Activation of the G₂/M-Specific Gene CLB2 Requires Multiple Cell Cycle Signals

Abstract

In budding yeast (Saccharomyces cerevisiae), the periodic expression of the G₂/M-specific gene CLB2 depends on a DNA binding complex that mediates its repression during G₁ and activation from the S phase to the exit of mitosis. The switch from low to high expression levels depends on the transcriptional activator Ndd1. We show that the inactivation of the Sin3 histone deacetylase complex bypasses the essential role of Ndd1 in cell cycle progression. Sin3 and its catalytic subunit Rpd3 associate with the CLB2 promoter during the G₁ phase of the cell cycle. Both proteins dissociate from the promoter at the onset of the S phase and reassociate during G₂ phase. Sin3 removal coincides with a transient increase in histone H4 acetylation followed by the expulsion of at least one nucleosome from the promoter region. Whereas the first step depends on Cdc28/Cln1 activity, Ndd1 function is required for the second step. Since the removal of Sin3 is independent of Ndd1 recruitment and Cdc28/Clb activity it represents a unique regulatory step which is distinct from transcriptional activation.

Transcriptional programs play important roles in the control of the eukaryotic cell cycle. For example, transcriptional regulators have been found as recipients of external signals that either promote or inhibit cell cycle progression. They are crucial targets of checkpoint signals as well as effectors of the Cdk-driven cell cycle clock, thereby timing and coordinating important cell cycle transitions (17, 25). Although in budding yeast (Saccharomyces cerevisiae), transcriptional regulation is not an absolutely crucial feature for the directionality of the cell cycle, the serial activation and inactivation of at least nine transcriptional activators and their dedicated sets of genes represent a notable and well-studied cellular feature that underlines progression through the different cell cycle phase transitions (4, 9). Among the different transcriptional signatures that follow progression through the cell cycle, the G₂/M-specific expression pattern in yeast has attracted interest because its transcription machinery appears to present one of the elusive targets of the mitotic kinase (27). Moreover, recent evidence also showed that the transcription of G₂/M-specific genes is the target of checkpoint signals as well as external stress signals, raising questions concerning the nature and function of these signal transduction systems (28).

Previous work identified several of the key players of G₂/M-specific transcription. The MADS box protein Mcm1 provides an essential protein anchor for the factors that specify the regulatory features of this system (16). In direct association with Mcm1, the forkhead-related protein Fkh2 seems thereby to permanently occupy G₂/M-specific promoters (10, 14, 19, 30). For activation, a third factor, Ndd1, is necessary that is recruited to the promoter through the forkhead-associated (FHA) domain of Fkh2 (6, 12, 22). Ndd1 is under complex cell cycle regulation that includes transcriptional oscillations, protein turnover (15), and protein modifications. It is thought that Ndd1 phosphorylation by the mitotic kinase Cdc28/Clb2 is an activating event necessary for optimal recruitment of Ndd1 to G₂/M-specific promoters, of which CLB2 is an important member (6, 22). Since Clb2 kinase is necessary for full transcriptional activation of its own promoter, it has been proposed that a positive-feedback loop maintains G₂/M cluster gene activity until cells undergo mitotic exit (2). Due to the central role of Clb2 as a transcriptional activator and CLB2 as its own target gene, the G₂/M cluster is also referred to as the CLB2 cluster of cell cycle-regulated genes.

Although both Fkh2 and Ndd1 are classified as activators of G₂/M promoters, there are crucial differences in the phenotypes of fkh2 and ndd1 mutants. Whereas Ndd1 is essential for the cell, Fkh2 function is dispensable (12). This phenotypic difference cannot simply be explained by the presence of a paralog, Fkh1, because even fkh1 fkh2 double mutants are viable, although they show highly abnormal morphology (10). Since fkh2 ndd1 double mutants are also viable, Fkh2 has been proposed to act as a sequence-specific repressor during early phases of the cell cycle (12, 30). Indeed, Northern blot analysis has shown that fkh2 ndd1 cells lose the periodic expression of CLB2 cluster genes, exhibiting enhanced transcriptional levels in G₁ and reduced transcriptional levels in G₂ phase (12). Low expression of Clb2 protein in cells has severe morphological consequences, as such cells are not able to execute a timely switch from polarized to uniform growth in the bud, resulting in highly elongated “pseudohyphal” cell shapes (10, 13). The extent of this morphology is a reflection of the severity of the defect and is one of the hallmarks of the MCM1 conditional mutant (1) and of the fkh2 fkh1 and fkh2 ndd1 mutants (12, 30). Whether and how Ndd1 might function in the derepression and/or activation of its target promoters remained largely unexplored. There has been much speculation about a positive-feedback loop involving Clb2 kinase and Ndd1; however, studies have not progressed much beyond showing that Ndd1 can be a substrate for the Cdc28 kinase (6). A protein kinase-independent function of the Cdc28/Clb2 complex has been proposed; this function might include a structural role in the recruitment of transcriptionally important protein complexes (18).

In this work, we investigated how Fkh2 might impose its repressive function during the G₁ phase of the cell cycle and further asked which signals are necessary for converting this repressor into an activator. With the Sin3/Rpd3 histone deacetylase (HDAC) we identified a crucial player in the repression of G₂/M-specific promoters. We also found that the initial steps of derepression depend on the function of G₁-specific cyclins and not on the function of the mitotic kinase or Ndd1. Finally, we present data implying that the function of Ndd1 at the CLB2 promoter is not associated with histone acetylation but more likely with chromatin remodeling.

