Subcellular Localization, Stoichiometry, and Protein Levels of 26 S Proteasome Subunits in Yeast*

The 26 S proteasome of eukaryotes is responsible for the degradation of proteins targeted for proteolysis by the ubiquitin system. Yeast has been an important model organism for understanding eukaryotic proteasome structure and function. Toward a quantitative characterization of the proteasome, we have determined the localization, cellular levels, and stoichiometry of proteasome subunits. The subcellular localization of two ATPase components of the regulatory complex of the proteasome, Sug2/Rpt4 and Sug1/Rpt6, and a subunit of the 20 S proteasome, Pre1, were determined by immunofluorescence. In contrast to findings in multicellular organisms, these proteins are localized almost exclusively to the nucleus throughout the cell cycle. We have also determined the cellular abundance and stoichiometry of these proteasome subunits. Sug1/Rpt6, Sug2/Rpt4, and Pre1 are present in roughly equal stoichiometry with an abundance of 15,000-30,000 molecules/cell, corresponding to a concentration of 13–26 μm in the nucleus. Also, in contrast to mammalian cells, we find no evidence of a p27-containing “modulator” of the proteasome in yeast. This information will be useful in comparing and contrasting the yeast and mammalian proteasomes and should contribute to a mechanistic understanding of how this complex functions.

The proteasome is a complex macromolecular assembly important for regulated protein degradation in eukaryotic organisms ranging in complexity from yeast to mammals and is also found in archaebacteria and some eubacteria (for reviews, see Refs. 1 and 2). Substrates of the proteasome include, but are not limited to, transcription factors, cyclins, proteins with destabilizing N termini, and misfolded or damaged proteins (2)(3)(4)(5). Most are targeted for degradation by components of the ubiquitin system through the covalent addition of a multiubiquitin chain (6). The core 20 S particle of the eukaryotic proteasome contains 14 different subunits arranged into four stacked, seven-member rings, which form a hollow cylinder (7). The two inner rings are made up of ␤-subunits, some of which have inward-facing protease active sites. The two outer rings are made up of ␣-subunits, the N termini of which cap the ends of the cylinder. In isolation, the 20 S proteasome is able to de-grade only peptides and unfolded proteins. In eukaryotes, a 700-kDa regulatory complex called 19 S, PA700, or regulatory particle can bind in an ATP-dependent fashion to one or both ends of the 20 S cylinder to form complexes commonly referred to as 26 S proteasomes (8 -10). These complexes are able to bind and degrade multiubiquitinated proteins in an ATP-dependent manner. Due to steric constraints on the entry of substrates into the interior of the 20 S proteasome (11,12), it has been widely speculated that the six related ATPase subunits of 19 S regulator use the energy derived from ATP hydrolysis to unfold proteins and perhaps translocate them into the core of the 20 S cylinder (12,13). In mammalian cells, additional regulators of the 20 S proteasome have been identified (14). PA28 stimulates the peptidase activity of the 20 S proteasome and may be important in production of peptides presented by major histocompatibility complex class I molecules (15,16). A second complex, modulator, augments activation of the 20 S proteasome by 19 S regulator, apparently by stimulating 19 S-20 S complex formation (17,18). There are no homologs of PA28 subunits in yeast and, as yet, no evidence for proteasome regulators other than the 19 S in this organism.
The yeast Saccharomyces cerevisiae has been an important model organism for the study of proteasome function. Many subunits of the proteasome have been identified in yeast through mutant strains that have defects in protein degradation (e.g. see Refs. 19 -21). The availability of these strains has also provided a simple means to determine whether a protein of interest is a substrate of the proteasome. Now that the yeast genome project is complete (22), all members of conserved gene families involved in the ubiquitin and proteasome system in a eukaryotic organism are known. Gene disruption experiments in yeast have been used to determine which subunits of the proteasome perform essential functions. In one notable case, a ubiquitin binding subunit of the 19 S regulatory complex identified in Arabidopsis and mammals was found not to be essential in yeast, calling into question its relative importance in vivo (23). In addition to the value of yeast in genetic studies, the yeast 20 S proteasome is the only eukaryotic proteasome for which a crystal structure is available. This structure is the prototype for eukaryotic proteasomes and differs in important ways from the archaebacterial proteasome (7). Recently, the 19 S regulator from yeast (regulatory particle) was purified, and all or nearly all of its subunits were identified (24), most of which correspond to subunits of the mammalian 19 S regulator. Despite the importance of experiments performed in yeast to the current understanding of the proteasome, important parameters for understanding proteasome function in this organism have not been determined. For instance, the subcellular localization of the yeast proteasome is not well established, and the cellular levels and stoichiometry of proteasome subunits are not known.
In a wide variety of multicellular organisms, the proteasome is found in both the cytoplasm and the nucleus (reviewed in Refs. 25 and 26). Subcellular fractionation experiments have generally found that the proteasome is more abundant in the cytosol than the nucleoplasm of animal cells, although the proportion varies with cell type and growth conditions (27)(28)(29). Immunofluorescence experiments in mammalian cells have found that the proportion of proteasome in the nucleus and cytoplasm varies in a complex pattern throughout the cell cycle (30 -32). There have been a number of attempts to determine the subcellular localization of proteasome subunits in yeast (33)(34)(35)(36)(37). The studies have used a variety of methodologies and have come to conflicting conclusions, sometimes even when the same methodology was used. We have addressed this issue using a conservative approach to localize subunits of the 20 S and 19 S subcomplexes of the proteasome and the putative modulator of the proteasome. We find that, in contrast to animal cells, yeast proteasome components are found almost exclusively in the nucleus throughout the cell cycle. We have also determined for the first time the absolute levels and relative stoichiometry of 19 S and 20 S proteasome components and have investigated the possibility of a modulator complex in yeast.

