The Human Cdc34 Carboxyl Terminus Contains a Non-covalent Ubiquitin Binding Activity That Contributes to SCF-dependent Ubiquitination*

Cdc34 is an E2 ubiquitin-conjugating enzyme that functions in conjunction with SCF (Skp1·Cullin 1·F-box) E3 ubiquitin ligase to catalyze covalent attachment of polyubiquitin chains to a target protein. Here we identified direct interactions between the human Cdc34 C terminus and ubiquitin using NMR chemical shift perturbation assays. The ubiquitin binding activity was mapped to two separate Cdc34 C-terminal motifs (UBS1 and UBS2) that comprise residues 206–215 and 216–225, respectively. UBS1 and UBS2 bind to ubiquitin in the proximity of ubiquitin Lys48 and C-terminal tail, both of which are key sites for conjugation. When bound to ubiquitin in one orientation, the Cdc34 UBS1 aromatic residues (Phe206, Tyr207, Tyr210, and Tyr211) are probably positioned in the vicinity of ubiquitin C-terminal residue Val70. Replacement of UBS1 aromatic residues by glycine or of ubiquitin Val70 by alanine decreased UBS1-ubiquitin affinity interactions. UBS1 appeared to support the function of Cdc34 in vivo because human Cdc34(1–215) but not Cdc34(1–200) was able to complement the growth defect by yeast Cdc34 mutant strain. Finally, reconstituted IκBα ubiquitination analysis revealed a role for each adjacent pair of UBS1 aromatic residues (Phe206/Tyr207, Tyr210/Tyr211) in conjugation, with Tyr210 exhibiting the most pronounced catalytic function. Intriguingly, Cdc34 Tyr210 was required for the transfer of the donor ubiquitin to a receptor lysine on either IκBα or a ubiquitin in a manner that depended on the neddylated RING sub-complex of the SCF. Taken together, our results identified a new ubiquitin binding activity within the human Cdc34 C terminus that contributes to SCF-dependent ubiquitination.

Ubiquitination is a complex protein modification reaction that produces a covalent isopeptide bond linking the ubiquitin Gly 76 carboxyl end to the ⑀-amino group of a lysine residue within a substrate protein (1). One of the hallmarks of this modification is that substrates are frequently attached with polyubiquitin chains, formed by isopeptide linkage joining the ubiquitin Gly 76 carboxyl end and one of its seven internal lysine residues. Three enzymatic activities typically participate in the ubiquitination reaction, including a ubiquitin-activating enzyme (E1) that forms an E1-Sϳubiquitin thiol ester complex in an ATP hydrolysis-dependent manner, a ubiquitin-conjugating enzyme (E2) capable of tethering a ubiquitin in thiol ester form that is transferred from E1, and a ubiquitin-ligase (E3). RING finger domain-containing E3s orchestrate the transfer of ubiquitin to a substrate by recruiting both the target protein and E2-Sϳubiquitin thiol ester and by helping position the attacking substrate lysine and the E2-bound donor ubiquitin optimally for conjugation (2). SCF (Skp1⅐Cullin⅐Fbox) is a four-subunit RING E3 ligase complex with CUL1 (cullin 1) as scaffold that at its N terminus tethers the Skp1⅐F-box protein complex to target a substrate and utilizes its C terminus to dock the ROC1/Rbx1/Hrt1 RING finger subunit for recruiting an E2 ubiquitin-conjugating enzyme (3)(4)(5)(6)(7)(8)(9). SCF-mediated polyubiquitination is activated by conjugation of Nedd8, a ubiquitin-like protein, to human CUL1 at lysine 720 (10,11), most likely due to Nedd8-imposed conformational changes in CUL1 (12), thereby relieving autoinhibitory interactions between the extreme C terminus of CUL1 and ROC1 (13).
