A Yeast Ubc9 Mutant Protein with Temperature-sensitive in Vivo Function Is Subject to Conditional Proteolysis by a Ubiquitin- and Proteasome-dependent Pathway*

The UBC9 gene of the yeast Saccharomyces cerevisiae is essential for cell viability and encodes a soluble protein of the nucleus that is metabolically stable. Products of mutant alleles selected to confer temperature-sensi-tive in vivo function were found to be extremely short- lived at the restrictive but long-lived at the permissive condition. An extragenic suppressor mutation was iso- lated which increased thermoresistance of a ubc9-1 strain. This suppressor turned out to stabilize the mu- tated gene product, indicating that the physiological activity of ubc9-1 protein is primarily controlled by con- ditional proteolysis. The labile ubc9-1 protein appears to be a substrate for ubiquitination, and its turnover was substantially reduced by expression of a ubiquitin derivative that interferes with formation of multi-ubiq- uitin chains. Stabilization resulted also from competitive inhibition of Ubc4-related ubiquitin-conjugating enzymes. Activity of the proteasome complex was crucial to rapid breakdown, whereas vacuolar proteases were dispensable. Thus, the heat-denatured ubc9-1 protein is targeted for proteolysis by the ubiquitin-protea- some pathway and may serve as a useful tool to further define the process by which a misfolded polypeptide is recognized. Whereas most cellular proteins are metabolically stable under normal growth conditions, a specific subset exhibits a high rate of turnover. A protein’s short half-life allows rapid adjust-ment of its concentration in response to changes in synthesis. And indeed,

The UBC9 gene of the yeast Saccharomyces cerevisiae is essential for cell viability and encodes a soluble protein of the nucleus that is metabolically stable. Products of mutant alleles selected to confer temperature-sensitive in vivo function were found to be extremely shortlived at the restrictive but long-lived at the permissive condition. An extragenic suppressor mutation was isolated which increased thermoresistance of a ubc9-1 strain. This suppressor turned out to stabilize the mutated gene product, indicating that the physiological activity of ubc9-1 protein is primarily controlled by conditional proteolysis. The labile ubc9-1 protein appears to be a substrate for ubiquitination, and its turnover was substantially reduced by expression of a ubiquitin derivative that interferes with formation of multi-ubiquitin chains. Stabilization resulted also from competitive inhibition of Ubc4-related ubiquitin-conjugating enzymes. Activity of the proteasome complex was crucial to rapid breakdown, whereas vacuolar proteases were dispensable. Thus, the heat-denatured ubc9-1 protein is targeted for proteolysis by the ubiquitin-proteasome pathway and may serve as a useful tool to further define the process by which a misfolded polypeptide is recognized.
Whereas most cellular proteins are metabolically stable under normal growth conditions, a specific subset exhibits a high rate of turnover. A protein's short half-life allows rapid adjustment of its concentration in response to changes in synthesis. And indeed, many regulatory proteins whose abundance is crucial for their proper in vivo function and certain rate-limiting enzymes are subject to selective degradation. Prominent examples of such short-lived proteins include ornithine decarboxylase, a key enzyme in polyamine biosynthesis (Murakami et al., 1992), the proto-oncogene products c-Mos (Watanabe et al., 1989) and c-Jun (Treier et al., 1994), the tumor suppressor protein p53 (Scheffner et al., 1990), and the cell cycle regulator cyclin B (Murray et al., 1989).
While metabolic instability is an intrinsic and natural property of specific proteins, normally stable polypeptides may be targeted for rapid degradation when their native conformation has been disrupted. Structural damage is frequently caused by external stress like heat or heavy metals. Occasional errors in transcription, translation, or folding will also give rise to abnormal protein variants. Similarly, unassembled subunits of multimeric complexes, such as free ribosomal proteins (Maicas et al., 1988), and misdirected proteins, like those that fail to be exported from the endoplasmic reticulum (Klausner and Sitia, 1990), were found to be substrates for proteolytic breakdown.
The ubiquitin-proteasome system constitutes a major pathway for the selective degradation of proteins in the nucleus and cytoplasm of eukaryotic cells (Ciechanover, 1994;Peters, 1994;Goldberg, 1995;Jentsch and Schlenker, 1995). Substrates of this proteolytic system are modified by the attachment of multiple molecules of ubiquitin, a small and highly conserved protein, prior to their destruction by the 26 S proteasome. The catalytic core of this macromolecular complex is the so-called the 20 S proteasome, a cylindrical particle composed of four rings with each seven protein subunits. This 20 S particle associates with two 19 S cap complexes, presumed regulatory subunits that confer energy dependence and specificity on the larger 26 S proteasome. Lysosomes in mammalian cells or yeast vacuoles, on the other hand, contain numerous proteases that act on proteins delivered by endocytosis or autophagy, a nonselective uptake process observed mainly under conditions of nutritional deprivation. Vacuolar protease activity was shown to be essential for the turnover of several plasma membrane proteins in yeast (Volland et al., 1994;Kölling and Hollenberg, 1994;Lai et al., 1995). A role of these proteases in catabolite inactivation of the cytoplasmic enzyme fructose-1,6bisphosphatase is controversial (Schork et al., 1994;Chiang and Schekman, 1994).
