DCNL1 Functions As a Substrate Sensor and Activator of Cullin 2-RING Ligase

DCNL1 Functions as a Substrate Sensor and Activator of Cullin 2-RING Ligase

Pardeep Heir, Roxana I. Sufan, Samantha N. Greer, Betty P. Poon, Jeffrey E. Lee, Michael Ohh Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, Ontario, Canada Downloaded from Substrate engagement by F-box proteins promotes NEDD8 modification of cullins, which is necessary for the activation of cullin-RING E3 ubiquitin ligases (CRLs). However, the mechanism by which substrate recruitment triggers cullin neddylation remains unclear. Here, we identify DCNL1 (defective in cullin neddylation 1-like 1) as a component of CRL2 called ECV (elongins BC/CUL2/VHL) and show that molecular suppression of DCNL1 attenuates CUL2 neddylation. DCNL1 via its DAD patch binds to CUL2 but is also able to bind VHL independent of CUL2 and the DAD patch. The engagement of the substrate hypoxia-inducible factor 1␣ (HIF1␣) to the substrate receptor VHL increases DCNL1 binding to VHL as well as to CUL2. Notably, an engi- neered mutant form of HIF1␣ that associates with CUL2, but not DCNL1, fails to trigger CUL2 neddylation and retains ECV in an inactive state. These findings support a model in which substrate engagement prompts DCNL1 recruitment that facilitates the initiation of CUL2 neddylation and define DCNL1 as a “substrate sensor switch” for ECV activation. http://mcb.asm.org/

ullin-RING ligases (CRLs), the largest family of E3 ubiquitin A235, and D241 in DCNL1), and it can also bind Ubc12 (12). It Cligases, are multiprotein complexes involved in protein deg- has been proposed that it is the concerted action of Dcn1 and Rbx1 radation (1, 2). They are assembled on a cullin scaffold (CUL1, that promotes cullin neddylation (14). However, the molecular CUL2, CUL3, CUL4A, CUL4B, and CUL5) and contain a RING events responsible for recruiting DCNL1 to cullin scaffolds for finger protein (RBX1 or RBX2) at the cullin C-terminal domain neddylation have remained unclear. (CTD) as well as a substrate receptor, which generally associates von Hippel-Lindau tumor suppressor protein (VHL) is the with the cullin N-terminal domain (NTD) via adaptor proteins substrate receptor of the CUL2-based CRL called ECV (elongins on December 14, 2020 at UNIV OF TORONTO (2). CUL3-based CRLs are the exception in which the Broad com- BC/CUL2/VHL) (15). VHL contains an ␣ domain, which binds plex, Tramtrack, and Bric-a-Brac (BTB)-domain containing pro- the elongin C adaptor protein that is critical for bridging VHL to tein mediates both substrate specificity and binding to CUL3 CUL2, and a ␤ domain, which recognizes substrates that have without the need of adaptor proteins (3). been posttranslationally modified for subsequent ubiquitylation The process of attaching ubiquitin onto a CRL substrate re- (16, 17). Hypoxia-inducible factor ␣ (HIF␣), the best-character- quires three enzymes. Using ATP, an E1 activating enzyme forms ized substrate of ECV, is targeted for polyubiquitylation upon a thioester bond between its catalytic cysteine residue and the hydroxylation of conserved proline residues within the oxygen- C-terminal carboxyl group of ubiquitin (4). The activated ubiqui- dependent degradation (ODD) domain, which is catalyzed by tin is then transferred onto the catalytic cysteine of an E2 conju- prolyl-hydroxylase enzymes in the presence of oxygen (18–21). gating enzyme (4). CRL through RBX1/2 helps transfer ubiquitin Under reduced oxygen tension or hypoxia, HIF␣ remains unhy- from E2 onto a lysine residue on the bound substrate (2). Neddy- droxylated, escapes ECV-dependent degradation, and het- lation, the process of conjugating the ubiquitin-like protein erodimerizes with HIF␤ to form an active transcription factor for NEDD8 onto a substrate, follows an analogous cascade (5, 6). the transactivation of numerous genes containing hypoxia-re- Cullins are modified by NEDD8 (7). Structural studies involving sponsive elements (HREs) to trigger various adaptive responses to CTD CTD the CTD of CUL5 (CUL5 )-RBX1 and NEDD8ϳCUL5 - hypoxia, including glycolysis, erythropoiesis, and angiogenesis RBX1 have demonstrated that in an unneddylated state, RBX1 has (22). Notably, mutations in VHL cause VHL disease, which is limited movement due to inhibitory cullin subdomains, thus re- commonly characterized by the stabilization of HIF␣, overexpres- straining its ability to recruit ubiquitin-charged E2 (8). Neddyla- sion of hypoxia-inducible genes, and the development of multiple tion causes a conformational change in cullin structure that re- tumors in several organ systems such as the central nervous system leases RBX1 from a confined configuration to enhance the binding and the kidney (23). of ubiquitin-charged E2 and thus CRL ubiquitylation capability Engagement of various substrates to their respective CRLs, in- (8, 9). The molecular events involved in the process of neddylation of cullins are not as straightforward as the resulting effects. For CUL1 Received 4 October 2012 Returned for modification 30 October 2012 neddylation to take place, rotation of the RING domain of RBX1 is Accepted 4 February 2013 needed in order to bring the catalytic Cys111 of UBC12 (NEDD8 Published ahead of print 11 February 2013 E2 enzyme) in close proximity to CUL1 Lys720, the NEDD8 ac- Address correspondence to Michael Ohh, [email protected]. ceptor site (10). Defective in cullin neddylation 1 (Dcn1) in Sac- Supplemental material for this article may be found at http://dx.doi.org/10.1128 charomyces cerevisiae (DCNL1 is the human homolog; also known /MCB.01342-12. as SCCRO or DCUN1D1) has been reported to act as an E3 Copyright © 2013, American Society for Microbiology. All Rights Reserved. NEDD8 ligase of cullins (11–13). Dcn1 is able to bind Cdc53 doi:10.1128/MCB.01342-12 (CUL1 homolog) through its DAD patch (amino acids D211,

April 2013 Volume 33 Number 8 Molecular and Cellular Biology p. 1621–1631 mcb.asm.org 1621 Heir et al.

