Analysis of a gene encoding Rpn10 of the fission yeast proteasome reveals that the polyubiquitin-binding site of this subunit is essential when Rpn12/Mts3 activity is compromised.

Substrates are targeted for proteolysis by the ubiquitin pathway by the addition of a polyubiquitin chain before being degraded by the 26 S proteasome. Previously, a subunit of the proteasome, S5a, was identified that was able to bind to polyubiquitin in vitro and thus proposed to act as a substrate recognition component. Deletion of the corresponding Saccharomyces cerevisiae gene, MCB1/RPN10, rendered cells viable indicating that other proteasomal polyubiquitin receptors must exist. In this study, we describe pus1(+), the fission yeast homologue of RPN10. This gene is also not required for cell viability; however, the Deltapus1 mutant is synthetically lethal with mutations in other proteasomal component-encoding genes, namely mts3, pad1, and mts4 (RPN12, RPN11, and RPN1). Overexpression of pus1(+) is able to rescue mts3-1 at 32 degrees C but overexpression of a cDNA encoding a version of Pus1 that does not bind to polyubiquitin cannot and leads to greatly reduced viability when used to rescue the mts3-1Deltapus1 double mutant. The Mts3 protein was unable to bind to polyubiquitin in vitro, but the Pus1 and Mts3 proteins were found to bind to one another in vitro, which taken together with the genetic data suggests that they are also closely associated in vivo.

The ubiquitin pathway is the major non-lysosomal protein degradation pathway in eukaryotes. It is responsible for the proteolysis of a wide variety of cellular proteins including important short-lived regulatory proteins such as p53, I-B, c-Myc, c-Fos, cyclins, and cyclin-dependent kinase inhibitors (reviewed in Refs. [1][2][3][4]. Another crucial function of this pathway is the elimination of misfolded or denatured proteins such as those that arise in cases of heat shock or other forms of cellular stress (5). Proteins are targeted for destruction by the addition of a polyubiquitin chain and then degraded by a large multisubunit protease known as the 26 S proteasome.
Ubiquitin is a highly conserved, 76 amino acid protein that is synthesized into chains by an enzymatic cascade. Through the actions of an activating enzyme, a ubiquitin-conjugating enzyme, and in most cases a separate ubiquitin ligase, a chain of ubiquitin monomers are joined to each other through a series of isopeptide bonds that are formed between the C-terminal glycine in one ubiquitin moiety and the side chain of an internal lysine in the next one (reviewed in Refs. 1, 3, and 6). A ubiquitin chain elongation factor has recently been identified, and the term E4 is used to describe this type of activity (7). The linkage between the polyubiquitin chain and the substrate is also via an isopeptide bond, usually utilizing a lysine residue within the target protein, forming a structure known as a polyubiquitinated protein conjugate. Although there is more than one lysine residue within the ubiquitin sequence, polyubiquitin chains linked through Lys-48 seem to be the most important in targeting substrates to the proteasome (8,9). Polyubiquitin chains linked by Lys-63 do not appear to have a role in proteolysis but are involved somehow in mediating post-replicative DNA repair (8,10,11).
After a ubiquitinated substrate conjugate has been generated, the next step in the ubiquitin-mediated proteolytic pathway is the recognition of the polyubiquitin chain by the 26 S proteasome. This large multisubunit protease consist of two subcomplexes, the 20 S catalytic core and the 19 S regulatory cap (reviewed in Refs. 6 and 12-14). The 20 S core consists of four seven-membered rings of subunits, grouped into two types, ␣ and ␤. The various catalytic activities of the proteasome are formed by the N-terminal threonine residues of some of the ␤-subunits and are housed in the interior of the 20 S core shielding the rest of the cell from the proteolytic activity (15)(16)(17). Crystallography studies have determined the three-dimensional structure of the 20 S core from Thermophilus acidophilum and Saccharomyces cerevisiae (18,19). At the ends of the 20 S core in T. acidophilum, there are narrow channels, through which it is presumed that substrates can enter. In the case of the yeast 20 S core, however, there are no openings at all suggesting that the channels are gated presumably by association with the 19 S cap. The finding that the 20 S core alone can degrade small peptides but not native proteins suggests that the recognition of polyubiquitinated conjugates, unfolding of substrates and their translocation into the 20 S core are functions performed by the 19 S regulatory cap sub-complex. The 19 S cap consists of 17 different proteins that fall into two groups (20). Six subunits are members of a family of ATPases, and it has been proposed that they function as reverse chaperones, unfolding substrates and translocating them into the core particle (21)(22)(23). A recent study demonstrates that the AT-Pases do, in fact, have the ability to act as chaperones and refold substrates in vitro (24). The remaining proteins are termed the non-ATPases. Their role is less clear but proposed functions also include unfolding and recognition of substrates as well as the recycling of ubiquitin moieties.
One of the major unanswered questions regarding the ubiq-uitin-mediated proteolytic pathway is how are polyubiquitinated substrates recognized by the 26 S proteasome? The simplest model would suggest that one particular subunit of the 19 S cap would be responsible and therefore possess polyubiquitin chain binding activity. One of the non-ATPase subunits of the human proteasome, S5a, was found to possess polyubiquitin chain binding activity in vitro (25). Homologues of this protein in Arabidopsis, Drosophila, and S. cerevisiae were also subsequently found to have in vitro chain binding activity (26 -28). Surprisingly, deletion of the corresponding yeast gene RPN10 (formerly MCB1) resulted in cells that were viable and that degraded the majority of short-lived proteins in a manner similar to wild type (28). Studies in yeast have shown that the degradation of ubiquitinated substrates is essential for the progression of the cell cycle (22,23), thus Rpn10 is presumably not the sole proteasomal recognition factor for polyubiquitin conjugates, although the strain lacking Rpn10 was sensitive to amino acid analogues suggesting that this protein does play a role in proteolysis. Intriguingly, the role of this subunit in degradation and its ability to bind to polyubiquitin appear to be mediated by different and distinct parts of the Rpn10 protein (29). Various studies have identified a region of S5a/Rpn10 that is necessary for binding to polyubiquitin (29 -31). It is a stretch of conserved hydrophobic amino acids usually LAL(M)AL, a finding consistent with reports that have implicated hydrophobic residues on the surface of ubiquitin as being important in the binding of polyubiquitin chains to the 26 S proteasome (32). Surprisingly, neither the complementation of the sensitivity to amino acid analogues nor the ability to degrade certain test substrates required a wild type polyubiquitin-binding site, but instead both were dependent upon an intact N-terminal sequence. It was unclear, therefore, whether the polyubiquitinbinding site identified in vitro has any relevance in vivo. Recent work has demonstrated that the 19 S regulatory cap of the yeast proteasome can be dissected into two discrete subcomplexes called the base and the lid (33). These sub-complexes were formed upon purification of the proteasome from the ⌬rpn10 strain, and their formation was then re-produced from wild type proteasomes using high concentrations of salt. The base consists of the six ATPases, Rpn10/S5a, and the two largest non-ATPases Rpn1 and Rpn2, whereas the lid is made up of the remaining non-ATPases. These sub-complexes were used to assign functions of the 19 S cap to various groups of subunits, and it was found that the base could activate the 20 S core in peptide hydrolysis but that for the degradation of polyubiquitinated conjugates both the base and lid were required. It would seem, therefore, that the ability of the proteasome to recognize polyubiquitinated substrates is due to either non-ATPase subunit(s) in the lid complex or by a combination of subunits in the lid and the base.
