The Nedd8-conjugated ROC1-CUL1 core ubiquitin ligase utilizes Nedd8 charged surface residues for efficient polyubiquitin chain assembly catalyzed by Cdc34.

Lysine 48-linked polyubiquitin chains are the principle signal for targeting proteins for degradation by the 26 S proteasome. Here we report that the conjugation of Nedd8 to ROC1-CUL1, a subcomplex of the SCF-ROC1 E3 ubiquitin ligase, selectively stimulates Cdc34-catalyzed lysine 48-linked multiubiquitin chain assembly. We have further demonstrated that separate regions within the human Cdc34 C-terminal tail are responsible for multiubiquitin chain assembly and for physical interactions with the Nedd8-conjugated ROC1-CUL1 to assemble extensive ubiquitin polymers. Structural comparisons between Nedd8 and ubiquitin reveal that six charged residues (Lys4, Glu12, Glu14, Arg25, Glu28, and Glu31) are uniquely present on the surface of Nedd8. Replacement of each of the six residues with the corresponding amino acid in ubiquitin decreases the ability of Nedd8 to activate the ubiquitin ligase activity of ROC1-CUL1. Moreover, maintenance of the proper charges at amino acid positions 14 and 25 are necessary for retaining wild type levels of activity, whereas introduction of the opposite charges at these positions abolishes the Nedd8 activation function. These results suggest that Nedd8 charged surface residues mediate the activation of ROC1-CUL1 to specifically support Cdc34-catalyzed ubiquitin polymerization.

Nedd8 (or its orthologue Rub1) is a small ubiquitin (Ub)-like molecule that modifies all members of the Cullin/Cdc53 protein family (1,2), resulting in the formation of an isopeptide bond linkage between the ⑀-amino group of a conserved Cullin lysine residue and the C-terminal carboxyl group of Nedd8 glycine 76. The conjugation is an ATP-dependent reaction that requires a Nedd8-specific E1 activating enzyme, composed of the APP-BP1 and Uba3 heterodimer, and Ubc12 as the E2 conjugating enzyme (3). In addition, the Cullin-interacting RING finger protein, ROC1/Rbx1/Hrt1, is required for the reaction as disruption of the RING domain abolishes the modification (4).
Two well characterized members of the Cullin family, CUL1 and CUL2, serve as subunits of the two multisubunit E3 Ub ligase complexes, SCF-ROC1 (5-8) and pVHL-elongin C/B-CUL2-ROC1 (9), respectively. It has now been well established that the ROC1-CUL1 subassembly acts as a core ubiquitin ligase, capable of supporting Ub polymerization (5,6,8,10,11). In addition, results from transient transfection experiments have shown that ROC1 and its homologue ROC2 interact with all members of the Cullin protein family and that the resulting ROC-Cullin complex is active in supporting Ub polymerization (6). Thus, while the biological roles of Cullin members including CUL3, CUL4A, CUL4B, and CUL5 remain elusive, it is highly likely that all of the ROC-Cullin based complexes are involved in cellular Ub-dependent proteolysis pathways. It is, therefore, conceivable that Nedd8, through its conjugation to Cullins, functions to regulate the stability of cellular proteins.
Accumulating genetic evidence has demonstrated a critical role for Nedd8 in the regulation of cell proliferation and development. In Saccharomyces cerevisiae, the Rub1 pathway is essential for cell growth when the function of the SCF is compromised by mutations in cdc34, cdc4, cdc53, or skp1 (12). In fission yeast, the Nedd8-modifying pathway is essential for cell viability (13). A dramatic cell cycle effect was also observed in the ts41 hamster cell line, where the Nedd8 pathway is defective (14). This cell harbors a temperature-sensitive allele of SMC1, a homologue of human APP-BP1. At a nonpermissive temperature, the ts41 cell undergoes multiple rounds of DNA replication without intervening mitoses. Furthermore, in Arabidopsis thaliana, recessive mutations in the AXR1 gene (a homologue of APP-BP1) resulted in a decreased response to auxin, a hormone that regulates diverse developmental processes by promoting changes in cell division and elongation (15). More recently, Deshaies and co-workers (16) and Deng and co-workers (17) have provided compelling genetic evidence that the COP9/signalosome promotes the cleavage of Nedd8 from CUL1, suggesting a complex mechanism required for the removal of Nedd8 from its substrate Cullin proteins.
Read et al. (18) have initially reported that Nedd8 modification activates the SCF ␤-TRCP -dependent ubiquitination of IB␣. They further demonstrated that Nedd8 does not affect the K m for SCF ␤-TRCP binding to IB␣, nor does it significantly alter the ability of CUL1 to form complexes with ROC1, Skp1, and ␤-TRCP. In addition, Podust et al. (19) and Morimoto et al. (20) have shown that the Nedd8 conjugation pathway is essential for proteolytic targeting of p27 by ubiquitination. Recently, we have provided biochemical evidence that the conjugation of Nedd8 to CUL1 at Lys 720 activates the Ub ligase activity of the ROC1-CUL1 complex (21), suggesting that Nedd8 is a novel regulator of the efficiency of polyubiquitin chain synthesis and, hence, promotes the rapid turnover of SCF-ROC1 substrates.