Accumulating evidence demonstrates a role for the Cdc34 E2 in SCF-dependent ubiquitination. Yeast Cdc34 (Ubc3) is essential for viability and is required for SCF Cdc4 -dependent degradation of the cyclin-dependent kinase inhibitor Sic1 (see Refs. 3 and 4 and references therein). This regulation has been recapitulated by in vitro experiments that reconstitute the SCF Cdcc4dependent and Cdc34-catalyzed ubiquitination of Sic1 and the degradation of the modified substrate by the 26 S proteasome (3,4,14). Human Cdc34 (hCdc34) 5 is highly conserved as it is * This work was supported, in whole or in part, by National Institutes of Health Grants GM61051 and CA095634 (to Z.-Q. P.). This work was also supported by the NMR Research Programs of the Korea Basic Science Institute (Grants T3022E (to K.-S. R.) and T29221 (to Y. H. J.)). □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Figs. S1-S6. 1 Supported by NCI, National Institutes of Health, Postdoctoral Training Grant T32 CA078207. 2 To whom correspondence may be addressed. Tel.: 212-659-5500; Fax: 212-849-2446; E-mail: zhen-qiang.pan@mssm.edu. 3 To whom correspondence may be addressed. Tel.: 82-43-240-5064; Fax: 82-43-240-5059; E-mail: ksryu@kbsi.re.kr. 4 To whom correspondence may be addressed. able to complement growth defects by a temperature-sensitive cdc34 mutant yeast strain (cdc34-2, harboring the G58R mutation) at non-permissive temperature (15,16). hCdc34 acts as an E2 for the ubiquitination of the human CDK inhibitor p27 Kip1 by SCF Skp2 and IB␣ by SCF ␤TrCP2 , respectively (5,16). Previous studies have shown that the yeast Cdc34 C-terminal tail is required for the G 1 -to-S phase transition of the cell cycle and that this activity is probably mediated by the ability of the yeast Cdc34 C terminus to interact with SCF (17,18). In humans, the C terminus of Cdc34 has been shown to interact with the SCF RING subcomplex (ROC1-CUL1(324 -776)), and is indispensable for the RING subcomplex-dependent and Cdc34-catalyzed ubiquitin chain assembly (19). Intriguingly, E2-25K (human homolog of Ubc1), an E2 containing a C-terminal extension, was found to harbor a ubiquitin-associating domain (UBA) (20). These observations have prompted us to investigate whether the hCdc34 C terminus possesses the ubiquitin-binding activity.

EXPERIMENTAL PROCEDURES
Preparation of Proteins for NMR-All DNA sequences encoding for the wild type and mutant forms of hCdc34 were cloned into the pGEX-4T-1 vector, and the human ubiquitin gene was cloned into pET-15b. Ub-linker-UBS1, consisting of ubiquitin, a linker peptide (SGG), and hCdc34 (residues 199 -215), was cloned into pET-15b. To create point mutants, sitedirected mutagenesis was carried out according to the manufacturer's instructions (Stratagene).
Cdc34 (wild type or mutant), ubiquitin, Ub-linker-UBS1 proteins, and E1 were expressed in Escherichia coli Rosetta (DE3). To prepare 15 N/ 13 C-or 15 4 Cl. His-ubiquitin or GST-Cdc34 was purified using Histrap or Hitrap-GST columns (GE Healthcare), respectively. The affinity tags were removed by thrombin digestion, and the protease was then removed by gel permeation chromatography using a Superdex 75 column (GE Healthcare).
Lys 48 -linked diubiquitin and triubiquitin chains, used in experiments as described in Table 3, Fig. 4, and supplemental Fig. S3, were synthesized by Ubc7 with the wild type ubiquitin at 25°C, which yielded predominantly di-and triubiquitin chains. To separate di-and triubiquitin, the reaction mixture was applied to ion exchange Mono-S chromatography that was developed on a linear gradient from 0 to 1 M NaCl at pH 4.5 (21). Coomassie staining analysis was performed to track di-or triubiquitin-enriched fractions. For further separation, di-or triubiquitin-enriched fractions were pooled and subjected to chromatography on Superdex 75 column (GE Healthcare). 13 C/ 15 N-labeled Lys 63 -linked diubiquitin was prepared through enzymatic reaction using E1, Ubc13, and Mms2. The reaction contained His-tagged 13 C/ 15 N-labeled ubiquitin(1-74) and ubiquitin K63C or contained His-tagged ubiquitin(1-74) and 13 C/ 15 N-labeled ubiquitin K63C and was incubated at 37°C for 2-h. The reaction mixture was applied to a Histrap column (GE Healthcare) to remove Ubc13 and Mms2, and then the eluted sample was directly applied to a Superdex 75 column (GE Healthcare) to separate the diubiquitin from ubiquitin, E1, and other chemicals. The His tag was removed by thrombin digestion, and then the sulfhydryl group of Cys 63 in the acceptor was aminoethylated by using a 25-fold molar excess of Aminoethyl-8 (Pierce) in 0.2 M Bistris propane (pH 8.5) (22).