The value of temperature-sensitive mutations (ts) 1 in the genetic analysis of cellular processes has been amply demonstrated (see Alberts et al. (1994), Lewin (1994), and Murray and Hunt (1993)). Temperature-sensitive mutations are presumed in most cases to be missense mutations resulting in specific amino acid replacements so that mutant gene products retain their proper structure and function only at relatively low temperatures. Despite their widespread use, the mechanisms underlying the conditional activity of ts mutant proteins have rarely been analyzed. This report describes temperature-sensitive mutations in the UBC9 gene of the yeast Saccharomyces cerevisiae. Products of ts alleles were found to undergo a temperature-dependent switch in metabolic stability, and experimental evidence suggests that efficient proteolysis accounts for the restricted physiological activity of ts ubc9p derivatives. Analysis of the degradation pathway indicated that breakdown of the thermolabile ubc9-1 protein in vivo is energy-dependent, involves multi-ubiquitination and depends on the integrity of the proteasome, but does not require vacuolar protease activity.
Yeast Ubc9p is a soluble nuclear protein and belongs to a family of ubiquitin-conjugating (Ubc) enzymes that catalyze the transfer of ubiquitin to cellular substrates (Jentsch, 1992;Hershko and Ciechanover, 1992;Finley and Chau, 1991). To date, 13 Ubc proteins have been identified in yeast and homologs are known to exist in other eukaryotic species, including plants and mammals. Sequence conservation and crystallographic studies indicate that in a core domain of about 150 amino acids, all Ubc proteins have a similar overall structure with four ␣-helices and a four-stranded antiparallel ␤-sheet (Cook et al., 1992(Cook et al., , 1993. Genetic analyses revealed that the cellular functions of these enzymes are diverse. Specific Ubc proteins are critically involved in stress resistance Jungmann et al., 1993), DNA repair (Jentsch et al., 1987), peroxisome biogenesis (Wiebel and Kunau, 1992), or cell cycle progression (Goebl et al., 1988;Schwob et al., 1994).
Mutant Alleles and Expression Constructs-The yeast UBC9 gene was cloned as a 1.5-kilobase pair XbaI-SspI genomic DNA fragment in plasmids pSE360 (ARS1, CEN4, URA3) and pSE362 (ARS1, CEN4, HIS3) to facilitate further manipulations. These constructs complement a chromosomal deletion of UBC9. Codons 68 and 69 were mutagenized by a dual step polymerase chain reaction (PCR) (Landt et al., 1990;Sharrocks and Shaw, 1992). Flanking primers WS42 (GCGGTATAC-TAACAAATCGATGAAC) and WS43 (TGAGAATTCTACTAACCAATA-GAT) were designed to allow exchange of mutated portions as ClaI-EcoRI fragments (sites are underlined). Two internal primers (WS59: GGTTTTGAIIIATATTCATTTGG and WS60: GTTTTGAAGGIIIT-TCATTTGG) were used together with WS42 in separate first step reactions (I: inosine). Products were cloned in pSE362-UBC9, and DNA was prepared from a pool of about 3000 E. coli transformants each. Temperature-sensitive alleles ubc9-2 and ubc9-3 were identified by transformation of yeast strain YWO90 (MATa, ubc9-⌬1::TRP1, pSE360-UBC9), plasmid shuffling (Sikorski and Boeke, 1991), and replica plating. Internal mutagenic primers WS56 (ACAATAGCTTTAAG-TATTTTAAATGAA) and WS65 (TTAGACGCTCCAAATCCAAATGC-CCCTGC) were used to construct alleles ubc9-4 and ubc9-5, respectively (altered codons are underlined). For fusion to the GAL1 promoter coding sequences of UBC9 and CDC28 alleles were inserted into plasmid pSE936 (Elledge et al., 1991) between the restriction sites EcoRI and XbaI. These fragments were obtained by PCR amplification with UBC9-specific primers WS38 (GGGAATTCGGAAGCAATATGAG-TAG) and WS39 (GCTCTAGAGGGAAAGATGGATTCCC) and CDC28specific primers WS67 (CTGAATTCGAACATGAGCGGTGAATTAGCA) and WS68 (ATACTAGTGCTTATGATTCTTGGAAGTAGGGG). To replace the active site cysteine residue in Ubc4 by a serine residue, the codon TGT at position 86 was changed to AGC by PCR-mediated mutagenesis. The ubc4-C 86 S allele was fused to the CUP1 promoter in pRS304 and integrated at the TRP1 locus in strain W9473. For deletion of the PRA1 gene, construct pra1⌬EM::URA3 was used in which the coding sequence between EcoRI (position 645) and MscI (position 1929) has been replaced by the URA3 marker. Yep96 and pUB204 are high copy (2 origin) plasmids with a TRP1 selectable marker that express ubiquitin or the dominant negative derivative Ub-K 48 R;G 76 A, respectively (Finley et al., 1994). For induction of these genes that are controlled by the CUP1 promoter, cells were treated with 0.1 mM CuSO 4 for 6 h.