It remains to be determined, however, how Nedd8 activates the Ub ligase activity of the ROC1-CUL1 complex. In this report, we show that the conjugation of Nedd8 to the ROC1-* This work was supported by Public Health Service Grant GM55059 (to Z.-Q. P). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18

Protein Expression and Isolation
Preparation of His-HA-Cdc34 and Its C-terminally Truncated Derivatives-pET-3a plasmids expressing the wild type and mutant Cdc34 proteins were transformed into the pJY2 (Affiniti)-containing BL21 (DE3) cells and grown in LB (0.5 liter) with 0.4% glucose in the presence of ampicillin and chloramphenicol at 37°C. Cultures were then cooled to room temperature when the optical density at 600 nm reached 0.5. Isopropyl-1-thio-␤-D-galactopyranoside at a final concentration of 0.8 mM was added to induce the culture overnight (12-14 h) at 25°C. Cells were pelleted at 5,000 ϫ g for 15 min at 4°C, and the pellet was resuspended in 1 ⁄25 culture volume of buffer A (20 mM Tris-HCl, pH 8.0, 1% Triton X-100, 0.5 M NaCl, 5 mM imidazole, 2 mM phenylmethylsulfonyl fluoride, 0.4 g/ml antipain, and 0.2 g/ml leupeptin). The resuspension was then sonicated (four repetitive 20-s treatments) and centrifuged at 17,000 rpm in an SS-34 rotor for 30 min at 4°C.
For the preparation of Cdc34, Cdc34 (aa 1-208), and Cdc34 (aa 1-194), soluble extracts (10 ml) were mixed with Ni 2ϩ -nitrilotriacetic acid-agarose beads (3 ml; Qiagen) for ϳ2 h at 4°C. The beads were then packed into a column and washed consecutively with buffer A (30 ml) and 9 ml of buffer B (20 mM Tris-HCl, pH 8.0, 10% glycerol, and 0.5 M NaCl) plus 5 mM imidazole. Bound protein was eluted with a 40-ml linear gradient of 5-250 mM imidazole in buffer B. Fractions containing peak levels of Cdc34, as judged by SDS-PAGE/Coomassie staining analysis, were pooled and dialyzed overnight at 4°C in buffer C (25 mM Tris-HCl, pH 7.5, 1 mM EDTA, 0.01% Nonidet P-40, 10% glycerol, and 1 mM DTT) plus 50 mM NaCl. The resulting material was then loaded onto a Q-Sepharose column (3 ml; Amersham Biosciences, Inc.). After washing the column with 30 ml of buffer C plus 50 mM NaCl, bound Cdc34 or Cdc34 (aa 1-208) was eluted by a 40-ml linear gradient of 50 -500 mM NaCl in buffer C, and the peak fractions were pooled. Cdc34 (aa  was recovered in the flow-through. All three Cdc34 derivatives were concentrated using centrifugal filters and further purified by FPLC using a Sephadex-75 gel filtration column (Amersham Biosciences). A single peak of protein corresponding to Cdc34, Cdc34 (aa 1-208), or Cdc34 (aa 1-194) monomer was pooled and used in the studies described in this report.
The majority of Cdc34 (aa 1-169) was found in the insoluble inclusion body. Prior to solubilization, the pellet was washed once with buffer A (25 ml) and twice with buffer D (50 mM Tris-HCl, pH 7.5, 20% sucrose, 1% Triton X-100, 2 mM phenylmethylsulfonyl fluoride, 0.4 g/ml antipain, and 0.2 g/ml leupeptin). The washed pellet was then resuspended in 20 ml of buffer E (50 mM Tris-HCl, pH 7.5, 500 mM NaCl, 5 mM imidazole) plus 8 M urea, and the resulting mixture was incubated at room temperature for 30 min. The solubilized material (10 ml) was mixed with Ni 2ϩ -nitrilotriacetic acid-agarose beads (3 ml) for 2 h at 4°C, followed by washing the beads with progressively decreasing concentrations of urea in buffer E. The bound protein was then eluted by imidazole and further purified using Q Sepharose and Sephadex-75 gel filtration chromatography as described above. A single peak of Cdc34 (aa 1-169) monomer was pooled and used in the studies described in this report.
Expression and Purification of Nedd8 Mutants-The various Nedd8 mutant proteins were prepared identically as the wild type protein using the procedure as previously described (21). The proteins were concentrated to ϳ0.3 mg/ml.
