INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

The transition from telophase at the end of mitosis into G1 of the next cell cycle is controlled at least partly by deactivation of the B cyclin protein kinase activity. This deactivation involves regulated degradation of the cyclin subunit by ubiquitin-mediated proteolysis (Surana et al., 1993; Irniger et al., 1995; Amon, 1997; Irniger and Nasmyth, 1997) directed by a conserved sequence in the cyclin molecule known as the `destruction box' (Glotzer et al., 1991). Another factor required for cyclin degradation is the cyclosome or anaphase-promoting complex (APC), which is conserved from yeast to humans (Irniger et al., 1995; King et al., 1995; Tugendreich et al., 1995; Peters et al., 1996; Zachariae et al., 1996). In budding yeast, APC becomes active at anaphase where proteolysis of Pds1 and the cohesins occurs (Cohen-Fix et al., 1996; Michaelis et al., 1997), leading to anaphase B. However, destruction of the Clb-Cdc28 kinase activity does not occur at this point but in telophase. Hence, APC function is regulated in some way. In budding yeast, there are many additional genes that have roles in the ending of mitosis. These include DBF2, DBF20, CDC5, and CDC15, which encode protein kinases, CDC14, which encodes a protein phosphatase, TEM1 and LTE1, encoding a guanosine triphosphate-binding protein and G-nucleotide exchange factor, respectively (Johnston et al., 1990; Schweitzer and Philippsen, 1991; Wan et al., 1992; Kitada et al., 1993; Shirayama et al., 1994; Toyn and Johnston, 1994). The multiple genetic interactions found among these genes (references above and our unpublished observations) suggest that the corresponding gene products act in a common signaling pathway regulating the M/G1 transition perhaps by controlling the deactivation of the B cyclin kinase.

We have used Dbf2 as a probe to study the events of late mitosis (Johnston et al., 1990; Parkes and Johnston, 1992; Donovan et al., 1994; Toyn and Johnston, 1993, 1994). One approach was to isolate dosage suppressors of dbf2^ts alleles, which normally cause arrest in telophase. One suppressor identified was SIC1 (Donovan et al., 1994) an inhibitor of the Cdc28 protein kinase (Mendenhall, 1993). High-copy SIC1 also suppresses a number of other mutants that arrest in telophase, suggesting an important role in the ending of mitosis (Donovan et al., 1994; Toyn et al., 1997). The SIC1 gene is expressed in late M phase, and the protein is unstable, persisting from late M phase to late G1 (Donovan et al., 1994; Schwob et al., 1994). This instability of the protein implies that the transcriptional regulation of the gene is a crucial level of control. Two transcription factors, Swi5 and Ace2, are active in cell cycle regulation of gene expression in late M/early G1. We and others have recently shown that Swi5 does indeed regulate SIC1 (Knapp et al., 1996; Toyn et al., 1997).

The SWI5 gene encodes a zinc finger motif transcription factor and is expressed under cell cycle control during G2 and M phase (Lydall et al., 1991). Swi5 protein enters the nucleus during late M phase and G1, and known targets include HO, CDC6, RME1, ASH1, and EGT2 (Nasmyth et al., 1990; Piatti et al., 1995; Bobola et al., 1996; Kovacech et al., 1996; Sil and Herskowitz, 1996; Toone et al., 1995). The identification of the SIC1 gene as a Swi5 target suggests that Swi5 might also control other genes with roles in the ending of mitosis. Indeed, we found swi5 strains to have phenotypes characteristic of mutants with defects in ending mitosis (Toyn and Johnston, 1994). The identification of further genes regulated by Swi5 might therefore reveal novel genes with roles in late M phase. Accordingly, we carried out a computer search for genes with Swi5-binding sites (Knapp et al., 1996) in their promoters. Surprisingly, we identified PCL9, a gene with high homology to Pcl2, a cyclin-like protein that associates with the cyclin-dependent protein kinase (Cdk) Pho85.

Pho85 functions in the control of phosphate metabolism, interacting with the cyclin Pho80 to regulate the transcription factor Pho4 (Ogawa and Oshima, 1990; Kaffman et al., 1994; O'Neill et al., 1996). When cells are grown in a phosphate-rich medium, Pho4 is phosphorylated by Pho80-85 and the Pho4 target gene Pho5 is repressed (reviewed in Lenburg and O'Shea, 1996). In addition, Pho85 interacts with two other cyclins, Pcl1 and Pcl2, when it appears to function in regulating cell cycle progression at START (Espinoza et al., 1994; Measday et al., 1994). START marks the point of commitment to the mitotic cell cycle, and execution of START requires activation of the Cdk Cdc28 by association with the G1 cyclins encoded by the CLN1 and CLN2 genes. Whereas Cdc28 is essential, Pho85 is dispensable but becomes essential for G1 progression when the CLN1 and CLN2 genes are deleted (Espinoza et al., 1994; Measday et al., 1994). Moreover, like CLN1 and CLN2, PCL1 and PCL2 are expressed under cell cycle control in late G1 by means of the SBF transcription factor (Measday et al., 1994). The Pcl2-associated kinase activity also apparently peaks in G1 (Measday et al., 1994). The above data clearly establish a role in the cell cycle for the Pho85-Pcl complexes, although their precise function remains unclear.

