Functional Analysis of Rpn6p, a Lid Component of the 26 S Proteasome, Using Temperature-sensitive rpn6 Mutants of the Yeast Saccharomyces cerevisiae*

Rpn6p is a component of the lid of the 26 S proteasome. We isolated and analyzed two temperature-sensitive rpn6 mutants in the yeast, Saccharomyces cerevisiae. Both mutants showed defects in protein degradation in vivo. However, the affinity-purified 26 S proteasome of the rpn6 mutants grown at the permissive temperature degraded polyubiquitinated Sic1p efficiently, even at a higher temperature. Interestingly, their enzyme activity was even higher at a higher temperature, indicating that once made mutant proteasomes are stable and have little defect in the proteolytic function. These results suggest that the deficiency in protein degradation observed in vivo is rather due to a defect in the assembly of a holoenzyme at the restrictive temperature. Indeed, both rpn6 mutants grown at the restrictive temperature were defective in assembling the 26 S proteasome. A striking feature of the rpn6 mutants at the restrictive temperature was that there appeared a protein complex composed of only four of the nine lid components, Rpn5p, Rpn8p, Rpn9p, and Rpn11p. Altogether, we conclude that Rpn6p is essential for the integrity/assembly of the lid in the sense that it is necessary for the incorporation of Rpn3p, Rpn7p, Rpn12p, and Sem1p (Rpn15p) into the lid, thereby playing an essential role in the proper function of the 26 S proteasome.

The 26 S proteasome is a multicatalytic protein complex that is highly conserved among eukaryotic organisms. The ubiquitin-proteasome pathway is essential for eliminating damaged or misfolded proteins and for degrading short lived regulatory proteins involved in cell cycle regulation, DNA repair, signal transduction, apoptosis, and metabolic regulation (1,2). Substrate proteins must be polyubiquitinated by E1 1 /E2/E3 enzymes prior to recognition and degradation by the 26 S proteasome (reviewed in Ref. 1). Ubiquitins (Ub) are linked to each other through isopeptidyl bonds at their 48th lysine (Lys 48 ), and this polyubiquitin chain is recognized by the proteasome, and the substrate protein is degraded in an ATP-dependent manner.
The 26 S proteasome consists of two complexes, the 20 S core particle (CP) and the 19 S regulatory particle (RP). It has the RP attached either on one end (RP 1 CP) or on both ends of the CP (RP 2 CP). The CP is composed of a stack of two ␣ rings and two ␤ rings in the order of ␣, ␤, ␤, ␣, each of which consists of seven related components (3). The proteolytic active sites are sequestered in the inner cavity of the CP (4). Crystallographic study revealed that the gate of the CP allows the entry of substrate proteins only upon binding with the RP (5). Upon high salt treatment, the RP can be divided further into two subunits, the base and the lid. The base consists of six AAA-ATPase subunits (Rpt1p-Rpt6p) and two non-ATPase subunits (Rpn1p and Rpn2p), whereas the lid is made of eight non-ATPase subunits (Rpn3p, Rpn5p-Rpn9p, Rpn11p, and Rpn12p) (6). Additional proteasome-associated proteins have also been identified and characterized (7,8). Recently, Sem1p (Rpn15p), a homologue of human Dss1p, was shown to be the ninth component of the lid (9), and Rpn13p was shown to be an additional base component (7,9). A non-ATPase subunit, Rpn10p, was found to associate with the base or the lid, depending on the purification method (6,10) and has been suggested to exist in the interface between the base and the lid (11).
To further understand the biochemical processes conducted by the 19 S RP, it is important to know what role the individual components of the RP are playing. The ATPases in the base are required for the unfolding of substrate proteins before their translocation into the CP cavity (12), and some other components have been studied in detail. Of the RP components, Rpn10p and Rpt5p have been suggested to be responsible for the recognition of substrate through the attached poly-Ub chain (13)(14)(15)(16)(17)(18)(19)(20). Rpn11p metalloisopeptidase, a lid component, was shown to play an essential role in cleaving off the Ub chain from polyubiquitinated substrate proteins (21)(22)(23). Lid components like Rpn3p, Rpn5p, Rpn6p, Rpn9p, and Rpn12p were also further characterized (14, 16, 24 -27). However, whereas functional characterization of the RP has considerably progressed, the mechanism of its assembly remains to be elucidated. To date, interactions among RP components were studied using yeast two-hybrid analysis (11,28), and Rpn10p was shown to strengthen the binding between the lid and the base (6). We have shown in a previous report that Rpn7p is indispensable for the incorporation of Rpn3p and Rpn12p into the lid (29). Obviously, more information concerning the individual components of the lid would be necessary for further understanding of the assembling process of the 26 S proteasome.
In this study, we focused on Rpn6p and carried out a detailed analysis of temperature-sensitive rpn6 mutants. Until now, the function of Rpn6p has only been analyzed by depletion mutants in yeast (26) and in Drosophila (30). These works suggested that Rpn6p was necessary for the proper assembly and activity of the 26 S holoenzyme (26) and that it plays a role in development (30). The data presented below show that Rpn6p is indeed essential for the structural integrity of the 26 S holoenzyme, especially for the proper construction of the lid, by incorporating four other components, Rpn3p, Rpn7p, Rpn12p, and Sem1p (Rpn15p), into the lid.

EXPERIMENTAL PROCEDURES
Strains, Media, and Genetic Methods-Yeast strains used in this work are listed in Table I and plasmids used for cloning and subcloning  various genes and their fragments are listed in Table II. Cells were cultured in omission medium prepared by removing appropriate nutrient(s) from synthetic complete (SC) medium, rich medium (YPDAU) (31), or SR-U in which 2% glucose of SC was replaced with 2% raffinose and uracil was omitted. Escherichia coli strain DH5␣ (supE44 ⌬lacU169 [80lacZ ⌬M15] hsdR17 recA1 endA1 gyrA96 thi-1 relA1) was used for construction and propagation of plasmids. Yeast transformations were performed as described previously (32).