Other Reagents-APP-BP1/UBA3 was affinity-purified as described previously (21). For preparation of Ubc12, GST-fused Ubc12 was expressed and isolated on glutathione beads as previously described (21). To cleave GST, the fusion protein that was immobilized to glutathione beads was incubated with biotinylated thrombin (45 units/ml of beads) for 2 h at 4°C. After centrifugation, the supernatant containing both Ubc12 and thrombin was mixed with streptavidin beads (Amersham Biosciences; 5 l of beads/unit of thrombin), and the resulting suspension was rocked for 1 h at 4°C to allow the absorption of the biotinylated thrombin to the beads. Ubc12, free of GST and thrombin, was obtained by centrifugation.
Human Ub E1 was prepared as described (5). The expression and purification of Ubc5c were carried out as described by Ohta et al. (6).
The GST-ROC1-CUL1 (aa 324 -776) complex, conjugated with the wild type or mutant Nedd8 protein, was then incubated in a reaction mixture (30 l) that contained 50 mM Tris-HCl pH 7.5, 5 mM MgCl 2 , 2 mM NaF, 10 nM okadaic acid, 2 mM ATP, 0.6 mM DTT, 5 g of 32 P-Ub, E1 (0.6 pmol), and the indicated E2 Ub-conjugating enzyme. The incubation was at 37°C for 30 min or, as otherwise specified, on a thermomixer (Eppendorf). The bound protein was released by boiling the beads with 20 l of Laemmli loading buffer in the presence of 12 mM DTT for 3 min. This treatment was sufficient to abolish the majority of the DTT-sensitive Ub-linked thiol-esters, resulting in an identical pattern of Ub ligation products as observed when the DTT concentration was increased to 0.1 M (data not shown). Thus, the 32 P-Ub conjugates, visualized by autoradiography following 10% SDS-PAGE, were predominantly those linked via isopeptide bonds.
Ub Conjugation Assay-The reaction mixture (20 l) contained 50 mM Tris-HCl, pH 7.4, 5 mM MgCl 2 , 0.2 mM DTT, 2 mM ATP, 3 g of 32 P-Ub, E1 (0.6 pmol), and 0.5 g of the wild type or C-terminally truncated Cdc34 protein. Reactions were incubated at 37°C for 30 min and terminated by the addition of Laemmli loading buffer. The final concentration of SDS was adjusted to 0.5%. The reaction products were then separated by 10% SDS-PAGE.

RESULTS
Conjugation of Nedd8 to CUL1 Activates Cdc34-catalyzed, but Not Ubc5c-catalyzed, Polyubiquitin Chain Assembly-We have recently demonstrated that Nedd8 modification activates the ability of ROC1-CUL1 to support Ub polymerization in the presence of Cdc34 (21). Previous studies have shown that both Ubc4 and -5 are also capable of mediating ROC1-dependent polyubiquitin chain assembly (6). To determine whether Nedd8 activation was E2-specific, we compared Cdc34 and Ubc5c in their ability to catalyze Ub polymerization in the presence of the Nedd8-conjugated ROC1-CUL1 complex. For this purpose, the unmodified GST-ROC1-CUL1 (aa 324 -776) complex, expressed and assembled in E. coli, was immobilized on glutathione-Sepharose 4B. To conjugate Nedd8 to the CUL1 (aa 324 -FIG. 1. Conjugation of Nedd8 to CUL1 activates Cdc34-catalyzed, but not Ubc5c-catalyzed, polyubiquitin chain assembly. GST-ROC1-CUL1 (aa 324 -776), immobilized on glutathione-Sepharose beads, was conjugated with (lanes 1 and 6 -9) or without (lanes 2-5) Nedd8, as described under "Experimental Procedures." Following treatment, the immobilized complex was measured for its ability to convert monomeric Ub into high molecular weight Ub conjugates in the presence of Cdc34 (A) or Ubc5c (B) using an established assay detailed under "Experimental Procedures." 0.5 g of the E2 protein was used. C, the effect of Nedd8 was analyzed using subsaturating levels of Ubc5c (amounts indicated). Aliquots of each reaction were separated by 10% SDS-PAGE followed by autoradiography. The high molecular weight Ub polymers (greater than 100 kDa) were quantitated using a PhosphorImager (Molecular Dynamics, Inc., Sunnyvale, CA). The results are presented in graphs.
Consistent with our previous observation (21), Nedd8 modification activated the Cdc34-catalyzed assembly of 32 P-Ub chains by 19-fold within 9 min of incubation (Fig. 1A). Surprisingly, Ubc5c-catalyzed Ub polymerization was not significantly affected by Nedd8 modification at all of the time points tested (Fig. 1B). Furthermore, Nedd8 activation was not observed with Ubc5c that was present in subsaturating levels (Fig. 1C). This excludes the possibility that the apparent absence of activation by Nedd8 with Ubc5c was due to the use of saturating levels of the E2 protein. Similarly, no Nedd8 activation was detected when Ubc4, which shares a high degree of sequence homology with Ubc5, was used as the E2 in the reaction (data not shown). These results demonstrate that the conjugation of Nedd8 to CUL1 selectively activates Cdc34-catalyzed, but not Ubc4/5c-catalyzed, polyubiquitin chain formation.