Despite a non-essential role in ubiquitin-mediated protein degradation, Rpn10/S5a would appear to contribute to the degradation of polyubiquitin conjugates in vivo as the steady-state levels of polyubiquitin conjugates were found to increase in a ⌬RPN10 strain (28). Furthermore, addition of exogenous Mbp1 (the Arabidopsis homologue of S5a) inhibits ubiquitin-dependent proteolysis in a cell-free system, presumably by competing with the proteasome for ubiquitinated conjugates (34). One possible way of identifying other possible polyubiquitin receptors would be to look for mutants that interact genetically with ⌬rpn10. Previously, we have identified mutations in a number of genes that encode subunits of the fission yeast 26 S proteasome (23,(35)(36)(37). Here, we describe the cloning and analysis of a fission yeast gene, pus1 ϩ , encoding the homologue of Rpn10/S5a.

EXPERIMENTAL PROCEDURES
Fission Yeast Strains and Techniques-All Schizosaccharomyces pombe strains used were derived from the wild type heterothallic strains 972 h Ϫ and 975 h ϩ . Standard genetic manipulations and media were as described (38). Minimal media plates containing canavanine were made using glutamate as a nitrogen source. Canavanine sulfate was made as a 6 mg/ml stock and added to media after autoclaving. Yeast transformations were performed by electroporation (39). Expression from the nmt1 promoter of pREP3X was repressed by the addition of thiamine to minimal medium at a concentration of 5 M (40).
Cloning of pus1 ϩ and the Generation of pus1 Constructs-The degenerate oligonucleotides used to clone a portion of the pus1 ϩ gene from an S. pombe cDNA library in the vector pREP1 (kindly provided by C. Norbury and B. Edgar) were designed against the peptide sequences MVLE(A/S)TM and F(D/E)FGVDP, respectively, 5Ј-ATGGTI(T/C)TIG-A(A/G)GCIACIATG-3Ј and 5Ј -IGG(A/G)TCIACICC(C/A)AA(C/T)TC(C-/A)AA-3Ј. PCR 1 was carried out for a total of 40 cycles using an annealing temperature of 37°C. The resulting DNA fragment was then amplified using the following oligonucleotides and cloned into the BamHI and SalI sites of pREP3X: 5Ј-GCTGGATCCATGGTI(T/C)TIG-A(A/G)GCIACIATG-3Ј and 5Ј-ACGCGTCGACTCAIGG(A/G)TCIACIC-C(C/A)AA(C/T)TC(C/A)AA-3Ј and sequenced using oligonucleotides directed against the nmt1 promoter and terminator sequences.
The 3Ј region of the cDNA was amplified from the library using the following oligonucleotides directed against the coding strand of pus1 and one directed against the nmt1 terminator region of the vector, respectively, 5Ј-TGGTGCTGGAAGCGACGATG-3Ј and 5Ј-TGGGCTTC-CATAGTTTGA-3Ј. A complete cDNA was then amplified and cloned into the BamHI and SmaI sites of the vector pREP3X (40) using the oligonucleotides 5Ј-ACTCAGGATCCAATGGTGCTGGAAGCGACGAT-G-3Ј and 5Ј-ATACTCCCGGGATCATTCTTGCATTTT-3Ј to give the vector pREP3Xpus1 ϩ . The genomic clone of pus1 ϩ was amplified from genomic DNA using the oligonucleotides 5Ј-ATGGTGCTGGAAGCGAC-GATG-3Ј and 5Ј-CGTTGCTGCATAAAGCATGGC-3Ј. Three introns were found at positions 26, 173, and 496 of the cDNA sequence. The pus1 ϩ sequence has since been sequenced as part of the S. pombe sequencing project and submitted to EMBL with the accession number O94444. In order to generate a pus1 null mutant, the following oligonucleotides were used to amplify the 5Ј region of the pus1 ϩ gene and subclone it into the BamHI and HindIII sites of pBluescript II KS (Stratagene) to give pBSpus1N. The 3Ј region of the pus1 ϩ gene was then amplified and subcloned into the HindIII and SalI sites of pBSpus1N to give pBSpus1NC. A 1.8-kilobase pair HindIII fragment containing the S. pombe ura4 ϩ gene was ligated into pBSpus1NC that had been digested with HindIII. The following oligonucleotides were used to amplify this deleted copy of the pus1 gene which was then transformed into the haploid S. pombe strain leu1-32ura4D-18 h Ϫ : 5Ј-CTAATTGATAATTCAGAA-3Ј and 5Ј-CGTTGCTGCATAAAGCAT-3Ј. Stable uracil prototrophs were isolated, and the correct deletion mutants were identified by Southern blot analysis and genomic PCR.
The human S5a gene was removed from pET26bS5a (a kind gift from Patrick Young) and subcloned into the NdeI and SalI sites of pREP1 (40). To generate the A5 and N5 mutants of Pus1, site-directed mutagenesis was carried out using the Quickchange site-directed mutagenesis kit (Stratagene) according to manufacturer's instructions using the oligonucleotides as follows: for the A5 mutant, 5Ј-AATCTTGA-CGTTGAAGCTGCGGCAGCCGCGGAACTTTCAATGGCG-3Ј and 5Ј-C-GCCATTGAAAGTTCCGCGGCTGCCGCAGCTTCAACGTCAAGATT-3Ј; for the N5 mutant, 5Ј-AATCTTGACGTTGAAAATAACAATAACAA-TGAACTTTCAATGGCG-3Ј and 5Ј-CGCCATTGAAAGTTCAATGTTA-TTGTTATTTTCAACGTCAAGATT-3Ј. The mutated versions were then subcloned back into pGEX-KG or pREP3X. The mts3 ϩ cDNA was amplified and subcloned into the NcoI and SalI sites of pGEX-KG using the oligonucleotides 5Ј-ACATGCCATGGGTATGAGTACATTAGACTTGA-AC-3Ј and 5Ј-TACTCGTCGACTTAAACAATCTGCTCTAATTC-3Ј. For sequencing the mts3-1 mutation, the mts3 coding region was amplified from genomic DNA isolated from the mts3-1 mutant using the oligonucleotides 5Ј-ATGAGTACATTAGACTTGAAC -3Ј and 5Ј-TTAAACAAT-CTGCTCTAATTC-3Ј. Both strands of the amplified region were se-quenced with oligonucleotides specific for the mts3 sequence. This truncated version of mts3 was subcloned into the NcoI and SalI sites of pGEK-GK using the 5Ј NcoI oligonucleotide described above and a 3Ј SalI oligonucleotide as follows: 5Ј-TACTCGTCGACTTACAGGTACAG-CAAACTTGT-3Ј.
