The ubiquitin/proteasome pathway plays a key role in regulating cell cycle progression. Previously, we reported that a conditional mutation in the Saccharomyces cerevisiae gene RPT4/PCS1, which encodes one of six ATPases in the proteasome 19S cap complex/regulatory particle (RP), causes failure of spindle pole body (SPB) duplication. To improve our understanding of Rpt4p, we created 58 new mutations, 53 of which convert clustered, charged residues to alanine. Virtually all mutations that affect the N-terminal region, which contains a putative nuclear localization signal and coiled-coil motif, result in a wild-type phenotype. Nine mutations that affect the central ATPase domain and the C-terminal region confer recessive lethality. The two conditional mutations identified, rpt4-145 and rpt4-150, affect the C terminus. After shift to high temperature, these mutations generally cause cells to progress slowly through the first cell cycle and to arrest in the second cycle with large buds, a G2 content of DNA, and monopolar spindles, although this phenotype can vary depending on the medium. Additionally, we describe a genetic interaction between RPT4 and the naturally polymorphic gene SSD1, which in wild-type form modifies the rpt4-145 phenotype such that cells arrest in G2 of the first cycle with complete bipolar spindles.
SELECTIVE proteolysis via the ubiquitin/proteasome pathway is fundamental to eukaryotic cell cycle progression (Kinget al. 1996; Koeppet al. 1999; Tyers and Jorgensen 2000). Protein degradation provides a rapid, irreversible method of inactivating regulatory molecules, including cyclins, cyclin-dependent kinase inhibitors, and anaphase inhibitors, at the appropriate time in the cycle. Ubiquitination (Ciechanover 1998; Hershko and Ciechanover 1998) targets proteins for degradation by 26S proteasomes, which are large, self-compartmentalizing proteases found in the cytoplasm and nucleus (Vogeset al. 1999; Zwicklet al. 2000). Each proteasome is assembled from a barrel-shaped 20S core particle (CP) and one or two 19S regulatory particles (RPs) linked to the ends of the CP. Proteolytic activity is sequestered inside the CP, and ubiquitinated substrates are believed to be recognized, unfolded, and translocated into the CP by the RP in an ATP-requiring process.
The RP contains ∼18 different subunits, 6 of which are ATPases (Ferrellet al. 2000). The subunits have been given a unified nomenclature that designates the 6 putative ATPases as Rpt1-6 (for regulatory particle triple-A) and the remaining subunits as Rpn1-12 (for regulatory particle non-ATPase; Finleyet al. 1998). The RP consists of two subcomplexes, the base and the lid (Glickmanet al. 1998a). Each RP base from Saccharomyces cerevisiae contains all 6 Rpt subunits assembled into a heteromeric complex (Glickman et al. 1998a,b), as well as Rpn1p, Rpn2p, and Rpn10p, and appears to bind directly to the CP. The lid, which is built of the remaining subunits, is located distal to the base. A chaperone-like ability to refold a denatured protein is exhibited by the base in vitro, suggesting that it could function in vivo to unfold substrates and promote their translocation into the CP (Braunet al. 1999). The base is also believed to play a role in opening the CP channel, which is blocked in yeast by portions of CP subunits (Groll et al. 1997, 2000).
The six ATPases appear to have distinct functions, since null mutations in each of the corresponding genes are lethal and the lethality is not suppressed by overproduction of other Rpt subunits (Ferrellet al. 2000). Experiments involving selective chemical inactivation of Rpt4/Sug2p and Rpt6/Sug1p support the idea that these subunits are nonredundant (Russellet al. 2001), as does the finding that analogous mutations in the ATP-binding regions of the six Rpt subunits produce distinct phenotypes (Rubinet al. 1998). Interestingly, Rpt2p appears to play a unique role in gating the CP channel (Köhleret al. 2001). Additionally, whereas rpt6/cim3 and rpt1/cim5 temperature-sensitive (ts) mutations cause cells to arrest at G2/M with a short bipolar spindle (Ghislainet al. 1993), a ts rpt4/pcs1td mutation causes cells to arrest with a monopolar spindle (McDonald and Byers 1997). Thus, we previously hypothesized that Rpt4p plays a specific role in the duplication of the spindle pole body (SPB), which functions as the centrosome equivalent and serves to organize microtubule arrays in yeast (Byers 1981a).
SPB duplication is a key event in the cell cycle, essential for the formation of a bipolar spindle. Electron microscopy (EM) reveals that the SPB is a trilaminar disk embedded in the nuclear envelope throughout the cell cycle; microtubules emanate from its outer and inner surfaces into the cytoplasm and nucleus, respectively (Byers 1981a). The single SPB that is present at the beginning of the cell cycle must duplicate to generate the two poles of the bipolar spindle. The first visible sign of duplication is the appearance of the satellite early in G1. This structure appears as an aggregate of material on the cytoplasmic surface of the half-bridge, a darkly staining region of the nuclear envelope adjacent to the extant SPB. The satellite apparently matures into a larger structure, the duplication plaque (Adams and Kilmartin 1999), which becomes embedded in the nuclear envelope as a second, mature SPB later in G1. rpt4/pcs1td cells fail to form a satellite at restrictive temperature, perhaps because they contain an abnormal half-bridge structure (McDonald and Byers 1997). Thus, these cells exhibit a failure of SPB duplication at the earliest known stage of the process. Since other early landmarks of the cell cycle (bud formation and DNA synthesis) are not prevented, we hypothesized that the Rpt4/Pcs1 subunit is specifically required for the proteolysis of substrate(s) that must be degraded for SPB duplication to occur (McDonald and Byers 1997).