We next explored the difference between Cdc34 and Ubc5c in the utilization of Ub lysine receptor residue(s) for the assembly of polyubiquitin chains. Consistent with our previous observation (5), when Ub K48R was used in place of the wild type Ub, Cdc34-catalyzed polyubiquitin chain formation was almost completely inhibited ( Fig. 2A), confirming that Cdc34 assembles Lys 48 -linked Ub chains. In contrast, when Ubc5c was used as the E2, polyubiquitin chains were still assembled efficiently (Fig. 2B, lanes 2-5). Conjugation of Nedd8 to CUL1 had no significant effect on the Ubc5c-catalyzed synthesis of non-Lys 48 -linked (most likely Lys 29 -linked; Refs. 22 and 23) Ub chains (lanes 6 -9). Taken together, these results show that Nedd8 specifically activates the Cdc34 catalyzed synthesis of Lys 48 -linked multi-Ub chains.
The C Terminus of Human Cdc34 Is Required for Mediating the Nedd8-stimulated and ROC1-CUL1-dependent Assembly of Polyubiquitin Chains-The above studies suggest a specific cooperation between Nedd8 and Cdc34 that promotes the Ub polymerization reaction. Previous studies have demonstrated that the C terminus of S. cerevisiae Cdc34 is required for cell cycle control (24,25) and that residues 171-209 constitute a minimal motif both necessary and sufficient for binding to the SCF components (26). These findings prompted us to examine the role of the C-terminal tail of human Cdc34 in mediating the Nedd8 stimulated and ROC1-CUL1-dependent Ub polymerization by deletion analysis. Based on sequence alignment to its S. cerevisiae counterpart, the human Cdc34 contains a C-terminal tail spanning residues 171-236. Both the wild type and the three C-terminally truncated Cdc34 proteins (Cdc34 residues 1-169, 1-194, and 1-208) were expressed in bacteria and purified to homogeneity as judged by Coomassie staining analysis (Fig. 3A).
Purified Cdc34 protein possesses a number of biochemical activities. These include the ability of Cdc34 to conjugate Ub in the presence of E1, to catalyze autoubiquitination (27), to directly interact with ROC1-CUL1 for assembling Lys 48 -linked multi-Ub chains (11), and to cooperate with Nedd8 for the activated synthesis of Ub polymers (Ref. 21; Fig. 1). In keeping with the notion that the Ubc domain (spanning amino acid residues 1-170 in human Cdc34; Ref. 28) is responsible for Ub conjugation activity, all three truncated Cdc34 proteins retained their ability to conjugate Ub in the presence of E1 (Fig.  3B). The observed DTT-insensitive Cdc34-Ub conjugates are monoubiquitinated species (see below).
Banerjee et al. (27) have previously shown that the purified S. cerevisiae Cdc34 protein catalyzes its own ubiquitination to assemble a multi-Ub chain on a lysine residue within its C terminus. As shown, human Cdc34 alone produced an array of Ub conjugates (Fig. 3C, lane 4). While the predominant reaction product was the monoubiquitinated Cdc34 that migrated as a doublet of ϳ47 kDa, other Ub conjugates included the E2 protein linked with chains composed of up to five Ub moieties. Cdc34 (aa 1-208) was more active than the wild type protein in catalyzing autoubiquitination, producing Ub conjugates with molecular masses up to 200 kDa (lane 6). Cdc34 (aa  catalyzed autoubiquitination with an efficiency comparable with that observed with the wild type protein (compare lanes 4  and 8). Interestingly, while Cdc34 (aa 1-169) was still capable of mono-and diubiquitination, it did not form multi-Ub chains (lane 10), suggesting a processivity deficiency in polymerizing FIG. 2. Ubc5c, but not Cdc34, assembles non-Lys 48 -linked polyubiquitin chains. GST-ROC1-CUL1 (aa 324 -776), immobilized on glutathione-Sepharose beads, was assayed for its ability to support Cdc34-catalyzed (A) or Ubc5c-catalyzed (B) polyubiquitin chain assembly using 32 P-labeled Ub K48R in place of the wild type Ub. 0.5 g of the E2 protein was used. The reactions were carried out and analyzed as described in Fig. 1. In the presence of ROC1-CUL1, but in the absence of Nedd8 modification, Cdc34 (aa 1-208) supported multi-Ub chain assembly, albeit with ϳ100-fold lower efficiency compared with the wild type (Fig. 3D, compare lanes 4 and 5 and lanes 10 and  11). Under these conditions, both the wild type and Cdc34 (aa 1-208) assembled Ub chains with molecular mass greater than 200 kDa, demonstrating a role for ROC1-CUL1 in promoting the assembly of extensive polyubiquitin chains. Cdc34 (aa 1-194) was only barely stimulated by ROC1-CUL1 (compare lanes 15 and 17). Furthermore, in contrast to both the wild type and Cdc34 (aa 1-208), Ub polymers formed by Cdc34 (aa 1-194) and ROC1-CUL1 were predominantly those of limited lengths that migrated in the range of 70 -100 kDa. Finally, the addition of ROC1-CUL1 could not activate Cdc34 (aa 1-169) to assemble multi-Ub chains (Fig. 3D, lanes 20 -23). These results demonstrate that the extreme C terminus of human Cdc34 (residues 209 -236) is required for its maximal activity in catalyzing ROC1-CUL1-dependent synthesis of Ub polymers. Furthermore, assembly of extensive polyubiquitin chains by ROC1-CUL1 requires Cdc34 residues 195-208. This suggests that Cdc34 (aa 195-208) may constitute a motif that interacts with the ROC1-CUL1 complex.