Purification of Recombination Proteins and the Preparation of Antiserum-The GST-Pus1, GST-Pus1N5/A5, and GST-Mts3 fusion proteins were all expressed in Escherichia coli and purified as described previously (37). Binding assays were carried out as described previously (37). Briefly, GST-Mts3 protein was purified on glutathione-Sepharose and then treated with thrombin to release Mts3. The thrombin was inactivated with anti-thrombin (Sigma). Equimolar amounts of Mts3 protein were added to either GST, GST-Pus1, or GST-Pus1N5 attached to glutathione-Sepharose. After 4 washes in binding buffer (150 mM NaCl, 50 mM Tris, pH 7.6, 10% glycerol), the beads were boiled in SDS sample buffer and the resulting eluted proteins analyzed by SDS-PAGE and Western blotting. For the production of antiserum, 100 g of purified Pus1 protein was emulsified in TiterMax (Hunters) and injected into rabbits. Four weeks later a second injection of 100 g of protein was given in saline and the serum collected after a further 10 days. The Mts3 antibody was made by Diagnostics Scotland and generated by injecting 1 mg of purified Mts3 protein into a sheep followed by a further boost of 300 g of protein. Antiserum was affinity purified using the corresponding recombinant protein as described (42). The following antisera were used as follows for Western blot analysis: affinity purified anti-Mts4 (37) diluted 1 in 5000; anti-20 S core complex (37) diluted 1 in 1000; affinity purified anti-Pus1 (this study) diluted 1 in 1000; affinity purified anti-Mts3 diluted 1 in 500; and anti-ubiquitin (Dako) diluted 1 in 1000.
Polyubiquitin Binding Assays and the Purification of the Fission Yeast Proteasome-Ubiquitin-lysozyme conjugates were prepared as described (43). The binding of these conjugates to proteins immobilized on nitrocellulose was as described (25). For the binding of ubiquitinlysozyme conjugates to GST fusion proteins bound to glutathione-Sepharose, 50 l of Sepharose with either purified GST, GST-Mts3, or GST-Pus1 bound were mixed with either 100 l of polyubiquitin 125 Ilysozyme conjugates, 100 l of 125 I-ubiquitin at 0.1 mg ml Ϫ1 or 100 l of 125 I-lysozyme. After incubation on a rotary wheel at room temperature for 1 h, the beads were washed three times with phosphate-buffered saline, and then the beads were boiled in SDS sample buffer and subjected to electrophoresis and Western blotting. For the binding of proteins to conjugates attached to Sepharose, 50 l of glutathione-Sepharose bound to polyubiquitin-GST were incubated with purified Pus1, Pus1N5, or Mts3 proteins that had been removed from GST fusions with thrombin and treated with anti-thrombin to inactivate it. The incubation and washing was as above. The GST-ubiquitin conjugates were made as described (43) using ubiquitin fused at its C terminus to GST (a kind gift from S. Jentsch). The purification of the 26 S proteasome from S. pombe was as described (37). Glycerol gradient centrifugation of fission yeast extracts was as described (42).
Fluorescence Microscopy-Cells were prepared and images collected exactly as described (42). Affinity purified Pus1 antiserum was used at a dilution of 1 in 20.

Cloning of the Gene Encoding S5a of the 26 S Proteasome in
Fission Yeast-Previously, we have isolated genes encoding several subunits of the 19 S cap complex of the fission yeast proteasome, namely mts2 ϩ , mts3 ϩ , mts4 ϩ , and pad1 ϩ (23, 36 -38). These genes were isolated as cDNA clones that were able to complement temperature-sensitive mutations in the corresponding genes. The gene encoding Rpn10/S5a in budding yeast was identified in the S. cerevisiae genome sequencing project as an open reading frame homologous to the S5a proteins previously identified in Arabidopsis, Drosophila, and humans (26,27,43). As this gene was found to be non-essential, it seemed possible that the fission yeast orthologue would also not be required for cell viability and therefore not be identified in our screen that has been highly specific for essential 19 S cap proteasomal subunits. For this reason we decided to clone the gene encoding S5a in fission yeast by degenerate PCR as detailed under "Experimental Procedures." The pus1 ϩ , for polyubiquitin subunit, gene is predicted to encode a protein of 243 amino acids with a predicted mass and pI of 27,089 and 4.79, respectively, and is to date the shortest S5a protein known (Fig. 1). Like the budding yeast protein, Rpn10, it is considerably shorter (between 119 and 153 amino acids) than its higher eukaryotic orthologues. Over its length it is 46% identical to human S5a, 47% identical to the Drosophila protein 54p, 49% identical to Arabidopsis Mbp1, and 51% identical to Rpn10. These levels of identity are similar to those exhibited between the other orthologues. A region based around the LALAL sequence has been identified that is necessary for binding to polyubiquitin in vitro (29,31), and disruption of the LALAL sequence itself was found to abolish binding. The higher eukaryotic S5a proteins possess two such sites, whereas Pus1, like Rpn10 from budding yeast, only has one located between residues 210 and 214. As seen with the other sequences, the N terminus of Pus1 is very highly conserved. A deviation from the consensus occurs in the Pus1 sequence immediately adjacent to the proposed polyubiquitin-binding site. In all other sequences cloned to date the LAL(M)AL region is followed by an arginine and then by V(L)SMEE. In Pus1 the arginine is replaced by a glutamic acid, and the first glutamic acid of the SMEE sequence is replaced by an alanine.