All synthetic peptides were purchased from PEPTRON Inc. The concentrations of proteins and peptides were determined based on their absorbance at 280 nm (23).
NMR Experiments-NMR spectra were acquired using Bruker 800-and 900-MHz spectrometers (KBSI) at 25°C. All chemical shift perturbation (CSP) experiments were performed in a buffer containing 10% D 2 O, NaCl at concentrations as specified and 50 mM MES (pH 5.5 and 6.0), BisTris (pH 6.5), HEPES (pH 7.0), or Tris-HCl (pH 7.5 and 8.0). The weighted CSP values were obtained using the square root of (⌬N 2 ϩ (⌬HN ϫ 6) 2 ), and the equilibrium binding constants were obtained by fitting the CSP data to the simple binding equation (24).
(1-Oxy-2,2,5,5-tetramethyl-D-pyrroline-⌬3-methyl)-methanethiosulfonate (MTSL) was purchased from Toronto Research Chemicals. Spin labeling was carried out by mixing peptide (UBS1 or UBS2) with a 5-fold molar excess of MTSL. The mixture was incubated for 3 h at 25°C. Non-reacted MTSL was removed by using desalting chromatography. Paramagnetic relaxation enhancement experiments were performed in a solution containing 100 mM NaCl, 10% D 2 O, and 50 mM MES (pH 5.5) or HEPES (pH 7.0). Where indicated, spectra were measured after incubating the NMR samples with 10 mM DTT overnight, which removed the attached MTSL group from the peptide.
Yeast Complementation Assay-The yeast expression vector pRS416-GAL1-FLAG was constructed from pRS416 by inserting a GAL1/GAL10 promoter and a FLAG tag into the multicloning site. The temperature-sensitive cdc34-2 yeast mutant strain (25) was transformed with the pRS416-GAL1-FLAG vector alone or expression vectors that encoded for the wild-type hcdc34 or deletion mutants. Yeast colonies were seeded in SD-Ura medium supplemented with 2% glucose, and the cultures were incubated at 30°C until the A 600 reached 1.0 -1.5. Equivalent amounts of the yeast cell culture, in which the volume of the culture was multiplied by the A 600 values, were serially diluted and spotted onto two SD-Ura plates that were supplemented with 2% galactose, with one incubating at the permissive temperature (25°C) and the other at non-permissive temperature (37°C) for 3 days. To confirm equal expression of of the various versions of hCdc34 in the yeast cells, an equivalent amount of the cell culture was grown in galactose medium, and the cells were harvested and then lysed in SDS-loading buffer with glass beads. The samples were analyzed by SDS-PAGE followed by immunoblotting with an anti-FLAG monoclonal antibody (Sigma).

IB␣ Ubiquitination-For the experiments shown in
For the experiment shown in Fig. 6C, the reaction was assembled as above with the exception that Ub K0 was used in the precharging reaction in place of PK-Ub. The production of monoubiquitinated species was quantified using phosphorimaging.
E2⅐Ubiquitin Thiolester Formation-The E2-Sϳubiquitin formation assay was carried out in a reaction mixture ( The Cdc34⅐S⅐[ 32 P]Ub mix was combined with or without the neddylation mix. The resulting mixture was then chased (in a final volume of 10 l) with bovine Ub (170 M) at 16°C for times as indicated. After separation by SDS-PAGE, the production of diubiquitin species was quantified using phosphorimaging.

