carrying Cand116 mutant clones were stained with anti-CiFL antibody. Clones carrying Cand116 mutant are marked by the lack of GFP (in green). A decrease in CiFL (red) is observed in the Cand1 mutant clones in the morphogenic furrow (open arrowhead) of the eye disc (G–I, filled arrowheads) and in the wing disc (J–L, filled arrowheads). (M–O) Third-instar eye disc carrying Cand116 mutant clones. Ci-LacZ expression (in red) was stained by anti-LacZ antibody. The Ci-LacZ expression level (red) is ubiquitous throughout the whole eye disc and remains unchanged in the Cand1 mutant clones (marked by lack of green). of CiFL (filled arrowheads in Figs. 4A–C) (Ou et al., 2002). The double- staining was elevated, indicating that the activity of the CRL complex mutant clones of Nedd8, Cand1 also induce a significant accumulation is decreased within the CSN5 mutant clones (Figs. 4G–K). In the of CiFL (filled arrowheads in Figs. 4D–F). A single mutation of Cand1 absence of Cand1, CSN5 mutant clones still displayed an accumulation caused a decreased level of CiFL (Figs. 2G–I), whereas double mutation of CiFL protein (Figs. 4L–P). The double-mutant of Cand1,CSN5 shows with Nedd8AN015 induced accumulation of CiFL similar to that seen in an accumulation of CiFL apparently similar to that of the CSN5 single the Nedd8 single mutant clones (Figs. 4A–F). These data show that, mutant than to that of the Cand1 mutant. This indicates that the Cand1 besides increased levels of HIB/rdx, elevated level of neddylated Cul3 mutant fails to affect the degradation of CiFL in the absence of CSN5. is also involved in the enhanced degradation of CiFL in the Cand1 These results suggest that CSN5 plays a more dominant role in the mutant. It suggests that regulation of Cul3 neddylation by Cand1 is process of degradation of CiFL than does Cand1. also important for Cul3-mediated substrate degradation. It has been proposed that CSN is required for CRL activity by Cand1 protects unneddylated Cul3 from proteasome-dependent removing Nedd8 from Cullin (Cope and Deshaies, 2003; Oron et al., degradation 2002). Cand1 is known to specifically bind to unneddylated Cullin and sequester it from the CRL complex (Hwang et al., 2003; Liu et al., 2002; It has been reported that Cand1 specifically binds to unneddylated Min et al., 2003; Zheng et al., 2002). To determine the effect of CSN on Cullins (Hwang et al., 2003; Min et al., 2003; Zheng et al., 2002). GST Cand1 for substrate degradation, we generated CSN5 null clones in the pull-down assays confirmed that Cand1 is able to associate with Cand1 mutant homozygous background and examined CiFL in the Drosophila Cul1 and Cul3 (Fig. 5A). Specific binding of Cand1 with Drosophila eye disc. A mutation in CSN5 caused a reduction of Cul1 and unneddylated forms of Cullins allowed us to consider the possibility Cul3 protein levels in Drosophila (Figs. 2A–F) (Wu et al., 2005). that Cand1 could induce stabilization of Cul1 and Cul3. APP-BP1 is Interestingly, although the unneddylated form of Cullins almost responsible for E1-like function in Nedd8 activation. The dAPP-BP1null disappeared, the CSN5 mutant displayed constant levels of neddylated (Drosophila APP-BP1 null mutant) showed a significant accumulation Cullins that is required for degradation of CiFL (lane 3 in Figs. 2A of unneddylated Cul1 (Fig. 5B, lanes 1 and 2) (Kim et al., 2007). The and B). In the eye disc containing CSN5 null mutant cells, however, CiFL Cul3 protein also displayed an accumulation pattern similar to that of 252 S.-H. Kim et al. / Developmental Biology 346 (2010) 247–257

Fig. 3. Loss of Cand1 stabilizes the Cul3 adaptor protein HIB/rdx, but not Slimb, F-box of Cul1. (A–I) Third-instar eye imaginal discs are oriented with anterior toward left, and the scale bar represents 20 μm. (A–F) Eye imaginal discs expressing myc-slimb were immunostained for Myc (red) and GFP (green). Protein level of Myc-tagged Slimb (red) was not changed in Cand1 mutant clones (marked by lack of GFP) (filled arrowheads in A–C), whereas the level of Myc (red) was reduced in CSN5 mutant clones (marked by the absence of GFP) (filled arrowheads in D–F). (G–I) Cand116 mutant clones (lacking GFP, green) were generated in the eye discs expressing the Flag-HIB/rdx under control of hs-GAL4. Protein level of Flag-HIB/rdx (red) was increased in Cand1 mutant clones (filled arrowheads), (J) Western blots with Myc and β-tubulin (control) antibodies. Protein extracts were prepared from larval brain and eye imaginal discs of wild-type (WT) and Cand116 homozygotes expressing Myc-Slimb from the tubulin promoter. (K) Western blots of larval extracts with HA and β- tubulin antibodies. Protein was extracted from larval eye disc and brain tissue. UAS-HA-HIB/rdx is expressed, driven by elav-GAL4,inw1118 (control) and the Cand116 mutant homozygote. The level of HIB is increased in the Cand116 mutant.