Apart from regulating phosphate metabolism and G1 progression, Pho85 clearly has additional functions. Strains deleted for PHO85 grow slowly, are unable to use lactate or glycerol as a carbon source, and accumlate high levels of glycogen (Huang et al., 1996; Lenburg and O'Shea, 1996; Timblin et al., 1996). Thus, Pho85 may associate with other cyclins in addition to Pho80, Pcl1, and Pcl2, which play distinct roles. In support of this hypothesis, no fewer than seven additional cyclins that associate with Pho85 have recently been identified (Measday et al., 1997). On the basis of homology these fell into two groups, and Pcl9 is in one of these groups (Measday et al., 1997).

Here we characterize PCL9 and show that it is cell cycle regulated in late M phase under control of Swi5. The protein is unstable, also accumulating in late M phase, and it has a Pho85-dependent kinase activity associated with it. We show further that its close homolog Pcl2 is not only controlled by SBF but additionally by Swi5. PCL9 and PCL2 are the only two cyclins at present known to be regulated at this point in the cell cycle. These results and genetic data presented in this paper suggest a novel role for Pho85 in late mitosis.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Media and Growth Conditions

Yeast was cultured in YP medium (1% yeast extract, 2% bactopeptone) supplemented with 2% glucose (YPD) or minimal medium made using yeast nitrogen base (Difco, Detroit, MI) according to the manufacturer's instructions. Synchronous cultures were prepared using the -factor method; synthetic -factor peptide-mating pheromone (final concentration 3.5 µg/ml) was added to log phase cultures of MATa cells growing in YPD. When the cells had arrested in G1 (usually after 3.5 h), the -factor was removed by filtration, and the cells were resuspended in fresh medium.

Strains and Plasmid Construction

The following strains were used: CG378 (MATa, ade5 leu2-3, 112 trp1-289 ura3-52), J252-9A (MATa leu2-3, 112 his6/7 trp1-289 ura3-52 swi5::LEU2), DTY87 (MAT leu2 his6 ura3 ace1 swi5::LEU2), DTY91 (MAT leu2 his6 ura3 ace1 ace2439::URA3), DTY92 (MAT leu2 his6 ura3 ace1 ace2439::URA3 swi5::LEU2), DTY59 (MAT leu2 his6 ura3 ace1), DTY7 (MAT leu2 his6 ura3), mycPCL9 (MATa, ade5 leu2-3, 112 trp1-289 ura3-52 mycPCL9::TRP1, mycPCL9 pho85 (ade leu trp ura mycPCL9::TRP1 pho85::LEU2), cdc5-1 (MATa his ura3), cdc14 (MAT ade1/2/5 trp1 ura3), cdc15 (MAT ade leu2 tyr1 ura3), tem1-3 (MAT his3 leu2 trp1 ura3), dbf2-1 (MAT ade1 trp1, 2 ura3).

For detection of Pcl9 protein in yeast crude extracts by Western blot and for immunoprecipitation of Pcl9 protein for kinase assays, three tandem copies of the c-myc epitope tag were inserted at the C terminus of the chromosomal copy of the PCL9 gene. Integration of the c-myc epitope tag into the yeast genome was carried out using a linearized plasmid carrying a c-myc epitope tagged C-terminal portion of the PCL9 gene. This integration procedure results in a full-length tagged version of the gene and a duplication of the PCL9 C terminus that is not expressed. The details of the plasmid construction are as follows. A 0.4-kilobase (kb) EcoRI-EcoRI fragment containing the 3'-end of PCL9 was made by PCR amplification using the oligonucleotides 5'-CCGGAATTCGGAGACAAGAAATGCTGTTG-3' and 5'-CCGGAATTCTTAGG ATCCTTGCTTGAAAAACGATGA-3' and ligated into the EcoRI site of YIplac204 (BamHI). This introduced a unique BamHI site immediately upstream of the PCL9 stop codon. Subsequently a 120-base pair (bp) BamHI fragment encoding the triple myc epitope tag from pUC119-3 myc (Dr. S. Kron, Whitehead Institute, Cambridge, MA) was ligated into the introduced BamHI site at the 3'-end of PCL9. This plasmid was linearized using NcoI, which cuts in the PCL9 fragment, before integration into the yeast genome. In all Trp⁺ transformants tested, Western blotting using 9E10 anti-c-myc monoclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA) detected the mycPcl9 protein with a molecular weight of 42 kDa, as expected.