EXPERIMENTAL PROCEDURES
Strains-All of the Saccharomyces cerevisiae strains used in this study are derived from W303a (MATa ade2-1 ura3-1 his3-11,15 trp1-1 leu2-3, 112 can1-100). Strains in which the genomic SUG1/RPT6 1 and SUG2/RPT4 loci express epitope-tagged proteins were generated by a two-step gene replacement strategy (details available on request). The final products are strains that are congenic to W303a and differ only in the presence of an epitope tag at the otherwise native SUG1 or SUG2 locus. This was confirmed in each case by PCR 2 amplification and sequencing of the entire locus, extending past the boundaries of the replacement fragment at both the 5Ј-and 3Ј-ends. Sug1/Rpt6 and Sug2/ Rpt4 are tagged at the N terminus with a single copy of the T7 epitope (Novagen) in Sc507 (T7-SUG1) (38) and Sc668 (T7-SUG2), respectively, and with three tandem copies of the T7 epitope in Sc669 (3XT7-SUG1) and Sc670 (3XT7-SUG2), which is identical to the previously described Sc562 (38). Strains in which the genomic PRE1 locus expresses epitopetagged proteins were produced by a "hit and run" strategy (details available on request). Briefly, a derivative of the URA3 integration plasmid pRS306 was produced that contained the 3Ј-half of the coding region, and 3Ј-untranslated region, of a PRE1 allele encoding an epitope tag at the C terminus. The plasmid was linearized by digestion within the PRE1 coding region and transformed into W303a. Uracil prototrophs expressing epitope-tagged Pre1 protein were counterselected on 5-fluorouracil. Strains were obtained that were able to grow on 5-fluorouracil, were uracil auxotrophs, and still expressed epitopetagged Pre1. The PRE1 locus was then PCR-amplified and sequenced to confirm that the pRS306 sequence had been lost to recombination and the native PRE1 locus was reconstituted except for addition of the epitope tag sequence. Pre1 is tagged at the C terminus with the T7 epitope in Sc702 (PRE1-T7) and with three tandem copies of the HA epitope (39) in Sc703 (PRE1-3XHA). In Sc691 (3XT7-SUG1 PRE1-3XHA), pRS306 vector sequence, including the URA3 marker, remains downstream of the tagged Pre1 locus. In all others, the native state of the genome has been restored except for the presence of the epitope tag(s). Sc582 is a derivative of Sc556 (which is identical to Sc668) in which the open reading frame YIL007C is replaced with URA3 (see below).
Antibodies-Anti-T7 monoclonal antibodies were from Novagen, rabbit anti-HA polyclonal antibodies were from Santa Cruz Biotechnology, and anti-Dpm1 antibodies were from Molecular Probes. Mouse anti-Sug2/Rpt4 antibodies have been described (38). Rabbit anti-Sug1/Rpt6 antibodies were raised against the N-terminal 156 amino acids of Sug1/ Rpt6. Rabbit anti-20 S proteasome antibodies were a gift from Keiji Tanaka. Rabbit anti-yeast cyclophilin antibodies were a gift from Kathryn Sykes. Immunopurified anti-Gal6 antibodies have been described (40). Alkaline phosphatase conjugated goat anti-rabbit and goat antimouse IgG (Bio-Rad) were used for detection in chemifluorescent Western blots. 35 S-Labeled goat anti-mouse and goat anti-rabbit antibodies (NEN Life Science Products) were used for detection in Fig. 4. For immunofluorescence, AffiniPure fluorescein-conjugated goat antimouse and lissamine rhodamine conjugated goat anti-rabbit IgGs (Jackson ImmunoResearch Laboratories) were used as secondary reagents. Antibodies were used at dilutions of 1:20,000 to 1:5,000 for Western blots and 1:100 to 1:50 for immunofluorescence.
Quantitative Western Blots-Yeasts were grown in yeast extract peptone medium (41) with 2% dextrose as the carbon source to an absorbance (600 nM) between 0.8 and 1.0. For Western blots, 1 ml of culture was harvested by centrifugation, suspended in 200 l of 2ϫ SDS loading buffer (60 mM Tris-HCl, pH 6.8, 10% glycerol, 2% SDS, 5% ␤-mercaptoethanol), frozen on dry ice, and stored at Ϫ80°C until use. Sodium azide (0.02%) was added to other aliquots of culture, which were stored on ice until counted with a hemocytometer. Aliquots containing a known number of cells in SDS loading buffer were loaded onto 10% Tricine SDS-polyacrylamide gels after heating at 95°C for 5 min. After electrophoresis, the entire gels, including the wells and stacking gels, were transferred to Immobilon-P (polyvinylidene fluoride, 0.45-M pores) membranes (Millipore) in a Genie blotting apparatus (Idea Scientific). Prestained molecular mass standards (Sigma) ranging from 33.5 to 125 kDa were quantitatively transferred. No immunoreactivity was detected at the well, at the interface between the stacking and separating gels, or on a second membrane placed behind the first. After transfer, membranes were dried and stored at 4°C until use. Membranes were rehydrated by shaking in methanol and then in water before incubation with antibodies. Chemifluorescent blots were developed with Vistra enhanced chemifluorescence substrate (Amersham Pharmacia Biotech) and scanned on the Storm system (Molecular Dynamics). Blots using radioactive secondary antibodies were exposed to phosphorimager screens and then read on the Storm system. Bands were quantitated with the ImageQuant program (Molecular Dynamics). Blots of mammalian cell extracts were prepared and developed the same way except that washed cells were suspended in extract buffer (25 mM Tris-HCl, pH 7.5, 2% SDS, 5 mM EDTA, 2ϫ complete protease inhibitors (Roche Molecular Biochemicals)) and boiled for 5 min. After centrifugation, the protein concentration of the supernatant was determined using BCA reagent (Pierce). Equivalent amounts of protein were loaded in each lane.
Recombinant Proteins-SUG1/RPT6 and SUG2/RPT4 were cloned into pQE60-His 6 (42) as NcoI-BamHI fragments to allow expression of the proteins with a six-histidine tag at the N terminus. Proteins were produced in Escherichia coli as inclusion bodies that were approximately 50 -75% pure, as estimated from Coomassie-stained SDS-polyacrylamide gels. The proteins were further purified by preparative SDS-PAGE using the Bio-Rad Prep Cell. 15 mg of crude protein was loaded onto a gel with a 10% acrylamide separating zone, and the apparatus was run at 15 W constant power and an elution buffer flow rate of 0.8 ml/min. Fractions of 3 ml each were evaluated by Coomassie staining, and fractions containing only the Sug protein were pooled. At this point, the proteins appeared to be homogenous by Coomassie staining. The proteins were precipitated from the pooled fractions with 10% trichloroacetic acid, washed with cold acetone and resuspended in denaturing buffer (8 M urea, 50 mM Tris-HCl, pH 7, 2 mM dithiothreitol). The solubilized proteins were quantitated by Bradford assay using BSA dissolved in denaturing buffer as standards. Aliquots containing 1.6 g of protein by Bradford were diluted into SDS loading buffer, run out by SDS-PAGE, and transferred as described above. The membranes were stained with Ponceau S, and the bands were excised. Three samples were sent for amino acid analysis (Howard Hughes Protein Sequencing Core, University of Texas-Southwestern Medical Center). BSA samples of known concentration were subjected to the same treatment. Amino acid analysis (corrected for losses of known BSA samples) gave values of 41 and 56% of the values obtained from the Bradford assay for His 6 -Sug1/Rpt6 and His 6 -Sug2/Rpt4, respectively. This could be due to losses during SDS-PAGE or transfer, or to inaccuracy of the Bradford assay done in the presence of urea, so calculations of protein levels using these standards are reported using both sets of values.