To investigate the structure and function of Rpt4p further, and to determine whether additional rpt4 mutations also affect SPB duplication, we systematically mutagenized RPT4 and analyzed the resulting phenotypes. Primarily, we converted clustered, charged amino acids to alanine. Since such residues are likely to be found at the surface of a protein, and since the alanine side chain is small and uncharged, this method can be an effective way of subtly perturbing surface domains and protein interactions without grossly distorting protein folding (e.g., Cunningham and Wells 1989; Basset al. 1991; Bennettet al. 1991; Wertmanet al. 1992; Reijoet al. 1994). We generated 53 charged-to-alanine mutations, as well as several others suggested by analysis of the Rpt4p sequence. Our results reveal an important role for the C terminus in Rpt4p function and indicate that much of the N terminus is dispensable. Phenotypic analysis of the two conditional rpt4 mutants recovered from this screen reveals roles for RPT4 in other aspects of early cell cycle progression, in addition to SPB duplication.
MATERIALS AND METHODS
Strains and culture techniques: All strains were of the S288C genetic background. Allele replacement was done in the haploid yeast strain Wx257-5c (MATa ura3-52 leu2-3, 112 trp1Δ his3Δ ssd1-d, originally constructed by Mark Winey). Strains containing the rpt4-145 and rpt4-150 alleles were crossed to Wx257-7b (MATα ura3-52 leu2-3, 112 trp1Δ ssd1-d) to obtain MATα rpt4-145 and MATα rpt4-150 strains, which were in turn crossed to BY4741 (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 SSD1; Brachmannet al. 1998) to move the rpt4 alleles to an SSD1 background. Standard yeast genetic methods were used (Mortimer and Hawthorne 1969) and yeast media were made as described (Hartwell 1967; Ausubelet al. 1994). 5-Fluoroorotic acid (5-FOA) was purchased from Toronto Research Chemicals (North York, Ontario) and cycloheximide was purchased from Sigma (St. Louis). For cell synchronization, cells were arrested with 10 μm α-factor (Sigma) and released as described (Wineyet al. 1991). Bacterial strains DH5α, XL10-Gold (Stratagene, La Jolla, CA), and TOP10 (Invitrogen, Carlsbad, CA) were used for plasmid construction; bacterial media were made as described (Sambrooket al. 1989).
Plasmid construction: Plasmid pHM59 was constructed for use as a template for RPT4 mutagenesis. It bears a 1.9-kb fragment of genomic DNA that contains the RPT4 coding sequence, with a selectable marker (TRP1) inserted into the RPT4 downstream flanking sequence. As a first step toward pHM59 construction, TRP1 was inserted into the Wx257-5c genome 93 bp downstream of the RPT4 stop codon using a hybrid-oligonucleotide polymerase chain reaction (PCR)-based method (Baudinet al. 1993). TRP1 was amplified from pRS314 (Sikorski and Hieter 1989) using forward primer HM-100 5′ TGCGAGAAAGCTTGCCTTATTTAAGCTCGAAA TGGCTATGGAAGCATTTAATAGAACAG 3′, reverse primer HM-99 5′ AAATTTAGCTTATTGAAAAAGCATTTAGAACTA GAGTTCACAGGCAAGTGCACAAACA 3′, and the eLON-Gase enzyme mix (Invitrogen Life Technologies). Underlined sequences anneal to TRP1; the nonunderlined portions represent sequences that flank the site 93 bp downstream of the RPT4 stop codon. The resulting product, a 975-bp TRP1-containing fragment flanked by RPT4 downstream sequences, was used to transform strain Wx275-5c to Trp+ by the Li-acetate method (Gietzet al. 1992). Amplification of genomic DNA derived from Trp+ transformants using primers HM-50 (5′ CACTCGAGGAGCTATCAACGGTGGGAACT 3′, which anneals 255 bp upstream of the RPT4 start codon) and HM-51 (5′ CACTCGAGGTTGGGATACACAGCGTGCC 3′, which anneals 373 bp downstream of the RPT4 stop codon), indicated that the fragment had integrated at the RPT4 locus in one case. [Insertion of TRP1 at this position relative to RPT4 in the Wx257-5c genome, to create strain YHM101, does not detectably affect yeast viability or growth characteristics (Figure 1, section 2).] The 2.9-kb PCR product was inserted into pCR4-TOPO (Invitrogen) to create pHM59. The TRP1-marked RPT4 gene can be excised as a single fragment from the vector using XhoI (site underlined in primers HM-50 and HM-51 above). Restriction enzymes were purchased from New England Biolabs (Beverly, MA).
Site-directed mutagenesis and construction of deletion mutants: The desired mutations in RPT4, together with silent mutation(s) that either create or destroy a restriction site at or very near the codon-altering mutation(s), were generated in pHM59 using the Stratagene QuickChange method. Site-directed mutagenesis was performed according to the manufacturer’s instructions, except that oligonucleotide primers were designed to have a melting temperature of ∼70°. Oligonucleotides were synthesized by MWG Biotech (High Point, NC), Sigma-Genosys (The Woodlands, TX), and Integrated DNA Technologies (Coralville, IA); sequences are available upon request. Eight of the alleles were made by remutagenizing a previously constructed allele such that two nearby sets of mutations were combined.
Two deletions were constructed in RPT4. To facilitate their construction, a BamHI site was engineered at position 15 (relative to the first nucleotide in the start codon) in RPT4. To create the first deletion (rpt4-1Δ), a naturally occurring BamHI site at position 138 was utilized. Following digestion with BamHI to release the appropriate fragment, the plasmid was reclosed such that an in-frame deletion in RPT4 was created. To create the second, larger deletion (rpt4-2Δ), a BglII site in RPT4 at position 246 was utilized, together with the engineered BamHI site, in the same manner.
Sequence analysis: DNA sequence reactions were performed using the Taq DyeDeoxy Terminator cycle sequencing kit or the BigDye Terminator cycle sequencing kit (both from Applied Biosystems, Foster City, CA) and were analyzed on an ABI 310 automated sequencer (Colgate University) or on an ABI 377XL automated sequencer (University of Washington Molecular Pharmacology Facility).