Pus1 Is a Component of the 26 S Proteasome in Fission Yeast and Can Bind to Polyubiquitin Conjugates in
Vitro-In order to analyze the Pus1 protein further, polyclonal antiserum was raised against the purified recombinant Pus1 protein. First we wanted to determine that Pus1 was in fact a component of the 26 S proteasome in fission yeast, as its sequence homology suggested. Previously, we have purified the 26 S proteasome from fission yeast (37,44). Nitrocellulose filters containing the glycerol gradients from these purification procedures were probed using the anti-Pus1 antiserum. As can be seen in Fig.  2A, the Pus1 protein was found in the fractions containing both the highest 26 S proteasome activity (37) and also the highest concentration of known proteasomal components. Although the predicted molecular mass of the Pus1 protein is 27 kDa, the Pus1 antiserum recognized a protein of approximately 32 kDa in both fission yeast extracts and in extracts of E. coli cells expressing recombinant Pus1. A similar discrepancy between the predicted and observed molecular masses was also seen with Rpn10 and the higher eukaryotic orthologues (27,34). Previous studies have found that S5a, unlike other proteasomal components, exists in a free form as well as in the complex bound form (27,34). In order to see if this was also the case with Pus1, glycerol gradient centrifugation of fission yeast extracts was carried out without any prior purification. As can be seen in Fig. 2B, Pus1 was found in both the low molecular weight fractions as well as in the higher sedimenting 26 S proteasome-containing fractions. These latter fractions were indicated by the presence of the known proteasomal components, Mts4/Rpn1 and the 20 S core complex.
The Pus1 protein was also analyzed for its predicted ability to bind to polyubiquitin. A filter binding assay was performed essentially as described (25) using Pus1 protein immobilized on a nitrocellulose filter. The wild type Pus1 protein was able to bind to the polyubiquitin conjugates but did not bind to either the ubiquitin monomers or to the lysozyme alone (Fig. 2C). In addition, mutant forms of the Pus1 protein were made and assayed for their ability to bind to polyubiquitin conjugates. Previous analysis has identified a binding site for polyubiquitin in Rpn10 orthologues in which the hydrophobic sequence LAL-(M)AL is critical. Two mutant versions of the Pus1 protein were made, one where the sequence was mutated to AAAAA (A5) and another to NNNNN (N5). Both the A5 and N5 mutant Pus1 proteins were found to migrate slightly slower than the wild type Pus1 protein. A similar discrepancy was observed with mutated versions of the Rpn10 protein (29). Interestingly the A5 version was found to bind as well as the wild type protein indicating that the overall hydrophobic nature of this patch is important and not the alternating LALAL sequence. The N5 version, however, as expected, was unable to bind and even upon prolonged exposure of the filters no signal was seen (data not shown). The regions of the filters that corresponded to the conjugates were excised from the filters, boiled in SDS sample buffer, and subjected to electrophoresis on a second SDS-polyacrylamide gel. PhosphorImager analysis revealed that the Pus1 protein had preferentially bound higher molecular weight conjugates (data not shown), a finding in keeping with the known properties of Rpn10 and its orthologues.
Pus1 Is Not Essential for Cell Viability-In order to determine whether or not pus1 ϩ encodes a protein that is essential for cell viability, a construct was engineered whereby most of the pus1 ϩ coding region was replaced by the selectable marker ura4 ϩ as shown in Fig. 3A. As the gene encoding Rpn10 in S. cerevisiae is not essential (34), the pus1::ura4 ϩ construct was transformed into a haploid strain of S. pombe. Haploid transformants were isolated and found to possess the pus1::ura4 ϩ construct integrated at the pus1 chromosomal locus as shown by genomic PCR and Southern blotting (data not shown). Furthermore, protein extracts were made from the pus1::ura4 ϩ strain and used in Western blotting analysis with the anti-Pus1 antiserum. As shown in Fig. 3B, the antiserum failed to detect a protein in these extracts, whereas with wild type extracts, a protein of 32 kDa was clearly detected. Microscopic examination revealed that the pus1::ura4 ϩ cells appeared to be wild type, and no differences in growth rates were observed in comparison with wild type cells, neither were any defects detected in mating or sporulation efficiency (data not shown). Cells lacking the S. cerevisiae orthologue Rpn10 were found to be sensitive to canavanine (34). This is an analogue of arginine which, when added to media, will cause the accumulation of abnormal proteins that are then degraded by the ubiquitin-dependent pathway (5). As shown in Fig. 3C, the pus1::ura4 ϩ cells grew considerably slower than wild type on media supplemented with 4 g/ml canavanine and on media containing 8 g/ml canavanine. This inability to grow on canavanine was rescued by the addition of a plasmid carrying the pus1 ϩ cDNA under the control of the exogenous nmt1 promoter or by the human orthologue, S5a, also under the control of nmt1 (Fig.  3D). This finding illustrates the conserved nature of the Rpn10/ S5a protein despite Pus1 being considerably shorter than its human counterpart. A subsequent study showed that a version of Rpn10 that was unable to bind to polyubiquitin was still able to complement the growth defect on canavanine (29). The same was found to be true for the pus1::ura4 ϩ mutant as the N5 version of the pus1 cDNA under the control of the nmt1 promoter was able to complement the growth defect.
Pus1 Is Localized around the Nuclear Periphery-Previously, we have shown that the fission yeast 26 S proteasome is localized predominantly around the nuclear periphery (42). Given that a proportion of the Pus1 protein appears to exist in a free form and is not associated with the 26 S proteasome, we decided to examine the cellular localization of the Pus1 protein by immunofluorescence microscopy. Pus1 was found to exhibit a localization pattern identical to those seen previously for other fission yeast proteasome components (Fig. 4) in that the protein was seen around the nuclear periphery in a punctate manner.
Genetic Interactions between Pus1 and Other Proteasomal Components-Previous attempts to identify polyubiquitinbinding proteins in the 26 S proteasome have relied on the use of filter binding assays whereby the complex is separated on a denaturing gel and transferred to a filter followed by the addition of polyubiquitin conjugates or chains (25). It is possible, therefore, that other polyubiquitin-binding subunits were unable to refold properly on the filters or that a combination of subunits is required for binding and that binding activity is therefore lost when the proteasome is dissociated. We decided to see if we could gain any information regarding other possible polyubiquitin-binding subunits by looking for genetic interactions between the ⌬pus1 mutant and other mutations in fission yeast proteasomal components. If a certain mutation proved to be synthetically lethal with ⌬pus1, it might indicate that the corresponding subunit was involved in polyubiquitin binding. This might be further examined by testing whether or not the synthetic lethality can be rescued by a non-binding version of Pus1. We found genetic interactions between pus1 and three other 26 S proteasomal mutants, namely mts3-1, pad1-1, and mts4 -1. The mts3 ϩ , pad1 ϩ and mts4 ϩ genes encode the non-ATPase subunits that are homologues of the S. cerevisiae 19 S cap subunits Rpn12 (Nin1), Rpn11, and Rpn1 (Nas1), respectively (35)(36)(37)(45)(46)(47). All these mutants are temperature-sensitive and were isolated as mutants which cannot grow at 36°C; however, they are unable to grow at lower temperatures as well. In the case of mts3-1 and pad1-1 the cells die at 32-33°C whereas mts4 -1 dies at 31°C.