Cul1 in dAPP-BP1 null mutant cells (Fig. 5C, lanes 1 and 2). To 2002; Schweisguth, 1999). Repression of proteasome function by determine whether Cand1 is responsible for the accumulation of expressing UAS-Prosβ21 with the ubiquitous Da-GAL4 driver resulted unneddylated Cul1 and Cul3 found in the dAPP-BP1 mutant, we in the accumulation of unneddylated Cul3 protein at the restrictive generated a Cand1, dAPP-BP1 double-mutant in Drosophila. Curiously, temperature (Fig. 5F, lanes 1 and 2). However, the dAPP-BP1 mutant, dAPP-BP1,Cand1 double-mutant cells had a decreased Cul3 protein which had a significant accumulation of unneddylated Cullins, was no level compared to that of dAPP-BP1 single mutant cells (Figs. 5C longer affected by inhibition of proteasome function (Fig. 5F, lanes 3 and E), whereas the Cul1 protein is still accumulated (Figs. 5B and D). and 4). These results indicate that the ubiquitin–proteasome This shows that loss of Cand1 fails to cause accumulation of Cul3 in machinery is involved in the degradation of unneddylated Cul3 and the dAPP-BP1 mutant, indicating that Cand1 is needed to stabilize that the dAPP-BP1 mutation interferes with this degradation effect by unneddylated Cul3 in the neddylation defective dAPP-BP1 null the proteasome, leading to the accumulation of unneddylated Cullins. mutant. Overexpression of Prosβ21 in Cand1 mutants still leads to the The ubiquitin–proteasome pathway is one of the most important accumulation of unneddylated Cul3 (Fig. 5F, lanes 5 and 6), systems for the targeted degradation of intracellular proteins. To demonstrating that unneddylated Cul3 is destabilized by the protea- study whether the proteasome is involved in Cul3 protein degrada- some in the absence of Cand1. tion, we used dominant temperature-sensitive (DTS) proteasome To further investigate the role of Cand1 in the protection of mutant flies. Prosβ21 has missense mutations in the 20S proteasome unneddylated Cul3 from proteasomal degradation, we expressed UAS- subunits β2 and the expression of UAS-Prosβ21 represses proteasome Prosβ21 using Da-GAL4 in Cand1, dAPP-BP1 double-mutant flies. The function at the restrictive temperature (29 °C) (Belote and Fortier, amount of Cul3 protein in the Cand1, dAPP-BP1 double-mutant was S.-H. Kim et al. / Developmental Biology 346 (2010) 247–257 253

Fig. 4. Mutations in Nedd8 or CSN5 induce accumulation of CiFL regardless of Cand1 mutation. (A–F) Eye disc carrying Nedd8AN015 clones (A–C) or Nedd8AN015, Cand116 double-mutant clones (D–F) were immunostained to examine the levels of CiFL (red). The mutant clones were marked by the absence of GFP (green). Merged images are shown in (C) and (F). For imaginal eye discs, anterior is to the left. Levels of CiFL protein (red) were accumulated in the Nedd8AN015 single- (filled arrowheads in A–C) or in the Nedd8AN015, Cand116 double-mutant (filled arrowheads in D–F). Scale bar, 20 μm. (G–J)EyedisccontainingtheCSN5null clone, where mutant clones fail to express GFP (shown in green), was stained for CiFL (red). Merged image is shown in (I). The level of CiFL protein (red) was increased in CSN5 null clones (filled arrowheads). (L–O) Eye disc containing CSN5 mutant clones (marked by lack of green) generated in the Cand116 homozygote background. Merged image is shown in (N). The CSN5null mutant clones still have accumulated levels of CiFL protein (red) in the Cand116 mutant background (filled arrowheads). (J, O) High magnification images of eye discs stained for CiFL (red). The magnified areas are outlined by the white boxes in (I) and (N), respectively. Scale bar, 20 μm. (K, P) Graphs showing averaged relative fluorescence intensities (y-axis) of CiFL (red) and GFP signal (green). The areas are indicated by the white arrows in (I) and (N), respectively. The intensity of CiFL (red) is increased in the region of low GFP expression (the mutant clones of CSN5). 254 S.-H. Kim et al. / Developmental Biology 346 (2010) 247–257

Fig. 5. Cand1 protects unneddylated Cul3 from proteasome-dependent degradation. (A) GST pull-down binding assay of [35S] labeled Cul1 (lanes 1–3) and Cul3 (lanes 4–6) with GST–Cand1. In vitro translated Cul1 and Cul3 bind to GST–Cand1. GST protein was used as a control. (B–C) Western blots with Cul1 (B) and Cul3 (C) antibodies. Protein extracts from wild-type (lane 1), dAPP-BP1null (lane 2), and double-mutant of dAPP-BP1null, Cand116 larvae (lane 3). Accumulated Cul3 in the dAPP-BP1null mutant is suppressed in dAPP-BP1, Cand1 double-mutant cells. (D–E) Quantification of relative total protein levels of Cul1 (D) and Cul3 (E). Error bars indicate S.D. of a minimum of three independent experiments. Asterisks indicate significance by t-test (*pb0.01, **pb0.001). (F) Western blots with Cul3 and β-tubulin antibodies. Proteins were extracted from third-instar larvae of the following genotypes: Da-GAL4/+ (lane 1); Da-GAL4/UAS-Prosβ21 (lane 2); Da-GAL4,dAPP-BP1null/dAPP-BP1null (lane 3); Da-GAL4,dAPP-BP1null/dAPP-BP1null,UAS-Prosβ21 (lane 4); Cand116;Da- GAL4/+ (lane 5); and Cand116;Da-GAL4/UAS-Prosβ21 (lane 6). All flies were raised at 29 °C. (G) Western blots with Cul3 and β-tubulin antibodies. Flies were incubated at 29 °C. Proteins were extracted from third-instar larvae of the following genotypes: Cand116;Da-GAL4,dAPP-BP1null/dAPP-BP1null (lane 1); and Cand116;Da-GAL4,dAPP-BP1null/dAPP-BP1null, UAS-Prosβ21 (lane 2). Unneddylated Cul3 in Cand116, dAPP-BP1null double-mutant cells was accumulated by overexpression of a dominant-negative form of the proteasome subunit. increased by the expression of Prosβ21 at the restrictive temperature proteins are present together with Cul3 protein in the same complex, (Fig. 5G). This shows that decreased Cul3 protein in the double- we performed the GST pull-down assay using a GST–Cand1 and His- mutant is reversed by inhibition of proteasome activity, confirming HIB/rdx, with or without Cul3. It shows that Cand1 associates with that the stabilization of unneddylated Cul3 by Cand1 is due to HIB/rdx in the presence of Cul3 (lane 5 in Fig. 6C). However, protection against proteasomal degradation. When we expressed surprisingly, we observed that HIB/rdx, Cul3 adaptor protein, can still Cand1 in the dAPP-BP1 mutant, there was no significant effect on Cul3 bind to Cand1 in the absence of Cul3 (lane 6 in Fig. 6C), indicating that protein levels (data not shown). This suggests that endogenous Cand1 these three proteins are able to bind to each other. We also found that could protect unneddylated Cul3 efficiently in the dAPP-BP1 mutant, the interaction of Cand1 with HIB protein was decreased by increasing not requiring more Cand1 to stabilize Cul3. Taken together, these the amounts of Cul3 (Fig. 6D), indicating that Cul3 competitively results support the possibility that Cand1 protects unneddylated Cul3 inhibits the binding of HIB/rdx to Cand1. These findings raise the from proteasome-dependent degradation. possibility that Cand1 might regulate HIB/rdx by interacting with Cul3, and increased level of HIB/rdx in Cand1 mutant (Fig. 3) might be Overexpression of Cand1 stabilizes the Cul3 protein the result of the direct interaction of Cand1 with HIB/rdx. Recent reports have suggested that neddylation of Cullin proteins is The aforementioned data in Fig. 5 indicate that Cand1 is needed to enhanced by binding of substrate and adaptor protein to Cullins stabilize the Cullin protein. To further evaluate the function of Cand1 (Bornstein et al., 2006; Chew and Hagen, 2007). Although HIB/rdx did for Cullin stability, we expressed Cand1 in mutant flies with a not compete with Cand1 to Cul3, it is possible that HIB/rdx may affect neddylation pathway defect and then detected Cullin protein levels. In Cand1 function by regulating Cul3 neddylation. When HIB protein the CSN5 null mutant, the amount of Cullins was markedly reduced, levels are down-regulated by multiple copies of UAS-HIB-RNAi driven and the remaining Cullins were present exclusively in the neddylated by elav-GAL4, overexpression of Cand1 induces stabilization of the form (lane 3 in Figs. 2A and B). Expression of Cand1 in CSN5null Cul3 protein (Fig. 6E). Our results suggest that overexpressing Cand1 homozygotes induced stabilization of both Cul1 and Cul3 proteins, induces stabilization of the Cul3 protein when the neddylation path- even though it was much less effective for Cul1 than for Cul3 (Fig. 6A). way is compromised. Moreover, Cand1 overexpression in dAPP-BP1null heterozygotes also increased the Cul3 protein level (Fig. 6B and Supplemental Fig. S2A). Discussion The Cul3 adaptor protein HIB/rdx interacts with Cul3 and then promotes CiFL degradation (Kent et al., 2006; Zhang et al., 2006). Since The neddylation pathway is highly conserved in many organisms, both Cand1 and HIB/rdx proteins were able to bind to Cul3, we and the neddylation step is essential for Cullin-mediated E3 ubiquitin performed an in vitro competitive binding assay to address the ligase activation. Cand1 is a highly conserved protein that binds to possibility of direct competition between Cand1 and HIB/rdx for unneddylated Cullins and sequesters Cul1 from the CRL complex (Liu association with Cul3. To investigate whether HIB/rdx would prevent et al., 2002; Zheng et al., 2002). It has been suggested that Cand1 the binding of Cand1 to Cul3 in a dose-dependent manner, GST- inhibits CRL activity in vitro. However, studies from Arabidopsis have tagged Cand1 and Cul3 protein were incubated with increasing shown that loss of Cand1 leads to decreased CRL activity, indicating amounts of HIB/rdx. We found that increasing the dose of HIB/rdx did that Cand1 is required for efficient CRL function (Chuang et al., 2004; not interfere with the interaction of Cand1 with Cul3 (Supplemental Feng et al., 2004). To elucidate this paradoxical effect of Cand1, we Fig. S2C), in spite of the ability of HIB/rdx to bind to Cul3 have used Drosophila as a model system. First, we found that loss of (Supplemental Fig. S2B). To examine whether Cand1 and HIB/rdx Cand1 increases the ratio of Nedd8 modified to unmodified Cul3 and S.-H. Kim et al. / Developmental Biology 346 (2010) 247–257 255

