Characterization of a dominant negative mutant of the cell cycle ubiquitin-conjugating enzyme Cdc34.

The yeast Saccharomyces cerevisiae CDC34 gene encodes a ubiquitin-conjugating enzyme that is required for the cell cycle G1/S transition. We show here that a dominant negative Cdc34 protein is generated by simultaneously replacing both Cys95 and Leu99 with Ser residues. Cys95 is an essential catalytic residue that forms a transient thiol ester with ubiquitin during catalysis, and Leu99 is highly conserved among all known ubiquitin-conjugating enzymes. Mutants that encode either an alanine or a serine at one or both of these two positions are inactive. Of these eight mutants, overexpression of CDC34-C95S,L99S in wild type strains was found to block cell growth. Although cells overexpressing Cdc34-C95S,L99S do not exhibit the characteristic multibudded phenotype of cdc34 temperature-sensitive or null mutants, this blockade is relieved by simultaneous overexpression of wild type Cdc34. Purified Cdc34-C95S,L99S protein can be shown to inhibit in vitro ubiquitination of the Cdc34-specific substrate, Cln2 protein. We suggest that Cdc34-C95S,L99S selectively sequesters a subset of Cdc34 substrates or regulators. These findings have implications for the structure/function relationships of ubiquitin-conjugating enzymes, and suggest a general method for identifying components and substrates of specific ubiquitination pathways of eukaryotes.

The yeast Saccharomyces cerevisiae CDC34 gene encodes a ubiquitin-conjugating enzyme that is required for the cell cycle G 1 /S transition. We show here that a dominant negative Cdc34 protein is generated by simultaneously replacing both Cys 95 and Leu 99 with Ser residues. Cys 95 is an essential catalytic residue that forms a transient thiol ester with ubiquitin during catalysis, and Leu 99 is highly conserved among all known ubiquitin-conjugating enzymes. Mutants that encode either an alanine or a serine at one or both of these two positions are inactive. Of these eight mutants, overexpression of CDC34-C95S,L99S in wild type strains was found to block cell growth. Although cells overexpressing Cdc34-C95S,L99S do not exhibit the characteristic multibudded phenotype of cdc34 temperature-sensitive or null mutants, this blockade is relieved by simultaneous overexpression of wild type Cdc34. Purified Cdc34-C95S,L99S protein can be shown to inhibit in vitro ubiquitination of the Cdc34-specific substrate, Cln2 protein.
We suggest that Cdc34-C95S,L99S selectively sequesters a subset of Cdc34 substrates or regulators. These findings have implications for the structure/function relationships of ubiquitin-conjugating enzymes, and suggest a general method for identifying components and substrates of specific ubiquitination pathways of eukaryotes.
The ubiquitin-conjugating enzymes (E2) 1 constitute a family of conserved proteins that participate either in an intermediate or in the final step of substrate ubiquitination (Hershko and Ciechanover, 1992;Finley and Chau, 1991). These enzymes form a thiol ester adduct with ubiquitin (Ub) in the presence of ubiquitin-activating enzyme (E1) and ATP in the following reactions: 1) E1 SH ϩ Ub ϩ ATP 7 E1 S-Ub ϩ AMP ϩ PP i , and 2) E1 S-Ub ϩ E2 SH 7 E1 SH ϩ E2 S-Ub .
Substrate proteins may be directly recognized by an individual E2 enzyme, resulting in the transfer of ubiquitin from an E2 S-Ub to a lysine on substrate proteins. Alternatively, substrate recognition may require the presence of another group of proteins known as E3 or ubiquitin-protein ligases (Reiss and Hershko, 1990;Bartel et al., 1990). A requirement for a specific E3 protein had been shown for the degradation of substrates in the N-end rule pathway (Bartel et al., 1990) and for p53 (Scheffner et al., 1993). In the N-end rule pathway, the Ubr1 (E3) protein contains separate sites for Rad6 (E2) and substrate bindings and confers specificity for one of the cellular Rad6-dependent ubiquitination pathways (Varshavsky, 1992). In the p53 degradation pathway, it has been further shown that ubiquitin from E2 S-Ub is transferred to a cysteine in the E3 protein (Scheffner et al., 1995), leading to the formation of an E3 S-Ub thiol ester.
CDC34 is one of 10 known ubiquitin-conjugating enzymeencoding genes in the yeast, S. cerevisiae (Goebl et al., 1988;Jentsch, 1992). This gene was initially identified on the basis of its requirement for cells to undergo the cell cycle G 1 to S transition (Byers and Goetsch, 1973). Under nonpermissive conditions, temperature-sensitive mutants of CDC34 develop numerous elongated buds, and the spindle pole body duplicates but fails to undergo the separation required for spindle formation (Byers and Goetsch, 1973). More recent studies have established a direct role of this ubiquitin-conjugating enzyme in targeting the degradation of specific regulators of the cell cycle. Known Cdc34-specific substrates in this category include the G 1 cyclins (Deshaies et al., 1995;Yaglom et al., 1995) and the Cdc28 kinase inhibitor Sic1 (Schwob et al., 1994). In addition, mutations in the CDC34 gene can lead indirectly to the abnormal accumulation of other cell cycle regulators such as the G 2 -specific B-type cyclins (Amon et al., 1994). Other than its cell cycle function, the Cdc34-encoded ubiquitin-conjugating enzyme has also been shown to target the degradation of the transcription factor GCN4 (Kornitzer et al., 1994), and it is likely that other functions of this enzyme may be uncovered by the identification of additional substrate proteins.