MATERIALS AND METHODS

Strains and plasmids.

Strains used in this study were derivatives of strain W303 (Table 1). Gene disruptions and epitope tags were created by integrational transformation of a PCR-amplified cassette as described previously (11). Double mutants were constructed either by standard genetic analysis or by a second step of integrational transformation. Strains disrupted with respect to ndd1 were constructed starting from a heterozygous diploid. This diploid strain was transformed with a YCplac33-derived plasmid expressing Ndd1 from a galactose-inducible promoter and sporulated on plates containing 0.1% galactose. Haploid ndd1 progeny were grown in the presence of 1% raffinose-1% galactose. Crosses of haploid ndd1 strains with other strains, in order to introduce modification at other loci, were always performed in the presence of galactose. Shifting growth conditions to include glucose to deplete Ndd1 was performed only during experiments (Northern blot, chromatin immunoprecipitation [ChIP], and viability assays). The plasmids used in this study (Table 2) were constructed using a PCR-based cloning strategy introducing the respective FKH2 fragments into the EcoRI and PstI cloning sites in pGBT9 or YCplac33. The GAL4_DBD-fkh2 1-306 R87A H123A construct (pJV292) was generated using a QuikChange site-directed mutagenesis kit (Stratagene) according to the instruction manual (primer sequences on request) and confirmed by sequencing. Standard methods for transformation techniques and genetic analysis were used.

TABLE 1.

Yeast strains

Strain	Genotype	Source
W303 1a	MATaho ade2-1 trp1-1 can1-100 leu2-3 his3-11 ura3-52 ssd1	Rodney Rothstein
K2771	MATacln1::cln2::cln3::LEU2 pGALpr-CLN1	Kim Nasmyth
K5875	MATacdc34-2ts pGALpr-SIC1V5V33A76-HA1	Kim Nasmyth
K7428	MATapGAL1-10 CDC20	Kim Nasmyth
MK155	MATaNDD1-HA6::HIS3	Koranda et al. (12)
MK195	MATapGAL1-10 CDC20 RPD3-HA6::HIS3 (from K7428)	This study
MK196	MATapGAL1-10 CDC20 SIN3-HA6::HIS3 (from K7428)	This study
MK257	MATaSIN3-HA6::TRP1	This study
MK267	MATaSIN3-HA6::TRP1 fkh2::HIS3	This study
JV305	MATabar1::SIN3-HA6::HIS3	This study
JV323	MATandd1::KanMX p[GAL1-10 NDD1 CEN URA3]	This study
JV1	MATabar1::	This study
JV361	MATabar1::sin3::HIS3	This study
JV363	MATabar1::rpd3::HIS3	This study
JV367	MATandd1::KanMX p[GAL1-10 NDD1 CEN URA3] SIN3-HA6::HIS3	This study
JV394	MATaRPD3-HA6::HIS3	This study
JV396	MATandd1::KanMX p[GAL1-10 NDD1 CEN URA3] RPD3-HA6::HIS3	This study
JV509	MATaSIN3-HA6::TRP1 fkh1::HIS3	This study
JV515	MATaSIN3-MYC18::TRP1 cdc34-2ts pGALpr-SIC1V5V33A76-HA1 (from K5875)	This study
HEL404	MATaSIN3-HA6::TRP1 cln1::cln2::cln3::LEU2 pGALpr-CLN1 (from K2771)	This study

TABLE 2.

Plasmids

Plasmid	Genotype	Source
pAS40	YCPlac33 NDD1-HA3 native promoter	This study
pJV251	pGBT9 GAL4BD-FKH2	This study
pJV287	pGBT9 GAL4BD-fkh2 1-306	This study
pJV292	pGBT9 GAL4BD-fkh2 1-306 R87A H123A	This study

Growth conditions.

Yeast precultures were grown in the appropriate selective media or in yeast extract-peptone media containing either 2% glucose or 1% raffinose and 1% galactose. Cells were diluted (optical density at 600 nm [OD₆₀₀] = 0.1 to 0.2) and grown to an OD₆₀₀ of 0.6 in rich media. To induce cell cycle arrest, bar1Δ strains (OD₆₀₀ = 0.4) were treated once with α-factor (1 μg/ml final concentration) during one replication. To arrest BAR1 strains, α-factor (0.5 μg/ml) was added four to five times (depending on the replication time of the strain) in intervals of 25 min. This resulted in a total concentration of 2 to 2.5 μg/ml. As presented in Fig. 2, additional synchronization of cells released from α-factor blocking was performed by the addition of nocodazole (5 μg/ml final concentration). In order to induce cell cycle arrest in GAL1-10 CDC20 strains, cells were grown in raffinose containing media for 2 h. In time course experiments, arrested cells were harvested by filtration, washed once with the respective medium, and released in permissive and/or restrictive conditions. In experiments in which cells were shifted from galactose to glucose conditions, 2% glucose was added 15 min before the release in order to suppress galactose-driven promoters. The cdc34^ts GAL1 SIC1V5V33A76 strain was grown in glucose-containing media and arrested at 27°C with α-factor during one replication. Cells were shifted to 37°C for 50 min (in order to adapt to higher temperature) and subsequently harvested by centrifugation, washed twice with prewarmed media, and released at 37°C in media containing galactose and raffinose.

FIG. 2.