Isolation of Temperature-sensitive Mutants-The adenine residue of ATG corresponding to the putative translation initiation codon was defined as the first (ϩ1) nucleotide. Temperature-sensitive rpn6 mutants were screened as follows (33). First, a BamHI integration site was introduced into the RPN6 ORF. PCR was conducted with two pairs of primers, Rpn6-Mut1 (SalI) 5Ј-GGAAGTCGACGAAATTAGAAGAGGC-CAGG-3Ј and Rpn6-Mut2 (BamHI) 5Ј-GGAAGGATCCAAGGAGTCCG-GCAC-3Ј, and Rpn6-Mut3 (BamHI) 5Ј-GGAAGGATCCCAGATCTTTG-TGTGCG-3Ј and Rpn6-Mut4 (NotI) 5Ј-GGAAGCGGCCGCACCTCTAC-AAAATCC-3Ј, using the genomic DNA of W303-1B as a template. The resulting PCR products were digested with appropriate restriction enzymes and inserted between the SalI and NotI sites of YIp vector pRS306 by three-fragment ligation to yield pNS102 harboring 5Ј-truncated RPN6 (nucleotides ϩ18 to ϩ659) with a BamHI site created near the 5Ј-terminal region. PCR mutagenesis was performed according to the method described previously (34) using a pair of primers, Rpn6-Mut1 and Rpn6-Mut4, and pNS102 as a template. The mutagenized PCR products were cloned into pRS306 (URA3) to yield a mutant library. The library DNA was digested at the BamHI site, introduced into W303-1B cells to replace the RPN6 gene with the mutated rpn6 gene. Ura ϩ transformants exhibiting temperature sensitivity of growth were collected as mutant candidates. The mutant candidates were crossed with the wild-type W303-1A, and the linkage between a putative mutant gene causing temperature-sensitive growth and the auxotrophic marker (URA3) was confirmed by tetrad analysis.
DNA Manipulation-DNA engineering and agarose gel electrophoresis were performed as described elsewhere (35). Nucleotide sequences were determined by the dideoxy chain terminating method (36) using an automated DNA sequencer, ABI 310 (Applied Biosystems).
␤-Galactosidase Assay-Cells carrying plasmids encoding Ub-Ala-␤galactosidase, Ub-Pro-␤-galactosidase, or Ub-Arg-␤-galactosidase were grown in 10 ml of SR-U medium to midlogarithmic phase at 30°C and then supplemented with a final concentration of 2% galactose after 2 h of incubation and incubated for a further 5 h at either 25 or 37°C. Cells were harvested and subjected to a ␤-galactosidase assay using o-nitrophenyl-␤-D-galactoside as a substrate following the protocol described elsewhere (42). Protein quantification was performed as described above.
Gel Filtration-Yeast total lysates were prepared as described above, and 2 mg of total proteins were run on a Superose 6 column (1.0 ϫ 30 cm) (Amersham Biosciences) as described previously (21). Fractions were collected every 2 min (500 l/tube). Peptidase activity of relevant fractions without the addition of SDS were measured as described previously (29). Peptidase activity was measured also with final concentration of 0.02% SDS in the reaction mixture to activate the 20 S CP. Proteasomal components in the fractions were detected by Western blotting using the indicated antibodies.
Affinity Purification of Proteasomes-Wild-type and mutant strains whose Rpn11p was tagged with 3ϫFLAG were constructed as described previously (43). A single colony from each strain was inoculated in 250 ml of YPDAU liquid medium and cultured overnight to yield 2-3 mg (wet weight) of cells. Cells were suspended in 5 ml of Buffer AЈ (50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 20% glycerol, 4 mM ATP, 10 mM MgCl 2 , 1ϫ ATP regeneration system) and broken using the multibead shocker (MB501; Yasui Kikai). After centrifugation for 20 min at 15,000 rpm at 4°C (CR22G; Hitachi), lysates were filtrated through a 0.45-m filter (Millipore Corp.) and mixed with 150 l of anti-FLAG antibody-immobilized M2 beads. The mixture was rotated for 90 min at 4°C. The beads were then washed extensively with Buffer A, and the proteasomes were 2 E. Isono, N. Kamata, and A. Toh-e, manuscript in preparation. eluted with 150 l of 0.2 mg/ml 3ϫFLAG peptides (Sigma). The method for affinity purification of the wild-type lid was described previously (29). Coomassie Brilliant Blue-stained gels were photographed using an image analyzer (LAS-3000; Fujifilm), and protein bands were quantified with the Image Gauge (Fujifilm) software.
In Vitro Degradation Assay-An in vitro degradation assay was performed using ubiquitinated Sic1 PY p as a substrate. Details of this method will be published elsewhere. 3 In short, a PY motif, a target sequence of the E3 Rsp5p, was introduced into the N-terminal region of Sic1p (designated Sic1 PY p), and the SIC1 PY gene was cloned into a pET vector (Novagen). The resulting plasmid was transformed to E. coli, and Sic1 PY p was expressed and purified as described previously (21). Purified Sic1 PY p was then incubated with Uba1p, Ubc4p, and Rsp5p, allowing polyubiquitination of Sic1 PY p. A 300 nM concentration of thus produced polyubiquitinated Sic1 PY p was mixed with 100 nM purified proteasomes and incubated in the existence of 1 mM ATP and 5 mM MgCl 2 for the indicated time at the indicated temperature. The reaction was terminated by the addition of the SDS-PAGE-loading buffer, and the mixture was subjected to SDS-PAGE, followed by Western blotting with anti-T7 antibody to detect the degree of degradation.