The 295-residue Cdc34 protein contains a 170-residue Nterminal domain that is conserved among all E2 proteins. This conserved domain is apparently sufficient for E2 S-Ub complex formation since the smallest E2 enzymes are comprised almost exclusively of this domain (Jentsch, 1992). In Cdc34 this conserved domain also contains an extra 12-residue sequence near the ubiquitin-accepting cysteine. This extra sequence is found only in one other yeast E2 protein, Ubc7 (Jungmann et al., 1993), and in both cases, the function of this extra sequence segment remains undefined. In the present study, we report the effect of mutations at the ubiquitin-accepting cysteine as well as at Leu 99 , a residue that is adjacent to this 12-residue segment. We show here that while both residues are essential for CDC34 functions, a unique dominant negative allele of this gene, CDC34 DN , could be generated by simultaneously substituting these two residues with serines. In addition to its potential utility in genetic analysis, CDC34 DN can be used to block Cdc34-dependent ubiquitination in vitro.

EXPERIMENTAL PROCEDURES
Materials-Ubiquitin was purchased from Sigma. The 125 I-labeled form of ubiquitin was obtained by radioiodination with the use of Iodogen (Pierce) according to the manufacturer's procedure. The specific activity of 125 I-ubiquitin was 5 ϫ 10 5 cpm/g. The ubiquitinactivating enzyme (E1) was purified to homogeneity from calf thymus as described previously (Ciechanover et al., 1982) and was stored at Ϫ80°C in 20% glycerol. E1 and Cdc34 concentrations were determined by first converting the proteins to E1 S-Ub and Cdc34 S-Ub , and the concentrations of these thiol esters were estimated from the known specific radioactivity in [ 125 I]ubiquitin.
Strains of Bacteria and Yeast, Plasmid Vectors, and Genetic Techniques-The bacterial plasmids and phage strains used in this work are listed in Table I. The S. cerevisiae strains are listed in Table II. All DNA manipulations were done according to Sambrook et al. (1989). Plasmid vectors that expressed Cdc34 from the galactose-inducible P GAL1 promoter were constructed using the vectors YEplac195 and YIplac211 (Gietz and Sugino, 1988) by isolating a ϳ0.7-kilobase BamHI-EcoRI fragment of pG12 that contained the P GAL1/GAL10 promoter region (Johnston and Davis, 1984) and ligating the fragment into the high copy (2-based) plasmids YEplac195 and YEplac181, and integrating plasmid YIplac211, yielding plasmid vectors YEp195GAL, YEp181GAL, and YIp211GAL. Fragments of ϳ0.9 kilobase encoding CDC34 and its mutants (without the native promoter) bordered by SalI and SphI sites (Banerjee et al., 1993) were then ligated into the SalI-and SphI-cut YEp195GAL and YIp211GAL, yielding various constructs used in this study. CDC34 was also cloned into the YEp181GAL vector. Proper induction by galactose of Cdc34 and mutant proteins for each of the yeast strains integrated/transformed with these plasmids was checked by immunoblotting with anti-Cdc34 polyclonal antibodies.
Growth and Induction of Yeast Strains-The S. cerevisiae strains used in this work were grown at 30°C (332 derivatives) or 23°C (MGG15 derivatives) in rich (YPD) or synthetic media (Sherman et al., 1986), with the latter containing 2% dextrose (SD medium), 2% raffinose (SR medium), or 2% galactose (SG medium). Transformation/ integration of S. cerevisiae was carried out by the methods of Sherman et al. (1986) or Shiestel and Gietz (1989). For galactose induction of yeast strains, cells were first grown to A 600 of ϳ0.1 in SR medium lacking uracil (and relevant amino acids when necessary), and galactose was added to a final concentration of 2%. Samples were withdrawn at time points described under "Results" and processed.
Construction of CDC34 Mutants-The CDC34 mutants were constructed by site-directed mutagenesis of the CDC34 gene in M13 mp19 (Banerjee et al., 1993). The Cys 95 TGT codon was changed to either a TCT for Ser or a GCT for Ala. The Leu 99 TTA codon was changed to either a TCA for Ser or a GCA for Ala. These changes were made by the procedure of site-specific mutagenesis (Taylor et al., 1985) with an Amersham mutagenesis kit. DNA sequencing by the chain termination method (Sanger et al., 1977) was used to verify all constructs. The CDC34-C95S,L99S sequence was also inserted into the P L -promoterbased plasmid as described previously for the wild type CDC34 gene (Banerjee et al., 1993), resulting in the plasmid pLCDC34-C95S,L99S.