The binding of Sin3 to the CLB2 promoter oscillates during the cell cycle. (A) Time course of SIN3-HA6 cells (JV305) released from a G₁ arrest induced by α-factor. To arrest cells in G₂ phase, nocodazole was added after 35 min postrelease. HA-ChIP was performed at the indicated time points. (B) CLB2 mRNA levels in wt (JV1) cells released from α-factor block. The transcriptional levels of CLB2 are compared with quantified Sin3-HA ChIP data derived from Fig. 2A. (C) Time course of GAL-CDC20 SIN3-HA6 (MK196) and GAL-CDC20 RPD3-HA6 (MK195) cells released from Cdc20-depleted G₂ arrest. HA-ChIPs were performed at the indicated time points. Positive and negative controls for Rpd3-HA recruitment were performed in logarithmically growing wt control strains (JV394 and w303). (D) Fkh2 presence at the GAL1-10 promoter is sufficient to mediate cell cycle-regulated recruitment of Sin3. The results of ChIP analysis of a SIN3-HA strain (JV361) expressing full-length Fkh2 fused to the DNA binding domain of Gal4 (pJV251) are presented. Data show the results of a time course experiment examining cells released from G₁ arrest induced by α-factor. The synchrony of cultures and the appropriate arrests (in G₁ and G₂ phase, respectively) were examined by fluorescence-activated cell sorter (FACS) analysis. Note that the strains used to produce the data shown in panels A, B, and D were bar1 strains, which exhibit a slight delay after release from α-factor block compared to BAR1 strains. WCE, whole-cell extract.

Northern blot analysis.

Samples were taken at indicated time points, diluted in ice-cold water, and pelleted by centrifugation. RNA isolation and separation were performed essentially as described in reference 7. Transferred RNA was simultaneously hybridized with two ³²P-labeled probes: (i) a fragment from the coding regions of CLB2 or SWI5 (EcoRV or HindIII and PstI digests, respectively) and (ii) a CMD1 fragment (PCR amplification) as an internal loading control. Probes were labeled using [α-³²P]dATP and a PrimeIt II random primer labeling kit (Stratagene). Blots were analyzed with a STORM 840 PhosphorImager (Molecular Dynamics), and signals were quantified with ImageQuant 5.0 software (Molecular Dynamics).

ChIP.

Assays were performed essentially as described previously (29). The antibodies used in ChIP experiments were as follows: (i) anti-hemagglutinin (anti-HA) (supernatant of 12CA5 hybridoma cells), (ii) anti-Myc antibody 4Ab, (iii) anti-acetyl histone H4 06-866 (Upstate), and (iv) anti-histone H4 ab10158-100 (Abcam). DNA was fragmented by sonication followed by centrifugation at 4°C to remove debris. Immunoprecipitation of chromatin by use of histone H4 antibodies required extended DNA fragmentation in order to restrict the length of DNA fragments. Coprecipitated DNA was analyzed with multiplex PCR, which included primer pairs amplifying DNA regions (ALD3, TEL, and FUS1) presumably unaffected by cell cycle-specific events. The PCR products were separated on an agarose gel. In cases where normalization was necessary the appropriate DNA bands were quantified using ImageQuant 5.0 software (Molecular Dynamics) and values were calculated using the unregulated DNA as an internal standard. Sequences of primer pairs used in chromatin analysis and detailed protocols are available on request.

RESULTS

Sin3 is involved in the negative regulation of CLB2 cluster genes.

The proposition of the idea of an active repression mechanism at G₂/M-specific promoters has its origin in the observation that loss of Fkh2 function can suppress the viability defect of ndd1 cells (12). To identify proteins involved in such an Fkh2-mediated repression mechanism, we tested loss-of-function mutants in several candidate repressor systems for their ability to rescue the proliferation of ndd1 cells. Whereas ssn6 or tup1 mutations did not give rise to viable ndd1 progeny after sporulation of heterozygous ndd1 diploids, we found that deletion of SIN3 allowed formation of ndd1 colonies (data not shown). The morphological characteristics associated with low-level expression of CLB2 cluster genes were less apparent in sin3 ndd1 mutants than in fkh2 fkh1 ndd1 mutants, suggesting a relatively high level of CLB2 expression in sin3 ndd1 strains (data not shown). Northern blot analysis also showed that mRNA levels of CLB2 cluster genes such as SWI5 and CLB2 were elevated in the G₁ phase of sin3 cells to a higher extent than in wild-type (wt) cells (Fig. 1A). To provide evidence that histone deacetylation is at the root of the repression mechanism in G₁ phase we performed a similar set of experiments with an rpd3 mutation. Again, cells that combined a deletion of NDD1 and RPD3 were viable (data not shown) and cells deleted for RPD3 exhibited derepressed CLB2 cluster genes in G₁ phase (Fig. 1A).

FIG. 1.