Immunofluorescence-Yeasts were grown as described for quantitative Western blots. Cells were fixed by adding buffered formaldehyde directly to growing cultures. One ml of culture was mixed with 50 l of 0.5 M potassium phosphate, pH 6.5, and 100 l of 37% formaldehyde and fixed at room temperature for 2 h. The cells were washed two times with 1 ml of 0.1 M potassium phosphate, pH 7.4, and then resuspended in 1 ml of phosphate-buffered sorbitol (1.2 M sorbitol, 0.1 M potassium phosphate, pH 6.5) and stored at 4°C for up to 12 h before spheroplasting. For spheroplasting, cells were collected by centrifugation at 800 ϫ g for 2 min and resuspended in 200 l of phosphate-buffered sorbitol. One l of 14.3 M ␤-mercaptoethanol and 6 l of 10 mg/ml zymolyase 20T (Seikagako Corp.) were added, and the cell suspensions were shaken at 200 revolutions/min for 30 min. Cells were collected by centrifugation, washed two times, and resuspended in 200 l of phosphate-buffered sorbitol. Fifteen l of cell suspension was settled onto wells of toxoplasmosis slides (Bellco Biotechnology) coated with poly-L-lysine. After 5 min, the liquid was aspirated, and the slides were incubated in methanol at Ϫ20°C for 6 min and in acetone at Ϫ20°C for 30 s. After drying, wells were blocked for 5 min with 1% bovine serum albumin in phosphate-buffered saline (PBS-BSA). This and all subsequent steps were performed at room temperature. PBS-BSA was also used for antibody dilution. Primary antibody incubations were for 2 h followed by 15 washes of each well with 20 l of PBS-BSA. Secondary antibody incubations were for 1 h followed by the same wash procedure. Each well was then washed 10 times with PBS. DNA was stained by incubation with 1 g/ml 4,6-diamidino-2-phenylindole in water for 5 min, followed by 10 washes with PBS. Slides were then air-dried, covered with Vectashield (Vector Laboratories) to minimize photobleaching, and sealed under coverslips secured with nail polish. Images were captured using a Leitz Laborlux-S epifluorescent photomicroscope and either an Optronincs VI470 CCD array camera and Signal Analytics IP Lab Spectrum capture and analysis software or a Wild Leitz MPS 52 35-mm camera back. In every case, congenic wild-type cells were processed in parallel with cells in which proteasome proteins were epitope-tagged.
YIL007C Disruption-The Saccharomyces Genome Data base (http:// genome-www.stanford.edu/Saccharomyces/) was searched using the tBLASTn basic local alignment search tool (43) with the predicted human p27 protein as query sequence. Only one predicted protein with significant homology, the product of open reading frame YIL007C, was identified. A disruption construct for YIL007C was generated by PCR amplifying a 264-base pair fragment of the 3Ј-untranslated region with oligonucleotides 5Ј-AAGAGCTACAAGCTTCCATGGGTACTTAAAAA-AACC-3Ј and 5Ј-ATCAACGGTACCGATTGGGAGCTCGAAG-3Ј. This product was digested with HindIII and KpnI and cloned into Hind-III/KpnI-cut pRS306 to generate pSJR144. A 369-base pair fragment of the 5Ј untranslated region was amplified with oligonucleotides 5Ј-CT-ACTATTACGGTTGGCAGTCTGTAGTAGAG-3Ј and 5Ј-CATTGTAGT-TCGTTCT TATTCTAAAGCTTGCTCC-3Ј, cut with HindIII and EcoRI, and cloned into HindIII/EcoRI-cut pSJR144 to generate pSJR147. pSJR147 was digested with HindIII and ligated with the HindIII fragment from YEp24 containing the URA3 gene to yield pSJR157. The EcoRI/KpnI fragment of pSJR157 (5Ј-and 3Ј-untranslated sequence of YIL007C flanking the URA3 marker) was transformed into Sc344 (W303a/␣) and Sc556 (W303a T7-SUG2). Approximately equal numbers of Ura ϩ colonies were obtained in each transformation. A Ura ϩ transformant of Sc556, Sc582, was further characterized by diagnostic PCR and fluorescent automated sequencing of the amplified locus, which verified that YIL007C had been deleted and replaced with URA3.