RESULTS AND DISCUSSION
The hCdc34 C Terminus Contains Two Distinct Motifs Capable of Direct Interactions with Ubiquitin-By employing NMR CSP, we identified direct interactions between ubiquitin and the hCdc34 C terminus. In this case, the purified hCdc34 C-terminal fragment spanning amino acids 170 -236, designated as hCdc34C, was 15 N-labeled and mixed with ubiquitin. Significant CSP was observed at pH 5.5 in two major hCdc34C areas, amino acid residues 205-215 and 216 -225 (Fig. 1A), which are named ubiquitin-binding site 1 and 2 (UBS1 and UBS2), respectively. The ubiquitin-binding constants (K d ) for UBS1 and UBS2 were determined at 1.6 and 1.7 mM (Fig. 1A), respectively, by CSP experiments performed with 15 N-labeled hCdc34C and increasing concentrations of ubiquitin. Notably, the third hCdc34C segment spanning amino acids 187-196 exhibited detectable CSP as well (Fig. 1A), but its magnitude was too weak to allow the determination of its K d . The UBS1-and UBS2mediated interactions were sensitive to both pH and salt (Table  1). Optimal interactions required acidic pH (Fig. 1, A and B,  left). The addition of NaCl at 500 mM abolished the interaction by either UBS1 or UBS2 (Fig. 1B, right). Of note, no CSP was detected in hCdc34C amino acids 226 -236, which are composed of a stretch of highly acidic residues (Asp/Glu), similar to those in UBS1 or -2 (Fig. 1A). It is thus unlikely that electrostatic interactions provided by acidic residues enriched in hCdc34C serve as the sole determinant dictating the binding to ubiquitin.
We mapped the hCdc34C-binding surface on ubiquitin by a CSP experiment with 15 N-labeled ubiquitin and increasing concentrations of hCdc34C. As shown (Fig. 1C), CSP signals were observed with ubiquitin residues (in order of descending magnitude): Lys 48 , Leu 71 , Leu 69 , Gln 49 , Val 70 , Leu 73 , Gly 47 , Thr 7 , Arg 74 , Arg 72 , and Gly 75 . These findings suggest a model for the hCdc34C-binding surface on ubiquitin (Fig. 1D). Although this interaction surface is similar to those mapped for other ubiquitin-binding domains (27,28), it is narrower and extended into the ubiquitin C-terminal tail (Leu 71 -Gly 75 ). In addition, these CSP titration experiments determined the K d for hCdc34C at 0.41 Ϯ 0.07 mM. Thus, the K d of hCdc34C is significantly lower than that determined for UBS1 or UBS2 (Fig.  1A), suggesting that UBS1, UBS2, and possibly hCdc34(187-196) act synergistically to enhance the affinity of hCdc34C for ubiquitin. Of note, sequence alignment analysis reveals that UBS1 is more conserved than UBS2 (Fig. 1E).
Given the roles for ubiquitin Ile 44 , His 68 , and Val 70 in mediating interactions with ubiquitin-binding domains (27), we sought to determine their contributions in binding to hCdc34C. For this purpose, CSP experiments were performed with 15 N-labeled hCdc34C in the presence of ubiquitin I44A, H68A, or V70A. Of these ubiquitin mutants examined, V70A exhibited a pronounced defect in binding to hCdc34C (Table  1). Of note, the plateau of the CSP of 15 N-labeled hCdc34C was not observed even in the presence of 3.5 mM V70A, precluding the precise measurement of its K d value. No significant effect was observed with I44A or H68A (Table 1). Thus, the hCdc34C-ubiquitin Val 70 interaction appears to have a significant role in mediating the binding of hCdc34C to ubiquitin.
The hCdc34C UBS1 and UBS2 Motifs Exhibit a Dual Ubiquitin Binding Orientation-To examine the mode of interactions between ubiquitin and UBS1 or -2, we employed a paramagnetic relaxation enhancement (PRE) assay, which is a highly effective NMR technique for measuring the orientation of a protein-protein interaction (29). In this system, 15 N-labeled ubiquitin was mixed with an MTSL-coupled UBS1 or -2 peptide (diagramed in Fig. 2A, top), followed by the acquisition of 1 H-15 N HSQC spectra. The results revealed that MTSL-UBS1 diminished the NMR peak intensity of ubiquitin residues Arg 74 , Leu 73 , Leu 71 , Gln 49 , and Gly 47 ( Fig. 2A, compare left and middle) and that this diminishing activity was reversed by treatment with DTT ( Fig. 2A, right), indicating a MTSL-UBS1-specific effect. In addition, the NMR spectrum at ubiquitin residues Ala 46 and Leu 8 was reduced by MTSL-UBS1 as well (data not shown). These results demonstrated interactions between the UBS1 peptide and ubiquitin residues Arg 74 , Leu 73 , Leu 71 , Gln 49 , Ala 46 , Gly 47 , and Leu 8 , which fall into two separate regions, with "parts 1 and 2" containing ubiquitin Ala 46 /Gly 47 / Gln 49 and Leu 8 /Leu 71 /Leu 73 /Arg 74 , respectively (Fig. 2B, middle). Similar observations were observed with the MTSL-UBS2 peptide (Fig. 2B, right). Given the short length of the peptides (11 amino acids) used in the PRE experiments, it is impossible for a single UBS peptide to cover both ubiquitin parts 1 and 2 in a single binding event (Fig. 2B). Thus, it is most likely that UBS1 and UBS2 bind to ubiquitin in two distinct orientations (Fig. 2C).