The Sin3/Rpd3 HDAC represses G₂/M promoters via Fkh2. (A) Deletion of SIN3 or RPD3 leads to promoter derepression in the G₁ phase. Wild-type (JV1), sin3 (JV361), and rpd3 (JV363) strains were either exponentially grown (log) or arrested in G₁ phase by α-factor. Expression levels of the G₂/M-specific genes SWI5 and CLB2 were determined by Northern blot analysis. Logarithmic growth and G₁ arrest data were determined by fluorescence-activated cell sorter (FACS) analysis. Blots were analyzed with a STORM 840 PhosphorImager (Molecular Dynamics), and signals were quantified with ImageQuant 5.0 software (Molecular Dynamics). (B) Sin3-HA recruitment to the CLB2 promoter requires Fkh2 but not Fkh1. ChIP directed against the HA tag was performed using a SIN3-HA (MK257) strain and tagged strains with either FKH2 (MK267) or FKH1 (JV509) deleted. Coprecipitated DNA regions were visualized by multiplex PCR using the indicated primer pairs for the DNA regions of interest. CLB2 cluster gene promoter regions are marked with an arrow. Unmarked DNA regions (e.g., ALD3 and FUS1) serve as internal controls. Lane 1: whole-cell extract (WCE) dilutions demonstrate equal amplification rates for all primer pairs comprised in the primer mix. (C) Sin3 is recruited by the N-terminal region of Fkh2, including the FHA domain. A SIN3-HA (JV305) strain was transformed with a series of vectors (pJV251, pJV287, and pJV292) expressing truncated Gal4_DBD-Fkh2 fusion proteins as illustrated. Chromatin IP with anti-HA antibodies was made in exponentially growing cultures, and the coprecipitated DNA was amplified via multiple PCR as described for panel B. Relative values for the GAL1-10 DNA (GAL1-10 increase with respect to ALD3) were calculated using ImageQuant software as described in the Material and Methods section. Values are derived from the results of three independent experiments.

Sin3 and Rpd3 recruitment at G₂/M promoters is dependent on the presence of Fkh2.

Our results clearly suggested the involvement of a Sin3/Rpd3 complex in G₂/M gene regulation. Yet we could not discount the possibility that these effects were of an indirect nature, as Sin3 complexes are part of many regulatory systems. We therefore measured promoter occupancy of CLB2 cluster genes by HA-tagged Sin3 or Rpd3 by use of a classical ChIP assay. PCR amplifications of CLB2 and SWI5 promoter regions indicated significant binding of Sin3-HA at these two G₂/M promoters in comparison to the results seen with either control promoters or the coding regions of CLB2 and SWI5 (Fig. 1B, lane 1, and data not shown). Moreover, the signal was absent in fkh2 cells but not in fkh1 cells (Fig. 1B, lanes 2 and 3). We obtained equivalent results by use of cells expressing Rpd3-HA (Fig. 2C and data not shown). Not only did these findings implicate Sin3 complexes as effectors of Fkh2 function, they also further emphasized the proposed functional differences between the two yeast forkhead proteins. To test whether the presence of Fkh2 at a promoter is sufficient for increased Sin3 recruitment we grafted the Gal4 DNA binding domain (Gal4_DBD) onto the amino terminus of Fkh2. The expression of the hybrid protein led to a significant increase in Sin3-HA binding at the GAL 1-10 promoter (Fig. 1C, schematic 1), an observation that provided further evidence for the intimate connection between Sin3/Rpd3 and Fkh2. Since Fkh2 seems to function as a common platform for both the transcriptional activator Ndd1 and the Sin3 complex, we wondered whether Ndd1 and Sin3 would compete for similar binding regions. In order to map the Fkh2 regions involved in Sin3 recruitment, we generated several truncations and mutant versions of Fkh2 fused to the Gal4_DBD and tested them for their ability to mediate Sin3-HA recruitment to the GAL1-10 promoter. Sequences amino terminal to the DNA binding domain of Fkh2 were able to recruit Sin3 to the GAL1-10 locus (Fig. 1C, schematic 2), suggesting that a region overlapping with the FHA domain was mediating the interaction. We also tested whether point mutations within the phosphothreonine recognition motif of Fkh2 that are known to prevent interaction between Ndd1 and Fkh2 (22) could also abolish recruitment of Sin3. An Fkh2 mutant containing alanine substitutions at positions R87 and H123 that completely abolishes Ndd1 recruitment still allowed Sin3-HA recruitment to the GAL1-10 promoter (Fig. 1C, schematic 3). Any deletion within amino acids located N-terminally to the FHA domain abolished interaction with Sin3 (data not shown). In summary, structures competent for recruiting Sin3 seem to be located in a region covering the first 194 amino acids of Fkh2 (data not shown). Although this region includes the complete FHA domain, its function as a phospho-amino acid acceptor is dispensable. Finally, we cannot discount the possibility that more than one site in Fkh2 contributes to Sin3 recruitment, as Fkh2 mutants missing the C-terminal domain seem to recruit Sin3 to a lesser extent. We thus propose that Sin3 recruitment to the N-terminal region of Fkh2 is mechanistically distinct from the Fkh2-Ndd1 interaction, a situation that might even allow the simultaneous binding of Sin3 and Ndd1.

Sin3 and Rpd3 occupation rates oscillate during the cell cycle.