Allele replacement: Following site-directed mutagenesis, plasmids were digested with XhoI to excise the TRP1-marked RPT4 fragment. The digested DNA was transformed into strain Wx257-5c containing pHM49, a URA3-marked CEN plasmid that bears the RPT4 coding region and sufficient flanking sequence (93 bp upstream and 102 bp downstream) to supply RPT4 function in haploid strains containing a disrupted genomic copy of the gene (McDonald and Byers 1997). Genomic DNA was isolated from Trp+ transformants and the RPT4 locus was amplified using primers HM-279 (5′ CAGCTGTA CTGTCTTTGTTC 3′) and HM-280 (5′ GATTGACCACATCT TTAGGC 3′), which lie 103 bp upstream and 104 bp downstream of the transformed fragment’s endpoints, respectively. A 3.2-kb product is obtained if the TRP1-marked fragment inserts into the RPT4 locus, whereas a 2.2-kb product is obtained if the RPT4 locus is unmodified. Between 20 and 90% of Trp+ transformants were generally found to incorporate the fragment at the RPT4 locus. Restriction analysis of PCR-amplified fragments and, in some cases, sequence analysis were then performed to test for the presence of the desired mutation(s) in the genome.
Phenotypic analysis of rpt4 mutants: To determine the effects of the inserted mutant rpt4 alleles in the absence of wild-type RPT4, strains were plated on media containing 5-FOA to select for loss of the URA3-marked pHM49 plasmid at 25° (Boekeet al. 1984). An inability to obtain 5-FOA-resistant colonies indicated that the respective mutations cause a recessive lethal (RL) phenotype. If 5-FOA-resistant colonies were obtained, they were tested on YEPD media for cold or heat sensitivity at 10° or 37.5°, respectively. At least two independent transformants were tested in this manner for each mutant allele, and following these analyses, the presence of the desired mutation in the genome was reconfirmed in each case. In two cases (rpt4-132 and rpt4-149), however, Trp+ transformants containing the constructed mutations were never recovered, although in each case 24 candidates with TRP1 inserted downstream from RPT4 were analyzed for the presence of the desired mutations. These mutations therefore may cause dominant lethality. To determine whether alleles causing reduced growth or temperature sensitivity were dominant or recessive, relevant strains were retransformed with pHM49 and tested for growth at 30° and 37.5°.
GFP-Rpt4-150 expression: To investigate whether the ts rpt4-150 mutation affects the level of Rpt4p at restrictive temperature, a LEU2-marked CEN plasmid (pHM54) that encodes a green fluorescent protein (GFP)-Rpt4p fusion was used (McDonald and Byers 1997). The expression of this fusion is driven by the RPT4 promoter. pHM54 was transformed into haploid yeast strain YHM10.1.49 (McDonald and Byers 1997), which contains a disrupted chromosomal copy of RPT4 and is made viable by the presence of pHM49. Leu+, 5-FOA-resistant transformants (which have lost pHM49) were found to be capable of growth at both 30° and 36.5°, indicating that the GFP-Rpt4p fusion supplies Rpt4p function at both temperatures.
The rpt4-150 mutations were reconstructed in the pHM54 RPT4 sequence to create pAH66. To test the functionality of this GFP-Rpt4-150 protein fusion, pAH66 was transformed into strain YHM10.1.49. In this case, Leu+ transformants were found to be capable of growth at 30° but not at 36.5° on medium containing 5-FOA, indicating that the GFP-Rpt4-150 fusion provides Rpt4p function at low but not high temperature. Therefore, the plasmid-borne construct encoding the GFP-Rpt4-150 fusion behaves similarly to the chromosomal rpt4-150 allele, leading to a restrictive temperature 1° lower than that of the non-GFP-fused, chromosomal allele.
Cytological techniques and Western analysis: Cells were prepared for immunofluorescence microscopy as described (Kilmartin and Adams 1984; Jacobset al. 1988). Microtubules were visualized using the DM1A mouse monoclonal anti-α-tubulin antibody (Sigma) and fluorescein-conjugated goat anti-mouse IgG (Molecular Probes, Eugene, OR), both at 1:100 dilution. DNA was visualized with 4′,6-diamidino-2-phenylindole (DAPI) at 1 μg/ml (Sigma).
Yeast cell extracts for Western analysis were prepared as follows. Cells derived from 10-ml cultures at a density of 4 × 107 cells/ml were harvested, rinsed, and resuspended in 400 μl loading buffer (125 mm Tris-Cl, pH 6.8, 20% glycerol, 4% SDS, 2% β-mercaptoethanol, 0.1% bromophenol blue) containing 2 mm Pefabloc SC (Boehringer Mannheim, Indianapolis). A total of 400 μl of glass beads (Biospec Products, Bartlesville, OK) were added to each sample, which were then alternately boiled and vortexed three times for 2 min and 30 sec, respectively. After centrifugation for 30 sec to pellet insoluble material, volumes of supernatant containing equivalent amounts of total protein (5-10 μl) were subjected to SDS-PAGE (Laemmli 1970) using a 6% polyacrylamide gel. Following transfer to nitrocellulose, GFP fusion proteins were visualized using anti-GFP antiserum (Invitrogen) followed by horseradish peroxidase-conjugated goat anti-rabbit IgG (Zymed, South San Francisco, CA), both at 1:2500 dilution. The signal was visualized using the enhanced chemiluminescence system (Amersham, Piscataway, NJ).
Overview: To improve our understanding of Rpt4p, we generated 58 novel rpt4 alleles and analyzed the resulting phenotypes. Fifty-three of these alleles were con-structed such that clustered, charged amino acids (Arg, Asp, Glu, His, and Lys) were converted to alanine; between one and four charged residues were mutated per variant. Several additional point mutations and deletions were made in regions of the protein with predicted structural features or functional roles (see below). The mutations are summarized in Table 1.