cluding HIF␣ to ECV, has been shown to induce the neddylation by GenScript and inserted into pcDNA3 using HindIII and BamHI. Con- of cullins, a phenomenon referred to as “substrate-mediated ned- structs that were cloned are described in Table S1 in the supplemental dylation” (24–30). Although the precise molecular mechanism by material. which substrate engagement triggers cullin neddylation and Immunoprecipitation and immunoblotting. Cells were harvested in thereby E3 enzymatic activity remains largely unknown, substrate EBC lysis buffer (50 mM Tris, pH 8, 120 mM NaCl, 0.5% NP-40) and receptors would be predicted to play an important, perhaps criti- supplemented with protease inhibitors (Roche) and MG132 (Peptides International) where indicated in the figure legends. Lysates were immu- cal, role in the substrate-induced CRL activation event, consider- noprecipitated using the indicated antibodies along with protein A-Sep- ing that the substrate receptors serve as the bridge connecting harose (Repligen). Bound proteins were washed five times in NETN buf- substrates to cullins. In this regard, components of the ECV com- fer (20 mM Tris, pH 8, 100 mM NaCl, 1 mM EDTA, 0.5% NP-40), eluted Downloaded from plex, namely, the VHL/elongins BC (VBC) subcomplex, have by boiling in sample buffer, and resolved by SDS-PAGE. Proteins were been crystallized in the presence and absence of a prolyl-hydroxy- electrotransferred onto polyvinylidene difluoride (PVDF) membrane lated HIF1␣ peptide (17, 31, 32). However, the addition of prolyl- (Bio-Rad), blocked, and probed with the antibodies indicated on the fig- hydroxylated HIF1␣ peptide did not cause an appreciable change ures. to the structure of VBC, leaving the mechanism of substrate-me- Luciferase assay. Cells were transfected in triplicate with pGL3- diated neddylation a mystery. 5ϫHRE-Luc (where Luc is luciferase) (kind gift from Richard P. Hill) and Here, we show that DCNL1 is a “substrate sensor switch” pRL-SV40 (where SV40 is simian virus 40) (Promega) Renilla luciferase ␮ whose incorporation into the ECV complex increases upon the control. Cells were treated with 10 M MG132 for4htohelp stabilize ␣ endogenous HIF1␣ in normoxia and lysed at 48 h posttransfection. A

recognition of HIF1 by VHL and activates the E3 activity via http://mcb.asm.org/ dual-luciferase reporter assay system (Promega) was used, and lumines- promoting CUL2 neddylation. cence was measured using a Mithras LB940 plate reader (Berthold Tech- MATERIALS AND METHODS nologies). Relative light units (RLU) from firefly luciferase were normal- ized against Renilla luciferase values. Values were standardized against the Cells. 786-O, HEK293A, HEK293T, PC3, and U2OS cells (American Type value for the pGIPZ control cell line, which was arbitrarily set to 100. Culture Collection) were maintained in Dulbecco’s modified Eagle’s me- Lentiviral production and generation of stable shRNA knockdown dium supplemented with 10% heat-inactivated fetal bovine serum cell lines. The following shRNA constructs were obtained from Thermo (Wisent) at 37°C in a humidified atmosphere with 5% CO . TF-1 cells 2 Scientific: pGIPZ (catalog number RHS4346), shRNA targeting DCNL1 (American Type Culture Collection) were maintained in RPMI 1640 me- clone 1 (shDCNL1-1; V3LHS_371768), shDCNL1 clone 2 (shDCNL1-2; dium supplemented with 10% heat-inactivated fetal bovine serum V3LHS_371770), shDCNL3 (V3LHS_360875), and shVHL (V2LHS_

(Wisent) and 2 ng/ml granulocyte-macrophage colony-stimulating factor on December 14, 2020 at UNIV OF TORONTO 202399). HEK293T cells were transfected with psPAX2, pMDG1.vsvg (GM-CSF; Invitrogen). HEK293A and U2OS cells stably expressing short (kind gifts from Linda Z. Penn), and shRNA-containing plasmid. Lenti- hairpin RNA (shRNA) were maintained in 2 ␮g/ml puromycin. virus containing supernatant was collected at 48 and 72 h posttransfection Antibodies and reagents. The following antibodies were obtained and was used for infection after being filtered. HEK293A and U2OS cells from Cell Signaling Technology: glutathione S-transferase (GST; catalog number 2622), hemagglutinin (HA) (mouse, 2367; rabbit, 3724), VHL were selected in puromycin 24 h after infection. Protein purification. pGEX-4T-1-DCNL1 was expressed in (2738), and NEDD8 (2745). The following antibodies were obtained from ␤ Sigma: FLAG-M2 (catalog number F1804), RBX1 (R4402), vinculin BL21(DE3) Escherichia coli and induced with isopropyl- -D-thiogalacto- (V9264), and ␤-actin (A5316). The following antibodies were obtained pyranoside (IPTG; Sigma). Pellets were resuspended in NETN buffer with from Santa Cruz Biotechnology: IgG (catalog number sc-2025), elongin B protease inhibitors (Roche), sonicated, cleared, and incubated with glu- (sc-11447), DCNL1 (sc-81835), His (sc-8036), and GAL4 (sc-510). FLAG tathione-Sepharose beads (Amersham Biosciences). The beads were (catalog number NB100-63146), DCNL3 (NBP1-55423), and HIF1␣ washed, and GST-DCNL1 was eluted using GST elution buffer (30 mM (NB100-105) antibodies were obtained from Novus Biologicals. Antibodies glutathione, 50 mM Tris, pH 8). pET-46 EK/LIC VHL(1–155) and pET- against VHL (catalog number 556347) and HIF1␣ (610958) were obtained 15b-DCNL1 were expressed in Rosetta-2 and BL21 cells, respectively, and from BD Biosciences. Anti-CUL2 (catalog number 51-1800), anti-T7 induced with IPTG. Pellets were resuspended in Ni-nitrilotriacetic acid ϫ (69522), anti-DCNL1 (ab57255), and anti-HA (12CA5) were obtained from (NTA) buffer (50 mM Tris, pH 8, 300 mM NaCl, 20 mM imidazole, 1 Invitrogen, Novagen, Abcam, and Boehringer Ingelheim, respectively. bacterial protein extraction reagent [B-PER]) with protease inhibitors MG132 (catalog number IZL-3175-v) was obtained from Peptides Interna- (Roche), sonicated, cleared, and incubated with Ni-NTA (Invitrogen). tional. Streptavidin-agarose resin was obtained from Thermo Scientific. His-tagged proteins were eluted using an imidazole gradient. Plasmids. The following plasmids have been previously described: siRNA transfection. An On-Target Plus Smart pool small interfering pcDNA3-T7-NEDD8, pcDNA3-T7-NEDD8⌬GG (where NEDD8⌬GG RNA (siRNA) targeting DCNL1 or scrambled nontargeting RNA control represents NEDD8 lacking the C-terminal glycine-glycine residues), siRNAs were obtained from Dharmacon. Cells were transfected with 150 pcDNA3-HA-CUL2, pcDNA3-HA-CUL2(K689R) (where CUL2 has the nM siRNA using X-tremeGENE siRNA transfection reagent (Roche) for mutation K689R), pcDNA3-T7-VHL, pcDNA3-HA-HIF1␣, pcDNA3- two consecutive days prior to lysis. HA-HIF2␣, pcDNA3-GAL4-HA, pcDNA3-GAL4-HA-HIF1␣ODD, In vitro binding assay. In vitro translation (IVT) was performed using pcDNA3-GAL4-HA-HIF1␣(555–575) [where HIF1␣(555–575) is HIF1␣ a TNT rabbit reticulocyte lysate in vitro transcription/translation system consisting of residues 555 to 575], pRc/CMV-HA-VHL, pRc/CMV-HA- (Promega) or a TNT-coupled wheat germ extract system (Promega). The VHL19, pRc/CMV-HA-VHL(1–155) [where VHL(1–155) is VHL translation products were mixed with purified protein at 30°C for 1 h. For consisting of residues 1 to 155], pRc/CMV-HA-VHL(C162F), pRc/CMV- binding assays not involving IVT, purified proteins were incubated to- HA-VHL(Y98H), and pcDNA3-HA-HIF3␣ (15, 16, 18, 27, 33, 34). HA- gether in 30 ␮l of binding buffer (50 mM Tris, pH 8, 120 mM NaCl, 0.1% SOCS1- and HA-HIF1␣(P564A)-encoding plasmids were generously NP-40, 5% glycerol) at 30°C for 1 h. Thereafter, 1 ml of binding buffer was provided by Robert Rottapel and William Y. Kim, respectively. pCMV6- added along with either Ni-NTA or glutathione-Sepharose. Bound pro- DCNL1-FLAG (catalog number RC203427), pCMV6-DCNL2-FLAG teins were washed five times in binding buffer prior to elution by boiling (RC208865), and pCMV6-DCNL3-FLAG (RC208082) were obtained in sample buffer. from Origene. Site-directed mutagenesis was performed using a In vitro neddylation and ubiquitylation assays. S100 extract prepa- QuikChange kit (Stratagene). Plasmids were verified by direct DNA se- ration and in vitro ubiquitylation assays were performed as previously quencing. Streptavidin-binding peptide (SBP) tag cDNA was synthesized described (16). In vitro neddylation was performed as previously de-