A GST-based pull-down assay was employed to test the ability of the wild type and C-terminally truncated Cdc34 proteins to interact with glutathione bead-immobilized GST-ROC1- FIG. 4. A, the effects of C-terminal truncation on the capacity of Cdc34 to bind to ROC1-CUL1. The binding reaction was carried out as described previously (11). Glutathione-Sepharose beads were bound with GST-ROC1-CUL1 (aa 324 -776) (ϳ5 g; lanes 1, 4, 7, 10, and 13) or GST (ϳ10 g; lanes 3, 6, 9, and 12). Following incubation with the wild type or C-terminally truncated Cdc34 protein (1 g), the beads were washed, and bound proteins were released and separated by 10% SDS-PAGE and transferred to a nitrocellulose membrane. Western blots were probed with an antibody (Novagen) recognizing the T7 tag that is present at the N terminus of the wild type and mutant Cdc34 proteins. Lanes 2, 5, 8, and 11 contained 10% of the input of each Cdc34 derivative. B, structural domains within the human Cdc34. Shown is a schematic representation of domains within the human Cdc34 that are responsible for catalysis, multi-Ub chain assembly, and interactions with ROC1-CUL1 for the efficient synthesis of Ub polymers.

Nedd8 Activates Cdc34-catalyzed Polyubiquitin Chain Assembly
CUL1 (aa 324 -776). As shown in Fig. 4A, the wild type Cdc34 was specifically bound to the complex (lanes 3 and 4), in keeping with our previous observation (11). It was estimated that under the conditions used, ϳ8% of the Cdc34 input was retained by the GST-ROC1-CUL1 (aa 324 -776) complex. While removal of residues 209 -236 reduced the binding by 60% (compare lanes 4 and 7), further deletions eliminated the interaction (lanes 9, 10, 12, and 13). For unknown reasons, Cdc34 (aa 1-169) was found to interact with GST substantially, whereas the other Cdc34 derivatives did not (lane 12). Since the replacement of GST by GST-ROC1-CUL1 (aa 324 -776) did not further increase the binding of Cdc34 (aa 1-169) (compare lanes 12 and 13), the observed interaction between Cdc34 (aa 1-169) and GST-ROC1-CUL1 (aa 324 -776) was due to an affinity of the truncated E2 protein for GST, but not ROC1-CUL1 (aa 324 -776). Taken together, these results suggest the presence of ROC1-CUL1-interacting residues within the region spanning amino acids 195-208 of Cdc34. However, the extreme C-terminal portion (residues 209 -236) plays a significant role in enhancing the interaction. This is entirely consistent with the observation that while Cdc34 residues 195-208 were essential for the ROC1-CUL1-dependent synthesis of Ub polymers, the maximal activity requires residues 209 -236 within the E2 protein (Fig. 3D).
A data base search identified putative Cdc34 orthologs from Drosophila and C. elegans. Sequence analysis reveals an expected conservation in the catalytic Ubc domain among human, Drosophila, C. elegans, and S. cerevisiae Cdc34 proteins (Fig.  5). Interestingly, an additional homologous region, corresponding to the human Cdc34 residues 170 -208, was found among these four Cdc34 orthologs (29% similarity at amino acid lev-els). This suggests that Cdc34 may utilize this evolutionary conserved region for interacting with the Nedd8-conjugated ROC1-CUL1 core Ub ligase to assemble multi-Ub chains.
Substitution of Nedd8 Charged Surface Amino Acids with the Corresponding Ub Residues Inhibits Nedd8 Activity-To further understand the role of Nedd8 in activating the ROC1-CUL1 mediated, Cdc34-catalyzed Ub polymerization, we sought to identify Nedd8 amino acid residue(s) that are required for this activity. Comparison of the crystal structures between Nedd8 and Ub identified unique amino acids within Nedd8 that may contribute to its function. Like Ub, Nedd8 displays an asymmetric distribution of charged residues that are organized to form "acidic" and "basic" faces (29). Most noticeably, some of these charged Nedd8 residues are conserved among the various Nedd8 orthologs but differ from Ub at the corresponding positions. As illustrated, on helix 1 of Nedd8, Glu 28 and Glu 31 form an electronegative surface that is not present on Ub (Fig. 6). There are also Glu 12 and Glu 14 on ␤ strand 2 that form another electronegative surface. Other notable differences between Ub and Nedd8 include an Asn to Arg substitution on residue 25 and a Phe to Lys substitution on residue 4. Intriguingly, these six charged Nedd8-specific residues (Lys 4 , Glu 12 , Glu 14 , Arg 25 , Glu 28 , and Glu 31 ) are arranged in two surface patches that lie along each side of the Nedd8 molecule (Fig. 6), suggesting a possible role for these residues involved in electrostatic interactions with other proteins.