To determine whether Sin3-HA binding at G₂/M promoters is in any way regulated, we looked at cells in different phases of the cell cycle. In cells arrested by pheromone exposure, the Sin3-HA protein efficiently coprecipitated with the CLB2 promoter region. This result correlated well with the fact that the promoter is transcriptionally inactive at the G₁ stage of the cell cycle (Fig. 2B). When released from the pheromone block, Sin3-HA interaction with the CLB2 promoter clearly diminished after 20 min, a time point at which bud emergence and DNA synthesis became noticeable and the CLB2 locus was activated (Fig. 2A). Interestingly, after the completion of S phase, promoter occupancy by Sin3 was reestablished. The timing of the Sin3 rerecruitment was validated by assaying cells treated with the microtubulus-depolymerizing agent nocodazole. This experiment showed that Sin3 is indeed associated with the promoters during G₂/M phase and that the initial observation is not an artifact emerging due to asynchrony of cell cultures (compare Fig. 2A and D). To confirm this binding behavior of Sin3, we also arrested cells in metaphase via Cdc20 depletion. Again, the Sin3 signal, as assayed by CLB2 promoter ChIPs, was substantial. We released the cells from metaphase arrest by using a galactose-inducible GAL1-CDC20 construct. A time course experiment showed that the Sin3-HA-derived signal became only slightly stronger after release and remained almost constant until it peaked during G₁ phase. Sin3 occupation levels then decreased, coinciding with the beginning of S phase (Fig. 2C, upper panel, last lane). We also monitored cell cycle-specific chromatin recruitment of a tagged version of Rpd3. These ChIP assays yielded a pattern virtually identical to that observed for Sin3. At the CLB2 promoter, Rpd3 recruitment also peaked during early G₁ followed by a rapid decline shortly before the onset of DNA synthesis. It also reappeared at the promoter during G₂/M phase (Fig. 2C, lower panel). S phase therefore seems to have been the only cell cycle phase in which the Sin3/Rpd3 complex was completely absent and perhaps actively removed from the promoter.

As mentioned before, the N-terminal region of Fkh2 is able to recruit Sin3 to a heterologous promoter, and we wondered whether such an artificial interaction is still cell cycle regulated. We analyzed synchronized cells expressing the Gal4_DBD-Fkh2 fusion construct. Our ChIP data clearly show that both events (the removal of Sin3 from the GAL1-10 promoter in S phase and its reappearance in G₂ phase) mimicked the regulatory pattern observed on the CLB2 promoter (Fig. 2D).

The decrease of Sin3 recruitment correlates with an increase of histone H4 acetylation at the CLB2 locus.

Due to the direct interaction between Sin3 and Fkh2, we expected to observe a difference in histone acetylation levels between wild-type cells and sin3 mutants. Indeed, SIN3-defective cells arrested in G₁ phase exhibited a significant increase in histone H4 acetylation at the CLB2 promoter (Fig. 3B). Interestingly, ChIP data derived from exponentially growing wt cells did not show any H4 acetylation signals (data not shown). A possible explanation is that the assumed acetylation event was transient. Consequently, the fraction of cells exhibiting acetylated promoter regions might be too small to be detected in a sample of cells comprising all stages of the cell cycle. In order to test this hypothesis, we monitored the acetylation status of histone H4 during the transition through S phase, as Sin3 was absent at this stage of the cell cycle. Cells were released from pheromone arrest, and samples for anti-acetyl-H4-based ChIP analysis were taken. As before, no or very little H4 acetylation was detectable with respect to the CLB2 promoter in pheromone-arrested cells. At the time of Sin3 dissociation, a strong but transient acetylation signal appeared in the promoter region of the CLB2 locus (Fig. 3C). The transient appearance of the acetylation signal was confined to the CLB2 promoter region, a region encompassing the multiple Mcm1 and Fkh2 binding sites of CLB2 (depicted as CLB2 in Fig. 3A). PCR amplification of three other CLB2 regions (depicted as CLB2 tata, CLB2 atg, and CLB2 orf in Fig. 3A) covering the CLB2 untranslated region and the CLB2 open reading frame resulted in a constitutive signal with little variation from G₁ phase to the beginning of S phase (Fig. 3D, time points 0′ to 20′). However, concomitant with the appearance of Ndd1 at the promoter (Fig. 3E, time point 30′), histone H4 acetylation signals were lost from the whole CLB2 locus (Fig. 3D, time points 30′ to 50′). Based on previous observations (21), we assumed that the disappearance of the acetylated histone signal was due to a general clearance of nucleosomes from the promoter rather than to an increase in deacetylation. ChIP experiments with antibodies against histone H4 showed that the level of the histone H4 signal at the CLB2 promoter decreased immediately after histone acetylation occurred and that it stayed low through S phase until G₂ (Fig. 4A). Interestingly, other CLB2 regions (CLB2 atg, CLB2 orf) were more efficiently immunoprecipitated by histone H4 than the CLB2 promoter region at these stages of the cell cycle (Fig. 4B). Thus, it seems that histone acetylation precedes the removal of nucleosomes from the region adjacent to the promoter (CLB2, CLB2 tata).

FIG. 3.

Sin3 relief from the CLB2 promoter leads to local H4 acetylation. (A) Scheme of the CLB2 locus, indicating binding regions of Mcm1 and Fkh2 and four different DNA fragments amplified by the indicated primer pairs. (B) H4 histones at the CLB2 promoter (CLB2) are acetylated in sin3 cells. Chromatin IP was performed with antibodies directed against acetylated histone H4 in G₁-arrested wt and sin3 cells (JV361). (C) At the CLB2 region H4 histones were transiently acetylated in S phase. A wt strain was released from pheromone-induced G₁ arrest and subjected to chromatin IP using anti-acetyl-histone H4 (upper panel). Multiplex PCR data from three independent experiments were quantified with ImageQuant software as described for Fig. 1C. (D) Chromatin regions downstream of the CLB2 promoter (CLB2 tata, CLB2 atg, and CLB2 orf) lose the histone H4 acetylation signature after S-phase completion, which correlates with Ndd1 recruitment. An NDD1-HA (MK155) strain was released from pheromone-induced G₁ arrest and subjected to chromatin IP using anti-acetyl-histone H4. (E) Ndd1-HA recruitment to the CLB2 promoter region. Cell lysate from the experiment whose results are presented in panel D was subjected to chromatin IP using anti-HA antibodies. Synchrony of cultures and cell cycle arrests were examined by fluorescence-activated cell sorter (FACS) analysis. WCE, whole-cell extract.