Fig. 6. Overexpression of Cand1 stabilizes the Cul3 protein. (A) Western blots with antibodies against Cul1 (lanes 1 and 2) and Cul3 (lanes 3 and 4). UAS-Cand1 was overexpressed by Da-GAL4. Proteins were prepared from larvae of the following genotypes: Da-GAL4,CSN5null/CSN5null (lanes 1 and 3); and Da-GAL4,CSN5null/CSN5null,UAS-Cand1 (lanes 2 and 4). Protein levels of Cullins in the CSN5null homozygote were elevated by overexpression of Cand1. (B) Western blots with anti Cul3 antibody. Beta-tubulin serves as a control. Protein extracts were prepared from brain lobes and eye discs of third-instar larvae. Da-GAL4 was used to overexpress three copies of UAS-Cand1. Proteins were prepared from larvae of the following genotypes: Da-GAL4,dAPP-BP1null/+ (lane 1); and 3XUAS-Cand1;Da-GAL4,dAPP-BP1null/+ (lane 2). Protein levels of Cul3 in the dAPP-BP1 heterozygote were elevated by overexpression of Cand1. (C) In vitro GST pull-down assay. Recombinant GST (7 μg) or GST–Cand1 fusion proteins (30 μg) were incubated with His purified HIB/rdx protein (10 μg) and in vitro translated Cul3 protein (15 μl). The GST fusion protein was pulled down by glutathione Sepharose 4B and detected by western blot using anti-Cul3 and anti-His antibodies. Inputs (5%) were loaded as control (lanes 1 and 2). GST pull-down experiments show that Cand1 interacts with HIB/rdx in the presence or absence of Cul3 protein. (D) Competitive GST pull-down assay. Recombinant GST (7 μg) or GST–Cand1 fusion proteins (21 μg) were incubated with His purified HIB/rdx protein (9 μg) and increasing amounts of in vitro translated Cul3 (0, 10 μl, 20 μl, and 50 μl). Five percent of inputs were loaded on lanes 1 and 2. Increasing the dose of Cul3 inhibit the interaction of Cand1 with HIB/rdx. (E) Western blot analysis of larval brain and eye disc extracts prepared from elav-GAL4 (lane 1); elav-GAL4N3XUAS-HIB-RNAi (lane 2); and elav-GAL4N3XUAS-HIB- RNAi,2XUAS-Cand1 (lane 3). When the expression of adaptor protein, HIB, was down-regulated, overexpression of Cand1 leads to stabilization of the Cul3 protein. (F) Proposed model for the role of Cand1. the level of Cul3 adaptor HIB/rdx, causing enhanced degradation of The absence of Cand1 increased the level of neddylated Cul3, and it CiFL, despite little effect on Cul1 (Figs. 2 and 3). Although Cand1 has suggests that Cand1 could inhibit the neddylation of Cul3. However, been reported to negatively regulate CRL activity by binding to the overexpression of Cand1 had no effect on Cul3 neddylation (data unneddylated Cul1 and dissociating the CRL complex in vitro (Hwang not shown). The amount of Cand1 seems to be sufficient to prevent et al., 2003; Liu et al., 2002; Zheng et al., 2002), accumulations of Cul3 neddylation in the wild-type background. However, neddylation neddylated Cullin and adaptor protein have never been observed in of Cul3 was decreased when Cand1 was expressed in the Cand1 studies of Cand1 depletion. These provide a better understanding of mutant background, indicating that Cand1 can suppress Cul3 the role of Cand1 in vivo, suggesting that the regulations of Cul3 neddylation (Supplemental Fig. S1C and D). neddylation and adaptor stability are important for Cand1 to control CiFL is processed by two different Cullins, Cul1 and Cul3, in the eye CRL activity. Unlike the results of Arabidopsis studies, in which Cand1 disc of Drosophila (Ou et al., 2002; Ou et al., 2003). In the posterior is required for optimal CRL activity, our study demonstrates that the area of the eye imaginal disc, CiFL is degraded by Cul3-mediated E3 Cand1 mutation of Drosophila stimulates the degradation of CiFL by activity, where loss of Cand1 affects the stability of CiFL. Because we enhancing Cul3-RING ligase activity. In addition, we provide a novel observed that mutation of Cand1 decreases the level of CiFL, we further insight into the role of Cand1 by which Cand1 is involved in the investigated the levels of adaptor proteins of Cul1 and Cul3. We found stabilization of unneddylated Cul3. We present evidence that Cand1 that the level of the Cul3 adaptor protein HIB/rdx is also increased in protects unneddylated Cul3 from proteasomal degradation. the Cand1 mutant, whereas the levels of Slimb, the F-box protein of 256 S.