1622 mcb.asm.org Molecular and Cellular Biology DCNL1 Drives Substrate-Mediated CUL2 Neddylation Downloaded from http://mcb.asm.org/ on December 14, 2020 at UNIV OF TORONTO

FIG 1 HIF␣ isoforms promote CUL2 neddylation. HEK293 cells transiently transfected with the indicated plasmids were treated with 10 ␮M MG132 protea- somal inhibitor and lysed. Equivalent amounts of protein lysates were immunoprecipitated with the indicated antibodies, resolved by SDS-PAGE, and immunoblotted with the indicated antibodies. WCE, whole-cell extract; IP, immunoprecipitation. scribed (35) with 2 pmol of GST-UBC12 (Boston Biochem) and 20 pmol dylation status (Fig. 1D). Considering that the addition of short of His-DCNL1 used to supplement the reaction mixture. prolyl-hydroxylated HIF1␣ peptide was not associated with an appreciable change to the structure of VBC subcomplex (17, 31, RESULTS 32), these results suggest that additional sequences within HIF1␣ Multiple HIF␣ isoforms trigger CUL2 neddylation. We have ␣ beyond the substrate-binding region promote the initiation of CUL2 previously shown that HIF1 promotes CUL2 neddylation (27). neddylation. It is, however, formally unclear at present whether a We asked whether the other HIF␣ isoforms induced a similar conformational change was induced upon the recruitment of full- modiﬁcation on CUL2. Ectopic expression of HIF2␣ or HIF3␣ length HIF1␣ or whether HIF1␣ permitted the recruitment of an bound to VHL was likewise associated with an increased level of additional factor(s) required for CUL2 neddylation. neddylated CUL2 (Fig. 1A and B; see also Fig. S1 in the supplemental material). Notably, the level of RBX1 did not appreciably DCNL1 promotes CUL2 neddylation. DCNL1 has been change, irrespective of substrate engagement to ECV (Fig. 1A). shown to cooperate with RBX1 as an E3 ligase in the neddylation Intriguingly, a short fragment of HIF1␣, HIF1␣(555–575), of CUL1 (12). We asked whether DCNL1 played a similar role in comparable in size to that which was used in the crystallization CUL2 neddylation. Increased ectopic expression of DCNL1 was with VBC complex (17, 31, 32), failed to promote robust CUL2 associated with elevated levels of neddylated CUL2 (Fig. 2A). neddylation in comparison to full-length HIF1␣ (Fig. 1C). The DCNL1-mediated neddylation of CUL2 was more pronounced in isolated ODD domain (residues 530 to 652) showed relatively the presence of ectopic wild-type (WT) NEDD8 and reduced modest induction of CUL2 neddylation (Fig. 1C and D). Notably, in the presence of NEDD8⌬GG, which is a form of NEDD8 that the core components of ODD domain- and HIF1␣(555-575)-en- lacks the C-terminal glycine-glycine residues and cannot be con- gaged ECV, including VHL, RBX1, elongin B, and CUL2, re- jugated (Fig. 2B)(5). As expected, the CUL2(K689R) mutant that mained comparable despite the marked difference in CUL2 ned- lacks the conserved NEDD8 acceptor site K689 (35, 36) was not

April 2013 Volume 33 Number 8 mcb.asm.org 1623 Heir et al. Downloaded from http://mcb.asm.org/ on December 14, 2020 at UNIV OF TORONTO

FIG 2 DCNL1 promotes NEDD8 modiﬁcation of CUL2. (A) U2OS cells were transiently transfected with increasing amounts of DCNL1-FLAG. At 48 h posttransfection, cells were lysed, and an equivalent amount of protein was resolved using SDS-PAGE. The resolved proteins were detected by immunoblotting using the indicated antibodies. (B to E) HEK293 cells were transiently transfected with the indicated plasmids. At 48 h posttransfection immunoprecipitation and immunoblotting with the indicated antibodies were performed. (F) HEK293, U2OS, or TF-1 cell lysate was immunoprecipitated using anti-DCNL1 or IgG control antibody, resolved, and immunoblotted with the indicated antibodies. WCE, whole-cell extract; IP, immunoprecipitation.