Subcellular Localization of 26 S Proteasome Subunits-In
order to determine the subcellular localization of the 20 S proteasome, we chose to epitope tag a ␤-subunit, PRE1 (44). A His 6 tag at the C terminus of this subunit has previously been used to purify active 26 S proteasome from yeast extracts (24,45). We produced strains in which the native Pre1 locus was reconstituted except for insertion of the sequence coding for a single T7 epitope tag or three tandem copies of the HA epitope at the C terminus of the protein. These tagged genes served as the only source of Pre1 in the modified strains. For immunolocalization of the 19 S regulator, we examined two ATPase subunits, Sug1/Rpt6 1 (46) and Sug2/Rpt4 (38). Both are components of the 19 S in yeast and mammals (38,45,47). Bovine Sug2/Rpt4 is also reported to be a component of another complex, the modulator, which stimulates the peptidase activity of the 20 S proteasome in a 19 S-dependent manner (17). Sug1/ Rpt6 and Sug2/Rpt4 were each tagged at the N terminus with either a single copy or three tandem copies of the T7 epitope tag. As before, the native Sug1/Rpt6 and Sug2/Rpt4 loci were reconstituted except for the sequence encoding the epitope tags, and these loci were the only source of Sug1/Rpt6 or Sug2/ Rpt4 in the cell. Western blotting using anti-epitope monoclonal antibodies (Fig. 1) demonstrated that there is no immunoreactivity in the parental strain, whereas a single protein of the expected size was detected in the tagged strains. Probing with polyclonal antibodies raised against Sug1/Rpt6, Sug2/ Rpt4, or multiple subunits of the 20 S proteasome showed that epitope tagging does not change the levels of these proteins, except in the case of triple-tagged Sug2/Rpt4 (Fig. 1A). As expected, triple tagging of Sug1/Rpt6 increases reactivity of the protein to anti-T7 antibody dramatically. In the case of Sug2/ Rpt4, triple tagging leads to some instability of the protein, so that a less dramatic increase in immunoreactivity of the fulllength protein to anti-T7 antibody was seen (Fig. 1A). Yeast cyclophilin served as a loading control. The tagged proteins are all essential for growth (38,44,46). Because the tagged alleles represented the only source of the proteins in the cell and were produced at the wild-type levels (Fig. 1), and these strains had no noticeable phenotype, we conclude that the tagged proteins must be correctly incorporated into proteasomes. Indeed, we have previously shown that epitope-tagged Sug1/Rpt6 is completely incorporated into proteasomes (38).
Immunofluorescence experiments using the anti-T7 monoclonal antibody show that single-tagged Sug2/Rpt4 co-localizes with nuclear DNA with little, if any, cytoplasmic staining detectable above the background seen in the untagged parental strain ( Fig. 2A). Immunofluorescence experiments with highly specific anti-Sug2/Rpt4 polyclonal antibodies (38) in the paren-FIG. 1. Sug1/Rpt6, Sug2/Rpt4, and Pre1 can be specifically detected using epitope tags. Yeast extracts from equal numbers of cells were separated by SDS-PAGE and assayed by Western blot with the antibodies indicated. Western blots using polyclonal antibodies raised against the untagged proteins were included to demonstrate that the abundance of Sug1/Rpt6, Sug2/Rpt4, and Pre1 in the strains used for all subsequent experiments is not altered by epitope tagging. Sequence encoding the tags was inserted at the relevant chromosomal loci so that these strains differ from the congenic parental strain only by the presence of the tag. Cyclophilin serves as a loading control. A, extracts from strains in which Sug1/Rpt6 and Sug2/Rpt4 were untagged (parental), single (T7), or triple (3XT7) epitope-tagged at their N termini. B, extracts from strains in which the 20 S proteasome subunit Pre1 was untagged (parental), single-tagged (T7), or triple-tagged (3XHA) at its C terminus.
FIG. 2. Immunolocalization of Sug1/Rpt6, Sug2/Rpt4, and Pre1 demonstrates that these 26 S proteasome subunits are predominantly nuclear. Nuclear and mitochondrial DNA was visualized by staining with 4,6-diamidino-2-phenylindole (DAPI). Epitope-tagged proteasomal proteins were localized by indirect immunofluorescence using mouse monoclonal anti-T7 epitope antibodies or rabbit anti-HA epitope antibodies. The untagged (parental) strains were treated with the same concentrations of primary and secondary antibodies as the tagged strains. Exposure times were the same for photographs of tagged and untagged strains. A, Sug1/Rpt6 and Sug2/Rpt4 were tagged at the N terminus with the T7 epitope. The 20 S ␤-subunit Pre1 was T7 tagged at the C terminus. B, Sug2 protein detected with mouse polyclonal anti-Sug2 antiserum. C, Sug1/Rpt6 was tagged at the N terminus with three tandem copies of the T7 epitope (3XT7). D, Pre1 was tagged at the C terminus with three tandem copies of the HA epitope (3XHA). E, co-localization of 3XT7-Sug1/Rpt6 and Pre1-3XHA in a double-tagged strain. F, nuclear envelopeendoplasmic reticulum protein Dpm1 detected with anti-Dpm1 monoclonal antibody. G, cytoplasmic Gal6 protein detected with immunopurified anti-Gal6 antibodies. tal, untagged strain gave very similar results (Fig. 2B), demonstrating that the T7 epitope tag does not cause a mislocalization of the protein. Staining in the Sug1/Rpt6 and Pre1 single-tagged strains was much weaker than in the tagged Sug2/Rpt4 strain, although the faint staining over background in these strains also co-localized with nuclear DNA (Fig. 2A). In order to more clearly visualize Sug1/Rpt6 and Pre1, immunofluorescence experiments were performed on the Sug1/Rpt6 (Fig. 2C) and Pre1 (Fig. 2D) triple-tagged strains. These experiments confirmed that these proteins are localized primarily to the nucleus with little, if any, specific staining above background detectable in the cytoplasm. The triple-HA tag has previously been used to visualize other proteins in the cytoplasm of yeast, indicating that the epitope tag does not itself cause nuclear localization (48). Co-localization of triple-tagged Sug1/Rpt6 and Pre1 was demonstrated by dual color immunofluorescence (Fig. 3E). Sug1/Rpt6, Sug2/Rpt4, and Pre1 were localized to the nucleus as shown in nearly all cells of unsynchronized cultures, including unbudded, small-budded, and large-budded cells and mitotic cells (those in which the nuclear DNA is found in an hourglass shape at the neck of a large bud), suggesting that these proteins are present in the nucleus throughout the cell cycle. Based on the intensity of staining and the relative volumes of the nucleus and cytoplasm, we estimate that less than 20% of the proteasome subunits studied here may be present in the cytoplasm. To confirm that we could distinguish localization to the nucleus from a nuclear envelopendoplasmic reticulum pattern, we used a monoclonal antibody against Dpm1, a nuclear envelope-endoplasmic reticulum integral membrane protein. As shown in Fig. 2F, the nuclear envelope was visualized as a bright ring around the nuclear DNA, quite distinct from the nuclear staining seen with proteasome subunits. As a control for the ability to detect a cytoplasmic protein we used antibodies against the Gal6 protein (40) (Fig.  2G). We conclude that, in contrast to mammalian cells, components of both the 19 S regulator and the 20 S particle of the 26 S proteasome are localized largely, if not exclusively, to the nucleus of yeast throughout the cell cycle.