Construction of an rpn6 Mutant Containing either F137L or L377P Amino Acid Substitution-PCRs were performed using primer pairs (a) Rpn6-Mut1 (SalI) and EK98 FLr, (b) EK97 FLf and Rpn6-Mut4 (NotI), (c) Rpn6-Mut1 (SalI) and EK100 LPr, and (d) EK99 LPf and Rpn6-Mut4 (NotI) using pNS102 as a template. PCR products by primer pairs a and b were mixed and used as a template for the second PCR with primers Rpn6-Mut1 (SalI) and Rpn6-Mut4 (NotI). The resulting products were cloned into the SalI-NotI site of pRS306 to yield pEK183FL. PCR products by primer pairs c and d were processed similarly to yield pEK183LP. These plasmids were digested at the BamHI site and introduced into W303-1B cells to yield strains YEK122 and YEK123, respectively. The T1394C and T2130C mutations in the integrants were verified by DNA sequencing.
In Vivo Cross-linking-Cells were grown to reach midlog phase and washed twice with 1ϫ phosphate-buffered saline (0.1 M phosphate, 0.15 M NaCl, pH 7.4) and then suspended in 5 ml of 1ϫ phosphate-buffered saline. 20 mM stock solution of DSP (Pierce) dissolved in dry Me 2 SO was added to a final concentration of 2 mM and left 30 min at a room temperature. The cross-linking reaction was quenched by adding 1 M Tris-HCl, pH 7.5, to a final concentration of 150 mM and incubating for 15 min at a room temperature. Cells were then washed twice with Buffer A and subjected to affinity purification of proteasomes. Samples were boiled in 5% 2-mercaptoethanol containing SDS-loading buffer to break the disulfide bond in DSP prior to SDS-PAGE.

RESULTS
Two Temperature-sensitive Mutants of RPN6 Were Isolated-To investigate the function of Rpn6p, we carried out a PCR mutagenesis-based screening of temperature-sensitive rpn6 mutants. A mutation library consisting of ϳ6,000 independent plasmid DNAs was created and introduced into W303-1B cells by selecting Ura ϩ transformants as described under "Experimental Procedures," and the URA3 transformants showing temperature-sensitive growth were saved. The candidates were crossed with wild-type haploid cells (W303-1A), and the resulting diploids were dissected. Bona fide rpn6 mutants should show complete co-segregation between Ura ϩ and the temperature sensitivity.
We obtained two mutants, which were named rpn6-1 and rpn6-2. Both mutants had a growth rate comparable with wildtype cells at the permissive temperature (25°C) but stopped growing after 6 -8 h of incubation at the restrictive temperature (37°C) (data not shown). The temperature-sensitive phenotype was suppressed by introducing a single copy plasmid carrying the RPN6 gene but not with the vector alone ( Fig. 1A), showing that the growth defect at a higher temperature was indeed due to the mutations introduced into the RPN6 gene. Both rpn6 mutations were recessive (data not shown). Next, we determined the mutation sites of the mutants by nucleotide sequencing. The rpn6-1 mutation was a single nucleotide substitution in the stop codon, leading to a 20-amino acid residue extension at the C terminus. rpn6-2 was found to bear two amino acid replacements, one of which resides in the N-terminal half (F132L) and the other in the C-terminal PCI domain (L377P) (Fig. 1B).
rpn6-1 and rpn6-2 Are Defective in Degrading Proteasomal Substrates in Vivo-To test the effect of the rpn6-1 and rpn6-2 mutations upon proteasomal function in vivo, we performed two experiments. First, we examined the accumulation of polyubiquitinated proteins under the restrictive condition. Overnight cultures of wild-type and rpn6 mutant cells were diluted in YPDAU liquid medium, incubated to reach early log phase, and then divided into two portions. One-half was cultured at 25°C, and the other was cultured at 37°C for a further 7 h. Total proteins were extracted from both culture and subjected to Western blotting to detect polyubiquitinated proteins ( Fig. 2A). A larger amount of polyubiquitinated proteins were 3 Y. Saeki, E. Isono, and A. Toh-e, manuscript in preparation.
accumulated in rpn6 mutant cells than in wild-type cells, a feature shared by mutants defective in the proteasome function in yeast (15,16,(25)(26)(27)29), Trypanosoma (44), and Arabidopsis (45). The accumulation was more prominent in samples prepared from rpn6 mutant cells cultured at 37°C, suggesting that proteolysis by the ubiquitin-proteasome pathway is not properly functioning in the rpn6 mutant cells, especially under the restrictive condition. Next, we examined the degradation of model substrates in the mutant cells. Wild-type and mutant cells were transformed with plasmid carrying the gene encoding one of the model substrates Ub-Ala-␤-galactosidase, Ub-Pro-␤-galactosidase (a ubiquitin fusion degradation pathway substrate), or Ub-Arg-␤-galactosidase (an N-end rule substrate) (46). Representative transformants were picked up and precultured overnight in SR-U liquid medium. The preculture was then diluted to A 600 ϭ 0.2 in SR-U medium, and the refreshed culture was incubated to reach A 600 ϭ 0.4, divided into two portions, and cultured at either the permissive or the restrictive temperature. After 2 h of incubation, a final concentration of 2% galactose was added to allow the production of the Ub-fused substrates. Cells were incubated for further 5 h and then collected and subjected to ␤-galactosidase assay. The amounts of Ub-Pro-␤-galactosidase and Arg-␤-galactosidase were shown as a ratio to the amount of Ala-␤-galactosidase that is a highly stable substrate. Ub-Pro-␤-galactosidase and Arg-␤-galactosidase, both of which were rapidly degraded in wild-type cells, were significantly stabilized in the rpn6-1 and rpn6-2 cells grown at the restrictive temperature (Fig. 2B). It should be noted that Ub-Pro-␤-galactosidase and Arg-␤-galactosidase in the rpn6-2 cells at 25°C were more stable than in other strains. These results indicate that the mutants have a defect in the 26 S proteasome function for degrading both ubiquitin fusion degradation pathway and N-end rule pathway substrates especially at the restrictive temperature.