As revealed by both CSP analysis (supplemental Fig. S1A) and CD experiments (supplemental Fig. S1B), neither UBS1 nor UBS2 displayed any defined secondary structures (␣-helix and ␤-strand) when bound to ubiquitin or triubiquitin. We have been unable to deduce the structure of UBS1 bound to ubiq-uitin from NMR data using the CYANA program (30) due to the dual ubiquitin binding mode of UBS1 (see below).
The UBS1 Aromatic Residues Participate in Binding to Ubiquitin-We analyzed the binding mode of UBS1 further by NOE cross-peak analysis, which reveals the close proximity of a pair of interacting amino acids by the appearance of their NOE cross-peaks. In this experiment, we employed a Ub-linker-UBS1 fusion protein (diagrammed in Fig. 3A) to enhance the binding affinity of UBS1 for ubiquitin. As shown (Fig. 3B), all of the UBS1 aromatic residues (Phe 206 , Tyr 207 , Tyr 210 , and Tyr 211 ) exhibited NOE cross-peaks with ubiquitin Val 70 and with Ile 44 , albeit at a lesser degree. Thus, it appears that when bound to ubiquitin, all of the UBS1 aromatic residues (Phe 206 , Tyr 207 , Tyr 210 , and Tyr 211 ) are positioned in the vicinity of ubiquitin Val 70 . In further support of these findings, UBS1 in the ubiquitin fusion exhibited relatively strong rigidity as revealed by the analysis of the order parameters from the observed T1, T2, The plot shows the CSP response at various ubiquitin residues (as specified) as a function of hCdc34C concentration. D, the hCdc34C-binding surface on 15 N-labeled ubiquitin is shown, with different colors denoting the CSP intensity. E, sequence alignment of the C-terminal regions of Cdc34 homologs. The red, blue, and green colors represent residues that are identical, strongly conserved, and conserved, respectively. Residues 187-196 and the UBS1 and UBS2 regions in Cdc34C are highlighted in yellow, cyan, and gray, respectively. and heteronuclear NOE values (supplemental Fig. S2B). By contrast, the UBS1 motif in the context of hCdc34C was flexible (data not shown). These results demonstrated an intramolecular interaction between ubiquitin and UBS1 in the fusion protein. In addition, analysis of the residual dipolar coupling values suggests that the fused UBS1 bound to ubiquitin with two different orientations (supplemental Fig. S2C). Of note, we also detected Ub-UBS1 intramolecular interactions with a fusion protein containing N-terminally located UBS1, followed by a linker of GGSGG and then ubiquitin. The overall binding mode of this fusion is similar to that of Ub-linker (SGG)-UBS1, albeit with lower levels of CSP (data now shown). Such discrepancy can be explained by the well known structural flexibility of the ubiquitin C-terminal tail that may allow the movement of the C-terminally located UBS1 for interactions with ubiquitin in the context of Ub-linker (SGG)-UBS1.
To assess the contribution by UBS1 residues in binding to ubiquitin, we measured and compared K d for the wild type and mutant peptides in the CSP assay using 15 N-labeled ubiquitin ( Table 2). Replacement of Phe 206 /Tyr 207 , Tyr 210 /Tyr 211 , or Asp 208 /Asp 209 , by GG all decreased binding affinity of UBS1 for ubiquitin. Taken together, these results suggest a role for UBS1 aromatic residues in participating in interactions with ubiquitin.