FIG. 4.

H4 histones lose contact to the CLB2 promoter region after S-phase completion. Samples of a pheromone-synchronized wt culture were withdrawn at the indicated time points and examined by ChIP with anti-H4 antibodies. The coprecipitated DNA was analyzed with two different PCR setups containing mixtures of primer pairs as indicated in the figure (ALD3, CLB2, and TEL, panel A; CLB2tata, CLB2atg, CLB2orf, and TEL, panel B). The presented data were normalized with ImageQuant software using the TEL DNA band as the internal standard. Note that in order to provide synchrony, cell lysates used for the anti-acetyl histone H4 ChIP experiments whose results are presented in Fig. 3C were also subjected to the anti-histone H4 ChIP experiments whose results are presented in this figure. Whereas the telomeric region (TEL) represents a negative control in anti-acetyl histone H4 ChIP experiments, it serves as a positive control for histone presence. WCE, whole-cell extract.

Ndd1 function is required for nucleosome clearance but not for Sin3 removal and histone acetylation.

Is Ndd1 involved in overriding Sin3/Rpd3-mediated promoter repression? To answer this question we first inquired whether cells depleted for Ndd1 could still remove the Sin3 complex from the CLB2 promoter. For this purpose we analyzed the promoter occupancy of Sin3 and Rpd3 in cells that expressed Ndd1 from a galactose-inducible promoter. The cells were pregrown in galactose-containing medium, arrested in G₁ by pheromone treatment, and subsequently released into glucose-containing medium. After progression through S phase they arrested in G₂ due to the lack of Ndd1. At this stage they displayed the typical elongated bud morphology of cells with low Clb kinase activity (Fig. 5A). Still, Sin3/Rpd3 removal from the CLB2 promoter was indistinguishable from wt cell results (Fig. 5A). Operating under the assumption that Ndd1 was effectively depleted, we suggest that Sin3 removal most likely does not depend on Ndd1 activity. Using a similar experimental setup, we now addressed the question of whether Ndd1 function affects the histone acetylation pattern. After release from G₁ arrest, histone H4 acetylation increases irrespective of the presence of Ndd1. After S-phase completion, however, Ndd1-depleted cells exhibited significantly higher levels of acetylated histone H4 at the CLB2 promoter than cells expressing Ndd1 (Fig. 5B). In order to confirm this observation, we measured the promoter occupancy by H4 histones. As expected, H4 histone levels at the CLB2 promoter region were significantly elevated in the absence of Ndd1 (Fig. 5C). We thus conclude that Ndd1 was neither antagonizing the function of the Sin3/Rpd3 HDAC nor triggering the increase in the acetylation of H4 histones. Instead, Ndd1 appears to function in the displacement of nucleosomes from the promoter region.

FIG. 5.

Sin3 relief and histone H4 acetylation at the CLB2 promoter region are followed by Ndd1-dependent histone H4 clearance. (A) The Sin3/Rpd3 complex is released from the CLB2 promoter in cells depleted for Ndd1 activity. ChIP-time course experiments were performed with SIN3-HA (JV367) and RPD3-HA (JV396) strains, both deleted for NDD1 and expressing Ndd1 under the control of a GAL1-10 promoter. Cells were grown under galactose conditions, arrested in G₁ phase with α-factor, and released into glucose-containing medium. Samples for chromatin IP using anti-HA antibodies were collected at the indicated time points. The synchrony of the culture was examined by fluorescence-activated cell sorter (FACS) analysis. Impaired Ndd1 activity is reflected by elongated bud morphology and cell cycle arrest in G₂. (B) The decrease of acetyl-histone H4 signal at the CLB2 promoter region in G₂ requires Ndd1 presence. An ndd1 GAL1-10 NDD1 (JV323) strain was arrested with α-factor in G₁ and released in either galactose (+Ndd1)- or glucose (−Ndd1)-containing medium. DNA was amplified by multiple PCR as described for Fig. 1 and quantified with ImageQuant software. Values for the CLB2 promoter region were calculated using TEL signals as an internal standard. (C) Promoter remodeling requires Ndd1. Strain and experimental setup were as described for panel B. Samples were subjected to anti-H4 IP, and coprecipitated DNA was amplified with the indicated primer pairs. Signals for the CLB2 promoter region are emphasized (arrow). The presented data were normalized with ImageQuant software using the TEL DNA band as an internal standard. Note that ALD and TEL signals represent positive controls for histone presence. (D) Samples for Northern blot analysis of the cultures described for panel C were probed for CLB2 transcripts and CMD1 as a reference. WCE, whole-cell extract.

The release of Sin3 is independent of Clb kinase but requires G₁ cyclins.