Each mutant allele, marked with TRP1, was used to replace the wild-type RPT4 gene in haploid strain Wx257-5c, which also contained a wild-type copy of RPT4 on a URA3-marked plasmid. After gene replacement at the RPT4 locus had been verified, we selected for loss of this plasmid on medium containing 5-FOA and determined the resulting phenotype (see materials and methods). We recovered 51/53 of the charged-to-alanine mutant alleles in vivo (the remaining two may cause dominant lethality; see materials and methods). Nine of the alleles cause RL phenotypes, whereas most (34) of the remaining alleles appear to result in wild-type (WT) phenotypes. Figure 1, section 3, shows rpt4-129, a representative WT strain, compared to the parent, Wx257-5c (Figure 1, section 1), or to its WT derivative, YHM101, which has incorporated the RPT4::TRP1 fragment at the RPT4 locus (Figure 1, section 2). Six alleles cause reduced growth (RG), especially at elevated temperature (37.5°); representative strains rpt4-123 and rpt4-137 are shown in sections 4 and 5. Two strains, rpt4-145 (Figure 1, section 6) and rpt4-150 (Figure 1, section 7), are ts, but none were found to be cold sensitive. Finally, because RPT4 was originally identified as the cycloheximide-resistant, ts lethal allele crl13 (McCusker and Haber 1988; section 8 in our Figure 1), we tested each of the strains for resistance to 1 μg/ml cycloheximide; none were found to be resistant (data not shown).
Mutational analysis of the N terminus: The N-terminal region of Rpt4p contains a very hydrophilic region (residues 13-96 are 49% charged, 77% charged and polar), a putative bipartite nuclear localization signal (NLS; residues 62-78), and a region (residues ∼69-101) that is predicted to participate in the formation of a short α-helical coiled-coil (Figure 2). We hoped to gain insight into the functional significance of these features with our mutational analysis.
We constructed a total of 21 point mutations or clusters of mutations in the N-terminal region (Table 1) and determined that 20 of them resulted in no detectable mutant phenotype. (The rpt4-107 allele confers a mild RG phenotype.) The ability of Rpt4p to function despite these mutations suggests that the N-terminal region, which represents the most divergent region of Rpt4p relative to other Rpt subunits (Figure 3), might be dispensable for function. To test this idea, we created two alleles, rpt4-1Δ and rpt4-2Δ, which are deleted for residues 6-46 and residues 6-82, respectively. Strains containing these alleles again display a WT phenotype, indicating that these residues play no essential role in Rpt4p function. As well as deleting a significant portion of the hydrophilic region, rpt4-2Δ also removes the potential NLS and half of the putative coil region.
Mutational analysis of the ATPase domain: The six putative ATPases of the proteasome RP are members of the AAA protein family, a group of ATPases that contain Walker A and B motifs characteristic of ATP-binding domains (Walkeret al. 1982) within an extended, conserved region of ∼230-250 residues known as the AAA cassette (Kunauet al. 1993; Beyer 1997). The region of highest sequence identity [∼60% in pairwise comparisons using the BLAST alignment program (Altschulet al. 1990)] in the Rpt subunits corresponds to regions D-H of the AAA cassette (Beyer 1997) and is defined by residues 213-352 in Rpt4p (Figure 3). Mutations in this region are much more likely to be deleterious than mutations in the N-terminal region. Of the 14 recovered charged-to-alanine mutations in the central region, 5 were determined to be RL. (An additional RL allele, rpt4-125, affects residues that lie in region A of the AAA cassette.) Four of the RL alleles affect residues either within or very near Walker motifs A and B (Figure 2). rpt4-132, which also affects residues located very near Walker motif B, represents one of the two alleles we were not able to recover in vivo, suggesting that it may exert a dominant lethal (DL) effect. None of the mutations we constructed in this region resulted in a ts phenotype, but we sequenced the previously identified crl13 allele (McCusker and Haber 1988) and determined that it contains two changes (G13R and L231R) compared to wild type. Given that residues 6-82 can be deleted from Rpt4p without discernible effect, the ts effect of the crl13 allele most likely derives from the L231R mutation. This conserved leucine residue is separated from Walker motif A by one residue. The remaining 9 mutations in the central region caused either RG or WT phenotypes (Table 1). Many of these mutations affect conserved residues (Figure 3).
Mutational analysis of the C terminus: We define the C terminal region of Rpt4p as residues 353-437 because this portion shows reduced conservation compared to the central ATPase region [∼31% identity in pairwise BLAST comparisons (Altschulet al. 1990)] among Rpt subunits (Figure 3). Although it contains no previously characterized motifs or predicted structural features, mutations we generated reveal that the C-terminal region of Rpt4p is quite important for its function. One of the two potential DL alleles (rpt4-149) that we were unable to recover in vivo affects residues in this region. Of the 14 recovered mutations, 3 were RL (rpt4-142, rpt4-143, and rpt4-152); all of these affect at least one conserved residue (Table 1, Figure 3). One allele (rpt4-144) causes a RG phenotype. Additionally, the only ts lethal mutations generated by this mutagenesis are found in this region: rpt4-145 (D375A, E377A, D390A, R392A) and rpt4-150 (D406A, R407A). The rpt4-145 strain has a restrictive temperature (36°) slightly lower than that of rpt4-150 (37°-37.5°). The remainder of our analyses focused on these two alleles.
We wished to determine whether the ts mutants rpt4-145 and rpt4-150 exhibit defects similar to our previously characterized rpt4 mutant (rpt4/pcs1td), which displays a specific failure in SPB duplication. Primarily in the second cycle after a shift to high temperature, the rpt4/ pcs1td strain arrests as large-budded cells with monopolar spindles and duplicated, unsegregated DNA (McDonald and Byers 1997). To ascertain whether the rpt4 ts alleles cause arrest at a specific stage of the cell cycle, asynchronous liquid cultures of rpt4-145 and rpt4-150 were shifted to 36° or 37.5°, respectively. After ∼7 hr at restrictive temperature, ∼80% of cells from both strains were found to be large budded, compared to 25-30% for the cultures grown at 30° (data not shown). Experiments using synchronous, α-factor-arrested cultures released into fresh YM-1 media (a buffered version of YEPD) at 37° revealed that the majority of cells arrest in the second cycle after the temperature shift.