In Vitro Degradation Assay of Wild-type and Mutant Proteasomes-Since both mutants showed defects in proteolysis in vivo, we thought the mutations might affect the proteolytic activity of the 26 S proteasome under the restrictive condition. To investigate the enzymatic activity of mutant proteasomes, we carried out an in vitro degradation assay with purified proteasomes. To affinity-purify proteasomes from wild-type and mutant cells, we generated strains in which the chromosomal RPN11, encoding one of the lid components Rpn11p, were replaced with RPN11-3ϫFLAG. Each strain was cultured in 250 ml of YPDAU medium at the permissive temperature to reach midlog phase, and then proteasomes were purified using anti-FLAG antibody-immobilized beads as described under "Experimental Procedures." Purified proteasomes were quantified on an SDS-polyacrylamide gel (Fig. 3A, left panel). rpn6 mutant proteasomes did not show an apparent difference in the overall band pattern on SDS-PAGE except that the Rpn6-1p band moved slightly slower than the wild-type Rpn6p due to the C-terminal extension (Fig. 3A, right panel). Our concern was that the purified rpn6-1 proteasomes had less CP attached to the RP, which might be a cause of the C terminus extension of Rpn6-1p. To see whether the mutations change the overall conformation of the holoenzyme, the purified proteasomes were run on a nondenaturing gel and visualized by overlaying a fluorogenic substrate (Fig. 3B). Both mutant proteasomes show the same migration pattern as the wild type and seem to have no significant aberration in the 26 S holoenzyme conformation and peptidase activity. We quantified the amount of the CP on the gel and used equal amounts (100 nM) to test the degradation of Ubn-Sic1 PY p by adding 300 nM Ubn-Sic1 PY p to the reaction mixture containing proteasomes and ATP-MgCl 2 . The reaction was terminated at either 3 or 10 min after incubation. The amount of the remaining ubiquitinated Sic1 PY p was measured by Western blotting with anti-T7 antibody (Fig. 3C) and shown as a function of time (Fig. 3D). Against our expectations, mutant proteasomes degraded the ubiquitinated substrate quite effectively, although enzyme activity was slightly lower than wild-type enzyme. Since mutant cells show temperaturesensitive growth, we then preincubated the proteasomes for 15 min at 40°C and performed the same degradation assay at 38°C to see whether the enzymatic activity would be affected by the higher temperature. Surprisingly, mutant proteasomes degraded the substrate even more efficiently at 38°C than at 25°C, as did the wild-type proteasomes (Fig. 3C). These results suggest that rpn6-1 and rpn6-2 mutations per se do not significantly affect the ubiquitin-dependent protease activity of the mutant holoenzyme at a higher temperature. Hence, the stabilization of model substrates as well as the accumulation of polyubiquitinated proteins in the rpn6 mutant cells should be due to the dysfunction of the 26 S proteasome at the restrictive temperature by a reason(s) other than the loss of enzyme activity.
rpn6-1 and rpn6-2 Cells Are Defective in the Assembly of the 26 S Proteasome-The facts that rpn6 mutant proteasomes were correctly assembled at the permissive temperature (Fig.  3A) and that, once made, they were structurally stable upon treatment at 38°C (data not shown) led us to a possibility that the mutants are likely to be defective in the process of the FIG. 1. Characterization of two temperature-sensitive rpn6 mutants. A, rpn6-1 (YNS1) and rpn6-2 (YNK7) cells either carrying a CEN vector (pRS314) or RPN6-CEN plasmid (pNS112) were streaked on YPDAU plates, and incubated for 2 days at the indicated temperature. B, mutation sites of rpn6-1 and rpn6-2. The nucleotide sequences of the rpn6-1 and rpn6-2 ORFs were determined. The asterisk indicates the stop codon. In the rpn6-1 allele, no amino acid change was found within the ORF, but the stop codon of the RPN6 gene was changed to the Tyr codon, and a new stop codon appeared after the 20-amino acid extension at the C terminus. The rpn6-2 mutant was found to bear two amino acid substitutions, one of which was placed in the N-terminal half (F137L) and the other in the PCI domain (L377P).
assembly of the 26 S proteasome at the restrictive temperature. To examine this possibility, we performed gel filtration chromatography of the whole cell extract of wild-type and mutant cells grown at 37°C. Wild-type and rpn6 mutant cells growing at 25°C logarithmically were shifted to and incubated for 7 h at 37°C, and cells were harvested by centrifugation. Whole cell extracts were prepared in the existence of ATP and MgCl 2 and resolved on a gel filtration column. Fractions were subjected to assay for peptidase activity and SDS-PAGE followed by Western blotting.