-H. Kim et al. / Developmental Biology 346 (2010) 247–257 the Cul1 RING ligase, remain constant (Fig. 3). It suggests that Cand1 dAPP-BP1 null mutant (data not shown), even if Cand1 overexpression could regulate Cul3-based E3 ligase activity by suppressing the level of increases Cul3 protein levels in the dAPP-BP1null heterozygote HIB/rdx. Several adaptor proteins are destabilized by autoubiquitina- background (Fig. 6B). tion of CRL activity (Galan and Peter, 1999; Wirbelauer et al., 2000). Although Cand1 affects mostly the Cul3 protein, it also influences CSN also maintains adaptor stability by deneddylating Cullin and the Cul1 protein. Cand1 can bind to Cul1 and the overexpression of recruiting deubiquitination enzymes (Wee et al., 2005). Interestingly, Cand1 induces the stabilization of Cul1 as well as Cul3 (Figs. 5 and 6). it has recently been observed that the CSN-associated deubiquitinat- However, the effect of Cand1 on Cullins seems to differ depending on ing enzyme Ubp12 maintains the stability of the Cul3 adaptor, but not the type of tissue. Immunoblot analysis of Cand116 extracts from third- the F-box, Cul1 adaptor (Schmidt et al., 2009). This provides a possible instar brain lobes and eye discs showed no distinguishable effect on clue that Cand1 may regulate the stability of HIB/rdx through the ratio of neddylated Cul1, but loss of Cand1 caused a reduction of deubiquitinating enzymes by working with CSN. Direct interaction CiFL protein in the anterior region of the wing disc, where the Cul1- of Cand1 with HIB/rdx (Figs. 6C and D) suggests another possibility dependent E3 ligase degrades CiFL protein (Fig. 2). that Cand1 might suppress the level of HIB/rdx through a direct Here, we propose that Cand1 contributes to the fine-tuning of association with HIB/rdx. Taken together, the evidence presented Cul3-mediated E3 ligase activity by acting on the neddylation state as here indicates that Cul3-dependent E3 ubiquitin ligase activity is well as on the stability of unneddylated Cul3 and adaptor protein increased by the loss of Cand1 function. (Fig. 6F). Binding of Cand1 to unneddylated Cul3 would shift the It has been suggested that Nedd8 covalent conjugation to Cullin equilibrium away from the neddylation of Cul3 that is required for causes instability of the Cullin protein (Wu et al., 2005). However, our substrate degradation and then cause sequestration of unneddylated results show that the neddylated form of Cul3 has maintained protein Cul3 from proteasomal degradation. Moreover, Cand1 could be stability in the Cand1 mutant, albeit at a slightly reduced Cul3 protein involved in the suppression of Cul3 adaptor protein, HIB/rdx, to level (Fig. 2B and F). This observation could be related to the function regulate CRL activity. Loss of Cand1 shifts the equilibrium toward the of CSN because there is a significant decrease in the total amount of neddylated form of Cul3 and increases the level of Cul3 adaptor HIB/ Cullins in CSN mutant cells (Gusmaroli et al., 2007; Stuttmann et al., rdx, which leads to enhanced degradation of CiFL, a substrate of CRLs. 2009; Wu et al., 2005). Both CSN and Cand1 proteins have been Neddylation of Cul3 is essential for CRL activity, so the mutation of proposed to be involved in the cycle of assembly and disassembly of Cand1 fails to down-regulate accumulation of CiFL in Nedd8 or CSN5 the CRL complex (Cope and Deshaies, 2003; Lo and Hannink, 2006; mutants. In the absence of dAPP-BP1, unneddylated Cul3 would tend Zhang et al., 2008). This model explains how Cand1 and CSN have to bind to Cand1, which protects unneddylated Cul3 from proteaso- paradoxical effects on CRL activity and insists that Cand1-mediated mal degradation and induces accumulation of Cul3. cycling is required for optimal CRL activity. However, our data do not The mechanisms underlying the Cullin neddylation pathway are support this cycling model, in which loss of Cand1 enhances the closely conserved in Drosophila and in mammals. Consequently, the degradation of CiFL as a result of increased activity of CRLs. The double- study of Drosophila Cand1 and Cullin provides a novel insight into the mutant analyses suggest that regulation of the neddylation pathway is regulation of Cullin based E3 ligases by Cand1. a major mechanism for CiFL degradation. Loss of Cand1 failed to suppress accumulation of CiFL protein in Nedd8AN015 or CSN5null Acknowledgments mutant clones (Fig. 4). The functions of Nedd8 and CSN with regard to Cullin seem to play a more dominant role in regulating CRLs We thank H. Suh for the transgenic flies and D. Segal, J. Jiang, than that of regulation by Cand1. This could explain why over- Genexel and the Bloomington Stock Center for providing the fly expression of Cand1 in CSN5null mutant causes an increase only in the stocks. The E7 and 2A1 antibodies were obtained from the neddylated forms of Cullin, although Cand1 stabilizes unneddylated Developmental Studies Hybridoma Bank developed under the Cullin (Figs. 5 and 6). auspices of the NICHD and maintained by The University of Iowa, Inhibition of proteasome function by overexpressing a dominant- Department of Biology, Iowa