Potential genetic interaction with SSD1: Further characterization of the rpt4 cell cycle arrest revealed different cell cycle behaviors (see below), depending on the strain background. One possible explanation for this observation is that the rpt4 phenotypes might vary depending on the state of the naturally polymorphic gene SSD1 (Suttonet al. 1991), which has been shown, in wild-type form, to suppress numerous different mutations (discussed in Doseff and Arndt 1995; Uesonoet al. 1997; Kaeberlein and Guarente 2002), indicating involvement in a variety of processes. For example, SSD1 suppresses the lethality of a cln1 cln2 double mutation in a W303 strain background, which carries an ssd1-d allele (Cvrcková and Nasmyth 1993). To determine which SSD1 allele is present in the rpt4 strains, we amplified and sequenced the gene and found a C2094G mutation in SSD1, which converts Tyr698 to a stop codon. The premature truncation occurs in approximately the middle of the 1250-residue wild-type Ssd1 protein and is in accordance with the finding that an 83-kD protein is detected with anti-Ssd1p antibodies in W303 (ssd1-d2), whereas an essentially full-length protein is ∼160 kD (Uesonoet al. 1997). We therefore conclude that the rpt4 strains carry an ssd1-d allele. For comparative purposes, we created rpt4-145 SSD1 and rpt4-150 SSD1 strains and performed analyses on both sets of strains in parallel, as described below.
Flow cytometry analysis: To determine whether DNA synthesis is complete at the time of rpt4-145 and rpt4-150 arrest, we performed flow cytometry experiments. Cells were initially grown at 30°, arrested in G1 with α-factor, and then released into fresh media (YEPD) without α-factor at 37°. Cytometry profiles taken at time points after the temperature shift are shown in Figure 4A and indicate that both rpt4-145 and rpt4-150 arrest with a 2N (G2) content of DNA, regardless of the SSD1 allele present in the strain (6 hr at restrictive temperature, columns b, c, and d in Figure 4A, and data not shown). The kinetics of DNA synthesis and budding differed for the rpt4 strains, however, depending on the SSD1 allele present. Both rpt4-145 ssd1-d (column d in Figure 4A) and rpt4-150 ssd1-d (not shown) display a significant lag phase prior to beginning DNA synthesis; this lag phase is completely (rpt4-145, column c in Figure 4A) or nearly (rpt4-150, column b in Figure 4A) abolished in the presence of SSD1.
Because SSD1 interacts genetically with genes involved in the protein kinase C pathway (see Kaeberlein and Guarente 2002 and references therein), which functions in maintaining cellular integrity in response to low osmolarity (Heinischet al. 1999), we tested whether osmotic stabilization would affect the behavior of rpt4 ssd1-d strains. Interestingly, the addition of 0.5 m KCl to the growth medium resulted in DNA synthesis with essentially wild-type kinetics for both rpt4-145 ssd1-d (column e in Figure 4A) and rpt4-150 ssd1-d (not shown), suggesting that osmotic constraints indeed caused the lag phases. Similarly, the rpt4 ssd1-d strains display a delay in budding and bud growth after release from α-factor compared to ssd1-d, SSD1, or rpt4 SSD1 strains (but similar budding behavior compared to an rpt4-145 ssd1::LEU2 strain; not shown), which can be rescued by the addition of 0.5 m KCl to the growth medium (Figure 4B and not shown). Finally, we note that the presence of the ssd1-d allele in the rpt4 strains is also associated with decreased viability after shift to high temperature (Figure 4C and not shown).
Cytological analysis: To examine the appearance of the microtubule cytoskeleton and DNA in the ts strains held at restrictive temperature, we visualized these structures using fluorescence microscopy. After ∼7 hr at restrictive temperature, the DNA was unsegregated and localized adjacent to the bud neck in ∼80% of the large-budded cells for both rpt4-145 ssd1-d and rpt4-150 ssd1-d (Figure 5, B and C). Large-budded wild-type cells (parental strain Wx257-5c) with DNA localized adjacent to the bud neck contain short intranuclear spindles (Figure 5A), visualized as a bright band of tubulin staining running across the nucleus. Short intranuclear spindles are characteristic of cells in G2 phase or mitotic metaphase. In contrast, the large-budded rpt4-145 or rpt4-150 cells rarely appeared to contain such spindle structures; few, if any, intranuclear microtubules were visible. Instead, a bundle of cytoplasmic microtubules was generally seen to extend away from the nucleus (Figure 5, B and C). Such a staining pattern is reminiscent of the monopolar spindles formed in the rpt4/pcs1td mutant (McDonald and Byers 1997).
To determine whether the rpt4-145 and rpt4-150 mutations indeed result in a monopolar arrest, we used serial section EM to examine the effects of these mutations in both ssd1-d and SSD1 backgrounds. In all cases, cells were initially synchronized by arrest with α-factor, after which they were released into fresh media at the restrictive temperature. We initially examined these strains after growth in YM-1 media, which generally results in better EM fixation; later experiments were done using YEPD after we observed differences in the arrest depending on the medium used.
Both rpt4-150 ssd1-d and rpt4-150 SSD1 cells arrest with monopolar spindles: When grown in either YEPD or YM-1, rpt4-150 ssd1-d cells primarily arrest as large-budded cells during the second cell cycle (as determined by monitoring budding) after shift to high temperature. The cells do not reach G2 of the second cell cycle until 6-8 hr after the temperature shift, whereas wild-type cells reach that point by 2.5-3 hr after the shift. After 7 hr at 37.5° in YM-1, we found that the majority (33/42 or 79%) of the rpt4-150 ssd1-d cells had monopolar spindles (Figure 6A), indicating that the process of SPB duplication had failed. For comparison, a normal, bipolar spindle (with SPBs found in adjacent sections) is shown in Figure 6, B and B′. Normal half-bridge structures were generally observed adjacent to the unduplicated SPBs (arrow, Figure 6A), but in no cases were satellites observed on the half-bridges. In addition to the cells displaying monopolar arrest, 7/42 (17%) had duplicated their SPBs and contained bipolar spindles. In three of these cases, imbalance in the size of the SPBs was observed, in that one SPB appeared to be about half the size of the other SPB and appeared to nucleate fewer microtubules (Figure 6C), suggesting a defect in SPB assembly or maintenance. We refer to spindles organized in such a manner as imbalanced. Similarly, we found that in YEPD, ∼60% of the cells slowly traverse the first cycle after the temperature shift and arrest at G2 of the second cycle. After 6.5 hr at 37.5°, 29/37 (78%) large-budded cells examined contained monopolar spindles.