neddylated in the presence of ectopic DCNL1 and NEDD8 CUL2 (Fig. 2) as well as other cullins (12), and the engagement of (Fig. 2C). The combined expression of DCNL1 and NEDD8 also HIF1␣ to ECV has been shown previously to promote CUL2 ned- resulted in the formation of multiple NEDD8 moieties on CUL2, dylation (27, 37). We asked whether substrate-mediated CUL2 which was most likely the result of polyneddylation on the single neddylation is coordinated with DCNL1. First, endogenous K689 residue rather than multimononeddylation on multiple DCNL1 coprecipitated endogenous CUL2, VHL, and HIF1␣ un- lysines since the CUL2(K689R) mutant failed to show this modi- der normoxia (in the presence of proteasome inhibitor to stabilize fication profile (Fig. 2C). The DAD patch of DCNL1 is known to the otherwise ECV-targeted HIF1␣ in the presence of oxygen) be crucial for binding to CUL1 since the triple point mutant (Fig. 3A). However, under hypoxia, a condition that prohibits the DCNL1(D211A A235R D241A)[DCNL1(ARA)] is defective in interaction between VHL and HIF1␣, the level of ectopic DCNL1- CUL1 binding (12). The DCNL1(ARA) mutant was likewise de- coprecipitated CUL2, VHL, and HIF1␣ dramatically decreased fective in binding to CUL2 (Fig. 2D) and showed reduced ability (Fig. 3B, compare lanes 9 and 10). These results suggest that to promote NEDD8 chain formation on CUL2 in the presence of DCNL1 is preferentially associated with ECV that has engaged its T7-NEDD8 (Fig. 2E). Notably, endogenous DCNL1 coprecipi- substrate. Consistent with this notion, ectopic expression of the tated with endogenous CUL2 in several cell lines examined HIF1␣(P564A) mutant, which fails to be prolyl-hydroxylated due (Fig. 2F). These results demonstrate that DCNL1 promotes to a conserved proline-to-alanine substitution and hence escapes NEDD8 modification of CUL2. VHL recognition, diminished the level of CUL2 and VHL that DCNL1 interacts with ECV and triggers CUL2 neddylation coprecipitated with DCNL1 in comparison to HEK293 cells ex- in the presence of HIF1␣. DCNL1 promotes neddylation of pressing wild-type HIF1␣ (Fig. 3C). Similarly, the absence of ec-

1624 mcb.asm.org Molecular and Cellular Biology DCNL1 Drives Substrate-Mediated CUL2 Neddylation Downloaded from http://mcb.asm.org/ on December 14, 2020 at UNIV OF TORONTO

FIG 3 DCNL1 engages the ECV complex. (A) PC3 cells were treated with 10 ␮M MG132 for 4 h, lysed, and immunoprecipitated using anti-DCNL1 or control antibodies. Bound protein was resolved by SDS-PAGE and immunoblotted with the indicated antibodies. (B and C) HEK293 cells were transfected with the indicated plasmids. At 4 h prior to lysis, cells were treated with MG132. Cells were lysed at 48 h posttransfection. Anti-FLAG immunoprecipitation and immunoblotting were performed using the indicated antibodies. Cells were maintained in hypoxia, where indicated, immediately after the addition of MG132 for 4 h preceding lysis. (D) S100 extracts generated from VHL-null or VHL-reconstituted cells were incubated with the indicated purified proteins or with rabbit reticulocyte lysate in vitro-translated protein products. Anti-HA immunoprecipitation was performed, and bound proteins were resolved by SDS-PAGE and detected by immunoblotting. (E) The indicated HEK293 stable cell lines were transiently transfected with HA-VHL or empty plasmid. Cells were treated with MG132 for 4 h and lysed, and anti-HA immunoprecipitation was performed. Bound protein was resolved by SDS-PAGE and detected through immunoblotting. WCE, whole-cell extract; IP, immunoprecipitation. topic expression of either wild-type HIF1␣ or VHL attenuated the mented with purified GST-UBC12 and His-DCNL1 with in vitro level of CUL2 coprecipitated with DCNL1 (Fig. 3C). These results translated HIF1␣ or empty plasmid. The lowest level of neddy- suggest that the process of HIF1␣ engagement to VHL promotes lated CUL2 coprecipitating with VHL was observed in reaction DCNL1 interaction with VHL and CUL2. mixtures devoid of HIF1␣, irrespective of the presence of DCNL1 In a complementary experiment, we asked whether DCNL1 (Fig. 3D). The addition of in vitro translated HIF1␣ promoted an facilitates the increased neddylation of CUL2 upon substrate en- increased level of neddylated CUL2, which further increased, al- gagement. An in vitro neddylation reaction was performed using beit modestly, upon the addition of purified exogenous His- VHL-null and VHL-reconstituted S100 cellular extracts supple- DCNL1 (Fig. 3D). Importantly, stable shRNA-mediated knock-

April 2013 Volume 33 Number 8 mcb.asm.org 1625 Heir et al. Downloaded from http://mcb.asm.org/