Expression of Cdc34, Cdc34-C95S,L99S, and Rad6 Proteins in Escherichia coli and Their Purification-The plasmids pNMCDC34 (Banerjee et al., 1993) and pLCDC34-C95S,L99S were transformed into E. coli host AR58, which contains a temperature-sensitive repressor. These cells, harboring pNMCDC34 and pLCDC34-C95S,L99S plasmids, were grown at 30°C to a density of 1 absorbance unit at 600 nm. The culture (2 liters) was shifted to 42°C for 2 h and incubated for an additional 3 h at 39°C. An SDS gel analysis of proteins in crude E. coli extracts revealed the induction of Cdc34 as a major protein.
The proteins were purified according to previously published protocols (Banerjee et al., 1993). Rad6 was purified from E. coli extracts that had the protein overexpressed from the RAD6 gene cloned in pKK223-3 vector (Haas et al., 1991).
Expression of M13 mp19-CDC34 Mutant Genes-To express Cdc34 mutant proteins, individual mutant genes in M13 mp19/18 were used to infect a 1:500 dilution of an overnight culture of E. coli host TG1 in LB to obtain a multiplicity of infection of 30 -40. After 3 h of growth at 37°C, the medium was adjusted to contain 1 mM isopropyl-1-thio-␤-Dgalactopyranoside, and incubation was continued for an additional 2 h. Cells from 3-ml cultures were harvested by centrifugation, washed with 50 mM Tris (pH 7.5), and suspended in 100 l of the same buffer and 0.5 mM dithiothreitol. Cells were lysed by sonication for 30 s at 30 watts, and after 5 min of centrifugation at 12,000 rpm in an Eppendorf centrifuge, the supernatant was used to assay for ubiquitin thiol ester and conjugate formation in a final reaction volume of 20 l using standard reaction conditions. The amounts of Cdc34 and mutant proteins were determined and normalized by immunoblotting using the ECL (Amersham Corp.) method of detection. Incubation times of 15 and 45 min were used in thiol ester and Ub-Cdc34 complex formation, respectively.
Cdc34 S-Ub and Ub-Cdc34 Complex Formation-Reactions were carried out in 50 mM Tris (pH 7.5), 10 mM MgCl 2 , 2 mM ATP, 50 M dithiothreitol, and 5 M ubiquitin at 30°C. Unless stated otherwise, 50 nM of E1 (from calf thymus) and 100 nM of Cdc34 or its mutant proteins were used in the reactions. The amounts of Cdc34 and mutant Cdc34 proteins in the assay were determined by quantitation with Cdc34specific antibodies using the Amersham ECL detection method and purified Cdc34 standards. Reactions were stopped by withdrawing aliquots of the reaction mixture into SDS-sample buffer in which ␤-mercaptoethanol had been omitted. When Ub-Cdc34 complexes were assayed, protein samples were adjusted to contain 5% ␤-mercaptoethanol, and samples were heated in a boiling water bath for 3 min. Protein samples were subjected to electrophoresis in a 14% SDS-polyacrylamide gel, and autoradiography was used to visualize radiolabeled bands.
Preparation of Polyclonal Anti-Cdc34 Antibodies and Western Blot Analysis-Anti-Cdc34 antiserum was prepared in New Zealand White rabbits using the recombinant Cdc34 protein produced and purified by the method of Banerjee et al. (1993). Immunization and antibody processing techniques were done according to the protocols of Harlow and Lane (1988). For visualization and determination of the amount of Cdc34 protein, yeast cells were harvested by centrifugation, washed once with 50 mM Tris-HCl buffer, pH 7.5, 1 mM dithiothreitol, and resuspended in 200 l of breakage buffer (50 mM Tris-HCl, pH 7.5, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 2 g/ml leupeptin, and 2 g/ml pepstatin A). Cells were broken by vortexing with glass beads. A 100-l aliquot of 3 ϫ PAGE sample buffer (without ␤-mercaptoethanol) was added, and the mixture was immediately boiled for 3 min. The glass beads and cell debris were removed by centrifugation. A 10-l aliquot was removed for protein assay by Pierce BCA reagent, and the remainder of the sample was adjusted to contain 5% ␤-mercaptoethanol. Cell extracts containing equal amounts of protein (300 g) were electrophoresed on a 15% SDS-PAGE gel (Laemmli, 1970). Proteins were transferred to polyvinylidene difluoride membrane and visualized by reaction with anti-Cdc34 polyclonal antibodies by protocols described by Harlow and Lane (1988). The enhanced chemiluminiscence (ECL) kit was purchased from Amersham Corp. and used according to their instructions. Alkaline phosphatase-conjugated anti-rabbit IgG was purchased from Sigma, and color reaction was done according CLN2 with T7 promoter Deshaies et al. 1995 to Blake et al. (1984).