To determine whether residues Lys 4 , Glu 12 , Glu 14 , Arg 25 , Glu 28 , and Glu 31 were required for Nedd8 function, point mutants were generated by site-directed mutagenesis, replacing charged amino acids with the corresponding Ub residues, designated as K4F, E12T, E14T, R25N, E28A, and E31Q. Mutant Nedd8 Activates Cdc34-catalyzed Polyubiquitin Chain Assembly proteins were expressed in bacteria, and the purified Nedd8 variants were compared with the wild type protein for their ability to conjugate to CUL1 and for their capacity to activate the Ub ligase activity of ROC1-CUL1.
Immunoblot analysis revealed that all six purified mutant Nedd8 proteins were conjugated to CUL1 (aa 324 -776) with similar efficiencies compared with the wild type protein (Fig.  7A, compare lanes 3 and lanes 4 -9). Consistent with our previous observation (21), the conjugation reaction converted ϳ50% of CUL1 (aa 324 -776) into two slow migrating species conjugated with one or two Nedd8 moieties. Of note, no conjugate was formed between CUL1 (aa 324 -776) and two Nedd8 E31Q moieties (Fig. 7A, lane 9). It is presently unclear whether the ROC1-CUL1 complex conjugated with two Nedd8 moieties is more efficient than that with one Nedd8 molecule in activating Ub polymerization. It should be noted that only single Nedd8 conjugates of CUL1 are detected in cells.
Next, the mutant Nedd8 variants were compared with the wild type protein for their ability to activate the ROC1-CUL1 Ub ligase. As shown in Fig. 7B, all of the mutants were less effective than the wild type, albeit at varying degrees, in activating ROC1-CUL1. While the effects of Nedd8 K4F and Nedd8 E14T were the most dramatic, reducing the Nedd8 activation function by 4-fold, Nedd8 E12T and Nedd8 E31Q appeared to possess modest effects, decreasing the Nedd8 activity by less than 25%. These results suggest that Nedd8 charged surface residues are critical in activating the ability of ROC1-CUL1 to promote Ub polymerization.
Maintenance of Proper Electrostatic Potential at Positions 14 and 25 Is Critical for Nedd8 Activity-We next determined whether the Nedd8 activity was critically dependent on the type of electrostatic potential possessed by the charged surface amino acid residues. For this purpose, Nedd8 residues Glu 14 and Arg 25 were each replaced by a residue of the same charge group, or a residue with the opposite electrostatic potential. Purified Nedd8 mutant proteins were compared with the wild type for their ability to conjugate to CUL1 as well as to activate the Ub ligase activity of ROC1-CUL1. As shown (Fig. 8A), Nedd8 variants containing aspartate (lane 5) or arginine (lane 6) in place of glutamate at position 14 were conjugated to CUL1 (aa 324 -776) with approximately equal efficiency as the wild type (lane 3) or Nedd8 E14T (lane 4). When examined for their ability to activate the Ub ligase activity of ROC1-CUL1, Nedd8 E14T or Nedd8 E14R reduced the level of activation by 5-or 30-fold, respectively, in comparison with the wild type protein (Fig. 8B, compare lane 3 with lanes 4 and 6). In contrast, Nedd8 E14D , maintaining the same charge at position 14, had a minimal effect, reducing the Nedd8 activity by 1.5-fold (lane 5). These results suggest that the presence of an acidic residue and, presumably, its negative charge at position 14 is critical for the Nedd8-mediated activation of the ROC1-CUL1 ubiquitin ligase.
When Nedd8 residue Arg 25 was mutated to lysine or glutamate, immunoblot analysis revealed that neither substitution affected the conjugation of Nedd8 to CUL1 (aa 324 -776) (Fig.  9A, lanes 3-6). While Nedd8 R25K , retaining the same charge as the wild type, was as active as the wild type protein (Fig. 9B,  compare lanes 3 and 5), Nedd8 R25E , containing the opposite charge at position 25, reduced the level of activation by 10-fold (Fig. 9B, compare lanes 3 and 6). These results strongly support the notion that specific types of electrostatic potential at proper positions on the surface of Nedd8 are involved in the activation of ROC1-CUL1 to support Ub polymerization.

Distinct Role of Cdc34 in the Assembly of Nedd8-stimulated
Polyubiquitin Chains-In this report, we have shown that Nedd8 selectively activates the Cdc34-catalyzed synthesis of Lys 48 -linked Ub polymers and that this effect is mediated by Nedd8 charged surface residues.