The peak of the 26 S proteasome was found in fraction 21, and the peak of the 20 S CP was at fraction 25, although wild-type extract contained very small amount of free 20 S CP. In wild-type extracts, all of the components of the lid, the base, and the 20 S CP so far tested were co-migrated at the position of the 26 S proteasome (Fig. 4B, top). In the rpn6 extracts, the peak of enzyme activity assayed without SDS, corresponding to the 26 S proteasome, shifted to the lighter fractions (approximately fraction 22) (Fig. 4A), suggesting an incomplete assembly of the 26 S holoenzyme in the mutants. Western blot using antibodies against lid components (Rpn3p, Rpn5p, Rpn7p, Rpn9p, Rpn12p), base (Rpt5p), and the 20S CP shows a dramatic change in the blotting pattern of the mutants compared with that of the wild-type strain (Fig. 4B). In the rpn6-1 mutant, signals of the 20 S CP and base subunits were detected in a lighter fraction in accordance with the peak shift of peptidase activity. Interestingly, signals of Rpn3p, Rpn7p, and Rpn12p were only slightly detected in the void fraction, if at all. Rpn5p and Rpn9p were detected in the same fractions (fractions 28 -32), but their peak did not coincide with that of either the CP or the base. These data suggest that in rpn6-1 cells, the binding of the 20 S CP and RP is deterred and that the assembly of the lid is incomplete. The data suggested that lid components including at least Rpn5p and Rpn9p formed a subcomplex. Essentially the same elution profile was seen with rpn6-2 extract. However, there are some differences between the two mutants in the distribution of components like Rpn3p or Rpn7p. In the case of Rpn3p, it was detected in the void fraction in the rpn6-2 extract, whereas no signals were detected in the rpn6-1 extract.
This may be a reflection of the difference of the two alleles in assembling process of the lid subcomplex.
Affinity Purification of the Lid from rpn6 Mutant Cells-To further investigate the nature of the subcomplex of the lid in rpn6-1 cells, we affinity-purified the lid from RPN11-3ϫFLAG wild-type and rpn6-1 strains cultured for 7 h at 37°C. The lid was separated from the rest of the complex by treating the beads with 1 M NaCl and was obtained as the material retained by the beads after extensive washing. The lid thus purified was resolved by SDS-PAGE along with the wild-type 26 S proteasome, the CP, and the base, followed by either Coomassie Brilliant Blue staining (Fig. 5A) or Western blotting using antibodies against lid components (Rpn3p, Rpn5p, Rpn7p, Rpn8p, Rpn9p, Rpn12p, and Sem1p), anti-AtRpn6p, and an anti-FLAG antibody detecting Rpn11p (Fig. 5B). Comparison of Coomassie Brilliant Blue-stained protein band patterns with the wild-type lid indicates that only four out of the nine lid components are affinity-purified with Rpn11p in the rpn6-1 strain under the restrictive condition. The results of the Western blotting confirm the subunit composition of the lid produced by rpn6-1 cells incubated at the restrictive temperature. In contrast to the wild-type lane, in which all lid components were present, some of the components were missing in the lid derived from rpn6-1 cells (lid rpn6-1 ); Rpn3p, Rpn6p, Rpn7p, Rpn12p, and Sem1p (Rpn15p) were missing in the rpn6-1 lane. The lid rpn6-1 was found to consist of only Rpn5p, Rpn8p, Rpn9p, and Rpn11p. In addition, peaks of the protein bands of the same gel were scanned by Image Gauge software (Fig. 5C), indicating that these four components forming the lid rpn6-1 are present in almost equal amounts. A subcomplex of the lid with the same composition was also seen in rpn6-2 cells (data not shown).
The lid rpn6-1 obtained by affinity purification could be an artifact produced during preparation of cell lysates. To show that the affinity purification really reflects the in vivo state of the mutant proteasome, we performed an in vivo cross-linking experiment. Wild-type and rpn6-1 cells grown for 7 h at 37°C were treated with DSP, a membrane-permeable cross-linking reagent, prior to affinity purification. To see the efficiency of FIG. 2. rpn6 mutants are defective in  proteolysis in vivo. A, accumulation of polyubiquitinated proteins. Wild-type (WT) (W303-1B) and rpn6-1 (YNS1) and rpn6-2 (YNK7) cells cultured for 7 h at 25 or 37°C were harvested, and total proteins were extracted by the alkali extraction method. Western blotting was performed using anti-polyubiquitin antibody (FK2), and Cdc28p was detected with anti-Cdc2p (human) antibody as a loading control. B, degradation of N-end rule pathway and ubiquitin fusion degradation pathway substrates. Wild-type (W303-1B), rpn6-1 (YNS1), and rpn6-2 (YNK7) cells were transformed with plasmids carrying a gene encoding one of the following: Ub-Ala-␤galactosidase, Ub-Arg-␤-galactosidase, or Ub-Pro-␤-galactosidase. Expression of Ub-X-␤-galactosidase (where X represents Ala, Arg, or Pro) was induced by adding 2% galactose to the medium. Cells were harvested after 7 h of induction either at 25 or 37°C, and steady state levels of ␤-galactosidase activity were assayed. Ub-Pro-␤-galactosidase and Arg-␤-galactosidase levels are indicated relative to the Ala-␤-galactosidase level. The averages of three independent experiments are shown. Open bar, wild type; light gray bar, Ub-Pro-␤-galactosidase; solid bar, R-␤-galactosidase.
cross-linking, half of the affinity-purified cross-linked proteasomes were treated with salt. Mock-treated proteasomes dissociated into the RP and the CP upon salt treatment (Fig. 5D,  lanes 1 and 2), whereas in the case of proteasomes purified from DSP-treated cells, a considerable amount of CP remained bound to the RP (Fig. 5D, lanes 3 and 4), verifying the efficient cross-linking reaction. Protein profiles of mock-treated and DSP-treated rpn6-1 cells did not show a difference, indicating that the subcomplex found in the rpn6-1 mutant exists free in vivo and was not a by-product of the purification procedure (Fig. 5D, lanes 5 and 6).