City, IA 52242. This work was supported negative form of a proteasome subunit causes accumulation of by the BK21 Research Fellowship from the Ministry of Education, unneddylated Cul3 (Fig. 5F). The neddylation defective dAPP-BP1 Science and Technology, Republic of Korea, and by a Korea Science and mutant also exhibits elevated levels of unneddylated Cul3, but Engineering Foundation (KOSEF) grant funded by the Korean repressed proteasomal activity in the dAPP-BP1null mutant fails to Government (MOST) (no. M103KV010002-08K2201-00210). causes Cul3 accumulation (Fig. 5). These results support the theory that unneddylated Cul3 is degraded by the proteasome, but this Appendix A. Supplementary data degradation effect is inhibited by mutation of the Nedd8 E1-activating enzyme, dAPP-BP1. Accumulation of Cul3 in the dAPP-BP1 mutant is Supplementary data associated with this article can be found, in suppressed by loss of Cand1, and decreased Cul3 in the dAPP-BP1, the online version, at doi:10.1016/j.ydbio.2010.07.031. Cand1 double-mutant is again accumulated by reducing proteasome activity (Fig. 5). This shows that Cand1 is responsible for the References accumulation of unneddylated Cul3 in the dAPP-BP1 mutant as a result of inhibition of proteasome-mediated degradation. Repression Belote, J.M., Fortier, E., 2002. Targeted expression of dominant negative proteasome mutants in Drosophila melanogaster. Genesis 34, 80–82. of proteasome function in the Cand1 mutant induces accumulation of Bornstein, G., Ganoth, D., Hershko, A., 2006. Regulation of neddylation and deneddyla- unneddylated Cul3, showing that neddylated Cul3 is destabilized by tion of cullin1 in SCFSkp2 ubiquitin ligase by F-box protein and substrate. Proc. the proteasome in the absence of Cand1. Natl. Acad. Sci. U. S. A. 103, 11515–11520. Recent reports indicate that supplementation of substrate and Chew, E.H., Hagen, T., 2007. Substrate-mediated regulation of cullin neddylation. J. Biol. Chem. 282, 17032–17040. adaptor to Cullin-RING ligases promotes Cullin neddylation and Chuang, H.W., Zhang, W., Gray, W.M., 2004. Arabidopsis ETA2, an apparent ortholog of dissociation of the Cullin–Cand1 complex (Bornstein et al., 2006; the human cullin-interacting protein CAND1, is required for auxin responses – Chew and Hagen, 2007). In agreement with the previous reports, our mediated by the SCF(TIR1) ubiquitin ligase. Plant Cell 16, 1883 1897. Ciechanover, A., 1998. The ubiquitin–proteasome pathway: on protein death and cell data also suggest that the neddylation process might regulate the life. EMBO J. 17, 7151–7160. dissociation of Cul3 from Cand1. If Cand1 is dissociated from Cullin by Cope, G.A., Deshaies, R.J., 2003. COP9 signalosome: a multifunctional regulator of SCF neddylation, a defect in the neddylation process might promote the and other cullin-based ubiquitin ligases. Cell 114, 663–671. Cope, G.A., Suh, G.S., Aravind, L., Schwarz, S.E., Zipursky, S.L., Koonin, E.V., Deshaies, R.J., interaction of Cand1 with unneddylated Cul3. This could explain why 2002. Role of predicted metalloprotease motif of Jab1/Csn5 in cleavage of Nedd8 the level of Cul3 was not affected by overexpression of Cand1 in the from Cul1. Science 298, 608–611. S.-H. Kim et al. / Developmental Biology 346 (2010) 247–257 257 del Pozo, J.C., Dharmasiri, S., Hellmann, H., Walker, L., Gray, W.M., Estelle, M., 2002. Ou, C.Y., Lin, Y.F., Chen, Y.J., Chien, C.T., 2002. Distinct protein degradation mechanisms AXR1–ECR1-dependent conjugation of RUB1 to the Arabidopsis Cullin AtCUL1 is mediated by Cul1 and Cul3 controlling Ci stability in Drosophila eye development. required for auxin response. Plant Cell 14, 421–433. Genes Dev. 16, 2403–2414. Dias, D.C., Dolios, G., Wang, R., Pan, Z.Q., 2002. CUL7: A DOC domain-containing cullin Ou, C.Y., Pi, H., Chien, C.T., 2003. Control of protein degradation by E3 ubiquitin ligases selectively binds Skp1.Fbx29 to form an SCF-like complex. Proc. Natl. Acad. Sci. U. S. A. in Drosophila eye development. Trends Genet. 19, 382–389. 99, 16601–16606. Pan, Z.Q., Kentsis, A., Dias, D.C., Yamoah, K., Wu, K., 2004. Nedd8 on cullin: building an Doronkin, S., Djagaeva, I., Beckendorf, S.K., 2002. CSN5/Jab1 mutations affect axis expressway to protein destruction. Oncogene 23, 1985–1997. formation in the Drosophila oocyte by activating a meiotic checkpoint. Develop- Petroski, M.D., Deshaies, R.J., 2005. Function and regulation of cullin-RING ubiquitin ment 129, 5053–5064. ligases. Nat. Rev. Mol. Cell Biol. 6, 9–20. Doronkin, S., Djagaeva, I., Beckendorf, S.K., 2003. The COP9 signalosome promotes Pintard, L., Willems, A., Peter, M., 2004. Cullin-based ubiquitin ligases: Cul3–BTB degradation of Cyclin E during early Drosophila oogenesis. Dev. Cell 4, 699–710. complexes join the family. EMBO J. 23, 1681–1687. Duda, D.M., Borg, L.A., Scott, D.C., Hunt, H.W., Hammel, M., Schulman, B.A., 