Protein Turnover Measurements-Yeast strains carrying alleles of UBC9 or CDC28 fused to the GAL1 promoter in pSE936 (Elledge et al., 1991) were grown at 22°C in synthetic complete medium with 2% galactose. Time courses were started by addition of 2% glucose to repress transcription from the GAL1 promoter except for the experiment shown in Fig. 7, where total protein synthesis was inhibited by addition of 0.1 mg/ml cycloheximide. Cultures were shifted immediately to 37°C or remained at 22°C, and samples were withdrawn at appropriate intervals. Cells were harvested by centrifugation, washed with cold water, and frozen in liquid nitrogen. Cell pellets were resuspended in lysis buffer (0.1% Nonident P-40, 150 mM NaCl, 50 mM NaF, 5 mM EDTA, and 50 mM Tris-Cl, pH 7.5) with inhibitors (1 mM phenylmethylsulfonyl fluoride, 1 g/ml N-tosyl-L-phenylalanine chloromethyl ketone, 0.6 mM dimethylaminopurine, 1 g/ml leupeptin, 1 g/ml pepstatin, 10 g/ml soybean trypsin inhibitor) and disrupted by vortexing with glass beads. Lysates were cleared by centrifugation at 12,000 ϫ g for 5 min, and protein concentration was measured by the Bradford assay. After boiling in sample buffer, 20 g of total protein per lane was separated on a 12% polyacrylamide-SDS gel and transferred to a nitrocellulose filter with a semi-dry apparatus. Equal loading and transfer was confirmed by staining the membrane with Ponceau S. After preincubation in 5% non-fat dry milk plus 0.1% Tween 20 in Tris-buffered saline, immunoblots were probed with rabbit antisera against Ubc9p (K4S3) or Cdc28p (kindly provided by Dr. C. Mann, Gif-sur-Yvette, France) at a dilution of 1:1000. Donkey anti-rabbit IgG conjugated to horseradish peroxidase and enhanced chemiluminescence reagents (Amersham Corp.) were used for immunodetection. For production of an antiserum in rabbits, a UBC9 cDNA fragment (Seufert et al., 1995) was inserted into the hexahistidine vector pQE9 (Quiagen). Ubc9p was expressed in the E. coli strain M15 and purified by nickel-chelate chromatography (Quiagen).

Thermosensitive Derivatives of Yeast Ubc9
Protein-The UBC9 gene from the yeast Saccharomyces cerevisiae encodes a ubiquitin-conjugating enzyme essential for cell viability (Seufert et al., 1995). To study its in vivo function we were interested to generate temperature-sensitive ubc9 alleles. Since a ts mutation in CDC34, another member of the UBC gene family, changes the codon of a conserved proline residue (Ellison et al., 1991), the corresponding and a neighboring codon of UBC9 were subjected to a PCR-mediated random mutagenesis. A pool of mutagenized plasmids was used to transform strain YWO90 that carries a deletion of the chromosomal UBC9 gene and is rescued by a plasmid-borne copy. Following plasmid shuffling (Sikorski and Boeke, 1991) transformants were assayed for temperature sensitivity. Two strains were isolated that grew well at 22°C, but failed to grow at 37°C and regained temperature resistance after introducing wild type UBC9. The mutant alleles of these strains were termed ubc9-2 and ubc9-3. DNA sequencing confirmed that the only alterations in these alleles map to the targeted codons. The predicted amino acid changes in the ts proteins are given in Fig. 1 together with the sequence of the previously constructed ts derivative ubc9-1p. These residues are located in a proline-rich turn region within an antiparallel ␤-sheet structure present in Ubc proteins (Cook et al., 1992(Cook et al., , 1993. Conditional Proteolysis of Thermosensitive ubc9 Proteins-In the course of further experiments we observed that these mutations, which interfere with the in vivo activity of Ubc9p at high temperature, apparently caused a conditional reduction in the abundance of the mutated proteins. To directly compare levels of wild type and mutant Ubc9 proteins at various temperatures, isogenic strains that merely differ in the allele of UBC9 were grown at 22°C, 30°C, or shifted to 37°C for 3 h. Protein extracts were prepared and used for a Western blot analysis with a rabbit antiserum against Ubc9p. Loading of equal amounts of protein was confirmed by reprobing the same blot with antibodies to Cdc28p. As shown in Fig. 2, levels of the ts derivatives of Ubc9p were comparable with the amount of wild type protein at 22°C, significantly lower, however, at 30°C and almost undetectable after a shift to the nonpermissive temperature. In contrast, mRNA levels of the ubc9 alleles were not affected by elevated temperature (not shown) indicating that post-transcriptional events account for the observed loss of ts proteins. We therefore investigated in a next step if differential protein stability is the underlying cause. To assess protein turnover, the coding sequences of wild type and mutant ubc9 alleles were fused to the repressible GAL1 promoter. Strains carrying these constructs were grown in galactosecontaining medium to induce expression of these gene products. After promoter shutoff by glucose addition, samples were withdrawn from the cultures at time intervals and levels of Ubc9p, and its mutant derivatives were followed on Western blots. At 22°C all four proteins were essentially stable with half-lives of more than 1 h (Fig. 3). At 37°C, however, the ts proteins ubc9-1p, ubc9-2p, and ubc9-3p were highly unstable with estimated half-lives of 10 min or less. Wild type Ubc9 protein, on the other hand, remained completely stable at 37°C (Fig. 3). Thus, thermosensitive in vivo function correlates with temperature-dependent instability of ubc9 mutant proteins. To further investigate this interrelation, we analyzed the stability of mutant proteins whose defects are temperature-independent. The ubc9-4 protein totally lacks Ubc9 in vivo functions due to a replacement of the presumed active site cysteine residue with alanine. The ubc9-5 protein, in which two potential phosphorylation sites have been mutated (substitution of serine residues 122 and 127 by alanines), exhibits reduced physiological activity at any temperature. Turnover measurements show that ubc9-4 and ubc9-5 proteins are indistinguishable from wild type Ubc9p in their stability at elevated temperature (Fig. 3), indicating that instability does not simply result from replacement of a conserved residue or from functional deficiency but is specific for mutant proteins whose in vivo function is temperature-sensitive.