Since several cell cycle-specific Cdks are activated in rapid succession during G₁/S transition, we wanted to know which of their forms might be required for the regulated removal of Sin3. Also, to assure that the effect is not just due to cellular growth and to cells surpassing a critical size we first tested a strain devoid of G₁ cyclin expression for its ability to initiate Sin3 removal. For this purpose we used a strain deleted for all the G₁ cyclin genes but kept alive by expressing CLN1 from a galactose-inducible promoter. Cells were pregrown on galactose, arrested in G₁ by pheromone, and released either in galactose-containing media or in glucose-containing media. We found that cells lacking Cln activity did not execute the removal of Sin3 from the CLB2 promoter. These cells continued to gain in cell size but were unable to bud and initiate S phase. However, when these cells were placed in galactose media, Sin3 dissociated normally from the CLB2 promoter (Fig. 6A). To assure that the persistence of Sin3 binding to the CLB2 promoter in cells depleted for G₁ cyclins was not a simple consequence of a prolonged pheromone arrest we monitored the decrease in the FUS1 expression, which is a sensitive indicator for the efficacy of pheromone signaling (Fig. 6B). The transcriptional levels of this gene decreased in G₁ cyclin-depleted cells with kinetics similar to those detected in cells expressing Cln1, indicating that the pheromone block was successfully inactivated in both cultures.

FIG. 6.

Sin3 removal from the CLB2 promoter requires CLN activity. (A) Sin3 dissociation requires cyclin activity. A ChIP experiment with anti-HA antibodies was performed using an Sin3-HA strain deleted for CLN1, CLN2, and CLN3, expressing Cln1 under the control of the GAL1-10 promoter (HEL404). Cells were grown in galactose-containing medium, arrested in G₁ phase, and released in either the absence (right panel) or the presence (left panel) of galactose. Cell synchrony was examined by fluorescence-activated cell sorter (FACS) analysis. (B) Cultures from the experiment described for Fig. 6A were successfully released from the α-factor (α-f.)-related cell cycle block. The results of Northern blot analysis of α-factor-dependent FUS1 expression are shown. post rel., postrelease; gal., galactose; gluc., glucose. (C) High-level Cln activity enables the dissociation of Sin3 from G2 promoters. ChIP-time course experiments were performed using a cdc34^ts strain carrying the nondegradable SIC1V5V33A76 allele under the control of the GAL1 promoter (JV515). Cells were grown under permissive conditions, arrested in G₁ phase, and released at 37°C in galactose-containing medium. Samples for ChIP analysis using anti-Myc antibodies were collected at the indicated time points followed by amplification of the DNA regions of interest with the indicated primer pairs. Cell synchrony was determined by FACS analysis. (D) Northern blot analysis of CLN1 expression levels for two time course experiments showing similar FACS profiles but different budding indices (A [right] and C). Cells were harvested simultaneously with cells used for ChIP analysis. WCE, whole-cell extract.

The previously described experiment indicated that Sin3 removal from G₂/M promoters is indeed one of the Start-dependent processes, and yet it did not resolve the question of whether any of the B-type cyclins were necessary for its execution. In order to answer this question, we analyzed Sin3 binding in a cdc34^ts strain expressing a nondegradable SIC1 allele under the control of a galactose-inducible promoter. At the G₁/S boundary, the Cdc4/SCF complex and the E2 enzyme Cdc34 initiate the ubiquitination and degradation of phosphorylated Sic1, which is the main inhibitor of B-type cyclin-associated Cdk activity. Impaired SCF activity combined with nondegradable Sic1 represses B-type cyclin activity while allowing further accumulation of G₁ cyclins (23). We arrested the cdc34^ts strain in G₁ and released it in galactose-containing media at restrictive temperature (37°C). Under these conditions, cells started to bud but failed to replicate their DNA. However, Sin3 binding to the CLB2 promoter decreased as cells started to bud (Fig. 6C), proving that B-type cyclin-dependent kinase activities are not involved in this event. The idea of the independence of Sin3 dissociation from B-type cyclin expression was further strengthened by time course experiments with cells conditionally inactivated for the Clb1-4 kinase (clb1Δ, clb2Δ, clb4Δ, and clb2^ts). Sin3 was still removed from the CLB2 promoter region in S phase (data not shown). We conclude that Sin3 removal from G₂/M promoters is solely triggered by the rise in Cln kinase activity and independent of the activation of any of the Clb kinases.

DISCUSSION

In this study, we identified the Sin3 histone deacetylase complex as a crucial negative regulator of G₂/M-specific gene expression in the yeast. While exploring the regulation of this repressor we uncovered previously unrecognized connections between the cyclin-dependent kinase activities and the function of the different G₂/M-specific transcriptional regulators. Up to now, most studies on this subject have focused either on a feedback loop involving Clb2 kinase and the activator Ndd1 (2, 22) or on Clb5 kinase and the carboxy terminus of Fkh2 (20). Most recently, another positive-feedback system was uncovered that is based on phosphorylation of Ndd1 by the polo kinase Cdc5 (5). The CDC5 gene is one of the essential members of the CLB2 cluster of genes. All of these studies dealt with the high expression of CLB2 cluster genes during G2/M phase but did not really shed light on the transcriptional initiation events at the beginning of S phase. This work now provides evidence that the proposed modifications of Ndd1 and Fkh2 by Clb kinases and Cdc5, respectively, cannot be the only connection between the cell cycle clock and G₂-specific transcription. According to our evidence, Sin3-mediated chromatin deacetylation provides an additional level of cell cycle control. Its reversal by the promoter-specific removal of the deacetylation complex actually seems to represent the first crucial step towards activation. Unexpectedly, this event happens prior to and independently of Ndd1 recruitment. It also does not require the function of any of the B-type cyclins. Instead, it seems to be part of the events orchestrated by G₁-specific cyclins at the G₁/S boundary. Thus, positive feedback is not part of the initial transcriptional activation choreography at G₂/M promoters. Instead, the previously proposed feedback mechanisms might just ensure that the expression of G₂/M-specific genes never decreases to levels that lead to premature rereplication in case of attenuated late cell cycle stages.