We found that overexpression of the pus1 ϩ cDNA was able to rescue the mts3-1 mutant at an intermediate restrictive temperature of 32°C but not at the fully restrictive temperature of 36°C (Fig. 5). In contrast, the N5 version of pus1 could not rescue. A previous study identified the RPN10 gene (also called SUN1) as being able to rescue the nin1-1 mutant in S. cerevi-siae (48). NIN1 ϩ is the orthologue of mts3 ϩ , and therefore this result was not surprising. In the previous study, however, overexpression of SUN1 was found to rescue the lethality of a nin1 disruptant strain. By way of contrast, overexpression of pus1 ϩ is unable to rescue the mts3-1 mutation at 36°C (Fig. 5) and is also unable to rescue the mts3 deletion mutant which results in complete loss of the Mts3 protein (35), indicating that these two fission yeast genes are not redundant (data not shown).
Further genetic evidence suggesting an interaction between mts3 and pus1 was provided by the finding that the mts3-1 allele and the pus1::ura4 ϩ mutant were found to be synthetically lethal at the permissive temperature of 25°C. Tetrad analysis was carried out on crosses between the pus1::ura4 ϩ and mts3-1ura4-D18 mutants. The presence of colonies with the genotype mts3 ϩ ura4 Ϫ in a tetrad, after crossing the two strains together, indicated that the tetrad also contained the pus1::ura4 ϩ mts3-1 double mutant. Non-parental ditype asci could be identified by the presence of two of the spores having the phenotype ts ϩ ura Ϫ . The remaining two spores in the tetrad would be the double mutant. In 16 NPD tetrads identified, only two of the four spores were viable. In each case the viable spores had the phenotype ts ϩ ura Ϫ ; therefore this meant that the pus1::ura4 ϩ mts3-1 double mutant was inviable and the two genes were synthetically lethal with one another. The double mutant spores were examined microscopically and were found to germinate and divide once or twice if at all (data not shown). By a similar analysis pus1::ura4 ϩ was also found to be synthetically lethal with the pad1-1 mutation and two mutant alleles of the mts4 ϩ gene, mts4 -1 (37), and mts4 -3. synthetically lethal with mutations in other proteasomal genes, namely mts2 ϩ (23) and mts1 ϩ3 which encode an ATPase subunit (Rpt2) and the non-ATPase subunit (Rpn9), respectively.
In order to try and characterize the causes of the lethality observed, the mts3-1, pad1-1, and mts4 -1 alleles were crossed to the pus1::ura4 ϩ mutant strain containing a multicopy plasmid with either the pus1 ϩ cDNA, the pus1N5 mutant cDNA, or the empty vector. The resulting tetrads were treated with glusulase to break down the walls of the asci, and the spores were plated out onto minimal media at a permissive temperature of 25°C so that the only spores that were ura ϩ and therefore carrying the pus1::ura4 ϩ mutation, and also leu ϩ and thus carrying the multicopy plasmid, would grow. The result-ing colonies were then replica-plated to the restrictive temperature of 36°C to determine if any of the colonies were temperature-sensitive (ts). The presence of such ts colonies would indicate that the corresponding plasmid was able to rescue the double mutant. The number of ts ura ϩ double mutants and pus1::ura4 ϩ single mutants that grew were counted, and the results are shown in Fig. 6A. For each cross 1000 colonies were 3 C. Gordon, unpublished results.  counted, and in no case were any ts colonies observed when the pus1::ura4 ϩ strain contained the empty vector confirming the synthetic lethality observed by tetrad analysis. In the case of pad1-1, almost half of the colonies were ts whether they carried the wild type pus1 cDNA or the pus1N5 non-binding mutant cDNA. Furthermore, the ts colonies were the same size regardless of the cDNA that they carried (data not shown) indicating that the region of Pus1 that is able to bind to polyubiquitin does not have to be intact in order to rescue the pad1-1pus1::ura4 ϩ double mutant, and therefore it seems unlikely that the synthetic lethality observed between these two mutations is a result of a decreased ability of Pus1 to bind to polyubiquitin conjugates. Similarly, the synthetically lethal combinations of mts4 -3 and ⌬pus1 were equally rescued by pus1 ϩ or pus1N5 with colonies of a similar size. Interestingly, in the case of the mts3-1pus1::ura4 ϩ double mutant, however, only a very small percentage of colonies containing the pus1N5 cDNA were ts, 3% compared with 25% of those carrying the wild type pus1 ϩ cDNA. Furthermore, the ts colonies expressing Pus1N5 were very small compared with the single mutant or the double mutant expressing the wild type pus1 ϩ cDNA (data not shown). In contrast, the pus1A5 mutated cDNA was equally able to rescue the double mutant with the colonies being the same size and in the same proportions as those containing the wild type cDNA (data not shown). These data suggested that the synthetic lethality observed between these two mutations might be due to a decrease in the ability of the proteasome to bind to polyubiquitinated substrates. Both the wild type Pus1 protein and the Pus1N5 protein were equally overproduced in the pus1::ura4 ϩ mutant cells and were expressed at levels in excess of wild type ruling out the possibility that the above data could be accounted for by differential expression of the wild type versus the mutated protein (Fig. 6B). The fact that the Pus1N5 mutant is able to complement the pad1-1pus1::ura4 ϩ or the mts4 -1pus1::ura4 ϩ double mutant equally well as the wild type cDNA also suggests that, apart from its ability to bind to polyubiquitin, this mutant protein would appear to be behaving as the wild type protein. The double mutant grew at the same rates with both cDNAs when placed at higher temperatures to try and exacerbate the effects of the pad1-1 mutation. At 30°C the pad1-1 mutant grows very slowly, and this was not affected by either version of pus1 (Fig. 6C). The version of the pus1 cDNA whereby the binding site was mutated to A5 instead of N5 and which was still able to bind to polyubiquitin conjugates was able to rescue the pad1-1, mts4 -1, and mts3-1pus1::ura4 ϩ double mutants as well as the wild type pus1 ϩ cDNA (data not shown). The mts3-1 mutant is viable at 25°C although there are consequences of this mutation even at this permissive temperature as illustrated by its ability to grow on media containing the microtubule de-stabilizing drug methyl-2-benzimidazole carbamide and by its appearance (35). The effects of the mutation are exacerbated as the temperature is raised, and at temperatures of 32°C and above, the mutant is not viable. Therefore, we tested the ability of the mts3-1pus1:: ura4 ϩ double mutant expressing either the pus1 ϩ or pus1N5 cDNAs to grow at the higher temperatures of 30 and 32°C. Overexpression of the wild type pus1 ϩ but not the pus1N5 cDNA was able to rescue the mutant at 32°C (Fig. 6D). The same result was seen at 30°C, a temperature at which the mts3-1 mutant is viable but not the double mutant.