To see whether the assembly of the base or the CP is normal FIG. 3. Purified rpn6-1 and rpn6-2 proteasomes have proteolytic activity. A, affinity-purified proteasomes. Wild-type (WT), rpn6-1, and rpn6-2 strains producing Rpn11p-3ϫFLAG instead of wild-type Rpn11p (YYS40, YNS4, and YNK9, respectively) were cultured at the permissive temperature to reach midlog phase, and proteasomes were affinity-purified using agarose-immobilized anti-FLAG beads. Part of the purified proteasomes were resolved by 12.5% SDS-PAGE and stained with Coomassie Brilliant Blue (left). Note that Rpn6-1p migrated slower than the wild-type Rpn6p because of the C-terminal extension. The wild-type Rpn6p and Rpn6-1p were detected with anti-AtRpn6p antibody to confirm the difference in the mobility (right). B, nondenaturing PAGE. Part of the purified proteasomes were resolved by 4% nondenaturing PAGE, and the positions of the 26 S proteasome were visualized by overlaying a fluorogenic substrate, succinyl-L-leucyl-L-leucyl-L-valyl-L-tyrosine 4-methylcoumaryl-7-amide (Suc-LLVY-MCA) (left). The same gel was transferred to polyvinylidene difluoride membrane, and the CP subunits were detected by Western blotting with anti-20 S CP antibody. C, in vitro degradation of polyubiquitinated Sic1 PY p by affinity-purified proteasomes. The 26 S proteasome (100 nM) purified from wild-type or rpn6 mutant strains were incubated with polyubiquitinated Sic1 PY p (Ubn-Sic1) (300 nM) at either 25°C (left panel) or 38°C (right). Reactions were terminated by the addition of SDS-loading buffer at the indicated times, and the reaction mixture was run on a 10% SDS-polyacrylamide gel and subjected to Western blotting with anti-T7 antibody. The lane at the far right of each panel is the control lane that contained the reaction mixture incubated for 10 min without proteasomes. The same membrane was reprobed with anti-20 S CP antibody to confirm the amount of the proteasomes subjected to each reaction. D, the signals on the gels in C were quantified by the Image Gauge software and plotted as a function of the incubation time. Dotted line, reaction at 25°C; solid line, reaction at 38°C. and whether the binding between these complexes is dependent on the correct assembly of the lid, we prepared wild-type and rpn6-1 strains in which either the 20 S CP or the base was tagged with 3ϫFLAG and performed immunoprecipitation experiments. The resulting strains, PRE1-3ϫFLAG (CP) and RPN1-3ϫFLAG (base), were cultured in YPDAU for 7 h at 37°C, and proteasomes were affinity-purified from the total extract (Fig. 5E). In contrast to the wild-type cells, a complete 26 S holoenzyme was not effectively purified from rpn6-1 cells by either tag. When purified using the CP tag, mainly the CP was recovered, and a lesser amount of RP was co-precipitated. Consistent with this, when purified using the base tag, only a small amount of the lid and the CP were co-purified. In the base lane of the rpn6-1 extract (Fig. 5E), the signal of Rpn1-3ϫFLAG was exceedingly strong compared with other components. This result indicates that the incorporation of 3ϫFLAGtagged Rpn1p into the base seems to be inefficient in the mutant strain, which is a tendency occasionally observed also in wild-type cells. However, band quantification confirmed all components of the CP or the base to be present (data not shown), indicating that the assembly of the CP and the base were not deterred by the rpn6-1 mutation. These results suggest that the defect in the assembly of the lid does not affect the assembly of the CP or the base itself but prevents their efficient assembly into the RP.
Two-hybrid Interaction of Rpn5p and Rpn6p Mutant Proteins-Both of the substituted amino acids in Rpn6-2p were conserved in Caenorhabditis elegans, Drosophila melanogaster, A. thaliana, and Homo sapiens, and there is a possibility that these amino acids are important for the interaction with Rpn5p (11). To test this possibility, we performed yeast two-hybrid analysis with constructs rpn6-2FL-GAD and rpn6-2LP-GAD bearing the F132L or the L377P mutation, respectively. These plasmid constructs, along with the wild-type RPN6-Y plasmid, rpn6-1-GAD, and the vector alone, were co-transformed with GBD-RPN5 into the two-hybrid host strain. Representative transformants were tested for their growth on SC-LWH plates (Fig. 6A). As we have expected, Rpn6-1p-GAD showed positive interaction with GBD-Rpn5p, suggesting that the C terminus extension does not affect the binding of Rpn6-1p with Rpn5p. Interestingly, the L377P mutation alone abolished the interaction between the Rpn6p mutant protein and Rpn5p, whereas the F132L mutation did not prevent the interaction.
Our next interest was whether the two-hybrid interaction between Rpn5 and Rpn6 has any causal relationship with the temperature-sensitive growth displayed by the rpn6-2 mutant. rpn6-1 cells. A, lids were affinity-purified from Rpn11p-3ϫFLAG tagged wild-type (WT) (YYS40) and rpn6-1 (YNS4) cells cultured for 7 h at 37°C. Samples were run along with the wild-type 26 S proteasome, the CP, and the base on a 12.5% SDS-polyacrylamide gel, and protein bands were stained with Coomassie Brilliant Blue. Positions of lid components are indicated on the right. n, Rpn. B, part of the same samples as in A were resolved by 12.5% SDS-PAGE, transferred to polyvinylidene difluoride membrane, and subjected to Western blotting using the indicated antibodies against lid components. C, densitometric analysis of the lid. Protein bands of the same gel photographed in A (lanes lid and lid rpn6-1 ) were scanned by the Image Gauge software. Lid components corresponding to the respective peaks are indicated. Dotted line, wild type; solid line, rpn6-1. Note that Rpn6-1p shows a band shift due to the C terminus extension. D, affinity purification was carried out as in A, after pretreatment of cells with a chemical cross-linker. Wild-type (YYS40) and rpn6-1 (YNS4) cells cultured for 7 h at 37°C were treated with DSP, washed, and lysed, and proteasomes were affinity-purified as described under "Experimental Procedures." To verify the cross-linking reaction, the wild-type sample was divided into two portions, and one portion was washed with buffer containing 0.5 M NaCl. Note that in the mock (Me 2 SO)-treated sample, the CP dissociated upon salt treatment (lane 2), whereas a considerable amount remained bound in the DSP-treated sample (lane 4), indicating the cross-linking to be successful. The profile of the affinity-purified lid subcomplex did not alter upon DSP treatment (lanes 5 and 6), suggesting that this subcomplex existed free from the 26 S proteasome in vivo. E, affinity purification of proteasomes from CP-and base-tagged stains. Wild-type and rpn6-1 strains producing Pre1p-3ϫFLAG and Rpn1p-3ϫFLAG instead of wild-type Pre1p (CP component; YYS 37 and YNS 2, respectively) and Rpn1p (base component; YYS39 and YNS3, respectively) were cultured for 7 h at 37°C, and proteasomes were affinity-purified using agarose-immobilized anti-FLAG beads. The purified proteasomes were resolved on a 12.5% SDSpolyacrylamide gel and stained with Coomassie Brilliant Blue. Bands corresponding to the tagged components are indicated with solid arrows. The approximate migrating positions of RP and CP components are indicated on the right.