The rpt4-150 SSD1 strain behaved similarly to rpt4-150 ssd1-d in that ∼60% of the cells traverse the first cell cycle and arrest in the second cycle in YEPD. After 6 hr at 37°, 3/33 (9%) large-budded cells examined contained complete spindles, one imbalanced, whereas the majority (30/33 or 91%) contained monopolar spindles. Thus, the SSD1 allele does not prevent the SPB duplication defect observed in rpt4-150 ssd1-d cells, although it does suppress the delay in DNA synthesis and budding that occurs in these cells. The unduplicated SPBs were generally associated with what appears to be an unusually curved, elongated half-bridge. This aberrant structure (indicated by the arrow, Figure 6D′) is always observed in an adjacent section in the series relative to the unduplicated SPB itself (Figure 6D). Satellites were not observed associated with these half-bridge structures, which appear similar to those seen in rpt4/pcs1td cells (McDonald and Byers 1997). Although unduplicated SPBs associated with elongated half-bridges were also observed in rpt4-150 ssd1-d cells after growth in YEPD medium, they were not generally observed in that strain after growth in YM-1.
rpt4-145 ssd1-d cells arrest with monopolar spindles or imbalanced spindles depending on the growth medium: EM analysis of the rpt4-145 ssd1-d strain grown in YM-1 revealed that this mutant, which arrests primarily in the second cell cycle after shift to restrictive temperature, displays a relatively uniform arrest. After incubation at 37° for 8 hr, 32/34 (94%) of the cells examined contained monopolar spindles. The unduplicated SPBs were generally associated with the elongated half-bridge described above (not shown, but similar to that shown in Figure 6, D and D′). rpt4-145 ssd1-d cells grown in YEPD medium, which is unbuffered, appear to be more compromised, in that they arrest primarily in G2 in the first cell cycle following the shift to high temperature and display a less uniform phenotype. Such differences have been observed in other cases; for example, apc1-1 cells also arrest in G2 during the first cell cycle after shift to high temperature if grown in YEPD and during the second cell cycle if grown in YM-1 (L. Goetsch, unpublished observations). After 3 hr at 37° (a time point at which these cells are still in the first cell cycle; see Figure 4A), we observed a mixture of phenotypes, including monopolar spindles and imbalanced spindles (the most common phenotype). Figure 6, F and F′, show two adjacent sections of a cell that appears to contain a broken spindle, with SPBs that are dramatically different in size and appear to nucleate different numbers of microtubules. A minority of cells contained duplicated, unseparated SPBs of unequal size (Figure 6E; arrow indicates smaller SPB).
rpt4-145 SSD1 cells rapidly arrest with complete bipolar spindles: During growth in YEPD medium, almost all (90%) of the rpt4-145 SSD1 cells arrest as large-budded cells during the first cell cycle after the shift to restrictive temperature. In striking contrast to rpt4-145 ssd1-d, virtually all cells arrest with complete bipolar spindles (N = 19 and 10, after 2.5 or 5.5 hr at 37°, respectively). Very few abnormalities were observed in these spindles and the SPBs were essentially always of equal size (not shown, but similar to the spindle shown in Figure 6, B and B′). These results indicate that SSD1 prevents the defects in SPB duplication that occur in rpt4-145 ssd1-d cells, in addition to essentially eliminating the delays in budding and initiation of S phase that occurs in these cells.
Rpt4-150 protein expression: The EM analysis described above reveals that the rpt4-145 and rpt4-150 phenotypes display some similarity to that conferred by the rpt4/pcs1td allele (McDonald and Byers 1997). This allele was constructed as a degron-RPT4/PCS1 fusion; the degron unit functions as a heat-inducible degradation signal that is degraded at high temperature, together with sequences fused downstream of it (Dohmenet al. 1994). Presumably, therefore, the rpt4/pcs1td phenotype results from a lack of Rpt4p at the restrictive temperature (although this was never proven directly). To investigate whether the rpt4-150 mutation affects the level of Rpt4p at restrictive temperature, we utilized two plasmids: pHM54, which encodes a GFP-Rpt4p fusion that is functional at both 30° and 36.5°, and pAH66, which encodes a GFP-Rpt4-150p fusion (engineered to contain the rpt4-150 mutation) that is functional at 30° but not at 36.5° (see materials and methods).
We determined the relative amounts of Rpt4p produced by strains containing pHM54 (YHM10.1.54) and pAH66 (YHM10.1.66) after growth at 30° and 36.5° for 6 hr. Equivalent amounts of whole-cell lysates from these strains were subjected to SDS-PAGE and blotted to nitrocellulose. The blot was probed with anti-GFP antibodies, which recognize a primary band of the expected size for the GFP-Rpt4p fusion (∼76 kD) in the lysates derived from strains YHM10.1.54 (Figure 7, lanes 2 and 5) and YHM10.1.66 (lanes 3 and 6). No band is detected in lysates derived from the control strain (Wx257-5c), which does not contain a GFP-Rpt4p encoding plasmid (lanes 1 and 4). Lysates run in lanes 1-3 were obtained from cells grown at 30° and those run in lanes 4-6 from cells grown at 36.5°. Approximately equivalent levels of GFP-Rpt4 and GFP-Rpt4-150 protein fusions were observed at both 30° and 36.5°. Although we were unable to obtain a functional GFP fusion with rpt4-145 to repeat this analysis for that allele, the results with rpt4-150 indicate that the arrest caused by at least one of the C-terminal mutations does not derive from an absence of Rpt4p at restrictive temperature.