Mts3 Does Not Bind to Polyubiquitin in Vitro-The above data could be explained by the fact that Mts3 (Rpn12) is also responsible for binding to polyubiquitin and that the inability of Pus1N5 to rescue mts3-1pus1::ura4 ϩ at 30 and 32°C, where the effects of the mts3-1 mutation will be greater than at 25°C, is due to the lack of polyubiquitin binding by Pus1N5 to bind conjugates. An alternative explanation is that Mts3 and Pus1 bind to each other in the proteasome and that the N5 mutation reduces the ability of Pus1 to bind to Mts3 especially at higher temperatures, when the effects of the mts3-1 mutation are enhanced, resulting in a decrease in the incorporation or retention of the truncated version of Mts3 in the proteasome. We investigated these possibilities. First of all, the ability of recombinant Mts3 protein to bind to polyubiquitin conjugates FIG. 6. Overexpression of pus1 ؉ but not pus1N5 can rescue mts3-1pus1::ura4 ؉ at 30 -32°C. A, results of the random spore analysis between pus1 ::ura4 ϩ and mts3-1, pad1-1  and mts4 -1. B, Western blot of fission yeast extracts prepared from wild type cells (lane 1), pus1::ura4 ϩ containing pREP3Xpus1N5 ϩ (lane 2), and pus1::ura4 ϩ containing pREP3Xpus1 ϩ (lane 3). The filter was cut in two. The lower and higher molecular mass halves were incubated with anti-Pus1 and anti-Mts4 antiserum, respectively. Molecular mass is indicated in kilodaltons. C, the pad1-1pus1::ura4 ϩ double mutant grows equally well with pus1 ϩ or pus1N5 at both 25 and 30°C. D, the pus1N5 cDNA is unable to rescue mts3-1pus1::ura4 ϩ at 30 or 32°C.
was tested in three ways, and in each case the Pus1 protein acted as a positive control. High concentrations of Mts3 and Pus1 were transferred to nitrocellulose filters, so that if only a small proportion of Mts3 was re-naturing correctly, it might be enough to bind to conjugates. These filters were incubated with 125 I-labeled polyubiquitinated lysozyme. Only the Pus1 protein bound the conjugates (data not shown). As incorrect re-naturation of proteins on nitrocellulose could have accounted for a lack of binding, the next assay utilized the GST-Mts3, GST-Pus1 fusion protein attached to glutathione-Sepharose. Radiolabeled ubiquitin, lysozyme, or lysozyme-polyubiquitin conjugates were passed over the proteins, and after washing, the beads were boiled in the presence of SDS sample buffer and the resulting proteins subjected to further analysis by Western blotting. As shown in Fig. 7A, GST-Pus1 was able to bind to polyubiquitinated lysozyme as expected but not to lysozyme or ubiquitin alone. In contrast, GST-Mts3 was unable to bind to the conjugates. It is possible that a protein might not assume its native conformation when attached to GST even though the two proteins should be well separated by a linker region. GST-Mts3 was found to be partially functional in vivo as it was able to rescue the mts3-1 mutant at 32°C but not at 36°C (data not shown). Therefore, a further assay was performed whereby GST-ubiquitin conjugates were formed so that the conjugates were immobilized on beads and then either recombinant Pus1, Pus1N5, or Mts3 were passed over. As seen in Fig. 7C, only Pus1 was able to bind to the conjugates in this assay.
Pus1 and Mts3 Are Able to Bind One Another in Vitro, and This Binding Is Not Affected by the N5 Mutation-As mentioned above, another explanation of the genetic data described above could be that Pus1 and Mts3 bind to one another in the 26 S proteasome and that disruption of the LALAL sequence required for polyubiquitin recognition in vitro also affects this binding. To test this possibility, GST, GST-Pus1, or GST-Pus1N5 were purified from E. coli and bound to glutathione-Sepharose 4B beads. GST-Mts3 was also purified from E. coli and the Mts3 released from the GST fusion by the addition of thrombin. Recombinant Mts3 appears as two bands, both of which are recognized by the Mts3 antiserum. After incubating Mts3 with the Sepharose-immobilized proteins, the beads were washed extensively and then the bound proteins eluted by boiling the beads in SDS sample buffer. These proteins were then analyzed by Western blotting. As can be seen in Fig. 8, Mts3 was able to bind to Pus1 but not to GST. Furthermore, Mts3 was also equally able to bind to the Pus1N5 protein.
These data suggest that the low viability of the mts3-1-pus1::ura4 ϩ double mutant and the inability of Pus1N5 to rescue it at 30 -32°C was not due to the inability of Pus1N5 to bind to Mts3 in the proteasome. It was not possible to test this interaction by the two-hybrid method as both Pus1 and Mts3 were found to activate transcription of reporter genes when fused to the DNA binding domain of Gal4 in the absence of the Gal4 activation domain containing plasmid (data not shown).