FIG. 5. Subcomplex of the lid is detected in
To examine this, we created strains in which the chromosomal RPN6 was replaced with the mutant rpn6 gene bearing either the F132L or L377P mutation and tested their growth at 25 and 37°C. Whereas the rpn6-2FL strain grew normally at both temperatures, the rpn6LP strain showed growth defect at 37°C (Fig. 6B). This result suggests that the L377P amino acid replacement in the PCI domain deters the interaction with Rpn5p, and this loss of interaction is responsible for the temperature-sensitive phenotype of the rpn6-2 mutant.

DISCUSSION
RPN6 is an essential gene in the yeast Saccharomyces cerevisiae and is a counterpart of the human proteasome subunit S9/p44.5 (47). For functional analysis of Rpn6p, three mutants have been employed: ⌬rpn6 (26) and two mutants, rpn6-1 and rpn6-2, described in this study. The rpn6-1 mutation occurred in the stop codon of the RPN6 ORF, creating a C-terminal extension consisting of 20 amino acid residues. The rpn6-2 mutant allele has two amino acid substitutions, one of which is the L377P mutation in the PCI domain, which we found to be responsible for the loss of interaction with a previously reported interaction partner Rpn5p (11) and for the temperaturesensitive growth of this mutant. Rpn6-1p has no mutations in the ORF and hence has the wild-type PCI domain, and indeed the Rpn6-1p-GAD fusion protein interacted with GBD-Rpn5p.
The PCI domain, which occurs in subunit polypeptides of several multiprotein complexes, consists of ϳ100 amino acid residues that are predicted to form an ␣-helical domain playing an important role in the incorporation of the subunit polypep-tides into the complex (48,49). Deletion of this domain leads to loss of interaction with other components in the complex (11,30) or causes instability of the whole complex (50). Supporting the significance of the PCI domain as a scaffold for the assembly of subunits, the rpn6-2 mutant was defective in the assembly of the lid at higher temperature probably due to the single amino acid change in the PCI domain of Rpn6p. However, as in the case of Rpn7p, in which not only amino acid changes in the PCI domain but also those occurring outside the PCI domain caused loss of interaction with the known partner (29), regions other than the PCI domain of Rpn6p might also play a role in the interaction with Rpn5p and the formation of the lid complex. In accordance with this consideration, the PCI domain of Rpn6p alone did not show yeast two-hybrid interaction with Rpn5p. 4 The advantage of employing temperature-sensitive mutants of the essential gene such as RPN6 would be that it is possible to purify proteasomes containing the mutant subunit by culturing the cells under the permissive condition. To see whether the defects seen in these rpn6 mutants are caused by deterioration in enzymatic activity of the mutant 26 S proteasome at a higher temperature, we purified the holoenzyme from rpn6-1 and rpn6-2 mutant cells grown at the permissive temperature and compared them with the wild-type holoenzyme. The 26 S proteasome from the rpn6 mutant cells were found to be structurally almost normal. However, the incorporation of Rpn6-1p with the C-terminal extension into the lid seems weaken the association of the 20 S CP and the RP (Fig. 3A). To our surprise, the mutant 26 S proteasomes showed enzyme activity toward polyubiquitinated Sicl PY p at both 25 and 38°C (Fig. 3D). The activities of rpn6 mutant proteasomes were slightly lower than that of the wild-type enzyme; however, they degraded the substrate more rapidly at 38°C, the restrictive temperature for the rpn6 mutant cells. This result indicates that the temperature sensitivity shown by these mutants is not due to lower enzyme activity at a higher temperature. From these results, we conclude that the rpn6-1 and rpn6-2 mutations do not seriously affect the activity of the holoenzyme, once the mutant 26 S proteasome is properly assembled. The mutant 26 S proteasomes, once made, did not disassemble or were as stable as wild-type proteasomes upon treatment at 38°C (data not shown).
The temperature sensitivity shown by the rpn6 mutants should be explained other than by the instability of the mutant holoenzyme upon exposure to a higher temperature. Therefore, we anticipated that the defect could be in the process of the assembly of the 26 S proteasome. Indeed, the assembly of the lid at a higher temperature was found to be incomplete in the rpn6 mutants, and among the examined components, only Rpn5p and Rpn9p were co-fractionated in gel filtration (Fig.  4B). Lid components like Rpn3p, Rpn7p, and Rpn12p were not detected throughout the gel filtration fractions, especially in the rpn6-1 extract, suggesting that they cannot exist stably in the absence of a wild-type Rpn6p. Since small amounts of these proteins are detected in the void fraction, they may form some kind of aggregate or be in some insoluble fraction under the restrictive condition.