Proteolysis as a Cause for Conditional in Vivo Function-These data raise the possibility that the conditional lethality of strains carrying ubc9 ts alleles is a direct consequence of the temperature-dependent instability of ubc9 mutant proteins. To address this hypothesis a ubc9-1 ts strain was subjected to UV mutagenesis, and revertants were isolated that had regained the ability to grow at 35°C. Revertant strains were further characterized by backcrossing and tetrade dissection. A mutation that we named sub1-1 (for suppression of ubc9-1 ts) was A partial amino acid sequence alignment of Cdc34 and Ubc9 proteins is shown (top). This region contains the -P-X 3 -P-X 2 -P-P-motif that is conserved among members of the Ubc protein family. The cdc34-1 allele was isolated in a screen for yeast cell cycle mutants (Hartwell et al., 1974). Construction of the ubc9-1 allele has been described (Seufert et al., 1995). ubc9-2 and ubc9-3 alleles were isolated after mutagenesis of codons 68 and 69 by a polymerase chain reaction with degenerate primers. Products were used to replace a portion of the UBC9 gene on a 1.5-kilobase pair XbaI-SspI fragment in plasmid pSE362(ARS1, CEN4, HIS3). Following transformation of strain YWO90 (MATa, ubc9-⌬1::TRP1, pSE360(ARS1, CEN4, URA3)-UBC9), plasmid shuffling (Sikorski and Boeke, 1991), and replica plating, ts alleles were identified by their ability to support growth at 22°C but not at 37°C, and their DNA sequence was determined. The deletion of codon 69 in ubc9-3 is indicated by a dash. Numbers above the sequence correspond to the amino acid positions in the respective proteins. Temperaturesensitive viability of ubc9 mutant strains (bottom). For growth comparison wild type UBC9 strain YWO92 (MATa, ubc9-⌬1::TRP1, pSE362-UBC9) and mutant strains ubc9-1 YWO100 (MATa, ubc9-⌬1::TRP1, pSE362-ubc9-1) and ubc9-2 YWO98 (MATa, ubc9-⌬1::TRP1, pSE362-ubc9-2) were spotted in serial 10-fold dilutions on YPD plates and incubated for 3.5 days at 22°C or 2 days at 37°C.  1) and YWO96 (MATa, ubc9-⌬1::TRP1, pSE362-ubc9-3) carrying either wild type UBC9 (9) or mutant alleles ubc9-1 (-1), ubc9-2 (-2), or ubc9-3 (-3) were grown at 22°C, 30°C, or shifted from 22 to 37°C for 3 h (37°). Protein extracts were prepared and 20 g of protein/lane was loaded onto a 12% SDS-polyacrylamide gel. Ubc9p levels in these strains were compared on Western blots using a rabbit antiserum to Ubc9p (batch K4S3 at a dilution of 1:1000). To confirm equal loading, the blot was reprobed with a rabbit antiserum to Cdc28p (kindly provided by Dr. C. Mann; Gif-sur-Yvette, France) at a dilution of 1:1000. Immunoblots were developed using donkey anti-rabbit IgG horseradish peroxidase conjugates and an enhanced chemiluminescence system (Amersham).

FIG. 3. Conditional in vivo degradation of temperature-sensitive ubc9 mutant proteins.
In the ubc9-4 allele the codon for cysteine at position 93 (TGT) has been changed to GCT specifying alanine by PCR-based mutagenesis. In the ubc9-5 allele codons for serine at position 122 (TCT) and 127 (TCC) have been changed to GCT and GCC specifying alanine. Wild type UBC9 and the indicated mutant alleles were fused to the GAL1 promoter in plasmid pSE936 (ARS1, CEN4, URA3) and transformed into strain K699. For turnover measurements strains were grown in the presence of galactose at 22°C. Cultures remained at 22°C (left panel) or were shifted to 37°C (right panel), and samples were withdrawn at indicated times after glucose addition to repress the GAL1 promoter. Detection of Ubc9p was done as described in legend to Fig. 2. The chromosomal UBC9 copy in strain K699 gave rise to a constant low level of Ubc9p that is seen in ubc9-1p, ubc9-2p, and ubc9-3p expressing strains at the 40-and 60-min time points at 37°C (right panel, lanes 3 and 4).
found to be recessive and to belong to a single locus unlinked to UBC9. Stability measurements showed that the otherwise rapid turnover of ubc9-1 protein at high temperature was significantly slowed down in the thermoresistant sub1-1 strain (Fig. 4). Interestingly, a certain degree of stabilization was observed in all 10 isolates tested and its extent correlated with the variable increase in thermoresistance of distinct revertant strains. Moreover, raising ubc9-1 protein levels by expression from the strong GAL1 promoter also conferred elevated thermoresistance (Fig. 4). These results demonstrate that low in vivo levels of the mutant protein due to temperature-dependent degradation are the primary cause for the thermosensitive viability of the ubc9-1 strain. Recently, a portable domain for heat-inducible protein degradation has been developed. Its successful application confirms that active turnover may abolish the in vivo function of a protein (Dohmen et al., 1994).