Cdc34 is a member of the class II E2 ubiquitin-conjugating enzymes, which are characterized by the presence of a C-terminal extension in addition to the N-terminally located conserved catalytic domain (Ubc domain) (28). The budding yeast Cdc34 is an essential gene product that primarily acts at the G 1 to S-phase transition by mediating the ubiquitin-dependent degradation of the cyclin-dependent kinase inhibitor Sic1 in a process that requires the participation of the SCF Cdc4 -ROC1/ Rbx1 E3 ubiquitin ligase (30 -33).
We have shown that in the presence of ROC1-CUL1, Ubc4/5 is capable of assembling Ub polymers in a reaction that is not stimulated by Nedd8 modification. The Ubc4/5 family of proteins, required for stress response, belongs to the class I subtype E2 conjugating enzymes, which are primarily composed of the Ubc core catalytic domain without a C-terminal tail (28). Cdc34 and Ubc4/5 differ in that while the former enzyme catalyzes Lys 48 -linked Ub polymer formation, the latter assembles both Lys 48 -and Lys 29 -linked multi-Ub chains (5,22,23,27). While the precise role of Lys 29 -linked chains remains to be FIG. 6. Nedd8 charged surface residues reside on two distinct surfaces. The three-dimensional structures of Nedd8 and Ub were obtained from the PDB TM data base (PDB ID numbers 1NDD for Nedd8 and 1UBI for Ub) at the Research Collaboratory for Structural Bioinformatics Web site and visualized using the Swiss-PDBViewer software. The final image was rendered using the POV-ray 3 software. The side chains for Nedd8 amino acids Lys 4 , Glu 12 , Glu 14 , Arg 25 , Glu 28 , and Glu 31 as well as the corresponding Ub residues are displayed as indicated.
Nedd8 Activates Cdc34-catalyzed Polyubiquitin Chain Assembly determined, it is evident, however, that uniform Lys 29 -linked chains cannot be broadly used in proteolytic targeting. This is because yeast cells expressing Ub K29R as the sole source of Ub exhibit unaltered proteolytic competence (35). In contrast, yeast cells induced to express Ub K48R are arrested in late G 2 or M phase of the cell cycle and are defective in the turnover of short lived proteins, demonstrating that the Lys 48 -linked chains are the principal signal for targeting proteins for degradation by the 26 S proteasome (reviewed in Ref. 34). It remains to be determined whether Cdc34 functions as the predominant E2 conjugating enzyme for SCF-ROC1 in vivo to assemble Lys 48 -linked polyubiquitin chains for targeting substrate degradation.
Our data is consistent with the hypothesis that Nedd8 specifically up-regulates the Cdc34-dependent proteolytic pathway. We demonstrate that Nedd8 acts to increase the rate and efficiency with which Cdc34 and ROC1-CUL1 polymerize Ub (Fig. 1). This is in keeping with the observed effects of Nedd8 on the Cdc34-catalyzed ubiquitination of IB␣ in vitro. Reed et al. (18) have shown that the SCF ␤-TRCP complex containing a CUL1 K720R mutant subunit exhibited a decreased efficiency in the ubiquitination of IB␣ with both Cdc34 and UbcH5a present. However, the overall pattern of polyubiquitin chains produced by this non-Nedd8-modified E3 complex resembles those formed by the Nedd8-conjugated complex. Consistent with this, we have observed that in the presence of Cdc34, while SCF ␤-TRCP-ROC1 conjugated with Nedd8 promotes the ubiquitina-tion of IB␣ more efficiently than the mutant SCF ␤-TRCP -ROC1 complex containing CUL1 K720R , the polyubiquitin chains generated by both complexes are similar (data not shown). These data suggest that in the Cdc34-catalyzed ubiquitination reactions, Nedd8 functions to increase the efficiency of polyubiquitin chain synthesis.
Role of the C-terminal Tail of Human Cdc34 in Polyubiquitin Chain Assembly-The distinct function of S. cerevisiae Cdc34 in cell cycle control has been attributed to its C-terminal tail (amino acids 170 -295). Both the Ellison (24) and Gonda (25) laboratories have shown that a chimeric Ubc2-Cdc34 protein, containing Ubc2 residues 1-151 (the Ubc domain) and Cdc34 residues 171-244, possesses both Ubc2 and Cdc34 activities in vivo. In subsequently published studies, the Ellison (36) and Goebl (26) groups have further defined residues 171-209 as a minimal motif (called CCD) that is required for Cdc34 function in vivo and for binding to the SCF components. These findings suggest a unique role for the C terminus of S. cerevisiae Cdc34 in mediating its cell cycle function by interacting with the SCF.