FIG 4 DCNL1 modulates HIF1␣ activity. (A) HEK293 cells were transfected with nontargeting scrambled siRNA (siSCR) or DCNL1-targeting siRNA (siDCNL1). An equivalent amount of protein lysate was resolved by SDS-PAGE and detected by immunoblotting using the indicated antibodies. (B) Stable cell lines in the indicated cellular backgrounds were generated through lentivirus-mediated infection of two different shRNA constructs targeting DCNL1. Cells were treated with MG132 for 4 h and lysed, and an equivalent amount of protein lysate was resolved by SDS-PAGE. The indicated proteins were detected by immunoblotting. (C) The indicated stable cell lines were transiently transfected with plasmids encoding firefly luciferase under the control of a hypoxia response element and a Renilla luciferase control. Cells were treated with MG132 for 4 h prior to a dual-luciferase assay. pGIPZ was arbitrarily set to 100 to represent on December 14, 2020 at UNIV OF TORONTO normal HIF1␣ activity. Error bars represent standard deviations of the mean. An unpaired t test was performed to assess the statistical significance. down of DCNL1 reduced the level of neddylated CUL2 DCNL1 engages VHL independent of CUL2. Although RBX1 coprecipitated with HA-VHL (Fig. 3E). These results are consis- and DCNL1 have been shown to cooperate to neddylate cullins, a tent with the notion that DCNL1 promotes CUL2 neddylation fundamental difference is that RBX1 appears to be constitutively under conditions of substrate engagement via VHL. associated with CUL2 irrespective of substrate presence (Fig. 1A). DCNL1 negatively regulates the level and activity of HIF1␣. These results raised the possibility that DCNL1, and not RBX1, We have previously shown in ts41 cells, which have a temperature- requires a molecular cue to interact with and promote the neddy- sensitive mutation in the NEDD8 E1 enzyme, that at nonpermis- lation of CUL2. sive temperature CUL2 neddylation is attenuated and HIF1␣ is The DCNL1(ARA) mutant that cannot bind CUL2 was able to stabilized (35). Consistent with this observation, siRNA-mediated engage HA-VHL and HA-HIF1␣, albeit at a reduced capacity in knockdown of DCNL1 in HEK293A cells decreased the level of comparison to wild-type DCNL1 (Fig. 5A), which raises an inter- neddylated CUL2, which was, as predicted, associated with an esting possibility of CUL2 playing a role in reinforcing DCNL1- increased level of HIF1␣ (Fig. 4A). Similarly, it has been shown VHL interaction. In addition, the VHL(1–155) (␤ domain) trun- that DCNL3 knockdown stabilizes the CUL3 substrate NRF2 (37). cation mutant lacking the canonical elongin C binding region DCNL1, DCNL2 (long isoform), and DCNL3 have been shown to (and thus does not interact with CUL2) (15) coprecipitated be promiscuous in their ability to bind multiple cullins (37). Al- DCNL1(ARA), the interaction of which further increased in the though there is promiscuity among DCNL members in binding to presence of HIF1␣ (Fig. 5B). The disease-causing VHL(C162F) CUL2, DCNL1 was most associated with neddylated CUL2, fol- mutant that does not form an ECV complex due to its inability to lowed by DCNL3 (see Fig. S2A in the supplemental material). bind elongin C (15) was capable, albeit at a lower capacity, of However, CUL2 is associated with a wide variety of substrate re- binding DCNL1 (Fig. 5C). Moreover, bacterially purified His- ceptors, in addition to VHL (38), and thus whether DCNL1 and VHL(1–155) pulled down bacterially purified GST-DCNL1 using DCNL3 have redundant roles in the context of VHL-E3 ligase nickel agarose (Fig. 5D). These results suggest that DCNL1 is able function was unclear. Therefore, we measured the activity of a to interact with VHL independent of CUL2. Moreover, DCNL1 is VEGF HRE-luciferase reporter in two polyclonal populations, able to bind to CUL2 independent of VHL, as shown in VHL-null shDCNL1-1 and shDCNL1-2, stably transduced with unique len- or VHL knockdown cells (see Fig. S3A and B, respectively, in the tivirus-driven oligonucleotide sequences against DCNL1 (Fig. 4B; supplemental material). Other substrate receptors of the CRL see also Fig. S2B in the supplemental material), and observed that family including SOCS1 (another CUL2-based CRL) and SOCS2 the attenuation of DCNL1 expression, but not that of DCNL3 via (CUL5-based CRL) (39, 40) were also capable of binding DCNL1 lentivirus-based shDCNL3, increased the HRE-driven reporter (see Fig. S3C and D). However, SOCS3 (CUL5-based CRL) (41) activity (Fig. 4C). These results suggest that HIF activity is, at a was not able to bind to SBP-DCNL1 (see Fig. S3D). Notably, minimum, preferentially influenced by DCNL1. SOCS1 engaged DCNL1(ARA) independent of its SOCS box,

1626 mcb.asm.org Molecular and Cellular Biology DCNL1 Drives Substrate-Mediated CUL2 Neddylation Downloaded from http://mcb.asm.org/ on December 14, 2020 at UNIV OF TORONTO

FIG 5 CUL2 is not required for DCNL1 binding to VHL. (A, B, and C) HEK293 cells were transfected with the indicated plasmids. Cells were treated with MG132 for 4 h. Cells were lysed, immunoprecipitated using an anti-FLAG antibody, resolved by SDS-PAGE, and immunoblotted using the indicated antibodies. (D) Bacterially purified GST or GST-DCNL1 was incubated with bacterially purified His-VHL(1–155). His pulldown (PD) was performed using nickel agarose; proteins were then resolved by SDS-PAGE and detected by immunoblotting using antibodies directed against GST and His. (E) Rabbit reticulocyte lysate in vitro translated HA-HIF1␣ and HA-VHL were incubated with bacterially purified GST or GST-DCNL1. GST pulldown was performed using glutathione-Sepharose. Proteins were resolved by SDS-PAGE and immunoblotted using the indicated antibodies. (F) HA-HIF1␣ was in vitro translated using either rabbit reticulocyte lysate or wheat germ extract and incubated with bacterially purified GST or GST-DCNL1. GST pulldown was performed using glutathione-Sepharose, and bound proteins were resolved by SDS-PAGE and detected using the indicated antibodies. WCE, whole-cell extract; IP, immunoprecipitation. which, similar to the ␣ domain of VHL, is required for CUL2 VHL. However, these findings could not exclude the possibility of interaction via elongin C (see Fig. S3C). These results suggest that a direct interaction between DCNL1 and HIF1␣. To test this no- certain, but not all, substrate receptors bind DCNL1. tion, HA-HIF1␣ was in vitro translated in either rabbit reticulo- Intriguingly, the addition of in vitro translated HIF1␣ in- cyte lysate or wheat germ extract, and an in vitro binding assay was creased the interaction between in vitro translated VHL and puri- performed with bacterially purified GST-DCNL1. GST-DCNL1 fied GST-DCNL1 (Fig. 5E), and the binding between the disease- was able to pull down only HIF1␣ that was in vitro translated in causing VHL(Y98H) that has a compromised ability to engage rabbit reticulocyte lysate (Fig. 5F), which contains enzymes that HIF␣ (16) showed diminished binding to DCNL1 in comparison prolyl-hydroxylate HIF1␣ to promote binding to VHL (18). These to wild-type VHL (Fig. 5C). These results suggest that a substrate- findings raise a potential model in which VHL upon engagement engaged VHL has a greater affinity for DCNL1 than unengaged of HIF␣ increases the recruitment of DCNL1, which subsequently