Further Cases of Mutant Protein Degradation-To see if instability of ts mutant proteins is an exceptional phenomenon restricted to Ubc9p, we analyzed the in vivo half-lives of temperature-sensitive derivatives of the well characterized Cdc28 protein, the major cell cycle regulatory kinase in budding yeast (Reed, 1992;Nasmyth, 1993). To this end, wild type and three ts alleles of CDC28 were cloned and placed under control of the GAL1 promoter. After terminating expression Cdc28 protein levels at the nonpermissive temperature were followed by Western analysis. Wild type Cdc28p was apparently long lived. Its level remained constant during the 90-min time course (Fig.  5). In contrast, a pronounced decrease in abundance of cdc28-4, cdc28-13, and cdc28-1N proteins was observed, indicating that the life span of these temperature-sensitive derivatives is considerably reduced (Fig. 5). It thus appears that metabolic instability is a recurrent property of ts mutant proteins that may contribute to their conditional in vivo function. This view is consistent with recent studies of yeast Sec61p, a component of the protein translocation apparatus in the endoplasmic reticulum membrane (Rapoport, 1992). Mutations in Sec61p that render its in vivo function temperature-sensitive were found to reduce Sec61p levels at the nonpermissive condition. Overexpression of Sss1p, which is part of the Sec61p complex, restored normal Sec61p levels as well as thermoresistance of the mutant strain (Esnault et al., 1994). Moreover, thermosensitivity of sec61-2 cells was suppressed by inactivation of the Ubc6 ubiquitin-conjugating enzyme that is associated with the endoplasmic reticulum membrane and has been implicated in protein degradation (Sommer and Jentsch, 1993).
Degradation of a Thermolabile Ubc9p Derivative Depends on Proteasome Activity-We were interested in characterizing the pathway that mediates proteolysis of a temperature-sensitive ubc9 mutant protein. Vacuolar proteases in yeast function in turnover of membrane proteins and global protein breakdown during starvation or sporulation. Furthermore, catabolite inactivation of a cytoplasmic enzyme has been reported to involve specific uptake of the substrate and destruction in the vacuole (Chiang and Schekman, 1991). Proteasomes, in contrast, are thought to catalyze the majority of selective proteolysis events in the cytoplasm and nucleus. To define the contribution of these systems, heat-induced degradation of ubc9-1p was analyzed in strains deficient in activities of either the vacuole or the proteasome complex. First, an isogenic pair of strains was constructed by deleting the PRA1 gene in W9490. Since the PRA1-encoded product, proteinase A, is essential for vacuolar zymogen activation, a pra1⌬ strain is defective in multiple hydrolytic activities of this organelle (Ammerer et al., 1986;Woolford et al., 1986). In addition, strain c13-ABYS-86 mutated in two key endopeptidases, proteinase A and B, and two exopeptidases of the vacuole, carboxypeptidase Y and S (Achstetter et al., 1984), were compared with wild type strain YWO2. Cultures were grown at low temperature in the presence of galactose to induce ubc9-1 transcription that is driven by the GAL1 promoter in these cells. Following glucose addition for promoter shutoff and shift-up to 37°C, samples were harvested and used to determine ubc9-1p levels by Western analysis. Strains with defects in vacuolar proteases apparently retained the full capacity to degrade ubc9-1p, because no reduction in the turnover kinetic could be observed (Fig. 6). Whereas a mutation of PRA1 is sufficient to stabilize several membrane proteins (Volland et al., 1994;Kölling and Hollenberg, 1994;Lai et al., 1995), the vacuolar system is obviously not required FIG. 4. Thermosensitivity of a ubc9-1 strain determined by abundance of the mutant protein. Stabilization of ubc9-1 protein by the sub1-1 mutation that suppresses thermosensitive viability of a ubc9-1 strain (top). For half-life comparison ubc9-1 protein was expressed from the GAL1 promoter in the wild type strain W9445 (MATa, ubc9-⌬1::TRP1, pSE936-GAL1-ubc9-1, SUB1) or sub1 strain W9467 (MATa, ubc9-⌬1::TRP1, pSE936-GAL1-ubc9-1, sub1-1). Protein levels at indicated times after promoter shutoff were determined by Western analysis with antibodies against Ubc9p as described in the legend to Fig. 2. Elevated expression of ubc9-1p increases thermoresistance (bottom). The ubc9-1 strain W9432 (MATa, ubc9-⌬1::TRP1, pSE362-ubc9-1) and the GAL1ubc9-1 strain W9445 (see above) that express the mutant protein from the moderately active UBC9 and strong GAL1 promoter, respectively, were spotted in serial 5-fold dilutions on YPgalactose plates and incubated for 4 days at 22°C or 2.5 days at 34°C.