Stoichiometry and Absolute Cellular Levels of 26 S Proteasome Subunits-Evidence presented by Glickman et al. (24) strongly suggests that each proteasomal ATPase, or Rpt protein, is present in one copy within partially purified 19 S regulatory complex. Therefore, we were surprised by the strik-ingly higher Western and immunofluorescence signals for Sug2/Rpt4 compared with Sug1/Rpt6 when both had a single epitope tag and were detected with the same anti-epitope antibody (Figs. 1A and 2A). Such an excess would be consistent with the presence of Sug2/Rpt4 in a second complex, perhaps a yeast equivalent to the modulator discovered in bovine red blood cell extracts. This prompted us to further investigate the stoichiometry of these 19 S subunits, as well as their stoichiometry in relation to the 20 S proteasome component Pre1. Using the single-tagged strains described above we performed quantitative Western blots. A representative experiment, performed in duplicate, is shown in Fig. 3. Because each protein was tagged with the same epitope tag and the proteins were run on the same gel and probed together, we would expect the differences in immunoreactivity to be proportional to the relative stoichiometry of these proteins in the cell. Consistent with the immunofluoresence data ( Fig. 2A), Sug1/Rpt6 is apparently present in the cells at approximately the same level as Pre1, whereas Sug2/Rpt4 seems to be present in a dramatic molar excess over both Sug1/Rpt6 and Pre1. Probing of the same membranes with antibodies specific for Sug1/Rpt6 and Sug2/ Rpt4 demonstrated that tagging of Sug1/Rpt6 and Sug2/Rpt4 does not lead to a change in their abundance. Although we do not have antibodies specific for Pre1, epitope tagging of Pre1 did not change the abundance of several 20 S proteasome subunits recognized by a polyclonal anti-20 S antibody (Fig. 3). However, we noted that the extent of the apparent stoichiometric excess of Sug2/Rpt4 varied with the treatment of the membrane after transfer. Membranes were routinely dried for overnight storage, then rehydrated before probing. When this protocol was followed (as in Fig. 3), the apparent Sug2/Rpt4 excess was large (approximately 10-fold). However, if the membrane was placed into buffer immediately after transfer was complete, without drying, the apparent stoichiometric excess of Sug2/Rpt4 over Sug1/Rpt6 and Pre1 was smaller (2-3-fold). In contrast, the apparent relative abundance of Sug1/Rpt6 and Pre1 did not change significantly. Side by side tests of the two protocols confirmed these phenomena (data not shown). This strongly suggested that most of the apparent stoichiometric excess of Sug2/Rpt4 is due to differences in availability of the epitope tag for antibody binding, and is not a true reflection of proteasome subunit stoichiometry.
To resolve the uncertainty associated with stoichiometry estimates based on the tagging strategy, and to determine the absolute levels of proteasome subunits within the cell, we performed quantitative Western blots using purified recombinant His 6 -Sug1/Rpt6 and His 6 -Sug2/Rpt4 protein as standards. Extracts were prepared as before, and aliquots containing 10 7 cell equivalents were loaded onto SDS-polyacrylamide gels. Different amounts of His 6 -Sug1/Rpt6 or His 6 -Sug2/Rpt4 (as determined by amino acid analysis) were also loaded. Proteins were detected with anti-Sug1/Rpt6 or anti-Sug2/Rpt4 polyclonal antibodies and appropriate 35 S-labeled secondary antibodies. Quantitation was performed using a phosphorimager. The experiment shown in Fig. 4 was performed independently six times from three different cell cultures. The Sug1/Rpt6 and Sug2/Rpt4 content in 10 7 cells was 10 and 14.4 ng, respectively, corresponding to 13,000 Ϯ 4600 and 17,000 Ϯ 8500 (mean Ϯ S.D.) molecules/cell, respectively. Bradford assays with BSA standards gave somewhat different protein concentrations for the His 6 -Sug1/Rpt6 and His 6 -Sug2/Rpt4 standards than amino acid analysis. If we calculate the number of molecules per cell using the protein concentrations determined with Bradford assays, the values are 32,000 and 31,000 molecules/cell for Sug1/Rpt6 and Sug2/Rpt4, respectively. Therefore, regardless of how the standards were quantitated, Sug1/Rpt6 and Sug2/ FIG. 3. Comparison of single epitope-tagged strains suggests that the stoichiometry of 26 S proteasome subunits is Sug2/Rpt4 > > Sug1/Rpt6 ‫؍‬ Pre1. Sug1/Rpt6, Sug2/Rpt4, and the 20 S proteasome ␤-subunit Pre1 were each tagged with a single copy of the T7 epitope in strains congenic to the parental strain as indicated. Yeast extracts from equal numbers of cells were separated by SDS-PAGE and assayed by Western blot with the indicated antibodies. For each strain, extracts from two independent cultures were assayed. Cyclophilin serves as a loading control. Note that intensities of bands can only be compared within a panel and not between blots as they are performed with different antibodies.
Rpt4 seem to be present at a stoichiometry of approximately 1:1 in the cell. Therefore the apparent stoichiometric excess of Sug2/Rpt4 observed using the tagging approach in both Western blots and immunofluorescence on formaldehyde fixed cells was almost certainly an artifact of epitope availability to antibody. We conclude that Sug1/Rpt6 and Sug2/Rpt4 are present in equivalent stoichiometry in yeast cells. Furthermore, we can conclude that the absolute levels of both proteins in the cell are between 15,000 and 30,000 molecules/cell. Because the relative reactivities of Sug1/Rpt6 and Pre1 were roughly equivalent independent of assay conditions, it seems likely that Pre1 is also present at close to the same absolute level as Sug1/Rpt6 and Sug2/Rpt4.