Previous studies have shown that the human Cdc34 cDNA can functionally substitute for the S. cerevisiae cdc34 gene (37), demonstrating a functional conservation between the two orthologs. In this study, we presented evidence implicating that the C terminus of human Cdc34 contains multiple biochemical activities (summarized in Fig. 4B). Based on results with both yeast (27) and human (this study) Cdc34, it is evident that this E2 enzyme possesses an intrinsic ability to assemble Ub FIG. 7. Substitution of Nedd8 charged surface residues with the corresponding Ub residues inhibits Nedd8 activity. (GST-ROC1)-FLAG-CUL1 (aa 324 -776), immobilized on glutathione-Sepharose 4B, was modified by purified Nedd8 wild type protein or mutant variants as indicated (0.6 g of protein used in each case) using the procedure described under "Experimental Procedures." The resulting beads were incubated with the ubiquitination components including 0.5 g of Cdc34, 20 ng of E1, 3 g of 32 P-Ub, and other components as described under "Experimental Procedures." Aliquots of the reaction products were separated by 10% SDS-PAGE followed by immunoblot analysis using anti-FLAG antibodies (A) or by direct autoradiography (B). Production of high molecular mass Ub polymers (Ͼ100 kDa) were quantitated using a PhosphorImager and shown in bar graphs. The activity shown in lane 2 was considered as the basal value that had been subtracted when the wild type and mutant Nedd8 were compared for their ability to activate the Ub ligase activity of ROC1-CUL1 (see "Results").
Nedd8 Activates Cdc34-catalyzed Polyubiquitin Chain Assembly chains. In humans, this activity appears to require Cdc34 amino acid residues 170 -194. The Ellison group has previously proposed a Cdc34 dimerization model to account for its ability to catalyze the Ub ligation reaction (36). It was suggested that CCD (residues 171-209) directly contacts the catalytic domain of the other monomer, bringing two conjugated Ub molecules into proximity for ligation. In support of this, cross-linking analysis indicates a critical role of CCD for the yeast Cdc34 oligomerization reaction (36). Whether the human Cdc34 residues 170 -194 are directly involved in an intermolecular interaction with the catalytic domain of another Cdc34 monomer remains to be determined.
However, in the absence of ROC1-CUL1, the Ub chains assembled by the human Cdc34 are both inefficient and of limited lengths. Evidence provided by this study and previous works strongly suggests that ROC1-CUL1 contacts Cdc34 at residues 195-208 and that this interaction is critical for the assembly of extensive polyubiquitin chains. First, while human Cdc34 (aa 1-194) catalyzed autoubiquitination with an efficiency compa-rable with the wild type and Cdc34 (aa 1-208) (Fig. 3C), it was not activated by ROC1-CUL1 to form Ub chains of extensive lengths (Fig. 3D). Second, results from a GST-based pull-down experiment indicated that human Cdc34 residues 195-208 were required for a stable association between ROC1-CUL1 and the E2 protein (Fig. 4A). Third, GST-fused CCD is sufficient to interact with the SCF in yeast cells (26). Fourth, physical analysis has shown that the CCD domain is proteolytically accessible and structurally distinct from the C-terminal portion of the tail of yeast Cdc34 (36), suggesting an availability of CCD for interactions with other protein(s), such as the ROC1-CUL1 complex. Last, a significant evolutionary conservation is found in the region corresponding to the human Cdc34 residues 170 -208 (Fig. 5).
Based on data presented in this report, while the human Cdc34 residues 209 -236 played no role in autoubiquitination, they were required for the efficient synthesis of Ub polymers in the presence of ROC1-CUL1. Intriguingly, while Cdc34 (aa 1-208) retained nearly 40% of the capacity of the wild type protein to bind to ROC1-CUL1 (Fig. 4A), it only possessed 1% of the wild type level of activity in promoting ROC1-CUL1-de- , immobilized on glutathione-Sepharose 4B, was modified by purified Nedd8 wild type protein or mutant derivatives as indicated (0.6 g of protein used in each case) using a procedure as described under "Experimental Procedures." The efficiency with which the various Nedd8 proteins were conjugated to FLAG-CUL1 (aa 324 -776) (shown in A by immunoblot) and the efficiency with which they activated the Ub ligase activity of the ROC1-CUL1 (shown in B by autoradiography) were determined as described in the legend to Fig. 7. The activity shown in lane 2 was considered as the basal value that had been subtracted when the wild type and mutant Nedd8 were compared for their ability to activate the Ub ligase activity of ROC1-CUL1 (see "Results").
FIG. 9. Requirement of a positively charged side chain on Nedd8 residue 25 for activating the Ub polymerization activity of ROC1-CUL1. Effects of Nedd8 Arg 25 mutants on conjugation to FLAG-CUL1 (aa 324 -776) (shown in A by immunoblot) and on the activation of the Ub ligase activity of the ROC1-CUL1 (shown in B by autoradiography) were determined as described in Fig. 7. The activity shown in lane 2 was considered as the basal value that had been subtracted when the wild type and mutant Nedd8 were compared for their ability to activate the Ub ligase activity of ROC1-CUL1 (see "Results").