How might Cln kinases regulate the function of Sin3? Because Sin3 complexes are involved in many different processes (26), the effects of this kinase activity must be restricted to a small subset of complexes relevant for cell cycle progression. One obvious solution to the problem would be that the factors mediating the recruitment of Sin3 are the target of the kinase. Although several transcriptional corepressors such as Ume6 or Ash1 have been found to be stably associated with the core members of the large Sin3 complex (4) we suggest that the recruitment of this complex is solely regulated via its interaction with Fkh2. Our data clearly show that Fkh2 is directing the recruitment, because fkh2 mutants lose the Sin3 signal and a Gal4-Fkh2 fusion is sufficient for an increased Sin3 signal at the GAL1-10 locus. Results with Gal4DB-Fkh2 chimeras also strongly suggest that the regulated recruitment of Sin3 is a property associated with Fkh2 and largely independent of promoter context. These data also suggest that the C-terminal domain of Fkh2, which is missing in Fkh1, might play only an accessory role and not an essential role in Sin3 recruitment. The idea that Fkh2 interacts with the Sin3 complex directly is further supported by a global study on yeast protein complexes (10) in which the association of the proteins was clearly documented. Still, a large number of scenarios explaining the question of Cln kinase targets are possible: it could be Fkh2 itself (20), an as-yet-unknown adaptor between Sin3 and Fkh2, or any component of the core Sin3 complex that is involved. Two observations would support the last contention. Mass spectrometry data identified Sin3 as coprecipitating specifically with Cln2 but not with B-type cyclins (3). Moreover, a phosphorylated SPXXK motif was identified within Sin3 during screening for Cdc28-specific phosphopeptides in the yeast proteome (8), although its connection to a specific version of the protein kinase and to a physiological process has not been pursued.

Another aspect of our work concerns the changes in the chromatin composition of the CLB2 promoter during the switch from the repressed to the activated state. Our data support a two-step model in which the H4 histones adjacent to the promoter region are acetylated in a first step due to the removal of the Sin3/Rpd3 complex. In a second step, these histones are then expelled (Fig. 7). It is only the second step that requires the function of Ndd1, implying that the factor is not part of the respective histone acetylase complex. In contrast, the transcribed region of the CLB2 locus shows only a slight decrease in H4 histone binding, although the histone acetylation signal seems to be entirely lost with similar kinetics as the nucleosome(s) is removed from the promoter. Currently we cannot determine whether this event is a consequence of a deacetylase activity or whether it reflects the selective expulsion of acetylated histones (Fig. 7). Recent work unraveled the positions and the cell cycle-specific remodeling of nucleosomes at the CLB2 locus by use of chromatin analysis based on micrococcal nuclease digestion patterns (24), and it is worthwhile to compare these results. Especially interesting is the fact that the promoter region is indeed hypersensitive with respect to the nuclease. It therefore seems likely that our ChIP data for the promoter region (CLB2) reflect modifications and the remodeling of mainly one nucleosome. A second interesting observation is the fact that the nucleosome patterns across the transcribed region of CLB2 undergo several distinct changes during transcriptional activation that might indicate the repositioning of nucleosomes towards the 3′ end (25). We propose therefore that Ndd1 might be the factor that directs the recruitment of a still-unidentified form of chromatin-remodeling machinery necessary for the high level of transcriptional activation observed during the S phase. These changes would have to be restored during the later stages of the cell cycle, and it is likely that the two remodeling factors Isw1 and Isw2 are involved in this process. This idea is supported by the observations that cells deleted for ISW2 and ISW1 can bypass ndd1 lethality (personal observation and reference 25), that isw2 and isw1 mutations affect the distribution of nucleosomes across the CLB2 locus, and that the cells exhibit transcript levels higher than normal for CLB2 (25).

FIG. 7.

Model of chromatin changes at the CLB2 locus during the transition from repression to activation. Ovals represent nucleosomes and their putative positions at the CLB2 locus as described by Sherriff et al. (24). Nucleosomes with acetylated H4 histones are depicted with small flags. The model is based on results presented in Fig. 1 to 6 and reflects data generated by primer pairs amplifying three different regions of the CLB2 locus (CLB2, CLB2 tata, and CLB2 atg). Vertical arrows signify different kinase activities triggering the transitions from G₁ phase to S phase and from S phase to G₂ phase. The question mark signifies that in G₂ phase the presence of nonacetylated H4 Histones might be a consequence of either histone deacetylation (detached flag) or histone removal (detached oval) combined with histone remodeling (small horizontal open arrows).

If Sin3 removal from the CLB2 locus occurs independently of Ndd1 and Cdc28/Clb kinase activity, why does the inactivation of the deacetylase complex suppress the ndd1 lethality? One possible explanation can be found in the observation that Sin3 has already rebound to the promoter during the end of S phase and continuously occupies the region until the next cycle. If the target nucleosomes are not removed from the promoter, then the activity of the Sin3 complex could prematurely induce the formation of a repressive structure. The idea of such an antagonistic role for Ndd1 and Sin3 in G₂ phase is supported by our observation that overexpression of Ndd1 in a sin3Δ background is lethal (data not shown), which might be due to an inability of such cells to repress mRNA production of G₂/M-specific genes during and after mitotic exit. In summary, our results strengthen the view that activation of the G₂/M-specific gene CLB2 is not due to one simple activation mechanism but is generated by a complicated choreography of steps that are sequentially linked to separate cell cycle signals.