Deletion of Yeast p27 Does Not Affect Sug2/Rpt4 Levels or Subcellular Localization-In bovine red blood cells, Sug2/Rpt4
(p42) is a component of a second complex, distinct from 19 S, called modulator. This complex contains another ATPase also present in the 19 S regulator, Tbp1, and a novel protein called p27 (17). Most of the Sug2/Rpt4 present in bovine red blood cells is part of the modulator and not of the 19 S regulator (17). Although our cellular stoichiometry results for Sug1/Rpt6 and Sug2/Rpt4 suggested that this is not the case in yeast, we looked for evidence of a yeast analog of the modulator. The cDNA encoding the human homolog of bovine p27 has been cloned and its sequence was generously provided by George DeMartino and Keiji Tanaka. A search of the Yeast Genome Data base revealed only one hypothetical open reading frame, YIL007C, which encodes a protein with significant identity to human p27 (30% identical). An alignment of the hypothetical yeast protein with human p27 is presented in Fig. 5. We refer to it as yeast p27 for consistency, although it encodes a protein of approximately 25 kDa. In order to test the hypothesis that Sug2/Rpt4 is present in a yeast analog to the modulator, we generated a disruption construct for YIL007C that replaces the entire open reading frame with the URA3 marker. This construct was transformed into a diploid strain congenic to the parental strain as well as the haploid T7-SUG2/RPT4 strain. Comparable numbers of Ura ϩ transformants were recovered from transformations of both the diploid and haploid strains. We confirmed that YIL007C had been deleted and replaced by URA3 in the T7-SUG2/RPT4 strain by diagnostic PCR and subsequent sequencing of the PCR products. Because this strain is haploid, yeast p27 is not essential for vegetative growth. Subsequent to the completion of these experiments, Watanabe et al. (49) also reported that p27 is not essential.
The levels of Sug2/Rpt4 protein were compared between the yeast p27-deleted strain and the congenic wild-type strain as described above (Fig. 6A). There is no significant difference in the Sug2/Rpt4 levels between these strains (Sug2/ Rpt4(⌬YIL007C)/Sug2/Rpt4 ϭ 0.8 Ϯ 0.2). Furthermore, the subcellular localization of Sug2/Rpt4 in the ⌬YIL007C strain is identical to that seen in the parental strain, i.e. almost exclusively nuclear (Fig. 6B). We conclude that little, if any, Sug2/ Rpt4 is present in a complex that requires p27 for stability or appropriate subcellular localization.
p27 and the modulator were discovered in red blood cells. Erythrocytes are unusual in many respects, prompting us to investigate the relative amounts of Trip1 (mammalian Sug1/ Rpt6), 20 S components, and p27 in other mammalian cell types (Fig. 7). Although p27 is expressed in other mammalian cells types, the ratio of p27 to the 19 S component Trip1 (human Sug1/Rpt6) and subunits of the mammalian 20 S protea- FIG. 6. Deletion of YIL007C does not affect Sug2/Rpt4 levels or subcellular localization. A, in the two T7-tagged Sug2/Rpt4 strains, open reading frame YIL007C (encoding yeast p27) is either intact (T7-Sug2/Rpt4) or has been replaced with the URA3 marker (T7-Sug2/ Rpt4 ⌬YIL007C). The congenic, untagged parental strain and a strain in which Pre1 is T7-tagged were included as controls. Yeast extracts from equal numbers of cells were separated by SDS-PAGE and assayed by Western blot with the indicated antibodies. For each strain, extracts from two independent cultures were assayed. Cyclophilin served as a loading control. This experiment was repeated two times, and the intensities of the bands were quantitated: Sug2/Rpt4(⌬YIL007C) /Sug2/ Rpt4 ϭ 0.8 Ϯ 0.2 (ratio Ϯ population S.D., n ϭ 4). Note that intensities of bands can only be compared within a panel and not between them, as blots were performed with different antibodies. B, immunolocalization of Sug2/Rpt4 in a strain in which YIL007C is deleted. T7-epitope-tagged Sug2/Rpt4 was detected with anti-T7 antibodies as in Fig. 2. some is much higher in red blood cells than in three tissue culture cell lines assayed. We speculate that high expression of p27 may be an adaptation to conditions of slow, or no, cell growth. Therefore, it may not be important for the viability of rapidly growing yeast. It remains possible that yeast p27 is important under unusual growth conditions, in sporulation or in germination.
Sug2/Rpt4 Co-fractionates Entirely with Sug1/Rpt6 from Yeast Extracts-Although p27 is not essential for the stability or proper localization of Sug2/Rpt4, other complexes containing Sug2/Rpt4 could be present. Multiple Sug2/Rpt4-containing complexes have been reported in extracts from the hawkmoth Manduca sexta (50). To test this idea, we compared the ratio of Sug2/Rpt4 to Sug1/Rpt6 in crude extract and semipurified preparations of 26 S proteasome. A difference would imply the presence of Sug2/Rpt4 not associated with the 26 S proteasome. In a previous study, we showed that Sug2/Rpt4 co-fractionated by gel filtration with components of the 26 S proteasome. Fractions containing the 26 S proteasome were subsequently refractionated after dissociation of the 19 S regulatory complex from the core 20 S proteasome (38). These separations were reexamined by Western blot with a mixture of anti-Sug2/Rpt4 and anti-Sug1/Rpt6 antibodies to determine the relative stoichiometry of these proteins in crude extract, semipurified 26 S proteasome, and 19 S regulator dissociated from the 20 S particle. For comparison, extract from cells boiled directly in SDS-loading buffer was loaded along with samples from the native fractionation (Fig. 8). Although the relative intensity of Sug1/ Rpt6 and Sug2/Rpt4 immunoreactivities detected with the antibody mixture is a function of both the relative abundance of the two proteins and the affinity of the two antibodies, a change in the ratio with purification would be informative. If Sug2/ Rpt4 protein in a putative second complex was lost during extract preparation and clarification or to proteolysis, we would expect there to be relatively less Sug2/Rpt4 immunoreactivity in the extract than from boiled cells. In fact, this ratio is the same in extract and boiled cells (Fig. 8). If a putative second Sug2/Rpt4 complex dissociated from the 26 S proteasome or the 19 S regulator during the fractionation and was subsequently degraded, we would expect to see the ratio of Sug2/Rpt4 to Sug1/Rpt6 to decrease in these fractions. No such change was seen (Fig. 8), implying that all of the Sug2/Rpt4 is stably associated with the 19 S regulator. It remained possible that the S400 column used for the fractionation lacked sufficient resolution to separate a putative modulator-like complex from the 19 S. Therefore, the 19 S-containing fractions were refractionated on a Superdex 200 HR 10/20 column. In our hands, this column allows near baseline resolution of thyroglobulin (669 kDa) and apoferritin (443 kDa). Both Sug1/Rpt6 and Sug2/ Rpt4 eluted in a single peak from this column centered very near the elution volume of thyroglobulin, as expected for a 700-kDa complex (data not shown). A putative modulator-like complex would run at 300 kDa, and no Sug2/Rpt4 immunoreactivity was seen in this size range. Taken together with the p27 deletion data, this suggests that there is no modulator equivalent or a second Sug2/Rpt4-containing complex at detectable levels in yeast under normal growth conditions.