The mts3-1 Mutation Results in the Loss of the C-terminal 73 Amino Acids-In order to try and determine the nature of the synthetic lethality between ⌬pus1 and mts3-1, the mutation in the mts3-1 gene was sequenced. Surprisingly, the mutation was found to be a GAA to TAA transition at position 592 of the cDNA which results in the conversion of a GAA glutamine codon to a TAA stop codon. This results in a truncated Mts3 protein of 197 amino acids compared with the wild type length of 270 and was confirmed by Western blotting analysis (Fig. 9). As the mts3-1 mutant is viable at 25°C, this C-terminal region is clearly dispensable for normal growth. A further possible explanation for the synthetic lethality between ⌬pus1pREP3Xpus1N5 mts3-1 at 32°C is that the combination of the N5 mutation and the mts3-1 mutations leads to an inability of one or the other of these components to be incorporated into or retained in the proteasome. In order to test this, the ⌬pus1pREP3Xpus1N5mts-3-1 strain was grown to mid-log phase at 25°C and then placed at the restrictive temperature of 32°C for 8 h. Extracts were then prepared, and the proteasomes were partially purified by glycerol gradient centrifugation. As can be seen in Fig. 9, the proteasomes were found to be concentrated predominantly in fractions 6 -8 as indicated by the presence of the 20 S complex, Mts4 and Mts2. Pus1 was found smeared across the gradient as it was being expressed from a strong heterologous promoter on a high copy number plasmid; however, the pattern was essentially the same for both Pus1 and Pus1N5. For the truncated version of Mts3, a smaller band was seen as expected at about 21 kDa, and it was also found across the gradient with the highest amount in the lowest molecular weight fraction indicating that this truncated form does not incorporate well into the 26 S proteasome or that it falls away from the complex after assembly. The pattern was, however, the same for both the Pus1-and Pus1N5-expressing cells indicating that dissociation from the proteasome by the truncated Mts3 protein alone is unlikely to explain the synthetic lethality between ⌬pus1-pREP3Xpus1N5 and mts3-1 at 32°C. In summary, it appears that the loss of a proportion of the Mts3 protein from the 26 S proteasome in combination with the inability of Pus1 to bind to polyubiquitin is a lethal event. DISCUSSION The discovery of a subunit of the 26 S proteasome that was able to bind to polyubiquitin led to the simple model that this was the mechanism by which the complex recognized and bound to its substrates (25). The possible importance of S5a in substrate recognition was enhanced by the identification of other eukaryotic S5a proteins with binding activity and by the ability of the Arabidopsis S5a protein to inhibit ubiquitinmediated degradation in cell-free systems (34). However, the surprising discovery that the budding yeast S5a protein, Rpn10/Mcb1, was not essential for cell viability indicated that other polyubiquitin recognition factors existed (28). We were interested to see if the fission yeast Rpn10 subunit was also dispensable for growth and therefore cloned the corresponding gene. Pus1, at 243 amino acids, is the shortest Rpn10 protein cloned to date and like its S. cerevisiae orthologue is not essential for cell growth. We demonstrated that this protein had the expected properties of the fission yeast Rpn10 orthologue, namely that it co-purified with the fission yeast proteasome and that the recombinant Pus1 protein was able to bind to polyubiquitin in vitro. Furthermore, Pus1 is able to bind to tetra-ubiquitin that is linked exclusively through Lys-48 residues (data not shown). Chains formed in this manner are believed to be responsible for targeting substrates to the 26 S proteasome, and other Rpn10 orthologues have been shown to possess Lys-48 chain binding activity. Pus1, like Rpn10, has only one polyubiquitin-binding site. Higher eukaryotic orthologues, on the other hand, have a second possible site located in their extended C termini. The potential role in binding of this second site is unclear. One study with the human S5a protein demonstrated that mutating the LALAL sequence to AAAAA (A5) in the first site resulted in a moderate decrease in polyubiquitin binding, whereas conversion of the second site to A5 resulted in a dramatic decrease in binding ability. When both sites were changed to A5, binding was abolished altogether. This study also suggested that the two sites might bind polyubiquitin in a cooperative manner (31). In contrast, analysis of the Arabidopsis Mbp1 protein found that the second potential binding site did not appear to have significant polyubiquitin chain binding activity (29). When the Pus1 LALAL sequence was mutated to N5, binding was abolished as expected from previous studies on Rpn10 and Mbp1. Interestingly, changing the sequence to A5 did not abolish binding, indicating that in Pus1 at least, the hydrophobic nature of this patch is the critical feature for polyubiquitin binding rather than the alternating size of the hydrophobic residues.
The Pus1 protein exists in a low molecular mass form as well as sedimenting in the 26 S proteasomal fractions at the high molecular weight end of a glycerol gradient. This has previously been seen with the S. cerevisiae and Drosophila Rpn10/ S5a proteins and suggests that this aspect of S5a is conserved (28). One possibility is that the free S5a arises due to dissociation of the complex during purification procedures; however, during glycerol gradient centrifugation after anion exchange chromatography, all the Pus1 protein migrated with the proteasome suggesting that the additional Pus1 had been removed during the first purification step and that the low molecular mass form did not arise due to dissociation during the glycerol gradient step. Given the ability of S5a/Rpn10 to bind to polyubiquitin, a possible role for the additional free form is to shuttle substrates to the proteasome. Previously, we have found that the fission yeast proteasome is localized predominantly around the inner side of the nuclear periphery (42). Similar findings have been reported for the S. cerevisiae proteasome (49). These results suggest that this location is a major site of proteolysis in the cell and could necessitate the use of shuttling proteins to bring substrates to the complex. We were therefore very interested to examine the localization of the Pus1 protein in fission yeast cells to see if the cellular distribution of Pus1 was different to the rest of the proteasomal components. The Pus1 protein, however, displayed an identical localization pattern as the proteasome. If the extra Pus1 was distributed nonspecifically throughout the cell, it would not be obvious from this type of analysis, and therefore cell fractionation techniques might be required to determine where the additional Pus1 is and if indeed it is present in the non-stoichiometric amounts, with respect to the other proteasomal subunits, that it appears to be. If Rpn10/Pus1 does shuttle substrates to the proteasome, there will need to be other factors involved given the non-essential nature of this gene. Possible candidates might be expected to display genetic interactions with the rpn10/pus1 null mutants. Therefore, it is of interest that loss of the proteasome-interacting factor Rad23 in combination with loss of Rpn10 leads to increased canavanine sensitivity and an increase in polyubiquitinated substrates (50). Other possible candidates include de-ubiquitinating enzymes and E4 factors as they obviously possess the necessary ability to recognize polyubiquitin chains. One such DUB enzyme, Doa4, has recently been shown to interact with the proteasome (51). A mutant version of the E4-encoding gene UFD2, in combination with the rpn10 null strain, was found to be hypersensitive to conditions known to induce increased substrates for the ubiquitin-mediated proteo- lytic pathway (7). Further genetic analysis using the ⌬rpn10/ ⌬pus1 strains may well identify more candidates.
The pus1 ϩ gene is not essential for cell viability, a finding that confirmed the surprising RPN10(MCB1) results. All other fission yeast proteasomal subunits cloned to date are essential, and the mutants display a mitotic arrest phenotype presumably reflecting the requirement for the degradation of certain key cell cycle substrates. There are, however, consequences of losing pus1. As with the rpn10 null strain, the cells are sensitive to amino acid analogues such as canavanine and also display a mild accumulation of polyubiquitin conjugates (data not shown). Rescue of the canavanine sensitivity did not require an intact polyubiquitin-binding site so these results and those described previously suggested that the polyubiquitinbinding site has no in vivo relevance. Recently, the gene encoding the Rpn10 orthologue in the moss Physcomitrella patens was cloned and deleted (52). Interestingly, although the strain was viable, it was developmentally abnormal, generating caulonema that were unable to form buds and gametophores. A mouse Rpn10 homologue has also now been cloned, so it will be interesting to see if there are any developmental abnormalities associated with a null phenotype in this case and also to assess if developmental functions are associated with the increased size of the higher eukaryotic Rpn10 proteins (53). Despite the considerable increase in size, it is clear that there is functional conservation among the Rpn10 orthologues by the ability of the human S5a gene to rescue the canavanine sensitivity of the pus1 null mutant, as illustrated here and other previous studies.