FIG. 5. Turnover of wild type Cdc28 protein and its temperature-sensitive mutant derivatives. CDC28 alleles were cloned from strains K699 (CDC28), BF421-3c (cdc28-4), BF387-3a (cdc28-13) and W9312 (cdc28-1N) through PCR amplification. The coding sequences were fused to the GAL1 promoter in pSE936, and plasmids were transformed into strain W9312. Cultures grown at 22°C in the presence of galactose were shifted to 37°C after glucose addition to repress the GAL1 promoter. Immediately and after 30, 60, and 90 min, samples were harvested and processed for immunoblot analysis. A rabbit antiserum to Cdc28p (see legend to Fig. 2), and enhanced chemiluminescence detection were used to determine Cdc28 protein levels.
for rapid proteolysis of the thermolabile ubc9-1 protein. Next, turnover rates of ubc9-1p were determined in pre1 pre2 and pre1 pre4 strains mutated in subunits of the 20 S proteasome (Heinemeyer et al., 1993;Hilt et al., 1993). Mutations of PRE genes have been found previously to interfere with degradation of various short-lived regulators and model substrates of the ubiquitin pathway (Yaglom et al., 1995;Chen and Hochstrasser, 1995;Seufert and Jentsch, 1992). Comparison with isogenic wild type strains revealed significant stabilization of ubc9-1p in pre1 pre2 and pre1 pre4 mutant cells (Fig. 6), suggesting a role of the 20 S proteasome in breakdown of the heat denatured ubc9-1 protein. The fact that the observed stabilization is incomplete does not necessarily indicate participation of a further proteolysis pathway. Rather, the measurable levels of proteasome activity retained in these pre mutant strains might account for the residual turnover of ubc9-1p.
The 20 S proteasome forms the catalytic core of the energydependent 26 S proteasome complex but does itself not require ATP for the degradation of peptides or unfolded proteins in vitro (Wenzel and Baumeister, 1995). To further define the proteolysis pathway of the thermolabile ubc9-1 protein, an involvement of cellular energy metabolism was analyzed. We first confirmed that ubc9-1p turnover could be followed after blocking total protein synthesis by cycloheximide, since glucose addition to terminate ubc9-1 expression from the GAL1 promoter would conflict with the use of energy inhibitors. As shown in Fig. 7, a comparable decrease of ubc9-1 protein was observed when cycloheximide was added instead of glucose, indicating that new protein synthesis is dispensable for ubc9-1p breakdown. In a subsequent experiment strain W9445 was treated with either cycloheximide alone or cycloheximide plus NaF and NaN 3 to inhibit in vivo ATP production. After shift to 37°C samples were withdrawn from the cultures at short intervals and taken to quantitate ubc9-1 protein levels by Western analysis. In control cells with active energy metabolism turnover of ubc9-1p was very rapid with an estimated half-life of 10 min (Fig. 7). Treatment of cells with the energy inhibitors NaF and NaN 3 , however, immediately destroyed the relevant proteolysis system and resulted in complete stabilization of ubc9-1p (Fig. 7). In vivo degradation of ubc9-1p is obviously energy-dependent, and the notion that a labile ubc9-1 protein devoid of its native structure at high temperature might be broken down directly by the 20 S proteasome in the absence of ATP is apparently not correct.
Proteolysis of ubc9-1p Requires Ubiquitination-Previous studies established a role of the ubiquitin system in the removal of abnormal proteins (Jentsch, 1992;Ciechanover, 1994). An obligatory step in this proteolysis pathway is the attachment of a multi-ubiquitin chain to substrates (Finley and Chau, 1991). To see if degradation of ubc9-1 protein involves ubiquitination, the effect of expressing a mutated form of ubiquitin was analyzed. Ub-K 48 R;G 76 A carries a lysine-to-arginine change at position 48 that destroys a major site for ubiquitin chain formation and, moreover, a replacement of the COOH-terminal glycine by alanine that interferes with its release from conjugates. As a consequence, Ub-K 48 R;G 76 A acts as a dominant inhibitor of ubiquitin-dependent proteolysis through irreversible conjugation to substrates and termination of chain growth (Hodgins et al., 1992;Finley et al., 1994). Turnover of ubc9-1 protein was compared in strains that express either wild type ubiquitin or Ub-K 48 R;G 76 A from the strong CUP1 promoter. After transfer of cultures to 37°C and glucose-mediated repression of ubc9-1p synthesis, samples were taken for immunoblot analysis. We found that Ub-K 48 R;G 76 A considerably decreased the heat-induced degradation of ubc9-1 protein and extended its otherwise short half-life of about 10 min approximately 3-fold (Fig. 8). Interestingly, stabilization was accompanied by the appearance of slower migrating forms of ubc9-1p. Since the attachment Ub-K 48 R;G 76 A to proteins is stable and impairs growth of a ubiquitin chain, these forms might be adducts of single molecules of Ub-K 48 R;G 76 A and ubc9-1p. Their heterogeneous mobility could be due to structural differences arising from the use of distinct lysine residues in ubc9-1p as acceptor sites. To define this material more precisely, its presence was  pre1-1, pre4-1)-[pSE936-GAL1-ubc9-1] carrying mutations in the Pre1 and Pre4 subunits of the 20 S proteasome (pre1 pre4). For measurements of ubc9-1p turnover, strains were grown in galactose medium at 22°C. Following glucose addition, cultures were shifted to 37°C, and samples were withdrawn at the indicated times. Protein extracts were prepared, and ubc9-1p levels were determined by Western analysis using the antiserum K4S3.