DISCUSSION
There are many reports that address the subcellular localization of proteasomes in higher eukaryotes. Although there is disagreement on the details, it is fairly clear that the proteasomal proteins are present in both the cytoplasm and the nucleus of animal cells in a pattern that may vary with the cell cycle (25, 26, 30 -32). However, the localization of these proteins in yeast is much less clear. Fusions of Sug1/Rpt6 to ␤-galactosidase (46) or Son1/Rpn4 to invertase (36) were overexpressed and found to be concentrated in the nucleus. However, when a fusion of Nin1/Rpn12 with ␤-galactosidase was overexpressed in yeast, it was found as hazy staining and small, bright dots in the cytoplasm (34). Indirect immunofluorescence on a wild-type strain gave the same result, but the antibody used for these experiments recognized one other protein as strongly as Nin1/Rpn12 (34). More recently, a fusion of Sug2/Rpt4 with green fluorescent protein (GFP) was found concentrated in the nucleus throughout the cell cycle, as was overexpressed 19 S component Sen3/Rpn2 (37). Most recently, GFP fusions with the 19 S ATPase Cim5/Rpt1 and the 20 S ␣-subunit Pre6 were observed at nuclear periphery by Enenkel et al. (33). These authors also found that an epitope-tagged Cim5/Rpt1 co-localized with the nuclear envelope-endoplasmic reticulum protein Kar2/Bip and that staining with an anti-20 S antibody co-localized with nuclear pore complex proteins. They concluded that proteasomes are located primarily in the nuclear envelope-endoplasmic reticulum network of yeast (33). Rinaldi et al. (35) fused Mpr1/Rpn11 to GFP and found it was distributed throughout the cytoplasm in a nonuniform fashion, suggesting that it was associated with cytoplasmic structures.
We report here that components of both the yeast 20 S proteasome and the 19 S regulator are localized primarily to the nucleus of yeast throughout the cell cycle. We believe that our protocol is likely to reveal the in vivo localization of the proteasome for the following reasons: 1) in our approach the system was minimally perturbed. We used small epitope tags instead of fusions to large protein tags, such as GFP, invertase, or ␤-galactosidase. The epitope-tagged proteins were expressed at wild-type levels from their native genomic loci and were verified by quantitative Western blotting to be present at normal levels (Fig. 1). We have previously shown that epitopetagged Sug1/Rpt6 is completely incorporated into 26 S proteasomes under these conditions (38). 2) Our use of the small FIG. 7. p27 is relatively more abundant in red blood cells than in tissue culture cell lines. Whole cell extracts from C2C12 myoblasts, XS106 dendritic cells, NS46 fibroblasts, and fresh human red blood cells (RBC) were separated by SDS-PAGE. Western blots were probed with antibodies against human p27, the human homolog of Sug1/Rpt6 (TRIP1), human 20 S proteasome, and human cyclophilin as indicated.
FIG. 8. The ratio of Sug2/Rpt4 to Sug1/Rpt6 in 26 S proteasome and 19 S regulator is the same as in whole cells. Extracts prepared by boiling yeast directly in SDS loading buffer (Boiled Cells) or native extract and fractions containing partially purified 26 S proteasome or 19 S were separated by SDS-PAGE. Western blots were probed with a mixture of anti-Sug1/Rpt6 and anti-Sug2/Rpt4 antibodies. epitope tagging strategy allowed us to control for the possibility of nonspecific cross-reaction with both primary and secondary antibodies by assaying an untagged parental strain in parallel (Fig. 2, A, C, and D). 3) We assayed multiple subunits of the proteasome, including a 20 S component and two 19 S components. We chose one of the 19 S components (Sug2/Rpt4) because it was suspected of being in another complex (the modulator), and it was possible that the putative modulator might not entirely co-localize with the 26 S proteasome. 4) Results using a very specific antibody for Sug2/Rpt4 (38) in a completely wild-type strain agree with results using the epitope tags (Fig. 2B), demonstrating that these tags do not inappropriately localize the proteasome. 5) A nuclear localization for the proteasome in yeast is consistent with the differences in physiology between yeast and mammalian cells. In a recent study, a fusion of GFP to a component of the 20 S proteasome was used to examine the dynamics of proteasome distribution within living mammalian cells (32). The proteasome was found to distribute between the cytoplasm and the nucleus upon breakdown of the nuclear envelope during mitosis. After reformation of the nuclear envelope, the proteasome moved unidirectionally into the nucleus until the next mitosis. In yeast, the nuclear envelope does not break down during mitosis. Therefore, if the mechanism for proteasome import into the nucleus is conserved in yeast, then the proteasome should be found primarily in the nucleus as we observed.
The most thorough and careful previous attempt to localize the proteasome is that of Enenkel et al. (33). Therefore, it is of concern that our conclusions are at odds with theirs. However, a recent study in fission yeast (Schizosaccharomyces pombe) provides a logical explanation for the discrepancy. Wilkinson et al. (51) used both a GFP tagging strategy and indirect immunofluorescence to localize two 19 S regulatory complex subunits and the 20 S proteasome to the nuclear periphery of fission yeast during much of the cell cycle. Like Enenkel et al. (33), they found that a 19 S proteasome subunit co-localized with nuclear pore complex proteins. However, they used immunogold staining and electron microscopy to determine that the proteasome was present on the inner side of the nuclear membrane and not within or on the outer surface of the nuclear envelope. Therefore, it is possible that the same is true of the proteasomal proteins localized by Enenkel et al. (33). We clearly show that the proteasome distributes throughout the nucleus and not at the periphery (compare Fig. 2, A-E, and Fig.  2F). This discrepancy may be due to the GFP tag and/or the fixation conditions employed by Enenkel et al. (33).