Chromosomal deletion of the PCL9 gene was achieved using the following `knock out' plasmid. The PCL9 gene was cloned into the EcoRI site of pUC18 using a DNA fragment obtained by PCR containing the PCL9 coding region plus 455 bp of the PCL9 upstream region and 380 bp of the PCL9 downstream region. The oligonucleotides used for PCR amplification were 5'-CCGGGTCACTTCGGGATGAATTCAG-3' and 5'-CCGGAATTCAGAGTCAGA TAGATTTTGA-3'. The PCL9 coding region was then replaced by a TRP marker, introducing the SmaI/PstI TRP-fragment from YDP-W into the NsiI/EcoRV cut plasmid. This plasmid was cut with EcoRI before integration into the yeast genome. Trp⁺ transformants were analyzed by Southern blot for chromosomal deletion of PCL9.

Determination of mRNA Levels

Total RNA was extracted from cells as described previously (Johnston et al., 1990). A 5-µg sample of total RNA was denatured with glyoxal, separated on a 1.2% agarose gel, and transferred to a GeneScreen hybridization membrane (Dupont NEN Research Products, Boston MA). Blots were probed with restriction fragments internal to the genes concerned. Levels of hybridization were quantified using either a Personal Densitometer PD-130 (Molecular Dynamics, Sunnyvale, CA) or a PhosphorImager (Molecular Dynamics) and were normalized to ACT1 or MET4 transcripts as a loading control.

Preparation of Cell Extracts, Immunoblotting, and Pho85/Pcl9 Kinase Assays

Yeast cell extracts were prepared by glass bead lysis, in 250 mM NaCl, 5 mM EDTA, 50 mM Tris-HCL (pH 7.5), 0.5% NP-40, and protease inhibitor mix (PI mix-100 µg/ml phenylmethyl sulfonyl fluoride, 2 µg/ml aprotinin, 2 µg/ml leupeptin, 2 µg/ml pepstatin A, 50 µg/ml TLCK and 100 µg/ml TPCK). After centrifugation, protein concentration was determined. For detection of mycPcl9 by Western blot, 60 µg of yeast cell extract were used. Immunoblotting was performed as described previously (Toyn and Johnston, 1994). Dilutions of the antibodies were 1:100 for the 9E10 monoclonal antibody and 1:10'000 for the secondary Ab (anti-mouse immunoglobulin horseradish peroxidase linked; Amersham, Arlington Heights, IL).

For kinase assays, mycPcl9 was immunoprecipitated from 0.5 mg of yeast extract using 2.4 µg of 9E10 monoclonal antibody. The immune complex was coupled to protein G beads and washed six times with lysis buffer and twice with kinase buffer (50 mM Tris-HCL [pH 7.5], 10 mM MgCl₂, 1 mM dithiothreitol). Beads were then incubated with 10 µl of kinase cocktail (50 mM Tris-HCL [pH 7.5], 10 mM MgCl₂, 1 mM dithiothreitol, 1 µM cold ATP, 10 µCi of [-³²P] ATP, 100 ng of purified Pho4 protein) for 30 min at 25°C. Reactions were stopped by adding SDS sample buffer, boiled for 5 min, and loaded onto a 10% SDS-acrylamide gel.

Gel Mobility Shift Assay

The GST and GST-Swi5 fusion proteins were expressed in Escherichia coli and batch purified according to the manufacturer's instructions (Pharmacia, Piscataway, NJ). Binding reaction mixtures (20 µl volume) contained 15 mM Tris (pH 8.0), 75 mM NaCl, 7.5% glycerol, 12.5 mM dithiothreitol, 0.375 mM EDTA, 750 µg/ml bovine serum albumin, 0.1 mM Na-fluoride, 0.1 mM Na-vanadate, 100 µg/ml phenylmethyl sulfonyl fluoride, 2 µg/ml aprotinin, 2 µg/ml leupeptin, 2 µg/ml pepstatin A, 50 µg/ml TLCK, 100 µg/ml TPCK, 100 ng/ml poly(dI-dC), and 1 ng ³²P 5'-end-labeled DNA probe. Probes were obtained by annealing two complementary oligonucleotides spanning the following upstream sequences: 418 to 480 of the PCL2 promoter and 330 to 270 of the PCL9 promoter. Annealed oligonucleotides were purified on a 12% polyacrylamide gel. Competitor DNA was present in a 500-fold excess over the labeled probe. The amount of protein used in each binding reaction is indicated in the figure legend. The reactions were incubated at 25°C for 5 min and then on ice for 1 h, when they were loaded directly onto a 4% (40:1) nondenaturing polyacrylamide gel and electrophoresed in 0.6% Tris-borate-EDTA buffer at 4%. Gels were dried onto Whatman 3 MM paper and autoradiographed.

Determination of Cell Numbers and the Proportion of Cells with Divided Chromatin

The concentration of cells in growing cultures was determined using a particle counter (Coulter Electronics, Luton, United Kingdom). Yeast culture samples were lightly sonicated to disperse clumps before dilution and counting. To determine the proportion of cells with divided chromatin, culture samples were sonicated to disperse clumps and fixed by addition of an equal volume of ethanol. The cells were resuspended in a solution of 0.1 µg/ml 4,6-diamidino-2-phenylindole in 150 mM NaCl, and then observed by fluorescence microscopy at 1000× magnification. At least 100 cells were observed in each sample to determine the percentage of cells with divided chromatin.