2008. Podust, V.N., Brownell, J.E., Gladysheva, T.B., Luo, R.S., Wang, C., Coggins, M.B., Pierce, J.W., Structural insights into NEDD8 activation of cullin-RING ligases: conformational Lightcap, E.S., Chau, V., 2000. A Nedd8 conjugation pathway is essential for proteolytic control of conjugation. Cell 134, 995–1006. targeting of p27Kip1 by ubiquitination. Proc. Natl. Acad. Sci. U. S. A. 97, 4579–4584. Feng, S., Shen, Y., Sullivan, J.A., Rubio, V., Xiong, Y., Sun, T.P., Deng, X.W., 2004. Rabut, G., Peter, M., 2008. Function and regulation of protein neddylation. ‘Protein Arabidopsis CAND1, an unmodified CUL1-interacting protein, is involved in modifications: beyond the usual suspects’ review series. EMBO Rep. 9, 969–976. multiple developmental pathways controlled by ubiquitin/proteasome-mediated Read, M.A., Brownell, J.E., Gladysheva, T.B., Hottelet, M., Parent, L.A., Coggins, M.B., protein Degradation. Plant Cell 16, 1870–1882. Pierce, J.W., Podust, V.N., Luo, R.S., Chau, V., Palombella, V.J., 2000. Nedd8 Furukawa, M., Zhang, Y., McCarville, J., Ohta, T., Xiong, Y., 2000. The CUL1 C-terminal modification of cul-1 activates SCF(beta(TrCP))-dependent ubiquitination of sequence and ROC1 are required for efficient nuclear accumulation, NEDD8 IkappaBalpha. Mol. Cell. Biol. 20, 2326–2333. modification, and ubiquitin ligase activity of CUL1. Mol. Cell. Biol. 20, 8185–8197. Rubin, G.M., Spradling, A.C., 1982. Genetic transformation of Drosophila with Galan, J.M., Peter, M., 1999. Ubiquitin-dependent degradation of multiple F-box proteins transposable element vectors. Science 218, 348–353. by an autocatalytic mechanism. Proc. Natl. Acad. Sci. U. S. A. 96, 9124–9129. Schmidt, M.W., McQuary, P.R., Wee, S., Hofmann, K., Wolf, D.A., 2009. F-box-directed Goldenberg, S.J., Cascio, T.C., Shumway, S.D., Garbutt, K.C., Liu, J., Xiong, Y., Zheng, N., CRL complex assembly and regulation by the CSN and CAND1. Mol. Cell 35, 2004. Structure of the Cand1–Cul1–Roc1 complex reveals regulatory mechanisms 586–597. for the assembly of the multisubunit cullin-dependent ubiquitin ligases. Cell 119, Schweisguth, F., 1999. Dominant-negative mutation in the beta2 and beta6 proteasome 517–528. subunit genes affect alternative cell fate decisions in the Drosophila sense organ Gong, L., Yeh, E.T., 1999. Identification of the activating and conjugating enzymes of the lineage. Proc. Natl. Acad. Sci. U. S. A. 96, 11382–11386. NEDD8 conjugation pathway. J. Biol. Chem. 274, 12036–12042. Shiyanov, P., Nag, A., Raychaudhuri, P., 1999. Cullin 4A associates with the UV-damaged Gusmaroli, G., Figueroa, P., Serino, G., Deng, X.W., 2007. Role of the MPN subunits in DNA-binding protein DDB. J. Biol. Chem. 274, 35309–35312. COP9 signalosome assembly and activity, and their regulatory interaction with Skaar, J.R., Florens, L., Tsutsumi, T., Arai, T., Tron, A., Swanson, S.K., Washburn, M.P., Arabidopsis Cullin3-based E3 ligases. Plant Cell 19, 564–581. DeCaprio, J.A., 2007. PARC and CUL7 form atypical cullin RING ligase complexes. Hershko, A., Ciechanover, A., 1998. The ubiquitin system. Annu. Rev. Biochem. 67, Cancer Res. 67, 2006–2014. 425–479. Smelkinson, M.G., Kalderon, D., 2006. Processing of the Drosophila hedgehog signaling Hori, T., Osaka, F., Chiba, T., Miyamoto, C., Okabayashi, K., Shimbara, N., Kato, S., Tanaka, effector Ci-155 to the repressor Ci-75 is mediated by direct binding to the SCF K., 1999. Covalent modification of all members of human cullin family proteins by component Slimb. Curr. Biol. 16, 110–116. NEDD8. Oncogene 18, 6829–6834. Smelkinson, M.G., Zhou, Q., Kalderon, D., 2007. Regulation of Ci-SCFSlimb binding, Ci Hwang, J.W., Min, K.W., Tamura, T.A., Yoon, J.B., 2003. TIP120A associates with proteolysis, and hedgehog pathway activity by Ci phosphorylation. Dev. Cell 13, unneddylated cullin 1 and regulates its neddylation. FEBS Lett. 541, 102–108. 481–495. Jiang, J., Struhl, G., 1995. Protein kinase A and hedgehog signaling in Drosophila limb Stuttmann, J., Lechner, E., Guerois, R., Parker, J.E., Nussaume, L., Genschik, P., Noel, L.D., development. Cell 80, 563–572. 2009. COP9 signalosome- and 26S proteasome-dependent regulation of SCFTIR1 Kamura, T., Maenaka, K., Kotoshiba, S., Matsumoto, M., Kohda, D., Conaway, R.C., accumulation in Arabidopsis. J. Biol. Chem. 284, 7920–7930. Conaway, J.W., Nakayama, K.I., 2004. VHL-box and SOCS-box domains determine Theodosiou, N.A., Xu, T., 1998. Use of FLP/FRT system to study Drosophila development. binding specificity for Cul2–Rbx1 and Cul5–Rbx2 modules of ubiquitin ligases. Methods 14, 355–365. Genes Dev. 18, 3055–3065. Wee, S., Geyer, R.K., Toda, T., Wolf, D.A., 2005. CSN facilitates Cullin-RING ubiquitin Kawakami, T., Chiba, T., Suzuki, T., Iwai, K., Yamanaka, K., Minato, N., Suzuki, H., ligase function by counteracting autocatalytic adapter instability. Nat. Cell Biol. 7, Shimbara, N., Hidaka, Y., Osaka, F., Omata, M., Tanaka, K., 