Recent work has suggested that the proteasome of yeast may have a different role in cellular physiology than its mammalian counterpart. Inhibitors of the proteasome, such as the peptide aldehydes MG132 and the natural product lactacystin, prevent degradation of both long and short lived proteins in mammalian cells (52,53). However, in yeast, these same agents inhibit mainly the degradation of short lived proteins. Phenylmethylsulfonyl fluoride, which inhibits vacuolar serine proteases but not the proteasome, inhibits the degradation of yeast proteins with long half-lives (54). Many short lived regulatory proteins, including transcription factors and cyclins, reside in the nucleus, and some have been shown to be proteasome substrates in yeast (55). Differences in the localization and types of substrates degraded by the yeast and mammalian proteasome may be due, at least in part, to the role of the mammalian proteasome in the immune system. Proper immune surveillance demands that peptides generated by the mammalian proteasome for presentation by major histocompatibility complex class I (56, 57) must be representative of both long and short lived proteins from the cytosol as well as the nucleus. Clearly, the yeast proteasome does not have to satisfy this requirement. Thus, the localization of the yeast proteasome primarily to the nucleus may be commensurate with a physiological role that differs from its mammalian counterpart.
It is not clear from our immunofluorescence data whether there is any proteasome present in the cytoplasm, as the signal from this compartment was near background. Based on our sensitivity of detection and the larger volume of the cytoplasm compared with the nucleus, we estimate that, at most, 20% of the proteasome could have been present in the cytoplasm and escaped detection in our experiments. Although it is possible that most natural substrates of the proteasome in yeast are nuclear, proteins thought to be localized to the cytoplasm are degraded by the ubiquitin-proteasome system (58 -60). It is possible that a small fraction of the proteasome is present in the cytoplasm and is sufficient to degrade cytoplasmic proteasome substrates under most circumstances. This is consistent with the finding that yeast cells continue to grow when 70 -80% of proteasome activity is inhibited (54), suggesting that the capacity of the proteasome may be in great excess over what is required under most conditions. More speculatively, multiubiquitinated proteins in the cytoplasm may be transported to the nucleus for degradation by the proteasome.
In bovine red blood cells, there is evidence for a second Sug2/Rpt4-containing complex, the modulator, which contains bovine Sug2/Rpt4 (p42), Tbp1, and a novel protein, p27 (17). Work by Glickman et al. (24) suggests that each proteasomal ATPase is present in one copy in the 19 S regulator, but these experiments could have missed a second population of Sug2/ Rpt4 not associated with the 19 S regulator. We show here that deletion of the yeast p27 homolog does not decrease viability of yeast, in agreement with the findings of Watanabe et al. (49), or significantly alter the steady state levels of Sug2/Rpt4 protein.
We found by quantitative Western blot that Sug1/Rpt6 and Sug2/Rpt4 are in fact present in the cell at very similar levels. Finally, in contrast to bovine red blood cells, in which more Sug2/Rpt4 is present in the modulator than in the 19 S regulator (17), we found no evidence for Sug2/Rpt4 anyplace but in the 19 S regulator. If there is a modulator in yeast, it may be present only under unusual growth conditions or during meiosis, sporulation, or germination. The tagging experiments also suggested that Sug1/Rpt6 is approximately equimolar with a ␤-subunit of the 20 S proteasome, Pre1. There are two Pre1 polypeptides in each 20 S proteasome (7). Therefore, if we assume that each 19 S contains one Sug1/Rpt6 polypeptide (24), our findings indicate that there are two 19 S regulators for each 20 S proteasome present in the yeast cell. Thus, each 20 S proteasome could be capped at both ends with 19 S regulators. This is in contrast to the situation in mammalian red blood cells, in which the 20 S proteasome is in 3-4-fold excess over the 19 S regulator, 3 perhaps due to the presence of other regulators of the 20 S proteasome not found in yeast.
Quantitative Western blots allowed us to determine that 15,000 -30,000 molecules of Sug1/Rpt6 and Sug2/Rpt4 are present in each yeast cell. The average volume of the G 1 and S phase yeast nucleus is 1.95 m 3 (61). Assuming that Sug1/Rpt6 and Sug2/Rpt4 are exclusively localized to the nucleus, they are present at a concentration of 13-26 M. Because much of the nucleus may be occluded, the actual concentration in the nucleoplasm may be much higher. As it is likely that each 19 S complex contains one copy of each protein (24), this probably represents the concentration of the 19 S complex as well. In light of our findings, consider the relationship between the 26 S proteasome and its substrate, multiubiquitinated proteins.
Although to our knowledge the concentration ubiquitin-protein conjugates in yeast has not been determined, in several mammalian cell types it can be estimated as 20 -60 M (expressed as the concentration of ubiquitin monomer) (62). If we assume that efficient association of protein substrates with the proteasome requires modification by a chain of at least four ubiquitins (63), then the relevant substrate would be present at less than or equal to 5-15 M. If these estimates are valid for yeast, we can make two observations: first, if ubiquitinated proteins are distributed evenly throughout the cell, the proteasome may be present in excess of the steady state concentration of its substrate in the nucleus, whereas the substrate would be in a large excess in the cytoplasm. Even if all of the substrate were concentrated in the nucleus, substrate would only be in 20-fold excess of enzyme, assuming that the nucleus is 5% of the volume of a yeast cell. This is consistent with the observation that yeast continue to grow even when a significant portion of proteasome activity is inhibited (54). Second, the K 0.5 for inhibition of protein degradation by the 26 S proteasome with a chain of four ubiquitins has been estimated as 28 M (64). Thus, even if substrate is distributed evenly throughout the cell, both substrate and enzyme are near K D , suggesting that this measured affinity is sufficient to ensure rapid degradation of ubiquitin conjugates in vivo.
In summary, we have found that the subcellular localization of the 26 S proteasome subunits is different in yeast than that reported for mammalian cells. This may be related to different roles played by proteasomes in yeast and mammalian cells. It may also be a consequence of a difference in cellular organization in yeast compared with mammals, i.e. a closed mitosis. In addition, we determined the cellular stoichiometry and absolute protein levels of several 26 S proteasome subunits. Finally, we found that although S. cerevisiae has homologs of each protein found in the mammalian modulator, the unique modulator component p27 is nonessential, and we found no evidence of a modulator in yeast under normal growth conditions. Understanding the subcellular localization of the proteasome and the makeup of its regulatory complexes is a prerequisite to a working knowledge of its role in yeast cellular physiology. In addition, it is important to understand differences between the yeast and mammalian systems so that lessons learned in the tractable model organism can be appropriately applied to understanding the more complex proteasome system in mammals.