Database Search, Sequence Alignment, and Phylogenetic Analysis

Database searches were performed using the computer program FINDPATTERNS from the GCG package. Pcl9 and Pcl2 protein sequences were aligned using the computer program CLUSTAL from DNASTAR.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

The PCL9 Promoter Contains Putative Swi5 Binding Sites

We are interested in the control of exit from mitosis and its implications for the next cell cycle. A transcription factor active at this time in the cell cycle is Swi5. We have previously demonstrated that Swi5-dependent activation of the SIC1 gene contributes to deactivation of the B cyclin kinase, and hence exit from mitosis (Toyn et al., 1997). To identify further genes regulated by Swi5, which might have a role in late M phase, we searched the EMBL database for genes containing putative Swi5 binding sites in their promoter. The motif used in our search was ACCAGC (Knapp et al., 1996). Interestingly, two putative binding sites were found in the promoter of PCL9 (Figure 1), which encodes a protein showing significant homology to Pcl2, a cyclin-like protein interacting with Pho85 (Measday et al., 1997). Compared with most of the other Pcls, Pcl9 and Pcl2 display a high degree of homology not only over the region of the cyclin box but over much of the N-terminal part of the gene (Measday et al., 1997). They exhibit 84% similarity across the conserved region and 45% similarity when the whole sequence is taken into account.

View larger version (25K): [in this window] [in a new window]   Figure 1.   The PCL9 and PCL2 promoter contain putative Swi5 binding sites. Comparison of possible Swi5 binding sites found in the PCL9 and PCL2 promoter with those found in the promoters of SIC1 and HO. The conserved CCAGC motif is indicated.

PCL9 Transcription Is Cell Cycle Regulated in Late M Phase under the Control of Swi5

The Swi5 transcription factor is active predominantly in the late M-early G1 phase of the cell cycle, which results in periodic expression of its target genes at this time. To investigate the role of Swi5 in PCL9 transcription, we looked at PCL9 mRNA levels in a culture synchronized using -factor (Figure 2A). PCL9 mRNAs are completely absent in G1 and S phase cells. They first appear as cells start to divide their chromatin at the end of the first cell cycle. The expression of PCL9 at the end of mitosis is almost identical to that of SIC1 (Figure 2A), a gene already known to be controlled by Swi5 (Toyn et al., 1997).

View larger version (29K): [in this window] [in a new window]   Figure 2.   Cell cycle regulation of PCL2 and PCL9 transcript levels and Pcl9 protein level. A wild-type strain containing a 3 myc epitope-tagged version of the PCL9 gene was arrested using -factor and reinoculated into fresh medium. Samples were taken 5 min and 15 min after the release from the -factor holding, and then at 15-min intervals. Samples were analyzed for (A) the percentage of budded cells and cells with divided chromatin; transcript levels of PCL9, SIC1, PCL2, CDC6, PCL1, CLN2, and MET4. Quantitation of transcript levels are relative to MET4; PCL9 (solid circles), SIC1 (open circles), PCL2 (solid diamonds), PCL1 (crosses), CDC6 (triangles). (B) mycPcl9 protein level. Solid circles, PCL9 transcript levels; crosses, Pcl9 protein levels.

To address the question directly of whether Swi5 controls PCL9 transcription, we analyzed PCL9 transcript levels in yeast strains carrying a deletion of SWI5. A second transcription factor, Ace2, is active at about the same time in the cell cycle as Swi5 (Dohrmann et al., 1992), and we therefore also included strains deleted for the ACE2 gene. Expression of PCL9 is absent in a swi5 and in a swi5ace2 strain, whereas in an ace2 strain PCL9 transcript levels seem not to be affected (Figure 3A). This result clearly demonstrates that Swi5, but not Ace2, is necessary for PCL9 transcription. The effect of Swi5 on PCL9 transcription becomes even more evident when PCL9 mRNA levels were determined in synchronous cultures of swi5 and an isogenic wild-type strain (Figure 3B). As expected, in the SWI5 control strain, PCL9 was strongly cell cycle regulated, whereas in the swi5 strain PCL9 expression was almost completely abolished.

View larger version (38K): [in this window] [in a new window]   Figure 3.   A. SWI5 is required for PCL9 transcription. Densitometric quantitation of a Northern blot analysis using mRNA from exponentially growing wild-type (DTY7), swi5 (DTY87), ace2 (DTY91), and swi5ace2 (DTY92) strains. Blots were hybridized with probes from PCL2 and SIC1 and normalized to ACT1 as a loading control. (B) Effect of deletion of SWI5 on the cell cycle regulation of PCL2 and PCL9 transcription. Cultures of swi5 strains containing either the plasmid YIpSWI5 or control plasmid YIplac211 were synchronized using -factor, and samples were taken for Northern blot analysis (Toyn et al., 1997). The Northern blot was quantitated by densitometry and normalized to the level of actin mRNA detected on the same blot. The first sample on the autoradiograph is from log-phase cells before addition of -factor and is not shown in the quantitations. Quantitations of transcript levels from the SWI5 (solid circles) and the swi5 experiment (open circles) are compared. The data points relating to the percentage buds and divided chromatin were described by Toyn et al. (1997).