Analysis of Cln2
Multiubiquitination-Expression of Cln2 in vitro was achieved by inserting a BamHI fragment containing the CLN2 gene into the BamHI site of pGEM2, yielding pRD68, where the CLN2 gene could be transcribed by the T7 promoter (Deshaies et al., 1995). This RNA was used for in vitro translation by the rabbit reticulocyte lysate system. Reagents for the in vitro translation were obtained from Promega Biotech. [ 35 S]Cln2 synthesized in vitro was incubated in 15 mg/ml pRD84/RD205-3A extract. Reactions were supplemented with ϳ1 M Cdc28 (Deshaies et al., 1995) and the indicated amounts of Cdc34 or Cdc34-C95S,L99S, all purified from E. coli. Reactions were directly evaluated by SDS-PAGE and fluorography.

Overexpression of Cdc34-C95S,L99S Inhibits Cell Growth-
The S. cerevisiae CDC34 gene encodes a ubiquitin-conjugating enzyme that is required for the cell cycle G 1 /S transition (Byers and Goetsch, 1973). The yeast strain MGG15 contains a temperature-sensitive cdc34 -2 allele, and these cells are inviable at nonpermissive temperatures (Goebl et al., 1988). In the course of using MGG15 cells to determine the functional competence of several CDC34 site-specific mutants, we noted that MGG15 cells failed to form colonies even at permissive temperatures when these cells are also expressing a CDC34-C95S,L99S mutant via a galactose-inducible promoter in a 2 plasmid (Fig. 1A). In contrast, similar expression of the two singly substituted mutants, cdc34-C95S and cdc34-L99S, did not affect the viability of MGG15 cells. CDC34-C95S,L99S is likely to encode an inactive protein since the singly substituted mutants, cdc34-C95S and cdc34-L99S, each failed to complement the temperature-sensitive growth defect of MGG15 cells (data not shown).
To test whether the dominant effect of CDC34-C95S,L99S mutant is uniquely dependent on the cdc34 -2 allele, we also introduced this mutant gene into a yeast strain that carries a wild type CDC34 allele. The ABY100 strain contains an integrated copy of CDC34-C95S,L99S whose expression is regulated by the P GAL1 promoter (Table II). These cells also failed to grow on galactose medium (Fig. 1B) and became inviable (data not shown). These results indicate that the dominant effect of CDC34-C95S,L99S on cell growth and viability is not restricted to the cdc34 -2 strain. Similar expression of the singly substituted mutants, cdc34-C95S and cdc34-L99S, did not alter cell viability (data not shown).
Since the unique effect of the CDC34-C95S,L99S mutant may be due to its higher level of accumulation than those of the  (Table II) cells carrying ts-cdc34 -2 at the permissive temperature of 23°C. ABY210 (cdc34 -2 ϩ YEpGALCDC34), ABY214 (cdc34 -2 ϩ YEpGAL-cdc34-L99S), ABY216 (cdc34 -2 ϩ YEpGALcdc34-C95S), and ABY212 (cdc34 -2 ϩ YEp GALCDC34-C95S,L99S were grown to an A 600 of 0.2 in SR medium at 23°C and streaked in duplicate onto medium containing galactose. Panel B, the GAL1-CDC34-C95S,L99S mutant was integrated into wild type yeast strain 332 (ABY100). Similarly integrated GAL1-CDC34 resulted in ABY200 (Table II). Spot assays were done to compare the viability of ABY100 and ABY200 cells upon galactoseinduced expression of CDC34 and CDC34-C95S,L99S alleles. ϳ25,000 cells (grown in SR medium to an A 600 of 0.2, and counted by hemocytometer) were spotted in the first spot on each panel. The next two spots in each panel are two successive 10-fold dilutions. Plates were incubated at 30°C. ABY200 (URA3::GALCDC34) spots in galactose medium show no deleterious effect or loss in viability, whereas ABY100 (URA3::GALCDC34-C95S,L99S) spots in galactose medium show lethal effect on wild type yeast cells by expression of the CDC34-C95S,L99S allele.  Deshaies et al. (1995) singly substituted mutants, cdc34-C95S and cdc34-L99S, we also used immunoblotting to assess the level of mutant proteins in cells expressing these various mutant genes. Immunoblots of Cdc34 proteins from cells before and after 2 or 8 h of induction in galactose indicated that the Cdc34-C95S,L99S protein did not accumulate to a higher level than the other mutants (data not shown). Thus, the unique effect of Cdc34-C95S,L99S on cell viability is not due simply to the overexpression of an inactive Cdc34 protein.