FIG. 7. Proteolysis of the thermolabile ubc9-1 protein depends on energy metabolism. Expression of ubc9-1p was terminated under various conditions. A culture of strain W9445 (MATa, ubc9-⌬1::TRP1, pSE936-GAL1-ubc9-1) grown in galactose medium at 22°C was split and either glucose (first panel) or cycloheximide (0.1 mg/ml) (second and third panels), or cycloheximide (0.1 mg/ml) plus NaF (10 mM) and NaN 3 (10 mM) was added prior to shift up to 37°C. Samples were harvested at the indicated times and processed for immunoblot analysis. For detection of ubc9-1p, the antiserum K4S3 was used at dilution of 1:1000. assayed in strains expressing combinations of wild type and mutated Ubc9 protein and ubiquitin. Following a brief heat treatment of cells, extracts were prepared and probed in a Western blot analysis with a Ubc9p-specific antiserum. Slower migrating forms were only detected in a strain that expresses both Ub-K 48 R;G 76 A and the thermolabile ubc9-1 protein (Fig.  8). This material was neither seen when Ub-K 48 R;G 76 A was coexpressed with wild type Ubc9p, which is a metabolically stable protein and therefore not expected to be a substrate for ubiquitination, nor when ubc9-1 protein was expressed together with wild type ubiquitin, which typically forms dynamic chains of multiple moieties. These results support our interpretation that the labile ubc9-1 protein is a substrate for ubiquitination, but further experiments such as sequential immunoprecipitation of epitope-tagged derivatives and mutagenesis of potential acceptor sites will be required for a definite proof. Nevertheless, the observed inhibition of ubc9-1p proteolysis by Ub-K 48 R;G 76 A highlights a critical role of the ubiquitin pathway in breakdown of this thermolabile mutant protein.
Transfer of ubiquitin to substrates is catalyzed by a family of Ubc enzymes. Ubc proteins generally differ in their substrate specificities and cellular functions, but certain family members exhibit overlapping activities (Jentsch, 1992). Two closely related enzymes, Ubc4 and Ubc5, have been found previously to play a major part in the removal of damaged proteins . Turnover of ubc9-1p was monitored in strains carrying single disruptions of the genes UBC1, UBC4, UBC5, or RAD6. Heat-induced breakdown of ubc9-1p was not detectably reduced in these mutant strains (data not shown).
To create a more stringent situation, the dominant negative effect of a catalytically inactive Ubc4 derivative was exploited. Ubc4-C 86 S whose active site cysteine residue has been replaced by serine appears to act by titrating out one or more essential co-factors of Ubc4 and other Ubc proteins, since expression of Ubc4-C 86 S is lethal in a ubc4⌬ strain, and viability is restored by increased gene dosage of UBC1, RAD6, or a novel UBC in addition to UBC4 and UBC5. 2 Degradation of ubc9-1 protein at an elevated temperature was analyzed in a ubc4⌬ strain induced to express Ubc4-C 86 S from the CUP1 promoter. A significant decrease in ubc9-1p turnover was observed in this mutant strain relative to a wild type control (Fig. 9). These results suggest that proteolysis of ubc9-1 protein requires ubiquitin conjugation activity but any one out of several related Ubc enzymes appears to be sufficient.

DISCUSSION
The yeast UBC9 gene is essential for cell viability and encodes a soluble protein of the nucleus that is highly stable. Specific amino acid replacements that reduce or even totally destroy Ubc9 activity did not affect the metabolic stability of the protein (Fig. 3). However, mutations in UBC9 selected to confer temperature-sensitive in vivo function were found to dramatically reduce the protein's half-life under restrictive but not at permissive conditions (Fig. 3). An extragenic suppressor named sub1-1 was isolated by its ability to increase the maximum growth temperature of a ubc9-1 strain and turned out to reduce the otherwise rapid turnover of ubc9-1 protein (Fig. 4). These results demonstrate that conditional proteolysis is the primary mechanism by which the temperature-dependent activity of ubc9-1 protein is controlled. Reduced metabolic stability relative to the wild type protein has been described for several other proteins with amino acid alterations (e.g. Mc-Cusker et al. (1991) and Esnault et al. (1994)) and is reported here for the products of three temperature-sensitive alleles of the CDC28 gene (Fig. 5), indicating that susceptibility of mutant proteins to degradation is a recurrent phenomenon. Despite this correlation, instability will only be a side effect in a number of cases and the main cause for dysfunction of the mutated protein may be a defect in another relevant property such as catalytic activity, subunit composition, or localization. ubc9-1 protein carries a single amino acid change that apparently disrupts its native conformation at elevated temperature and targets the protein for rapid destruction. Several observations support the conclusion that heat-induced breakdown of ubc9-1 protein is mediated by the ubiquitin-proteasome pathway. First, expression of a dominant negative ver-2 R. Albang and W. Seufert, unpublished data.  -ubc9-1) was transformed with plasmids Yep96 or pUB204 that direct CUP1 promoter controlled expression of, respectively, wild type ubiquitin (Ub) or a mutant ubiquitin in which the lysine residue at position 48 has been replaced by arginine and the glycine residue at position 76 by alanine (Ub-K 48 R;G 76 A). Cells were grown in galactose medium at 22°C, and expression of plasmid-borne ubiquitin was induced by CuSO 4 (0.1 mM) for 6 h. Samples were taken at the indicated times after glucose addition and shift to 37°C. Extracts were prepared and ubc9-1p levels were determined by immunoblot analysis with the antiserum K4S3. Lower part, presumed ubiquitin conjugates of ubc9-1 protein. In a Western blot analysis with the Ubc9-specific antiserum K4S3, protein extracts of the following strains were compared. Lefthand lane, K699-[pSE936-GAL1-UBC9, pUB204] expressing wild type Ubc9 protein (Ubc9p) and mutant ubiquitin (Ub-K 48 R;G 76 A). Middle lane, W284 (MATa, ubc9::HIS3, pSE936-GAL1-ubc9-1)-[pUB204] expressing the thermolabile ubc9-1 protein (ubc9-1p) and mutant ubiquitin (Ub-K 48 R;G 76 A). Right-hand lane, W284 (MATa, ubc9::HIS3, pSE936-GAL1-ubc9-1)-[Yep96] expressing the thermolabile ubc9-1 protein (ubc9-1p) and wild type ubiquitin (Ub). Cells were grown as described above and harvested after a 15-min shift to 37°C.