Both SBF and Swi5 Contribute to the Expression of PCL2 in the Cell Cycle

PCL2 transcription has been shown previously to be cell cycle regulated at G1/S, due to the transcription factor SBF (Measday et al., 1997). However, the structural similarity between Pcl2 and Pcl9 and the strong regulation of PCL9 by Swi5 led us to reexamine PCL2 expression. In an -factor holding and release experiment, the peak in PCL2 mRNA levels in fact occurs about 15 min before the PCL1 and CLN2 transcripts, which are controlled solely by SBF (Figure 2A), and this is also apparent in existing data on PCL2 expression (Measday et al., 1994, 1997), although no conclusions were drawn from this discrepancy (see also Measday et al., 1997). To address more precisely when PCL2 mRNAs appear during the cell cycle, we looked at PCL2 transcription in yeast cells arrested at different stages of the cell cycle. In late mitotic mutants that were held at the restrictive temperature, PCL2 is expressed and transcript levels increase in amount. For instance, in dbf2 mutants, a 3.5-fold induction of PCL2 expression was observed (Figure 4). In sharp contrast, transcript levels of genes controlled solely by SBF, such as CLN2 and PCL1 (not shown), were reduced in amount (Figure 4). In yeast cells treated with nocodazole, which blocks cells earlier in M phase, PCL2 transcripts gradually decline (Figure 4). Thus, PCL2 expression occurs in late mitosis where Swi5 is active, suggestive of a dual regulation of PCL2 by this transcription factor as well as SBF.

View larger version (15K): [in this window] [in a new window]   Figure 4.   PCL2 is expressed in late mitosis. The late mitotic cell cycle mutant dbf2 was grown at 25°C, sampled, and transferred to 37°C, and further samples were taken at 40-min intervals for 2 h. For the nocodazole-treated culture, the drug was added to midlog cell of strain CG378 to a final concentration of 25 µg/ml, sampled immediately, and then after 45-min intervals for 3 h. Samples were analyzed for PCL2, CDC6, and CLN2 transcript levels by Northern blot analysis. The Northern blot was quantitated by densitometry and normalized to the level of actin mRNA detected on the same blot. PCL2 (square), CDC6 (diamonds), CLN2 (circles).

To further explore the regulation of PCL2 by Swi5, we examined PCL2 expression in the synchronized isogenic wild-type and swi5 strains (Figure 3B). In the wild-type, PCL2 expression occurs approximately 15 min earlier than in the swi5 strain. Importantly, the peak of PCL2 mRNA levels seen in the swi5 strain now coincide more closely to the timing of CLN2, an SBF-regulated gene. To confirm this difference in the timing of PCL2 expression, we repeated the above experiment with W303-1a and an isogenic swi5 strain (kindly provided by David Stillman) by analyzing samples taken at 10-min intervals. We obtained an essentially identical result, again observing the 15-min difference in PCL2 expression (our unpublished results). It should be noted that 15 min is a significant difference representing some 20% of the cell cycle in these cultures. These results indicate that two types of cell cycle signals control PCL2 transcription: one operating via Swi5 at the end of mitosis and a second operating via SBF in late G1. The Swi5-dependent control is, however, clearly the more minor component (Figure 3B), at least in rapidly growing laboratory cultures.

One other example of a gene with dual regulation by Swi5 and the G1 transcription machinery is CDC6 (Piatti et al., 1995), a gene involved in the initiation of DNA replication. Although CDC6 is expressed in a similar manner to PCL2 (Figures 2A and 4), there are some minor differences. For instance, CDC6 is expressed slightly earlier in the cell cycle than PCL2 (Figure 2A), and induction of CDC6 transcription in dbf2-arrested cells is much lower than PCL2 expression (Figure 4). The reasons for the difference between PCL2 and CDC6 expression is not clear but they are, of course, regulated by different G1 transcriptional regulators, namely SBF and MBF.

Swi5 Binds Directly to the PCL2 and PCL9 Promoters

The data described above suggest a role for Swi5 in PCL2 and PCL9 transcription. We searched the PCL2 upstream region for possible Swi5 binding sites. Like PCL9, two conserved sites were found, at 458 and 484 bp upstream of the ATG (Figures 1 and 5A). To test whether Swi5 does indeed bind to the PCL2 and PCL9 promoters, we carried out gel retardation analysis. The sequences of the promoter fragments used in these assays are shown in Figure 5. In both cases the fragment includes the two potential Swi5 binding sites. Incubation of these probes with Gst-Swi5 resulted in formation of two protein-DNA complexes with both the PCL2 and PCL9 promoter fragments (Figure 5, A and B; lanes 2 and 3), whereas no complex was formed by using Gst only (Figure 5, A and B; lanes 4-6). The relative mobilities of the two protein-DNA complexes suggest that the two Swi5 binding sites in the PCL2 and PCL9 promoter fragment are both recognized and bound by Swi5 (Knapp et al., 1996). To determine whether complex formation was specific to PCL2 and PCL9 promoter DNA, unlabeled competitor DNA was added to the binding reaction. Complex formation was competed by addition of competitor PCL2 or PCL9 promoter DNA (Figure 5, A and B; lane 9) but not by addition of competitor CLN1 or YG100 promoter DNA (Figure 5, A and B; lanes 10 and 11). Thus Swi5 can bind directly and specifically to the PCL2 and PCL9 promoter.