CDC34-C95S,L99S Acts on the Same Pathway as CDC34 -Immunoblots, probed with ubiquitin-specific antibodies, showed that the expression of CDC34-C95S,L99S, cdc34-C95S, and cdc34-L99S mutants did not cause detectable changes in the overall level of ubiquitin-protein conjugates (data not shown). However, two lines of evidence indicated that CDC34-C95S,L99S affects cell viability by disruption of CDC34-dependent protein ubiquitination. First, the effect of CDC34-C95S,L99S in ABY100 cells was suppressed by the concomitant expression of wild type CDC34 via either its natural promoter or a galactose-inducible promoter in 2 plasmids (data not shown). Similar plasmids carrying either the singly substituted cdc34-C95S or cdc34-L99S mutant did not restore growth of ABY100 cells on galactose medium (data not shown). Second, purified Cdc34-C95S,L99S protein inhibited the in vitro ubiquitination of the G 1 cyclin Cln2, a reaction that has recently been shown to require Cdc34 (Deshaies et al., 1995). Yeast extract derived from wild type CDC34 cells catalyzes the ubiquitination of Cln2 obtained by in vitro translation in reticulocyte lysate (Deshaies et al., 1995; Fig. 2, lane 2). This reaction could be further stimulated by the addition of purified Cdc34 protein (Fig. 2, lane 3). Ubiquitination of this substrate was inhibited by additions of purified Cdc34-C95S,L99S protein in a dose-dependent manner (Fig. 2, lanes 4 and 5).
The inhibition of Cln2 protein ubiquitination is not due to general inactivation of ubiquitin conjugation pathways since the overexpression of this mutant protein did not affect ubiquitin conjugation to other endogenous proteins (Fig. 3A). We have also assayed the effect of Cdc34-C95S,L99S on purified ubiquitin-activating enzyme by monitoring the catalytic transfer of ubiquitin from the ubiquitin-activating enzyme to another yeast ubiquitin-conjugating enzyme, Rad6 (Fig. 3B). As shown in Fig. 3B, the level of ubiquitin-Rad6 thiol ester was not detectably affected by 2-20 M of Cdc34-C95S,L99S. These results, taken together, indicate that Cdc34-C95S,L99S inhibits Cln2 ubiquitination via specific inhibition of the Cdc34-dependent pathway.
While the above results indicated that CDC34-C95S,L99S exerts its effect by interfering with an essential CDC34-dependent process, cells overexpressing this mutant did not ex- . Panel A, extracts were prepared from 10-h galactose-induced ABY200 (URA3::GALCDC34), and ABY100 (URA3::GALCDC34-C95S,L99S) in breakage buffer; ϳ300 g of total protein in 60 l volume was supplemented with 200 ng of 125 I-Ub and incubated at 30°C for 30 min. 30 l of 3 ϫ SDS-PAGE sample buffer was added, samples were boiled for 3 min, and the reactions were analyzed by 14% Laemmli gels. Lanes 1 and 3, ABY100 reaction at 10 and 30 min; lanes 2 and 4, ABY200 reaction at the same time points. The radioactive protein ladder indicating ubiquitination of yeast proteins is similar in intensity and pattern for extracts from both of these cell types. The positions of Life Technologies, Inc. prestained high molecular mass markers (myosin, 215 kDa; phosphorylase B, 105 kDa; bovine serum albumin, 70 kDa; ovalbumin, 43 kDa; carbonic anhydrase, 29 kDa; ␤-lactoglobulin, 18 kDa; and lysozyme, 14 kDa) are marked on the left. Panel B, formation of Rad6 S-Ub thiol ester complexes was assayed by incubation with 250 nM purified Rad6 protein and indicated amounts of Cdc34-C95S,L99S protein with ubiquitin-activating enzyme (E1), magnesium, ATP, and 125 I-ubiquitin for 20 min at 30°C. The reaction products were electrophoresed in a 14% SDS gel in the absence of thiol-reducing agents, and radiolabeled proteins were visualized by autoradiography.
hibit the morphological phenotype of previously characterized loss-of-function mutants. The cdc34 null mutant strain, as well as the temperature-sensitive cdc34 -1 and cdc34 -2 strains arrest as multibudded cells (Goebl et al., 1988). This morphology is absent in ABY100 cells that assume aberrant morphology after switching cells to a galactose-containing medium. A majority of these aberrant cells were found to have a single elongated bud (data not shown). The absence of multibudded cells is not due to a strain difference since MGG15 cells that are overexpressing Cdc34-C95S,L99S also took on, at both permissive and nonpermissive temperatures for cdc34 -2, morphologies similar to those of ABY100 cells in galactose (data not shown). This difference in morphology is consistent with the notion that Cdc34-C95S,L99S does not simply inactivate endogenous wild type Cdc34 protein. Possible mechanisms of Cdc34-C95S,L99S action are described further under "Discussion." The Role of Cys 95 and Leu 99 in Cdc34-C99S,L99S-Since neither cdc34-C95S nor cdc34-L99S affects cell viability, it appears that both mutations must be present to confer the dominant negative phenotype. Cys 95 is expected to be an essential residue since it is the only cysteine in the Cdc34 protein, and the presence of this amino acid is required for all ubiquitin-conjugating enzymes to form a thiol ester complex with ubiquitin. Leu 99 is a highly conserved residue among ubiquitin-conjugating enzymes, and the inability of cdc34-L99S to complement cdc34 -2 suggests that this residue is also essential. We have also generated a set of mutants in which Cys 95 and Leu 99 were replaced by alanines or by alanine/serine pairs. None of these mutants was able to complement cdc34 -2 (data not shown), suggesting that these two residues are indeed essential. We also constructed yeast strains with wild type CDC34 that had an integrated copy of these mutant genes under the control of a P GAL1 promoter (ABY110, ABY120, ABY130, ABY140, ABY150, ABY160, and ABY170 in Table II) and tested their effect on cell growth. Unlike CDC34-C95S,L99S, none of these mutants was found to block cell growth (data not shown). These results suggest that the dominant negative effect may have a stringent requirement for serines at these two residue positions.