The ability of pus1 ϩ to rescue the mts3-1 mutant at 32°C when overexpressed is probably due to the fact that the two subunits bind to one another in the proteasome. Most likely is that overexpression of pus1 ϩ helps to force the physical interaction between these two proteins thus increasing the incorporation of the mutant version of Mts3 into the complex. This close physical association is also apparent from studying the mts3-1 mutant at 25°C, and as shown in this case, the Pus1 protein smears across a glycerol gradient more than in wild type extracts (data not shown). The fact that we have shown that the two proteins can interact in vitro substantiates this idea. Furthermore, we have determined the nature of the mts3-1 mutation and shown that it results in the deletion of more than a quarter of the protein. As the mts3-1 mutant is viable at 25°C, this C-terminal part of the protein is dispensable for cell viability but appears important in mediating incorporation into the complex as increasing the temperature results in an increasing smearing of the protein across a glyc-erol gradient with the highest amount being in the low molecular weight fraction. Previously, it has been shown that at 36°C, this mutation results in the truncated version of Mts3 being totally lost from the proteasome. One reason why overexpression of pus1 is unable to rescue the mutant at 36°C could be that the binding between the two proteins is greatly reduced at this higher temperature and that in this case it is not possible to force the equilibrium of binding sufficiently in favor of the direction in which the truncated Mts3 will become incorporated into the proteasome. Also, overexpression of pus1 ϩ does not rescue a mts3 null mutant indicating that these two genes are not functionally redundant. This is in contrast to a previous study where overexpression of the S. cerevisiae RPN10 gene (also known as SUN1) was found to rescue a rpn12 (nin1) disruptant. RPN12 is the homologue of mts3 (48). This difference is intriguing although it is not clear if there was a complete loss of RPN12 in the disruption constructs. If partial RPN12 sequence remained and gave rise to a partial Rpn12 protein, overexpression of RPN10 might rescue a truncated version.
The inability of the N5 version of Pus1 to rescue the ⌬pus1mts3-1 or just the mts3-1 mutant (data not shown) at 32°C is intriguing. Presumably the N5 mutation doesn't cause any great perturbation in structure as it is able to rescue the canavanine sensitivity of the deletion strain and also the ⌬pus1pad1-1 and ⌬pus1mts4 -1 double mutants as well as the wild type Pus1 protein. Furthermore, the Pus1N5 protein is equally able to bind to Mts3 in vitro and also to the truncated version of Mts3 (data not shown). Moreover, the distribution of the C-terminally truncated Mts3 is identical in gradients made from cells overexpressing either the wild type or N5-mutated pus1 cDNA as is the distribution of these two versions of Pus1. If the LALAL region were involved in direct binding to Mts3, then mutating it to N5 would be expected to disrupt this binding and alter the gradient profiles. This was not observed. Also, one might expect in the case of direct binding involving LALAL that the addition of increasing amounts of Mts3 to Pus1 in vitro before the addition of polyubiquitin would diminish binding of the chains, but this was also not observed (data not shown). So what is the reason behind the failure of Pus1N5 to rescue mts3-1? One possible explanation is that Mts3 is another polyubiquitin-binding protein. However, our in vitro assays failed to detect any activity. This could be due to improper folding of the recombinant protein or perhaps Mts3 has to be assembled into the complex before assuming the right conformation to bind polyubiquitin. Alternatively, perhaps, the combined loss of Mts3 from the complex at 32°C in combination FIG. 9. Glycerol gradient sedimentation analysis of ⌬pus1mts3-1 with either pREP3Xpus1 ؉ or pREP3Xpus1N5 at 32°C. ⌬pus1mts3-1 cells transformed with either pREP3Xpus1 ϩ or pREP3Xpus1N5 were grown to mid-log phase at 25°C and then shifted to 32°C for 8 h. Extracts were then made and subjected to sedimentation on a 10 -40% glycerol gradient. Equal amounts of protein from alternate fractions were separated by 10 -15% SDS-PAGE and transferred to nitrocellulose filters. These were then probed with antiserum as indicated. Fraction numbers are indicated at the top of the panel and molecular mass is indicated at the right-and side. Fractions 2 and 18 correspond to the high and low molecular weight fractions, respectively. with a version of Pus1 that cannot bind to polyubiquitin results in too little ubiquitin-mediated protein degradation, and therefore the cells die. The recently developed strategy to produce sub-complexes of the proteasome should help in determining exactly what role certain subunits and combination of subunits play in the mechanism of degradation (33).
The observed synthetic lethality between ⌬pus1 and pad1-1 and mts4 -1 was equally rescued by the wild type and N5 versions of the Pus1. A possible explanation for these effects is that Pus1 also binds to these two proteins in the 19 S cap, although overexpression of pus1 ϩ was not able to rescue either the pad1-1 or the mts4 -1 mutations at intermediate restrictive temperatures (data not shown). The base sub-complex of the 19 S cap was found to contain the six ATPases, believed to form a ring structure, and the non-ATPases Rpn10, Rpn1 (Mts4), and Rpn2. We have previously shown that Mts4 binds to the ATPase Mts2 (Rpt2); therefore, if Mts4 also binds to Pus1, this would be consistent with the composition of the base (37). Interestingly, the human orthologue of Mts4 (S2) contains a LALAL sequence, although in Mts4 this is LALGL, and in the S. cerevisiae protein, Rpn1, it is LAMGI. The Pad1 protein and its homologues displays homology to the isopeptidase class of de-ubiquitinating enzymes, although our attempts to demonstrate that either Mts4 or Pad1 recombinant proteins bind polyubiquitin in vitro have not been successful (data not shown). If these proteins were involved in binding polyubiquitin then it might be possible to isolate other alleles in these genes, which in combination with ⌬pus1 cannot be rescued by the N5 version. In summary, we have confirmed the surprising Rpn10 results in fission yeast by showing that the pus1 ϩ gene is non-essential and that its ability to bind to polyubiquitin is not required to rescue sensitivity to canavanine. More importantly, however, we have demonstrated for the first time that there is a role for the LALAL polyubiquitin binding motif in vivo, and we have shown that, in combination with a compromised version of Rpn12/Mts3, it is essential.