View larger version (81K): [in this window] [in a new window]   Figure 5.   Swi5 binds to the PCL2 and PCL9 promoters. Gel retardation assays were carried out using a DNA fragment consisting of the PCL2 and PCL9 promoter sequences shown in the figure; possible Swi5 binding sites are underlined. For the binding assay, bacterially produced Gst-Swi5 and Gst were used: lane 1, 10 ng Gst-Swi5; lane 2, 100 ng Gst-Swi5; lane 3, 500 ng Gst-Swi5; lane 4, 10 ng Gst; lane 5, 100 ng Gst; lane 6, 500 ng Gst; lane 7, no added protein; lane 8, 500 ng Gst-Swi5; lane 9, 500 ng Gst-Swi5 and unlabeled PCL2, respectively, PCL9 competitor DNA; lane 10, 500 ng Gst-Swi5 and unlabeled competitor DNA containing bp 634 to 496 of the CLN1 promoter region; lane 11, 500 ng Gst-Swi5 and unlabeled competitor DNA containing bp 261 to 130 of the YG100 promoter.

Pcl9 Protein Levels Fluctuate in the Cell Cycle

Since PCL9 transcription is strongly cell cycle regulated we determined whether Pcl9 protein levels displayed similar kinetics. We therefore replaced the endogenous PCL9 gene by a version containing three myc epitope-tags at its C-terminal end and analyzed Pcl9 protein levels in a synchronized culture (Figure 2B). Following -factor block and release, Pcl9 protein is not detectable in G1 and S phase cells. It appears suddenly in late M phase, and the peak of accumulation occurs slightly later than that for cells containing divided chromatin, a late mitotic event. Pcl9 then accumulates strongly in the second cycle, the peak coinciding almost exactly with that for divided chromatin. A comparison of PCL9 transcript and protein levels in the same culture suggests that Pcl9 protein is produced shortly after the gene is transcribed (Figure 2B), but this may simply reflect the relative sensitivities of the Northern and Western techniques, plus the time taken for translation. As expected, the protein continues to accumulate while the transcript is present and probably reaches a peak around the M/G1 boundary, with some protein almost certainly persisting into G1.

Pcl9 Is Associated with a Pho85-Dependent Protein Kinase Activity

Based on its sequence homology with Pcl2, Pcl9 seemed likely to interact with and form a functional cyclin-cdk complex with Pho85. A specific interaction between Pho85 and Pcl9 was detected in vitro using affinity chromatography, essentially identical to the result obtained by Measday et al. (1997) (our unpublished results). To test whether this interaction produces a kinase activity in vivo, c-myc epitope-tagged Pcl9 was immunoprecipitated from lysates of wild-type, mycPCL9, or mycPCL9pho85 strains and assayed for associated kinase activity. In these assays purified Pho4, a known target of the Pho85 kinase, was used as a substrate (kindly provided by Colin Goding). mycPcl9 immunoprecipitates from mycPCL9 yeast lysates clearly phosphorylate Pho4 in vitro (Figure 6) showing an increase of some eightfold above background. As expected for a genuine kinase activity, phosphorylation of the Pho4 substrate is further increased if mycPCL9 is carried on a high-copy episomal plasmid (Figure 6, lane 3). In contrast, immunoprecipitates derived from strains lacking either Pho85 (mycPCL9pho85) or the myc epitope tag have only background kinase activity (Figure 6, lanes 4 and 5) (note that similar background kinase activity was observed previously with Pho4 as a substrate: Kaffman et al., 1994; Measday et al., 1994). Thus, the kinase activity associated with Pcl9 is Pho85-dependent. Furthermore, like Pcl2-Pho85 (Measday et al., 1994) and Pcl1-Pho85 (Espinoza et al., 1994), Pcl9 showed only limited associated kinase activity toward histone H1 and none toward myelin basic protein (our unpublished results). These results strongly suggest that Pcl9 and Pho85 form a functional cyclin-cdk complex in vivo.

View larger version (30K): [in this window] [in a new window]   Figure 6.   Pcl9 is associated with a Pho85-dependent kinase activity. mycPcl9 was immunoprecipitated from yeast extracts prepared from exponentially growing wild-type, mycPCL9, and mycPCL9pho85 cells, as well as from wild-type cells carrying either a YEpmycPCL9 plasmid or the empty YEp vector. In the control lane, the kinase reaction was carried out in the absence of substrate. Immunoprecipitates were coupled to protein G beads and analyzed for mycPcl9-associated kinase activity using Pho4 as a substrate. Bar diagram shows corresponding densitometric quantitation of the assay.