The requirement for a serine at the Cys 95 residue may be due to the unique ability of a serine to form a more stable linkage with ubiquitin (Sung et al., 1991). In a normal ubiquitin-conjugating enzyme, the active site cysteine forms a thiol ester linkage with the C-terminal carboxyl of ubiquitin. This thiol ester-linked ubiquitin is subsequently transferred either to a E3 protein or to substrates directly. Substitution of the active site cysteine by a serine has been shown to inactivate other ubiquitin-conjugating enzymes (Sung et al., 1991;Seufert et al., 1995) presumably because the more stable oxygen ester-linked ubiquitin is not further transferred. In the case of Cdc34, we have previously shown that the thiol ester-linked ubiquitin can also be transferred to a lysine within this enzyme in an intramolecular reaction to form a Lys 48 -specific multiubiquitin chain (Banerjee et al., 1993). This autoubiquitination reaction was used here to test the effect of Cys 95 substitution. For these experiments, the cdc34 mutants were expressed in E. coli and assayed for their ability to accept [ 125 I]ubiquitin in the presence of added ubiquitin-activating enzyme (Fig. 4). Under our assay conditions, the reaction with wild type Cdc34 leads to products that migrated as a set of discrete bands on SDS gels (Fig. 4A, lane 1). These discrete bands are due to the linkage of multiple ubiquitin groups, in the form of a ubiquitin chain, to a lysine residue in Cdc34 (Banerjee et al., 1993). In contrast, only a single ubiquitin-Cdc34 adduct was detected with the mutants Cdc34-C95S, Cdc34-C95S,L99A, and Cdc34-C95S,L99S. This single adduct was not present in any of the C95A-substituted mutants, consistent with the notion that this single adduct is due to the linkage of ubiquitin to the substituted serine at position 95. The absence of additional adducts of lower electrophoretic mobilities in these C95S-substituted mutants indicates that autoubiquitination did not occur with these mutants. These results indicate that the C95S-substituted cdc34 mutants can indeed form a more stable oxygen ester with ubiquitin and raise the possibility that the effect of CDC34-C95S,L99S on cell viability may require the linkage of ubiquitin to Ser 95 .
We have also monitored the formation of ubiquitin-Cdc34 thiol ester with shorter reaction time (15 min), using a nonreducing SDS gel (Fig. 4B, lanes 1-3). Both Cdc34-L99S and Cdc34-L99A were found to retain partial activity as indicated by the presence of a 42-kDa radiolabeled band that corresponds to ubiquitin-Cdc34 thiol ester. This result indicated that unlike FIG. 4. Formation of Ub-Cdc34 complexes by the mutant Cdc34 proteins. Panel A, Cdc34 or its mutants were expressed in E. coli harboring the appropriate M13 mp19/18-CDC34 constructs (see "Experimental Procedures"). Amounts of the recombinant protein in the bacterial extracts were normalized for the assays. Thiol-insensitive Ub-Cdc34 complexes were assayed by incubating E. coli extracts with ϳ100 nM overexpressed Cdc34 or its mutant proteins with ubiquitinactivating enzyme, magnesium, ATP, and 125 I-ubiquitin for 45 min. Samples were adjusted to contain 5% ␤-mercaptoethanol and heated at 90°C for 3 min prior to SDS-PAGE. The uppermost band corresponds to ubiquitin linked to a lysine on E1. The ladder bands with the wild type Cdc34 are due to the linkage of a multiubiquitin chain to a lysine on Cdc34 (Banerjee et al., 1993). The single band with the C95S mutants is presumably due to the formation of a Ub-Cdc34 oxygen-ester at Cdc34-Ser 95 (position indicated by arrow on the right). Panel B, formation of Cdc34 S-Ub thiol ester complexes was assayed by incubating E. coli extracts with ϳ100 nM overexpressed Cdc34 or its mutant proteins with ubiquitin-activating enzyme, magnesium, ATP, and [ 125 I]ubiquitin as in panel A, except the reaction time was reduced to 15 min. The reaction products were electrophoresed in a 14% SDS gel in the absence of thiol-reducing agents, and radiolabeled proteins were visualized by autoradiography. Open arrow indicates Ub-Cdc34 complex; closed arrow indicates E1 S-Ub complex. mutations at Cys 95 , mutants of Leu 99 are inactive at a step subsequent to ubiquitin-thiol ester formation. Thus, mutations at Leu 99 appear to interfere with Cdc34 interactions with either substrate or E3 protein(s). Interestingly, a higher activity was found for the L99S-substituted mutant than the L99A mutant. Similarly, the shorter reaction time also revealed a faster ubiquitin linkage to Ser 95 in Cdc34-C95S,L99S as compared with Cdc34-C95S and Cdc34-C95S,L99A (Fig. 4B, lanes 4, 9, and 5, respectively). Whether this difference in reactivity could account for the stringent requirement for the L99S substitution remains to be determined.