To study the subunit composition of the lid subcomplex in detail, we chose rpn6-1 that showed a more distinct phenotype and purified the lid from the Rpn11-3ϫFLAG-tagged strain grown at the restrictive temperature. We previously showed that five components (Rpn5p, Rpn6p, Rpn8p, Rpn9p, and Rpn11p) were forming a subcomplex in the rpn7-3 mutant (29). The lid FIG. 6. Two-hybrid interaction of Rpn5p and Rpn6p is abolished with the rpn6-2LP mutation. A, plasmids carrying either RPN6, RPN6-2FL, RPN6-2LP, or Rpn6-1p fused to the N terminus of the GAD vector (ScRpn6-GAD, pEK228, pEK226, and pEK227, respectively) were co-transformed with a GBD-ScRPN5 bearing plasmid into the YRG2 strain, and transformants were selected on SC-LW plates. Representative colonies were picked up, streaked on SC-LWH and SC-LW plates, and incubated for 5 days at 25°C. B, the amino acid substitution L377P is responsible for the rpn6-2 mutant phenotype. The two mutations in rpn6-2 causing amino acid replacements, F132L and L377P, were separately introduced into the wild-type strain and designated rpn6-2FL and rpn6-2LP, respectively. Wild type, rpn6-2FL (YEK122), and rpn6-2LP (YEK123) were cultured in YPDAU liquid medium and diluted to A 600 ϭ 1, and 10-fold serial dilutions were spotted on YPDAU plates to test the growth at 25 and 37°C. C, a model of the role of Rpn6p in the assembly of the lid. Rpn6p is necessary for the incorporation of Rpn3p, Rpn7p, Rpn12p, and Sem1p (Rpn15p) to the cluster of Rpn5p. purified from the rpn6-1 mutant did not contain Rpn6-1p and consisted of only four components: Rpn5p, Rpn8p, Rpn9p, and Rpn11p. The gel filtration pattern of the rpn6-2 extract implied the existence of a lid subcomplex containing Rpn5p and Rpn9p, which was similar to the one produced in the rpn6-1. Indeed, when the same affinity purification experiment was performed with rpn6-2 cells, we obtained a subcomplex with the same composition as in rpn6-1 cells (data not shown), indicating that the C terminus extension and the amino acid substitutions of the rpn6-2 mutations have basically the same effect upon lid assembly. However, it seems that rpn6-1 mutation has a severer effect in that it shows a slight defect in the binding of the CP to the RP even at the permissive temperature.
Although the base and the CP were co-fractionated by gel filtration of rpn6-1 extract, immunoprecipitation experiments revealed that they were not bound to each other (Fig. 5E). Interestingly, a larger amount of CP was immunoprecipitated in the rpn6-1 mutant compared with the wild-type strain, probably due to the existence of an increased amount of free CP in the soluble fraction. It is an interesting question how the mutant lid affects the binding between the base and the CP. Apparently, the mutant lid could not bind the base to form a complete RP, and the lid rpn6-1 was not co-fractionated with the base or the CP. The assembly of the base and the lid to the RP may be a prerequisite for the stable binding of the base to the CP.
In previous studies and the present study, we found partially assembled lid complexes in extracts prepared from temperature-sensitive lid mutants incubated at the restrictive temperature. We have shown that Rpn3p, Rpn7p, and Rpn12p were not incorporated into the lid due to mutations in Rpn7p (29). The lid rpn6-1 consists of Rpn5p, Rpn8p, Rpn9p, and Rpn11p. Rpn9p is known to be dispensable, and we succeeded in purifying the 26 S proteasome from ⌬rpn9 cells, 5 suggesting that Rpn9p is not the center for the lid assembly. Therefore, Rpn5p, Rpn8p, and Rpn11p could be the core of the lid. Under the condition when the mutant Rpn6p did not interact with Rpn5p, a component of the core cluster, Rpn3p, Rpn7p, Rpn12p, and Sem1p (Rpn15p) in addition to Rpn6p were no longer incorporated into the lid. This result is in good accordance with the yeast two-hybrid interaction map presented by Fu and coworkers (11), that shows the existence of a cluster consisting of Rpn5p, Rpn6p, Rpn8p, Rpn9p, and Rpn11p and another consisting of Rpn3p, Rpn7p, and Rpn12p. According to the gel filtration pattern, it is likely that the lid components other than found in lid rpn6-1 exist in some kind of an insoluble aggregate that could not be solubilized by merely breaking the cells with glass beads.
Are those lid subcomplexes found in rpn6 mutants degradation products from the 26 S proteasome, or are they produced de novo? At present, we do not have an answer to this question. However, two lines of unpublished evidence favor the latter possibility; (a) heat treatment of the mutant 26 S proteasome did not produce the lid subcomplex, and (b) cycloheximide in the mutant cultures incubated at a higher temperature prevented the formation of the subcomplex.
Although we cannot exclude the possibility that Rpn6p has other functions than merely serving as a scaffold for the incorporation of Rpn3p, Rpn7p, Rpn12p, and Sem1p (Rpn15p), by assuming that lid rpn6-1 is formed de novo, we can make a scenario of the formation of the lid and the function of Rpn6p. Interaction between Rpn5p and Rpn6p and the correct conformation at the C terminus of Rpn6p is important for the incor-poration of Rpn6p into the core of the lid or lid rpn6-1 . Probably only after the correct and stable incorporation of Rpn6p, Rpn7p could be incorporated into the complex, which in turn would recruit Rpn3p, Rpn12p, and Sem1p (Rpn15p) to form a complete lid (Fig. 6C).