It has previously been shown that Pho85 protein levels are constant throughout the cell cycle (Measday et al., 1994). However, a common level of regulation of Cdk activity is through cell cycle-dependent association with the cyclins. The strong cell cycle regulation of the Pcl9 protein therefore suggests a similar cycling of the Pcl9-Pho85 kinase activity. Preliminary results confirm this, a fluctuation of the Pcl9-associated kinase activity being clearly apparent in the second cell cycle, coincident with the Pcl9 protein levels.

pho85 Shows Genetic Interactions with sic1, spo12, and swi5

The expression patterns of PCL9 and PCL2 suggest a role for these cyclins in late M and the G1 phase of the cell cycle. However, strains deleted for PCL9 and PCL2, either alone or in combination, were indistinguishable from wild type. High-copy PCL9 was also unable to suppress the temperature sensitivity of late mitotic mutants such as cdc5, cdc14, cdc15, dbf2, and tem1.

The inability to detect any phenotype associated with deletions of PCL9 (and PCL2) is probably due to the known redundancy among budding yeast cyclins. We, therefore, decided to look at possible late mitotic phenotypes and genetic interactions associated with loss of PHO85, the Cdk interacting with Pcl9 and Pcl2. Again no late mitotic defects could be detected in pho85 strains. We then looked for synthetic interactions between pho85 and deletions of other, nonessential genes functioning in late mitosis, such as dbf2, spo12, lte1, swi5, and sic1. The only success was with sic1, spo12, and swi5 (Table 1). From a cross between pho85 and swi5 five of six double-mutant spore clones were spontaneously temperature-sensitive and failed to grow at 37°C. Swi5 regulates SIC1, which also has a role in ending mitosis. The cross of pho85 with sic1 yielded 32 spontaneous temperature-sensitive spore clones out of 37 double mutants for pho85 and sic1. The genetic interaction between pho85 and spo12 seems to be slightly weaker since only 7 of 27 double-mutant spores were temperature-sensitive. The very high proportion of temperature-sensitive spore clones from the pho85 double mutants with sic1 and swi5 argues strongly that these are significant. In the case of the spo12pho85 this was less certain and we therefore reintroduced the PHO85 genes into four of these. In each case this reversed the temperature sensitivity so that this is a genuine defect associated with loss of both SPO12 and PHO85. Since both Sic1 and Spo12 function in the ending of mitosis, the synthetic interaction with pho85 indicates a function for this Cdk in late mitosis-early G1.

View this table: [in this window] [in a new window]   Table 1.  Genetic interactions between pho85 and genes acting in late mitosis

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

In a search for genes with promoters containing Swi5 binding sites we identified PCL9, encoding a protein with marked homology to Pcl2, a cyclin-like protein interacting with Pho85 (Measday et al., 1994). Pcl9 was shown to interact with Pho85 using both the two-hybrid technique and affinity chromatography (Measday et al., 1997). Our data also strongly supports a Pho85-Pcl9 interaction. We too detected an interaction using affinity chromatography (our unpublished results) and, more important, we find that a protein kinase activity that phosphorylates a Pho85 target, Pho4, coimmunoprecipitates with Pcl9. This kinase activity is Pho85-dependent, being absent in pho85 strains. Moreover, we find the Pcl9 protein to be unstable, the level fluctuating in the cell cycle as occurs with cyclins. Specific proteolysis of cyclins is a key mechanism controlling progression through the cell cycle, degradation occurring by means of the ubiquitin pathway. Recognition by this destruction machinery frequently involves the `destruction box,' found in B-type cyclins (Glotzer et al., 1991). A related motif occurs in both Pcl9 and Pcl2 (our unpublished data), suggesting that cell cycle-regulated degradation of these Pcls may be conferred by the same mechanism. The kinase activity associated with Pcl9 also shows evidence of cell cycle regulation. Pcl9 therefore has the properties of a cyclin that interacts with Pho85 and activates the kinase in a cell-cycle dependent manner.

The impetus for our identification of PCL9 was an interest in pathways controlling the end of mitosis. We previously showed the Sic1 Cdk inhibitor to have a role in ending mitosis (Donovan et al., 1994; Toyn et al., 1997) and the SIC1 gene to be controlled by the Swi5 transcription factor (Knapp et al., 1996; Toyn et al., 1997). We therefore expected Swi5 to control other genes with roles in late M phase; indeed we found swi5 strains to have phenotypes associated with defects in ending of mitosis (Toyn et al., 1997). PCL9 transcript levels were sharply reduced in swi5 strains. Use of synchronous cultures and cells blocked in late M phase confirmed that PCL9 expression occurs largely in telophase under the control of Swi5. Partially purified Swi5 protein also bound to the PCL9 promoter in vitro. The peak level of the Pcl9 protein in the cell cycle occurred late, coincidentally with the peak of divided chromatin (telophase), consistent with Pcl9 functioning in late M phase and early G1. Hence, Pcl9 is a Pho85-interacting cyclin functioning in late M phase/early G1, the only cyclin so far known to act at this point in the cell cycle.