2001. NEDD8 recruits E2- 387–391. ubiquitin to SCF E3 ligase. EMBO J. 20, 4003–4012. Wertz, I.E., O'Rourke, K.M., Zhang, Z., Dornan, D., Arnott, D., Deshaies, R.J., Dixit, V.M., Kent, D., Bush, E.W., Hooper, J.E., 2006. Roadkill attenuates Hedgehog responses 2004. Human De-etiolated-1 regulates c-Jun by assembling a CUL4A ubiquitin through degradation of Cubitus interruptus. Development 133, 2001–2010. ligase. Science 303, 1371–1374. Kerscher, O., Felberbaum, R., Hochstrasser, M., 2006. Modification of proteins by Willems, A.R., Schwab, M., Tyers, M., 2004. A hitchhiker's guide to the cullin ubiquitin ubiquitin and ubiquitin-like proteins. Annu. Rev. Cell Dev. Biol. 22, 159–180. ligases: SCF and its kin. Biochim. Biophys. Acta 1695, 133–170. Kim, H.J., Kim, S.H., Shim, S.O., Park, E., Kim, C., Kim, K., Tanouye, M.A., Yim, J., 2007. Wirbelauer, C., Sutterluty, H., Blondel, M., Gstaiger, M., Peter, M., Reymond, F., Krek, W., Drosophila homolog of APP-BP1 (dAPP-BP1) interacts antagonistically with APPL 2000. The F-box protein Skp2 is a ubiquitylation target of a Cul1-based core during Drosophila development. Cell Death Differ. 14, 103–115. ubiquitin ligase complex: evidence for a role of Cul1 in the suppression of Skp2 Ko, H.W., Jiang, J., Edery, I., 2002. Role for Slimb in the degradation of Drosophila Period expression in quiescent fibroblasts. EMBO J. 19, 5362–5375. protein phosphorylated by Doubletime. Nature 420, 673–678. Wu, K., Chen, A., Pan, Z.Q., 2000. Conjugation of Nedd8 to CUL1 enhances the ability of Lammer, D., Mathias, N., Laplaza, J.M., Jiang, W., Liu, Y., Callis, J., Goebl, M., Estelle, M., the ROC1–CUL1 complex to promote ubiquitin polymerization. J. Biol. Chem. 275, 1998. Modification of yeast Cdc53p by the ubiquitin-related protein rub1p affects 32317–32324. function of the SCFCdc4 complex. Genes Dev. 12, 914–926. Wu, K., Chen, A., Tan, P., Pan, Z.Q., 2002. The Nedd8-conjugated ROC1–CUL1 core Liakopoulos, D., Doenges, G., Matuschewski, K., Jentsch, S., 1998. A novel protein ubiquitin ligase utilizes Nedd8 charged surface residues for efficient polyubiquitin modification pathway related to the ubiquitin system. EMBO J. 17, 2208–2214. chain assembly catalyzed by Cdc34. J. Biol. Chem. 277, 516–527. Liu, J., Furukawa, M., Matsumoto, T., Xiong, Y., 2002. NEDD8 modification of CUL1 Wu, J.T., Lin, H.C., Hu, Y.C., Chien, C.T., 2005. Neddylation and deneddylation regulate dissociates p120(CAND1), an inhibitor of CUL1–SKP1 binding and SCF ligases. Mol. Cul1 and Cul3 protein accumulation. Nat. Cell Biol. 7, 1014–1020. Cell 10, 1511–1518. Wu, J.T., Chan, Y.R., Chien, C.T., 2006. Protection of cullin-RING E3 ligases by CSN– Lo, S.C., Hannink, M., 2006. CAND1-mediated substrate adaptor recycling is required for UBP12. Trends Cell Biol. 16, 362–369. efficient repression of Nrf2 by Keap1. Mol. Cell. Biol. 26, 1235–1244. Xu, L., Wei, Y., Reboul, J., Vaglio, P., Shin, T.H., Vidal, M., Elledge, S.J., Harper, J.W., 2003. Min, K.W., Hwang, J.W., Lee, J.S., Park, Y., Tamura, T.A., Yoon, J.B., 2003. TIP120A BTB proteins are substrate-specific adaptors in an SCF-like modular ubiquitin ligase associates with cullins and modulates ubiquitin ligase activity. J. Biol. Chem. 278, containing CUL-3. Nature 425, 316–321. 15905–15910. Yeh, E.T., Gong, L., Kamitani, T., 2000. Ubiquitin-like proteins: new wines in new bottles. Ohh, M., Kim, W.Y., Moslehi, J.J., Chen, Y., Chau, V., Read, M.A., Kaelin Jr., W.G., 2002. An Gene 248, 1–14. intact NEDD8 pathway is required for Cullin-dependent ubiquitylation in Zhang, Q., Zhang, L., Wang, B., Ou, C.Y., Chien, C.T., Jiang, J., 2006. A hedgehog-induced mammalian cells. EMBO Rep. 3, 177–182. BTB protein modulates hedgehog signaling by degrading Ci/Gli transcription factor. Ohta, T., Michel, J.J., Schottelius, A.J., Xiong, Y., 1999. ROC1, a homolog of APC11, Dev. Cell 10, 719–729. represents a family of cullin partners with an associated ubiquitin ligase activity. Zhang, W., Ito, H., Quint, M., Huang, H., Noel, L.D., Gray, W.M., 2008. Genetic analysis of Mol. Cell 3, 535–541. CAND1–CUL1 interactions in Arabidopsis supports a role for CAND1-mediated Oron, E., Mannervik, M., Rencus, S., Harari-Steinberg, O., Neuman-Silberberg, S., Segal, cycling of the SCFTIR1 complex. Proc. Natl. Acad. Sci. U. S. A. 105, 8470–8475. D., Chamovitz, D.A., 2002. COP9 signalosome subunits 4 and 5 regulate multiple Zheng, J., Yang, X., Harrell, J.M., Ryzhikov, S., Shim, E.H., Lykke-Andersen, K., Wei, N., pleiotropic pathways in Drosophila melanogaster. Development 129, 4399–4409. Sun, H., Kobayashi, R., Zhang, H., 2002. CAND1 binds to unneddylated CUL1 and Osaka, F., Kawasaki, H., Aida, N., Saeki, M., Chiba, T., Kawashima, S., Tanaka, K., Kato, S., regulates the formation of SCF ubiquitin E3 ligase complex. Mol. Cell 10, 1998. A new NEDD8-ligating system for cullin-4A. Genes Dev. 12, 2263–2268. 1519–1526.