We undertook a systematic mutagenesis of RPT4 that focused on converting clustered, charged residues to alanine. One result to emerge from this study is that the N-terminal region of Rpt4p is quite resistant to the effects of such mutations. Like Rpt4p, the other Rpt subunits contain stretches of sequence near their N termini that are predicted to participate in α-helical coiled-coil formation (e.g., see Richmondet al. 1997). Intersubunit coiled-coil formation has been proposed to mediate the specific pairwise interactions between Rpt subunits (which might function in the assembly of the RP) that have been observed using in vitro assays (Richmondet al. 1997). In support of this hypothesis, deletion experiments show that the N-terminal region of Rpt2/S4 is necessary and sufficient for its interaction with Rpt1/S7 in vitro (Richmondet al. 1997). However, our mutational analysis suggests that the Rpt4p coil region may not play a critical role in its function. Half of the five heptad repeats in the coiled-coil motif can be deleted (rpt4-2Δ) without discernible effect. Furthermore, strains containing the rpt4-115 (I81D), rpt4-117 (L84E), and rpt4-121 (L98E) alleles, which convert hydrophobic residues at position a or d to charged residues and may thereby disrupt coiled-coil formation (Flemington and Speck 1990; Huet al. 1990; Pellegrinoet al. 1997), also result in WT phenotypes. Whereas, by the COILS program (Lupaset al. 1991), the wild-type region is given a probability near 1.0 of participating in coiled-coil formation, the I81D change in particular dramatically reduces the probability (to ∼0.3).
The sequence encoding the putative bipartite NLS is also deleted in the functional rpt4-2Δ allele. In higher eukaryotic cells, GFP-tagged proteasomes have been detected in the cytoplasm and nucleus and enter the nucleus either during reassembly of the nuclear envelope following mitosis or via unidirectional transport across the nuclear membrane (Reitset al. 1997). In yeast, there is some ambiguity about proteasome localization. We previously detected nuclear localization of a GFP-Rpt4p fusion (McDonald and Byers 1997). In a different study, however, both a CP subunit and an RP subunit were detected primarily in the nuclear envelope-rough endoplasmic reticulum membrane network in S. cerevisiae (Enenkelet al. 1998). In Schizosaccharomyces pombe, Wilkinson et al. (1998) have localized 26S proteasome subunits specifically to the inner face of the nuclear envelope. These authors suggest that our observation of whole nuclear localization of a GFP-Rpt4p fusion may have been a result of its plasmid-based expression, although a more recent study, in which tagged subunits (including Rpt4/Sug2p) were encoded by chromosomal alleles, also found general nuclear localization (Russellet al. 1999b). In any case, the potential NLS of Rpt4p is not required for its function, nor presumably for its localization, although the latter point remains to be determined. In the absence of such a signal, Rpt4p may be recruited to the nucleus via interactions with other subunits.
We previously hypothesized that Rpt4p might mediate the degradation of distinct substrates, such as inhibitors of SPB duplication. Our hypothesis was based on the observation that strains containing the rpt4/pcs1td allele arrest as large-budded cells with monopolar spindles at restrictive temperature (McDonald and Byers 1997). This phenotype differs markedly from that seen in strains containing conditional rpt6/cim3 or rpt1/cim5 mutations, which cause cells to arrest with short, bipolar spindles (Ghislainet al. 1993). Phenotypic analysis of strains containing the C-terminal rpt4-145 and rpt4-150 mutations supports a role for the Rpt4p C terminus in SPB duplication and in other aspects of G1/S progression. Although the nature of that role is unknown, one possibility is that this region interacts with specific substrate(s), which potentially include inhibitors of SPB duplication, as they are translocated into the proteasomal core. This idea contradicts the “helix shuffle” model (Rechsteineret al. 1993), in which the coiled-coil regions of different Rpt subunits are postulated to interact with similar regions of distinct substrates targeted for degradation. In agreement with this model, it has been shown that murine Rpt6/Sug1p and c-Fos interact through their respective leucine zipper/coiled-coil regions and that c-Fos is present in 26S proteasomes, suggesting that Rpt6/Sug1p might function in the degradation of c-Fos (Wanget al. 1996). The N-terminal mutations in Rpt4p described in the present study argue against such a role for the Rpt4p coil region.
A mutation that affects the C-terminal region of Rpt1p has been shown to be synthetic lethal in combination with a mutation in RPN12 (Takeuchi and Toh-e 1999). The mutation results in an A446V change, affecting a conserved residue (Figure 3, arrow) near the position that is mutated in the rpt4-150 allele. It is possible that the A446V mutation identifies a region of Rpt1p that interacts with Rpn12p (Takeuchi and Toh-e 1999). By analogy, a comparable region in the C terminus of Rpt4p might interact with a different lid component, and disruption of this interaction might result in the defects observed in the rpt4-145 and/or rpt4-150 strains.
Further insight into the structure/function relationship of Rpt4p should come from the analysis of additional conditional alleles. For example, it will be interesting to determine whether mutations in the ATPase domain ever lead to defects in SPB duplication and other aspects of G1/S progression. The crl13 allele (McCusker and Haber 1988) contains such a mutation. Our preliminary analysis indicates that at the restrictive temperature, the crl13 strain arrests as large-budded cells with short, bipolar spindles. The SSD1 allele present in the crl13 strain is not known, and we do not yet know whether this phenotype will vary depending on the status of SSD1. If crl13 causes a bipolar spindle arrest under all conditions, this finding would support the idea that the structural features of Rpt4p required for SPB duplication map to the C-terminal region.
Although we have hypothesized that the SPB duplication and other defects in rpt4 mutants ultimately derive from deficient proteolysis, recent evidence suggests that RP subunits function in additional, nonproteolytic roles. For example, a direct role in transcription is indicated by the finding that the RP base is recruited to an activated promoter (Gonzalezet al. 2002). The 19S RP (but not the 20S CP) has also been implicated in transcription elongation (Ferdouset al. 2001) and in the regulation of nucleotide excision repair (NER), a role that appears to be modulated by the NER protein Rad23 (Russellet al. 1999a; Gilletteet al. 2001). Although the exact function of the RP in these processes is unknown, one possibility is that the chaperone-like activity of the RP base (Braunet al. 1999) acts in the assembly or disassembly of NER machinery (Russellet al. 1999a; Gilletteet al. 2001). The SPB duplication defect in rpt4 mutants might reflect loss of this chaperone-like activity, for example, rather than defective proteolysis.