UBS1 Binds to Lys 48 -Ub Chains Preferentially-UBS1 displayed a higher affinity for Lys 48 -linked diubiquitin than for monoubiquitin, with a drop of K d by 7-fold for UBS1 and 4-fold for UBS2 (compare Table 1 and Table 3). In the form of hCdc34C, UBS1 and UBS2 exhibited different K d values for Lys 48 -diubiquitin (0.22 Ϯ 0.03 versus 0.40 Ϯ 0.05 mM) ( Table 3), suggesting that each site independently binds to one diubiquitin molecule. CSP analysis revealed that the UBS1-binding surface on either the donor or acceptor ubiquitin in a diubiquitin molecule was similar to that mapped on monoubiquitin (supplemental Fig. S3, A and B). Moreover, CSP located UBS1binding to the interface of the Lys 48 -diubiquitin (Fig. 4, A and  B). Intriguingly, the K d for 15 N-labeled diubiquitin to UBS1 is 0.20 Ϯ 0.05 mM, which is similar to that determined for 15 Nlabeled hCdc34C to diubiquitin (0.22 Ϯ 0.03 mM) ( Table 3). It is thus likely that UBS1 interacts with both the acceptor and donor ubiquitin of diubiquitin in a single binding event because

analysis of K d (mM) for interactions between 15 N-labeled hCdc34C and ubiquitin or mutants
Unless otherwise stated, CSP was carried out in a solution containing 50 mM MES, pH 5.5, and 100 mM NaCl.

TABLE 2 CSP analysis of K d (mM) for interactions between 15 N-labeled ubiquitin and UBS1 or mutants
CSP was carried out in a solution containing 50 mM MES, pH 5.5, and 100 mM NaCl.

TABLE 3 CSP analysis of K d (mM) for interactions between diubiquitin and hCdc34C or UBS1
Unless otherwise stated, CSP was carried out in a solution containing 50 mM MES, pH 5.5, and 100 mM NaCl.  uitin molecule (Fig. 4, C and D, and supplemental Fig. S3D). Lys 48 -diubiquitin is characterized by antiparallel 2-fold symmetry and a highly basic interface (31). Taken together, these findings support a model in which UBS1 binds to the interface of Lys 48 -diubiquitin in a dual orientation fashion (Fig. 4E).

Non-labeled proteins
It was observed that UBS1 bound to Lys 48 -diubiquitin with affinity much higher than to Lys 63 -diubiquitin (Table 3). Specifically, the affinity of UBS1 for the acceptor Ub in the Lys 48diubiquitin is at least 14-fold higher than that in the Lys 63diubiquitin. On the donor ubiquitin, UBS1 exhibited affinity for the Lys 48 -diubiquitin that is at least 6-fold greater than for the Lys 63 -diubiquitin (Table 3). In addition, a GST-based pulldown assay revealed strong interactions between hCdc34 and a Lys 48 -tetraubiquitin chain (supplemental Fig. S4).
In all, these findings suggest that UBS1 and UBS2 bind to a Lys 48 -linked diubiquitin molecule specifically with a stoichiometry of 1:1, occupying both the donor and acceptor ubiquitin moieties. It should be cautioned that the double side ubiquitin binding property of UBS1/UBS2 might have resulted in an inaccurate determination of K d for ubiquitin because during CSP titration, one UBS1/UBS2 site may be engaged in interactions with two ubiquitin molecules simultaneously.
UBS1 Contributes to the Function of hCdc34 in Vivo and in Vitro-We employed a yeast complementation assay to evaluate the function of hCdc34 UBS1 and UBS2 in vivo. Previous studies have shown that the full-length hCdc34 but not its truncated form (hCdc34 ) is able to support growth in the temperaturesensitive cdc34-2 yeast strain at non-permissive temperature (37°C) when the endogenous Cdc34 is disabled (16). As shown (Fig. 5A), hCdc34(1-215) appeared to allow temperature-sensitive cdc34-2 to grow at 37°C, whereas hCdc34(1-200) was completely inactive. However, the full-length hCdc34 supported the growth far better than hCdc34 . Note that under these conditions, the wild type and truncated versions of hCdc34 were expressed equally (Fig. 5B). These results suggest that the Cdc34 UBS1 motif (spanning residues 205-215) alone is capable of supporting the function of the Cdc34 C terminus in vivo, albeit to a limited extent. hCdc34(216 -236) containing the UBS2 motif is clearly required for the full function of Cdc34 in cells.