DISCUSSION
The ability of Cdc34-C95S,L99S to inhibit the cell cycle function of CDC34 is indicated by its in vivo effect on cell viability and by its in vitro effect on Cln2 ubiquitination. Proteins that are known to be targeted by Cdc34 for ubiquitin-mediated proteolysis include G 1 cyclins (Deshaies et al., 1995, Yaglom et al., 1995, the yeast transcription factor GCN4 (Kornitzer et al., 1994) and the Cdc28 kinase inhibitor Sic1 (Schwob et al., 1994). Since overexpression of Cln2 does not lead to cell inviability (Lew and Reed, 1993), it is likely that Cdc34-C95S,L99S also inhibits the degradation of other Cdc34-dependent substrates. One likely candidate is Sic1, which is normally degraded prior to cell entry into the S phase (Schwob et al., 1994), and a moderate overexpression of this protein has previously been shown to produce cellular morphology (Nugroho and Mendenhall, 1994) similar to those found for cells overexpressing Cdc34-C95S,L99S. Consistent with this notion is the recent demonstration that the human homolog of Cdc34-C95S,L99S could also inhibit Cdc34-dependent degradation of the cyclindependent kinase inhibitor, p27 (Pagano et al., 1995). The ability of Cdc34-C95S,L99S to inhibit the in vitro ubiquitination and/or degradation of two dissimilar substrates in two different species raises the possibility that this mutant may be used in analogous manner to establish the identity of additional Cdc34-specific substrates.
Cells expressing Cdc34-C95S,L99S exhibit a morphology that differs significantly from the multibudded morphology of previously characterized loss-of-function cdc34 mutants. A significant proportion of these cells contain a single, elongated bud, while multibudded cells are conspicuously absent. As the mechanism for multibudding in the cdc34 null mutant has not been defined, the morphological difference here could not be readily addressed. Nonetheless, this difference suggests that the effect of Cdc34-C95S,L99S is not equivalent to a straightforward loss of CDC34 functions. Previous studies have indicated that Cdc34 is capable of self association, and this process requires a region in the sequence that is apparently essential for its cell cycle function (Ptak et al., 1994). Conceivably, Cdc34-C95S,L99S could exert its effect by sequestering endogenous Cdc34. However, this mechanism is incompatible with the absence of multibudded cells. Furthermore, the effect of Cdc34-C95S,L99S could not be obtained with the other seven inactive cdc34 mutants that contain the same determinant for self-association. In addition, we have obtained preliminary results indicating that purified Cdc34-C95S,L99S does not inhibit a previously characterized in vitro autoubiquitination of Cdc34 (Banerjee et al., 1993) or the conjugation of ubiquitin to histone proteins (Haas et al., 1991). Thus, it is unlikely that the effect of Cdc34-C95S,L99S is due to the sequestration of endogenous Cdc34.
Since Cdc34-C95S,L99S encodes an inactive ubiquitin-conjugating enzyme, the mechanism of inhibition is likely to reside in a binding step where this mutant could compete effectively with Cdc34. A key unanswered question here is whether substrate recognition in this pathway also requires E3 proteins. A requirement for E3 has been shown for several other ubiquitination pathways. For example, the Ubr1 protein is required in the N terminus rule pathway (Bartel et al., 1990), and a protein known as E6AP is required for the ubiquitination of p53 (Scheffner et al., 1993). The ubiquitination of mitotic cyclins appears to require a large protein complex consisting of several distinct proteins (Sudakin et al. (1995); reviewed in Murray (1995)). Thus, the effect of Cdc34-C95S,L99S could result from the sequestration of Cdc34-specific substrates or the required E3 protein(s).