The similarity in structure between Pcl9 and Pcl2 led us to reexamine the regulation of the PCL2 gene. Although it is clearly controlled by SBF (Measday et al., 1994), we showed that it is also regulated by Swi5. Deletion of SWI5 affects the timing of PCL2 expression, and the gene is also expressed in cells arrested in late M phase when SBF is not active. PCL2 has two Swi5 binding sites in the promoter, and Swi5 bound to this promoter in vitro. Thus like CDC6, PCL2 is controlled by both Swi5 and the G1 transcriptional machinery. The relative contributions of Swi5 and SBF to PCL2 expression is difficult to establish precisely. PCL9 is obviously more strongly expressed in late M than PCL2. In addition, comparison of the synchronized swi5 and wild-type cultures suggest that SBF is responsible for the bulk expression of PCL2.

Strains deleted for PCL9 or both PCL2 and PCL9 showed no obvious perturbations of the cell cycle. This is not surprising given the known redundancy among both the Cln and Clb cyclins in yeast. However, it is well established in budding yeast that different classes of cyclins have different cell cycle stage-specific roles. Moreover, these classes of cyclins are expressed under cell cycle control in successive bursts of transcription (Figure 7), the timing of which coincides roughly with their function. For example, the G1 cyclins CLN1 and CLN2 are expressed in late G1 and function in START, whereas CLB1 and 2 are expressed in G2 and M and act in mitosis. The novel wave of cyclin synthesis in late M phase might therefore imply a function for Pcl9-Pho85 and Pcl2-Pho85 in the pathways controlling the end of mitosis. So far, our efforts to detect late mitotic defects associated with pho85 have been confined to genetic interactions with only three genes functioning at M-G1, sic1, spo12 and swi5 (and of course Swi5 regulates SIC1). Together with the absence of any late mitotic defects in pho85, this might indicate that Pho85/Pcl9/Pcl2 are not associated directly with the ending of mitosis. Instead, this Cdk activity may be associated with preparative events for the next cell cycle. It is well established that the preliminary events of S phase occur in late mitosis (for example see Cocker et al., 1996). Deletion of SPO12 and SIC1 both lead to a slight protraction of M phase (Donovan et al., 1994; Parkes and Johnston, 1992) and perhaps, in the absence of Pho85, defects occur in the M-G1 transition leading to the observed temperature sensitivity in the double mutants.

View larger version (13K): [in this window] [in a new window]   Figure 7.   The cell cycle timing of cyclin mRNAs in Sacharomyces cerevisiae. A schematic diagram indicating the relative timing of PCL9 and PCL2 in comparison with CLN1, CLN2, CLB1, CLB2, CLB3, CLB4, CLB5, CLB6, and PCL1 mRNAs throughout the cell cycle.

The Pcl9 protein, in fact, very likely persists into G1 consistent with some function spanning the M-G1 boundary. Given the high amino acid identity between Pcl2 and Pcl9, it is probable they are involved in the same or closely related cellular functions. Measday et al. (1997) concluded that both Pcl2 and Pcl9 may be involved in determining cellular morphology. We have found no morphological or cell wall defects in strains deleted for PCL2 and PCL9, either alone or in combination. On the other hand, at 37°C the spo12 pho85 strains failed to bud (our unpublished results) which could be consistent with a morphological role for Pho85.

The dual regulation of the PCL2 gene leading to its expression in late M phase and/or G1 is shared by CDC6. In the case of CDC6, this is likely to be due to its essential role in establishing the prereplication complex (pre-RC) that normally occurs in late M phase (Cocker et al., 1996). In stationary phase, this pre-RC is lost and then needs to be reformed in G1 for the initial cell cycle after refeeding. Since Cdc6 is unstable, it therefore also needs to be expressed in G1, and this occurs under control of the MBF transcription factor, which is active coincidentally with SBF. It is tempting to speculate that, like Cdc6, Pcl2 and Pcl9 carry out a role, normally executed in late M phase, that is necessary for the subsequent cell cycle. Since both are unstable, at least one protein needs the capacity to persist into G1; hence the regulation of PCL2 by both the Swi5 and SBF transcription factors. Whether this role lies in S phase control is not clear at present.

This regulation of two cyclins in late M phase along with CDC6 and SIC1 adds to the complexity of events known to occur in telophase. A large number of signaling proteins are already known to act at this time in the cell cycle (see INTRODUCTION). This is the last point at which satisfactory completion of preceding cell cycle events can be assessed, and these signaling proteins may be involved in this. Any control they exert may ultimately impinge upon regulation of APC and degradation of the B cyclins. In addition, the pre-RC is established in late mitosis, and here we have shown that the Cdk Pho85 is activated at this time. This complexity supports the notion that the M/G1 switch is a major cell cycle transition along with the G1/S and G2/M transitions.