To evaluate the contribution by hCdc34 UBS1 and UBS2 in ubiquitination, we initially employed a reconstituted IB␣ ubiquitination assay to determine whether mutations within UBS1 or UBS2 might impair the conjugation activity of Cdc34. This assay contains IKK␤phosphorylated GST-IB␣(1-54) as substrate, affinity-purified SCF ␤TrCP2 as E3, hCdc34 in wild type or mutant form as E2, E1, and ubiquitin (26). Of four mutants tested (F206G/Y207G, D208G/D209G, Y210G/Y211G, and V216/F224G), Y210G/Y211G exhibited defects in ubiquitination in comparison with the wild type E2 (Fig. 6A, compare   1-4 with lanes 9 -12). F206G/Y207G appeared defective as well, albeit to a lesser degree when compared with Y210G/ Y211G (Fig. 6A, lanes 5-8). Neither D208G/D209G nor V216/ F224G impaired ubiquitination (data not shown). As revealed by subsequent ubiquitination analysis, Y210G displayed a more pronounced decrease in ubiquitination than Y211G (Fig. 6B). Of note, neither Y210G/Y211G nor Y210G impaired the ability of Cdc34 to form E2-Sϳubiquitin (Fig. 6C). Thus, substitution of UBS1 aromatic residues by glycine appears to inhibit the ability of Cdc34 to catalyze the ubiquitination of IB␣, albeit at varying degrees. Among these aromatic residues, Cdc34 Tyr 210 is likely to have the most pronounced role in catalysis. In addition, GST-based pull-down experiments revealed that replacement of UBS1 aromatic residues by glycine did not alter the interactions between Cdc34 and the SCF RING subcomplex (supplemental Fig. S5), thereby excluding the possibility that altered E2-RING interactions are attributable to the observed decrease in ubiquitination.
We next determined whether Y210G/Y211G or Y210G inhibits the ability of Cdc34 to catalyze the ligation of a single ubiquitin to GST-IB␣(1-54) using lysineless ubiquitin. The results showed that either single or double mutation decreased monoubiquitination by as much as 4-fold (Fig. 6D). These findings pointed to a role for Cdc34 Tyr 210 in positioning the donor ubiquitin (thiol ester linked to Cdc34) for chemical reactions with the receptor lysine (on either the substrate or ubiquitin). To test this possibility, we compared the wild type Cdc34 and Y210G/Y211G or Y210G to catalyze diubiquitin synthesis in the absence or presence of the SCF RING subcomplex, ROC1-CUL1(324 -776), using an assay as described previously (26). The results showed that in the absence of ROC1-CUL1(324 -776), Y210G/Y211G exhibited activity for diubiquitin synthesis at levels comparable with the wild type (Fig. 7A). In keeping with this observation, none of the Cdc34 UBS1/UBS2 mutants (F206G/Y207G, D208G/D209G, Y210G/Y211G, V216/F224G, and F206G/Y207G/Y210G/Y211G) showed defects in ubiquitination catalyzed by E2 only (supplemental Fig. S6). Consistent with previous observations (26), the addition of the neddylated ROC1-CUL1(324 -776) stimulated the wild type Cdc34 for the production of diubiquitin (Fig. 7, compare A (lanes 1-4) with B (lanes 1-4); also see graphs). By contrast, Y210G/Y211G or Y210G was poorly stimulated by the neddylated ROC1-CUL1(324 -776) for diubiquitin synthesis (Fig. 7B, compare  lanes 1-4 with lanes 5-12; graph). Of note, under the conditions used, 70% of CUL1(324 -776) was neddylated as revealed by immunoblot analysis (Fig. 7C). Taken together, these findings support a hypothesis that Cdc34 Tyr 210 may help orient the donor ubiquitin for conjugation in a manner that depends on the SCF neddylated RING subcomplex.