Cys 95 and Leu 99 are located within a sequence region that is highly conserved among ubiquitin-conjugating enzymes. This conserved sequence region has been termed the catalytic core domain and is conserved in tertiary folding as shown by the crystal structures of Arabidopsis thaliana Ubc1 and Saccharomyces cerevisiae Ubc4 (Cook et al., 1993). A structural model of the Cdc34 catalytic core could be constructed by aligning residues 10 -100 of Cdc34 with the N-terminal 91 residues of Ubc4. FIG. 5. Panel A, model of the ternary complex formation between E3 and E2 S-Ub ubiquitin-E2 thiol ester complex docks on an E3 by noncovalent interactions to form II. Both ubiquitin and E2 contribute to the stability of II. Transthiolation (Scheffner et al., 1995) leads to the attachment of ubiquitin to a cysteine in E3, and ubiquitin no longer contributes to the retention of E2 in the ternary complex III. In the case of Ub-Cdc34-C95S,L99S oxygen ester, ubiquitin is not transferred to E3, leading to the sequestration of E3 in an inactive complex. Panel B, alignment of the Cdc34 catalytic site sequence with other ubiquitin-conjugating enzyme sequences. The yeast Cdc34 (Goebl et al., 1988) sequence is aligned with those of human Cdc34 (Plon et al., 1993), yeast Ubc7 (Jungmann et al., 1993), Rad6 (Jentsch et al., 1987), and Ubc4 (Seufert and Jentsch, 1990) to show the positioning of the 12/13-residue segment in Cdc34 and Ubc7. Positions of mutant residues in Cdc34-C95S,L99S are indicated below by closed circles. The starting and end residue numbers for each sequence in the alignment are given in parentheses.
Both Cys 95 and Leu 99 could be placed within this structural model at positions that are occupied by identical amino acids in Ubc4. In this model, substitution of Cys 95 by either alanine or serine would not introduce other structural perturbations. Thus, the difference between Cdc34-C95S,L99S and Cdc34-C95A,L99S is unlikely to be structural but rather in the ability of Ser 95 to form a stable oxygen ester with ubiquitin. This suggests that the inhibitory effect of Cdc34-C95S,L99S may require prior formation of the ubiquitin-Cdc34-C95S,L99S ester.
A model that could account for the inhibitory effect of Cdc34-C95S,L99S is depicted in Fig. 5A. In this model, ubiquitin contributes partly to the energetics of the ternary complex formation between the ubiquitin-E2 thiol ester and E3. Once ubiquitin has been transferred to E3, the ubiquitin-conjugating enzyme would presumably bind less tightly since it is no longer linked to ubiquitin. The reduced affinity may then facilitate the dissociation of the ubiquitin-conjugating enzyme, which could be recharged with ubiquitin by the ubiquitin-activating enzyme (E1). The existence of a ubiquitin binding site on E3 is supported by studies on a reticulocyte E3 in the N-end rule pathway (Reiss and Hershko, 1990). This model makes the prediction that E2 mutants containing a stably linked ubiquitin would be better inhibitors than inactive enzymes that cannot be linked with ubiquitin and explains the unique requirement for the C95S mutation. The requirement for the L99S mutation could be explained by the observation that this mutation causes ubiquitin to be linked to Ser 95 at a faster rate (Fig. 4). A structural basis of this effect could not be readily assessed using the two known structures of E2 enzymes since Cdc34 contains an extra 12-residue segment beginning at residue 101, and this extra segment could not be accommodated in a structural model. A similar sequence segment is also present in the yeast Ubc7 protein (Fig. 5B), and the crystal structure of this ubiquitin-conjugating enzyme has recently been determined, 2 and work is in progress to determine the mutational effect of the corresponding leucine residue in this enzyme.
Dominant negative mutants in other yeast genes have proved useful for the identification of interacting proteins via suppresser analyses. If the action of Cdc34-C95S,L99S depends on its interaction with E3 proteins, one class of suppressers is expected to be comprised of these proteins. The ability of purified Cdc34-C95S,L99S to block the in vitro ubiquitination of Cln2 suggests a further utility of this mutant for the identification of additional Cdc34-specific substrates. Although substrates in a specific ubiquitin-dependent pathway could usually be identified in yeast by showing their increased stability in a specific ubc mutant, this approach may be insufficient to provide unambiguous identification of CDC34-specific substrates. For example, while the G 2 -specific B-type cyclins are stabilized in CDC34 mutants (Amon et al., 1994), degradation of these cyclins is apparently mediated by UBC9 (Seufert et al., 1995), and the effect of cdc34 in this case could be attributed to the abnormal accumulation of Cln proteins in cdc34 mutants. Thus, an unambiguous identification of a CDC34-specific substrate may also require the use of in vitro approaches to demonstrate a direct requirement of this ubiquitin-conjugating enzyme. One such approach may be by using the Cdc34-C99S,L99S mutant protein to inhibit the ubiquitination or the degradation of a candidate substrate protein in a cell-free system. This approach has been used recently to help in establishing a role of Cdc34 in degradation of the human cyclin-dependent kinase inhibitor, p27. Dominant negative mutants of yeast UBC genes could be readily identified by genetic screens. The creation of an analogous dominant negative mutant of human Cdc34 with mutations identified in the yeast enzyme points to the important possibility that other dominant negative mutants of mammalian ubiquitin-conjugating enzymes could be obtained by a similar approach. Such mutants could then be used for exploring substrates and/or regulators of protein ubiquitination in mammalian systems.