April 2013 Volume 33 Number 8 mcb.asm.org 1627 Heir et al. binds CUL2 via its DAD patch to accentuate CUL2 neddylation in reduced ability to promote an interaction between DCNL1 and cooperation with RBX1. VHL (Fig. 7). These results suggest that despite the proper capacity DCNL1 binding to VHL is required for CUL2 neddylation to form an ECV complex, the HIF1␣(555–575) peptide fails to and HIF1␣ ubiquitylation. A prediction based on the above-pro- promote efficient CUL2 neddylation due, at least in part, to atten- posed model is that VHL that is specifically defective in binding uated recruitment of DCNL1. DCNL1 would be ineffectual in promoting ubiquitin-mediated degradation of HIF1␣ even if recruited to ECV. In an effort to test DISCUSSION this notion, the HIF1␣ ODD domain with a P564A mutation The notion that cullin neddylation precedes CRL-mediated sub- [HIF1␣ODD(P¡A)] fused to the ␣ domain of VHL(156–213) strate ubiquitylation has been extensively studied using a combi- Downloaded from was generated (Fig. 6A). HIF1␣ODD(P¡A)/VHL␣ chimeric nation of biochemical and structural studies. However, the ques- protein is unable to recruit endogenous VHL due to the conserved tion of why and how substrate engagement is seemingly linked to proline-to-alanine substitution within the ODD domain, but is the initiation of cullin neddylation has remained unclear. First, it able to recruit CUL2 via the fused ␣ domain of VHL, which is not would be an energetically unfavorable process that allows NEDD8 required for binding DCNL1 (see Fig. S4 in the supplemental ma- conjugation to CRLs that have not engaged a substrate since the terial). The generation of the chimeric protein was necessary as we process of NEDD8 activation requires ATP (4). Second, NEDD8 were unsuccessful in identifying a VHL mutant that retained modification of substrate-unengaged CRLs would compete HIF1␣ binding while being defective in DCNL1 binding, which against the substrate-engaged CRLs for active ubiquitin-carrying ␣ reinforces the notion that both the HIF1 and DCNL1 contact E2s. Thus, it seems reasonable that substrate engagement to the http://mcb.asm.org/ sites are within the ␤ domain of VHL. DCNL1 pulled down VHL substrate-conferring component of CRL be coordinated with the in the presence of the wild-type HIF1␣ ODD domain ubiquitin E3 activity-triggering event, the NEDD8 modification [HIF1␣ODD(WT)], which was difficult to visualize due to its in- of cullins. Here, we show that HIF1␣ binding to VHL promotes herent instability in mammalian lysate but not in the presence of the recruitment of DCNL1 to initiate CUL2 neddylation. DCNL1 HIF1␣ODD(P¡A) or HIF1␣ODD(P¡A)/VHL␣ (Fig. 6B). in essence acts as a substrate sensor and switch that activates ECV These results suggest that HIF1␣ODD(WT) is able to strongly upon substrate engagement. promote the interaction between DCNL1 and VHL and further The observations that both HIF1␣ and DCNL1 engage the ␤ support the notion that unhydroxylated HIF1␣ and the ␣ domain domain of VHL and that the interaction is strengthened in the of VHL are not able to bind to DCNL1. As expected, presence of all three components are intriguing. These results sug- HIF1␣ODD(WT), but not HIF1␣ODD(P¡A), bound endoge- gest that HIF1␣ binding to VHL causes a conformational change on December 14, 2020 at UNIV OF TORONTO nous CUL2 via endogenous VHL (Fig. 6C; see also Fig. S4). As in the ␤ domain that increases the affinity for DCNL1 binding. predicted, HIF1␣ODD(P¡A)/VHL␣ associated with endoge- However, careful examination of the crystal structures of VHL nous CUL2 via the fused ␣ domain of VHL since bound to a small HIF1␣ peptide and unbound indicates that there HIF1␣ODD(P¡A) is defective in binding endogenous VHL is not an appreciable variation between the two structures (17, 31, (Fig. 6C; see also Fig. S4). Importantly, the chimeric protein ex- 32). It is, however, possible that the short peptide used for crystal- clusively associated with unneddylated CUL2 (Fig. 6C; see also lization did not faithfully capture the potential conformational Fig. S4). changes introduced by a full-length HIF1␣ protein. Alternatively, Cullin neddylation is required for the efficient polyubiquityla- HIF1␣ may provide additional contacts for DCNL1 to generate a tion of CRL substrates. Consistent with this notion, GAL4-HA- higher-affinity ternary structure between VHL, HIF1␣, and HIF1␣ODD(WT) was efficiently polyubiquitylated in the pres- DCNL1. Consistent with these notions, the short HIF1␣ peptide ence of S100 cellular extract (devoid of proteasome) generated failed to promote CUL2 neddylation (Fig. 1B) despite being from VHL-null 786-O clear cell renal cell carcinoma (CCRCC) bound to VHL and recruited to the ECV complex. Following the cells reconstituted with wild-type VHL, while GAL4-HA- substrate-induced recruitment of DCNL1, the DAD patch allows HIF1␣ODD(P¡A)/VHL␣ showed an attenuated polyubiquityla- DCNL1 to interact with the cullin CTD to facilitate neddylation of tion pattern (Fig. 6D). These results suggest that the impaired CUL2 in cooperation with RBX1 and the E2 NEDD8-conjugating polyubiquitylation of GAL4-HA-HIF1␣ODD(P¡A)/VHL␣ is a enzyme, UBC12. Subsequently, ubiquitin-charged E2 is recruited result of its inability to trigger CUL2 neddylation (Fig. 6D). As via RBX1 to ultimately polyubiquitylate HIF1␣ substrate. expected, S100 extract produced from VHL-null 786-O CCRCC The vastly diversified and multitudinous CRL substrates are cells stably transfected with empty plasmid failed to polyubiqui- unlikely to share a structural element with affinity for DCNL1; tylate GAL4-HA-HIF1␣ODD(WT) (Fig. 6D). These results sup- rather, a possibility is that the CRL substrates would induce a port the notion that the increased recruitment of DCNL1 via VHL conformational change to their cognate substrate recognition upon HIF1␣ engagement promotes CUL2 neddylation and con- components that increases their affinity for DCNL1 or proteins sequential ECV activation. with an analogous function. We show here that VHL, SOCS1, and HIF1␣(555–575) peptide has reduced capacity to promote SOCS2, but not SOCS3, interact with DCNL1, suggesting that all VHL-DCNL1 interaction. Although there were marked differ- substrate-conferring proteins do not bind to DCNL1. In this re- ences between the small HIF1␣ peptide spanning the residues 555 gard, there are additional DCNL proteins in humans (DCNL2-1, to 575 and the ODD domain in their capacities to promote CUL2 DCNL2-2, DCNL3, DCNL4-1, DCNL4-2, and DCNL5), and neddylation (Fig. 1D), the integration of the CUL2 holoenzyme, DCNL1 and DCNL2-1 have been shown to bind CUL1, CUL2, including VHL and RBX1, was not appreciably affected (Fig. 1D). CUL3, CUL4A, CUL4B, and CUL5, while DCNL3 has been shown Therefore, we asked whether the HIF1␣(555–575) peptide has a to stimulate CUL3 neddylation at the plasma membrane to possi- diminished capacity to promote VHL-DCNL1 interaction. In bly facilitate ubiquitylation of membrane-associated substrates comparison to HIF1␣ODD, HIF1␣(555–575) showed markedly (37). This would suggest that certain CRL3 substrates upon bind-

1628 mcb.asm.org Molecular and Cellular Biology DCNL1 Drives Substrate-Mediated CUL2 Neddylation Downloaded from http://mcb.asm.org/ on December 14, 2020 at UNIV OF TORONTO

FIG 6 DCNL1 is recruited through VHL to initiate CUL2 neddylation and HIF1␣ degradation. (A) Schematic diagram of the indicated proteins and their capacity for recruitment of CUL2 and DCNL1. (B) The indicated constructs were in vitro translated in rabbit reticulocyte lysate and incubated with bacterially puriﬁed GST or GST-DCNL1. A GST pulldown (PD) was performed using glutathione-Sepharose, and bound proteins were resolved by SDS-PAGE and detected using the indicated antibodies. (C) HEK293 cells were transfected with the indicated plasmids. Cells were harvested at 48 h posttransfection and immunoprecipitated using an anti-GAL4 antibody. Proteins were separated by SDS-PAGE and immunoblotted using the indicated antibodies. (D) The indicated plasmids were in vitro translated, and in vitro ubiquitylation was performed in the presence (ϩ) or absence (Ϫ) of FLAG-ubiquitin (FLAG-Ub) and S100 extracts generated from VHL-null or VHL-reconstituted cells. The reaction mixtures were immunoprecipitated using anti-GAL4 antibody, resolved by SDS-PAGE, and immunoblotted using the indicated antibodies. WCE, whole-cell extract; IP, immunoprecipitation; IB, immunoblot.