RAD23 has a redundant function in SPB duplication, in addition to its role in NER. Specifically, dsk2Δ rad23Δ double mutants display a ts failure in SPB duplication (Bigginset al. 1996). We have also observed genetic interactions among dsk2Δ, rad23Δ, and rpt4-145. While neither rad23Δ nor dsk2Δ single mutations are ts, rad23Δ rpt4-145 and dsk2Δ rpt4-145 strains arrest as large-budded cells with a G2 content of DNA ∼2.5 hr after shift to 37° (YEPD medium). These cells arrest with monopolar spindles (100% for dsk2Δ rpt4-145 and 62% for rad23Δ rpt4-145) in an ssd1-d background and with complete bipolar spindles (100% for each strain) in an SSD1 background (L. Goetsch, unpublished observations). Further work is needed to define the functional relationship of these proteins in cell cycle progression.
Additional work is also required to determine the nature of the relationship between SSD1 and RPT4. The function of Ssd1p is unknown, although it has been shown to bind RNA and is localized primarily in the cytoplasm (Uesonoet al. 1997). SSD1 prevents the delays in DNA synthesis and budding seen in both rpt4 ssd1-d mutants and dramatically modifies the rpt4-145 ssd1-d phenotype (monopolar or imbalanced spindle arrest) such that rpt4-145 SSD1 cells arrest at G2/M with short bipolar spindles during the first cycle at restrictive temperature. This finding reveals a role for Rpt4p at a second point in the cell cycle, perhaps similar to that seen for Rpt1/Cim5p and Rpt6/Cim3p. The ability of SSD1 to modify the rpt4 phenotypes appears to derive at least partially from its role in the maintenance of cellular integrity, which it promotes in parallel with the protein kinase C pathway (see Kaeberlein and Guarente 2002). The observation that osmotic stabilization can rescue the delays in DNA synthesis and budding in rpt4 ssd1-d strains, as does SSD1, supports this idea. Defective cellular integrity does not appear to explain the origin of the SPB duplication defects in rpt4 ssd1-d strains, however, as they are not prevented by osmotic stabilization (data not shown). Although the rpt4-145 SSD1 strain displays no defects in SPB duplication, the rpt4-150 SSD1 strain exhibits a monopolar arrest at restrictive temperature.
The as-yet-undefined role of Rpt4p in SPB duplication appears to be required early in this process. Analysis of other conditional mutations that specifically affect SPB duplication have previously revealed several distinct patterns of failure. For example, mps2 (Wineyet al. 1991), ndc1 (Wineyet al. 1993), and bbp1 (Schrammet al. 2000) mutants display failure relatively late, such that a duplication plaque is assembled on the half-bridge, but is not inserted into the nuclear envelope. In contrast, cdc31 (Byers 1981b), kar1 (Rose and Fink 1987), mps1-1 (Wineyet al. 1991), dsk2Δ rad23Δ (Bigginset al. 1996), and rpt4 mutants fail earlier, such that the satellite, the earliest visible precursor to the nascent SPB, does not form or is not maintained. Interestingly, multiple genetic interactions have been observed between several members of this latter group of SPB duplication genes (DSK2, RAD23, KAR1, CDC31) and the protein kinase C pathway, suggesting that this pathway is involved in SPB duplication (Khalfanet al. 2000). This finding might ultimately provide insight into the rpt4 SPB duplication defects, given that SSD1 interacts genetically with RPT4 and the protein kinase C pathway.
An additional later role for Rpt4p in SPB assembly or maintenance is suggested by the presence of imbalanced spindles in rpt4-145 ssd1-d and rpt4-150 ssd1-d strains. These spindles are organized by SPBs that can vary quite dramatically in size and appear to organize different numbers of microtubules (Figure 6), suggesting that the smaller SPB was not fully formed or maintained at restrictive temperature. Mutations in SPC29, which encodes a component of the SPB and the satellite (Wiggeet al. 1998; Adams and Kilmartin 1999), can also result in SPB duplication defects (Elliottet al. 1999) or complete/broken spindles containing SPBs of unequal size (Adams and Kilmartin 1999).
SPB duplication in yeast serves as a model for understanding centrosome duplication in higher eukaryotic cells (Adams and Kilmartin 2000). Although morphologically dissimilar, the SPB and centrosome share some common components, and the regulatory mechanisms governing centrosome and SPB duplication show similarity as well. For example, a mouse ortholog of Mps1p has been shown to regulate centrosome duplication in cultured mouse cells (Fisk and Winey 2001). Links between the proteasome degradation system and centrosome duplication have also been established. For example, proteasomal components (Wigleyet al. 1999) and ubiquitin ligase components (Freedet al. 1999; Gstaigeret al. 1999) are associated with the centrosome and regulate its duplication cycle in Xenopus (Freedet al. 1999). Mutations in ubiquitin ligase components result in centrosome overduplication in mouse (Nakayamaet al. 2000) and Drosophila cells (Wojciket al. 2000). In the former case, this defect appears to arise at least partially from a lack of cyclin E degradation. In light of these results, it will be interesting to determine whether the mammalian Rpt4 subunit, p42 (Fujiwaraet al. 1996), in which the residues corresponding to those affected by the rpt4-145 and rpt4-150 mutations are largely conserved, has a specific role in centrosome duplication or assembly.
We thank Dr. Breck Byers for support and useful discussion, Dr. Robert Coyne for useful comments on the manuscript, Dr. James Haber for providing the crl13 strain, and Melanie Randall for participating in early phases of this project. The work at Colgate University was supported by a National Science Foundation grant (MCB-9727240) to H. B. McDonald. L. Goetsch’s work was supported by National Institutes of Health grant GM-18541 to Dr. Breck Byers.
Communicating editor: M. Johnston
- Received May 29, 2002.
- Accepted July 24, 2002.
- Copyright © 2002 by the Genetics Society of America