FIG. 9. A dominant mutation in the Ubc4 ubiquitin-conjugating enzyme affects degradation of ubc9-1 protein. Turnover of ubc9-1 protein was followed in strains K699-[pSE936-GAL1-ubc9-1] (wild type) and W9473 (MATa, ubc4-⌬1::HIS3, CUP1-ubc4-C 86 S:: TRP1)-[pSE936-GAL1-ubc9-1] expressing an active site mutant of Ubc4 protein from the CUP1 promoter (ubc4-dn). Cells were grown in galactose medium and treated with CuSO 4 (0.1 mM) for 6 h. Levels of ubc9-1 protein at the indicated times after repression of ubc9-1p synthesis by glucose were determined by Western immunoblot analysis with the antiserum K4S3. sion of ubiquitin, Ub-K 48 R;G 76 A, that is stably bound to substrates and interferes with formation of a multi-ubiquitin chain (Hodgins et al., 1992;Finley et al., 1994) caused a decrease of ubc9-1p turnover and resulted in accumulation of presumed conjugates with ubc9-1 protein (Fig. 8). Second, efficient proteolysis required Ubc4 or related ubiquitin-conjugating enzymes, since expression of a catalytically inactive Ubc4 protein that acts as a competitive inhibitor impaired ubc9-1p turnover (Fig. 9). Third, activity of the proteasome was critical for rapid breakdown but vacuolar proteases were dispensable: ubc9-1 protein was stabilized by mutations in subunits of the 20 S proteasome complex but remained short-lived in cells that lack functional proteases of the vacuole (Fig. 6). Finally, treatment of cells with energy poisons totally blocked degradation of ubc9-1 protein (Fig. 7). The complete and immediate response is consistent with an involvement of multiple ATP-dependent steps. Even though a contribution of an unidentified proteolysis pathway cannot be fully excluded, these results show that ubiquitination and breakdown by the proteasome is the major route for removal of this mutated form of the soluble and nuclear Ubc9 protein.
As it is true for most proteolytic substrates, the precise mechanisms how an aberrant protein is initially recognized and selected for degradation remains to be determined. Substrate specificity of the ubiquitin pathway is thought to be largely defined by the repertoire of conjugating enzymes in cooperation with so-called E3 proteins or ubiquitin ligases. Studies on oncogenic human papillomaviruses indicated that a cellular protein termed E6-AP, in complex with viral E6, directly binds to the tumor suppressor protein p53 and catalyzes ubiquitination of the substrate (Scheffner et al., 1993. Through the presence of a conserved carboxyl-terminal domain, a number of presumed ubiquitin ligases have been identified in various organisms . Ubiquitin appears to be transferred to E6-AP-related ligases specifically by homologs of yeast Ubc4 (Scheffner et al., 1994). Given the previously reported function of Ubc4 in the removal of damaged proteins ) and its role described here (Fig. 9), one might speculate that it could be one of the E6-APlike ligases that serves as a recognition factor for abnormal proteins. On the other hand, it is conceivable that certain heat shock proteins that are known to bind polypeptides in nonnative conformations (Gething and Sambrook, 1992) participate in the recognition process. In fact, a function of the Hsp70 homolog DnaK together with DnaJ in the degradation of a mutant form of alkaline phosphatase in E. coli has been reported (Sherman and Goldberg, 1992). The yeast S. cerevisiae possesses several members of the Hsp70-and DnaJ-related families of heat shock proteins that catalyze protein folding, complex assembly, and translocation across membranes (Cyr et al., 1994). But it is unclear so far, if these molecular chaperones are also involved in proteolysis and interact in some way with components of the ubiquitin system. The large extend and conditional character of the observed stability switch make ubc9-1 protein a suitable model substrate to study these open questions.