April 2013 Volume 33 Number 8 mcb.asm.org 1629 Heir et al.

K, Kato S. 1998. A new NEDD8-ligating system for cullin-4A. Genes Dev. 12:2263–2268. 7. Hori T, Osaka F, Chiba T, Miyamoto C, Okabayashi K, Shimbara N, Kato S, Tanaka K. 1999. Covalent modiﬁcation of all members of human cullin family proteins by NEDD8. Oncogene 18:6829–6834. 8. Duda DM, Borg LA, Scott DC, Hunt HW, Hammel M, Schulman BA. 2008. Structural insights into NEDD8 activation of cullin-RING ligases: conformational control of conjugation. Cell 134:995–1006. 9. Saha A, Deshaies RJ. 2008. Multimodal activation of the ubiquitin ligase SCF by Nedd8 conjugation. Mol. Cell 32:21–31.

10. Calabrese MF, Scott DC, Duda DM, Grace CR, Kurinov I, Kriwacki Downloaded from RW, Schulman BA. 2011. A RING E3-substrate complex poised for ubiquitin-like protein transfer: structural insights into cullin-RING ligases. Nat. Struct. Mol. Biol. 18:947–949. 11. Kurz T, Ozlu N, Rudolf F, O’Rourke SM, Luke B, Hofmann K, Hyman AA, Bowerman B, Peter M. 2005. The conserved protein DCN-1/Dcn1p is required for cullin neddylation in C. elegans and S. cerevisiae. Nature 435:1257–1261. 12. Kurz T, Chou YC, Willems AR, Meyer-Schaller N, Hecht ML, Tyers M, Peter M, Sicheri F. 2008. Dcn1 functions as a scaffold-type E3 ligase for cullin neddylation. Mol. Cell 29:23–35. 13. Kim AY, Bommelje CC, Lee BE, Yonekawa Y, Choi L, Morris LG, http://mcb.asm.org/ Huang G, Kaufman A, Ryan RJ, Hao B, Ramanathan Y, Singh B. 2008. SCCRO (DCUN1D1) is an essential component of the E3 complex for neddylation. J. Biol. Chem. 283:33211–33220. 14. Scott DC, Monda JK, Grace CR, Duda DM, Kriwacki RW, Kurz T, Schulman BA. 2010. A dual E3 mechanism for Rub1 ligation to Cdc53. Mol. Cell 39:784–796. 15. Lonergan KM, Iliopoulos O, Ohh M, Kamura T, Conaway RC, Con- away JW, Kaelin WG, Jr. 1998. Regulation of hypoxia-inducible mRNAs by the von Hippel-Lindau tumor suppressor protein requires binding to FIG 7 HIF1␣(555–575) peptide has reduced capacity to promote VHL- complexes containing elongins B/C and Cul2. Mol. Cell. Biol. 18:732–741. DCNL1 interaction. The indicated constructs were in vitro translated in rabbit 16. Ohh M, Park CW, Ivan M, Hoffman MA, Kim TY, Huang LE, Pavletich on December 14, 2020 at UNIV OF TORONTO reticulocyte lysate and incubated with bacterially purified GST or GST- N, Chau V, Kaelin WG. 2000. Ubiquitination of hypoxia-inducible factor DCNL1. GST pulldown (PD) was performed using glutathione-Sepharose, requires direct binding to the beta-domain of the von Hippel-Lindau pro- and bound proteins were resolved by SDS-PAGE and detected using the indicated antibodies. tein. Nat. Cell Biol. 2:423–427. 17. Stebbins CE, Kaelin WG, Jr, Pavletich NP. 1999. Structure of the VHL- ElonginC-ElonginB complex: implications for VHL tumor suppressor function. Science 284:455–461. ing to their cognate substrate receptors would increase recruit- 18. Ivan M, Kondo K, Yang H, Kim W, Valiando J, Ohh M, Salic A, Asara ␣ ment of DCNL3, but not necessarily of DCNL1, for subsequent JM, Lane WS, Kaelin WG, Jr. 2001. HIF targeted for VHL-mediated destruction by proline hydroxylation: implications for O2 sensing. Science neddylation of CUL3. Perhaps, similar to six cullins serving as 292:464–468. scaffolds for potentially hundreds (approximately 500) of CRLs 19. Jaakkola P, Mole DR, Tian YM, Wilson MI, Gielbert J, Gaskell SJ, (2), few DCNL members and possibly other yet-undefined pro- Kriegsheim A, Hebestreit HF, Mukherji M, Schofield CJ, Maxwell PH, teins with analogous function to DCNL1 could promote distinct Pugh CW, Ratcliffe PJ. 2001. Targeting of HIF-alpha to the von Hippel- cullin neddylation upon engagement of vastly diversified CRL Lindau ubiquitylation complex by O2-regulated prolyl hydroxylation. Science 292:468–472. substrates. 20. Epstein AC, Gleadle JM, McNeill LA, Hewitson KS, O’Rourke J, Mole DR, Mukherji M, Metzen E, Wilson MI, Dhanda A, Tian YM, Masson ACKNOWLEDGMENTS N, Hamilton DL, Jaakkola P, Barstead R, Hodgkin J, Maxwell PH, Pugh This work was supported by funds from the Canadian Institutes of Health CW, Schofield CJ, Ratcliffe PJ. 2001. C. elegans EGL-9 and mammalian homologs define a family of dioxygenases that regulate HIF by prolyl hy- Research (MOP-77718 to M.O. and MOP-115066 to J.E.L.). P.H. is the droxylation. Cell 107:43–54. recipient of the Natural Sciences and Engineering Research Council of 21. Bruick RK, McKnight SL. 2001. A conserved family of prolyl-4- Canada Postgraduate Scholarship. hydroxylases that modify HIF. Science 294:1337–1340. 22. Semenza GL. 2011. Oxygen sensing, homeostasis, and disease. N. Engl. J. REFERENCES Med. 365:537–547. 1. Petroski MD, Deshaies RJ. 2005. Function and regulation of cullin-RING 23. Kaelin WG, Jr, Ratcliffe PJ. 2008. Oxygen sensing by metazoans: the ubiquitin ligases. Nat. Rev. Mol. Cell Biol. 6:9–20. central role of the HIF hydroxylase pathway. Mol. Cell 30:393–402. 2. Zimmerman ES, Schulman BA, Zheng N. 2010. Structural assembly of 24. Chew EH, Hagen T. 2007. Substrate-mediated regulation of cullin ned- cullin-RING ubiquitin ligase complexes. Curr. Opin. Struct. Biol. 20:714– dylation. J. Biol. Chem. 282:17032–17040. 721. 25. Read MA, Brownell JE, Gladysheva TB, Hottelet M, Parent LA, Coggins 3. Watson IR, Irwin MS, Ohh M. 2011. NEDD8 pathways in cancer, Sine MB, Pierce JW, Podust VN, Luo RS, Chau V, Palombella VJ. 2000. Quibus Non. Cancer Cell 19:168–176. Nedd8 modification of Cul-1 activates SCF␤TrCP-dependent ubiquitina- 4. Schulman BA, Harper JW. 2009. Ubiquitin-like protein activation by E1 tion of I␬B␣. Mol. Cell. Biol. 20:2326–2333. enzymes: the apex for downstream signalling pathways. Nat. Rev. Mol. 26. Kawakami T, Chiba T, Suzuki T, Iwai K, Yamanaka K, Minato N, Cell Biol. 10:319–331. Suzuki H, Shimbara N, Hidaka Y, Osaka F, Omata M, Tanaka K. 2001. 5. Liakopoulos D, Doenges G, Matuschewski K, Jentsch S. 1998. A novel NEDD8 recruits E2-ubiquitin to SCF E3 ligase. EMBO J. 20:4003–4012. protein modification pathway related to the ubiquitin system. EMBO J. 27. Sufan RI, Ohh M. 2006. Role of the NEDD8 modification of Cul2 in the 17:2208–2214. sequential activation of ECV complex. Neoplasia 8:956–963. 6. Osaka F, Kawasaki H, Aida N, Saeki M, Chiba T, Kawashima S, Tanaka 28. Bornstein G, Ganoth D, Hershko A. 2006. Regulation of neddylation and