A Role for the UBS1 Aromatic Residues in SCF-dependent and hCdc34-catalyzed Ubiquitination-In summary, the results of NMR, genetic complementation, and ubiquitination experiments suggest strongly a role of the Cdc34 UBS1 aromatic residues in SCF-dependent ubiquitination. First, the PRE experiments have mapped the UBS1-binding surface on ubiquitin to the proximity of ubiquitin Lys 48 and C-terminal tail (Fig. 2B), both of which are key sites for catalysis. Second, the NOE crosspeak analysis suggests that when bound to ubiquitin in one orientation, the Cdc34 UBS1 aromatic residues (Phe 206 , Tyr 207 , Tyr 210 , and Tyr 211 ) are probably positioned in the vicinity of ubiquitin Val 70 (Fig. 3). Third, replacement of UBS1 aromatic residues by glycine or of ubiquitin Val 70 by alanine decreased UBS1-ubiquitin affinity interactions (Tables 1 and 2), underscoring the significance of these residues in mediating the bind- . The UBS1 Y210G/Y211G mutation is defective in diubiquitin synthesis that requires the neddylated RING subcomplex of SCF. The diubiquitin synthesis reaction was carried out in the absence (A) or presence (B) of the neddylated ROC1-CUL1(324 -776). The influence of the hCdc34 mutation on the production of diubiquitin was quantified and is presented graphically. C, immunoblot analysis of an aliquot of the neddylation reaction mixture (as described under "Experimental Procedures").
ing of UBS1 to ubiquitin. Fourth, yeast complementation experiments have suggested a role for the hCdc34 UBS1 motif in supporting the function of this E2 in vivo (Fig. 5). Fifth, reconstituted IB␣ ubiquitination analysis revealed a role for each adjacent pair of UBS1 aromatic residues (Phe 206 /Tyr 207 and Tyr 210 /Tyr 211 ) in conjugation, with Tyr 210 exhibiting the most pronounced catalytic function (Fig. 6, A and B). Last, Cdc34 Tyr 210 appears to be required for the transfer of the donor ubiquitin to a receptor lysine on either a substrate (Fig.  6D) or a ubiquitin in a manner that depended on the SCF neddylated RING subcomplex (Fig. 7). Based on these findings, it is tempting to speculate that the Cdc34 UBS1 aromatic residues help orient the donor ubiquitin C-terminal tail for conjugation.
Previous studies suggest a role for hCdc34 catalytic core residues Ser 95 and Glu 108 /Glu 112 in the function of the donor ubiquitin, because mutations at these sites abolish IB␣ mono-and polyubiquitination as well as diubiquitin synthesis in the presence or absence of the neddylated ROC1-CUL1(324 -776) (26). Intriguingly, the Cdc34 Y210G/Y211G mutation did not affect unassisted diubiquitin synthesis by Cdc34 alone (Fig. 7A). These observations suggest that both the hCdc34 catalytic core and C-terminal UBS1 motif contribute to the positioning of the donor ubiquitin. However, the engagement of the Cdc34 UBS1 in this activity is strictly dependent on the neddylated SCF RING subcomplex (ROC1-CUL1(324 -776)). It can be speculated that UBS1 itself is incapable of productive interactions with the hCdc34 Cys 93 -bonded ubiquitin (within the Cdc34⅐ S⅐ubiquitin complex) due to its low affinity (with K d measured at greater than millimolar range; see Table 1). However, the hCdc34 UBS1 could be brought into the vicinity of the donor ubiquitin by the SCF RING complex, which has been shown to interact with the Cdc34 C terminus (19). Recently, the nature of Cdc34 C tail-RING complex interaction was uncovered; Kleiger et al. (32) demonstrates that the acidic tail of Cdc34 interacts with the basic canyon located on the CUL1 C terminus. Once brought near the donor ubiquitin, the hCdc34 UBS1 aromatic residues, Tyr 210 in particular, may be in a position to engage in interactions with the ubiquitin C-terminal tail. Consequently, the proper orientation of the donor Ub Gly 76 is established, aligning with Lys 48 of the acceptor ubiquitin for conjugation.
In addition to its postulated role for positioning the donor ubiquitin for conjugation as stated above, the hCdc34 UBS1 (and probably UBS2) may have a role in the assembly of polyubiquitin chains. Given their capability to bind ubiquitin in two distinct orientations (Fig. 2B) and high affinity for Lys 48 -linked diubiquitin with a stoichiometry of 1:1 (Table 3 and Fig. 4), both UBS1 and UBS2 could conceivably help align the donor and the receptor ubiquitin for conjugation or provide sites for accommodating ubiquitin chains. However, given the role for UBS1 in the function of donor ubiquitin in catalysis and the possible redundancy of UBS1 and UBS2, we have been unsuccessful in dissecting the role of UBS1/UBS2 in the receptor ubiquitin for chain assembly using mutagenesis approaches.