1630 mcb.asm.org Molecular and Cellular Biology DCNL1 Drives Substrate-Mediated CUL2 Neddylation

deneddylation of cullin1 in SCFSkp2 ubiquitin ligase by F-box protein and Jr. 2002. An intact NEDD8 pathway is required for Cullin-dependent substrate. Proc. Natl. Acad. Sci. U. S. A. 103:11515–11520. ubiquitylation in mammalian cells. EMBO Rep. 3:177–182. 29. Liakopoulos D, Busgen T, Brychzy A, Jentsch S, Pause A. 1999. Con- 36. Wada H, Yeh ET, Kamitani T. 1999. Identiﬁcation of NEDD8- jugation of the ubiquitin-like protein NEDD8 to cullin-2 is linked to von conjugation site in human cullin-2. Biochem. Biophys. Res. Commun. Hippel-Lindau tumor suppressor function. Proc. Natl. Acad. Sci. U. S. A. 257:100–105. 96:5510–5515. 37. Meyer-Schaller N, Chou YC, Sumara I, Martin DD, Kurz T, Katheder 30. Wada H, Yeh ET, Kamitani T. 1999. The von Hippel-Lindau tumor N, Hofmann K, Berthiaume LG, Sicheri F, Peter M. 2009. The human suppressor gene product promotes, but is not essential for, NEDD8 con- Dcn1-like protein DCNL3 promotes Cul3 neddylation at membranes. jugation to cullin-2. J. Biol. Chem. 274:36025–36029. Proc. Natl. Acad. Sci. U. S. A. 106:12365–12370. 31. Hon WC, Wilson MI, Harlos K, Claridge TD, Schoﬁeld CJ, Pugh CW, 38. Bennett EJ, Rush J, Gygi SP, Harper JW. 2010. Dynamics of cullin-RING

Maxwell PH, Ratcliffe PJ, Stuart DI, Jones EY. 2002. Structural basis for ubiquitin ligase network revealed by systematic quantitative proteomics. Downloaded from the recognition of hydroxyproline in HIF-1 alpha by pVHL. Nature 417: Cell 143:951–965. 975–978. 39. Kamizono S, Hanada T, Yasukawa H, Minoguchi S, Kato R, Minoguchi 32. Min JH, Yang H, Ivan M, Gertler F, Kaelin WG, Jr, Pavletich NP. 2002. M, Hattori K, Hatakeyama S, Yada M, Morita S, Kitamura T, Kato H, Structure of an HIF-1␣-pVHL complex: hydroxyproline recognition in Nakayama K, Yoshimura A. 2001. The SOCS box of SOCS-1 accelerates signaling. Science 296:1886–1889. ubiquitin-dependent proteolysis of TEL-JAK2. J. Biol. Chem. 276:12530– 33. Maynard MA, Qi H, Chung J, Lee EH, Kondo Y, Hara S, Conaway RC, 12538. Conaway JW, Ohh M. 2003. Multiple splice variants of the human HIF-3 40. Vesterlund M, Zadjali F, Persson T, Nielsen ML, Kessler BM, Norstedt alpha locus are targets of the von Hippel-Lindau E3 ubiquitin ligase com- G, Flores-Morales A. 2011. The SOCS2 ubiquitin ligase complex regulates plex. J. Biol. Chem. 278:11032–11040. growth hormone receptor levels. PLoS One 6:e25358. doi:10.1371/journal 34. Stickle NH, Chung J, Klco JM, Hill RP, Kaelin WG, Jr, Ohh M. 2004. .pone.0025358.

pVHL modification by NEDD8 is required for fibronectin matrix assem- 41. Kamura T, Maenaka K, Kotoshiba S, Matsumoto M, Kohda D, Con- http://mcb.asm.org/ bly and suppression of tumor development. Mol. Cell. Biol. 24:3251– away RC, Conaway JW, Nakayama KI. 2004. VHL-box and SOCS-box 3261. domains determine binding specificity for Cul2-Rbx1 and Cul5-Rbx2 35. Ohh M, Kim WY, Moslehi JJ, Chen Y, Chau V, Read MA, Kaelin WG, modules of ubiquitin ligases. Genes Dev. 18:3055–3065. on December 14, 2020 at UNIV OF TORONTO

April 2013 Volume 33 Number 8 mcb.asm.org 1631