|Home | About | Journals | Submit | Contact Us | Français|
Schizosaccharomyces pombe GATA factor Ams2 is responsible for cell cycle-dependent transcriptional activation of all the core histone genes peaking at G1/S phase. Intriguingly, its own protein level also fluctuates concurrently. Here, we show that Ams2 is ubiquitylated and degraded through the SCF (Skp1-Cdc53/Cullin-1-F-box) ubiquitin ligase, in which F box protein Pof3 binds this protein. Ams2 is phosphorylated at multiple sites, which is required for SCFPof3-dependent proteolysis. Hsk1/Cdc7 kinase physically associates with and phosphorylates Ams2. Even mild overexpression of Ams2 induces constitutive histone expression and chromosome instability, and its toxicity is exaggerated when Hsk1 function is compromised. This is partly attributable to abnormal incorporation of canonical H3 into the central CENP-A/Cnp1-rich centromere, thereby reversing specific chromatin structures to apparently normal nucleosomes. We propose that Hsk1 plays a vital role during post S phase in genome stability via SCFPof3-mediated degradation of Ams2, thereby maintaining centromere integrity.
► Cell cycle-regulated histone transcription factor Ams2 is stable only during G1/S ► Ams2 gets phosphorylated by DDK during S phase and presumably G2 phase ► Phosphorylated Ams2 is ubiquitylated by SCFpof3 and degraded via the proteasome ► Ectopic Ams2 expression interferes with histone homeostasis and centromere function
The ubiquitin-proteasome system plays a pivotal role in various cellular processes, including cell cycle progression, cell proliferation, and differentiation (Hershko et al., 2000). Substrate proteins are ubiquitylated by the enzymatic cascade consisting of ubiquitin-activating enzyme (E1), ubiquitin-conjugating enzyme (E2), and ubiquitin ligase (E3) (Ravid and Hochstrasser, 2008). These ubiquitin transferase reactions result in the formation of polyubiquitin chains on substrates, which are recognized by the 26S proteasome, followed by rapid irreversible degradation. Among the three enzymes, the E3 ubiquitin ligases are central to determining the timing and specificity of substrate proteolysis.
There are two conserved ubiquitin ligases that regulate cell cycle progression: anaphase promoting complex/cyclosome (APC/C) and Skp1-Cdc53/Cullin-1-F-box (SCF) (Peters, 2006; Petroski and Deshaies, 2005). Although these two E3s appear to have diverged from a common ancestor, they show one important difference in substrate recognition; whereas APC/C recognizes a small motif within individual substrates, called the “destruction box” or “KEN box,” SCF binds specifically to phosphorylated substrates. Therefore, it is generally accepted that the main spatiotemporal regulation of SCF-mediated proteolysis occurs via substrate phosphorylation by cell cycle-regulated protein kinases, rather than activation and inactivation of this ligase complex itself.
Hsk1/Cdc7 protein is a serine/threonine protein kinase, which is conserved from yeast to human (Masai et al., 1995; Sclafani, 2000; Snaith et al., 2000). Mutants in hsk1+ in fission yeast and CDC7 in budding yeast show defects in the initiation of DNA replication, indicating that the activity of the protein is essential for S phase progression. Hsk1/Cdc7 has a regulatory subunit, Dbf4/Dfp1 (Him1), and as such this kinase complex is called Dbf4-dependent protein kinase (DDK). In budding yeast, Dbf4 is a substrate of APC/C, whereby Cdc7 kinase activity is low during G1 and becomes activated coincident with the initiation of S phase (Ferreira et al., 2000; Weinreich and Stillman, 1999). Although it remains to be established whether Dfp1/Him1 is under the control of APC/C in fission yeast, the expression of Dfp1/Him1 is known to be regulated by the cell cycle, peaking at the G1/S transition when the protein kinase activity is maximal, such that its presumptive substrates, the minichromosome maintenance (MCM) proteins, are phosphorylated in an Hsk1-dependent manner (Brown and Kelly, 1999; Takeda et al., 2001). Six MCMs compose a DNA helicase complex that is recruited to replication origins and unwinds DNA duplex to promote replication fork progression. Although some other proteins, such as Mrc1 and Swi6, are suspected to be phosphorylated in a DDK-dependent manner (Bailis et al., 2003; Shimmoto et al., 2009), it is still unclear how DDK-dependent phosphorylation of these substrates is involved in the initiation of DNA replication. It remains, thus, to be established whether the regulation of S phase progression is the sole role of the DDK. For example, Hsk1 kinase is known to play an important role in chromosome segregation and gene silencing at heterochromatin independent of its requirement for S phase regulation (Bailis et al., 2003).
Canonical histone gene transcripts in lower eukaryotes and plants are polyadenylated and their accumulation is limited to S phase. In budding yeast, histone mRNA levels are short lived, resulting in their elimination when transcription is terminated at the end of S phase (Osley, 1991). Various factors are known to be required for periodic transcription of histone genes. One such factor is HIRA, which is a highly conserved histone chaperone. Budding yeast HIRA is essential for repression of histone transcription outside S phase (Sherwood et al., 1993; Xu et al., 1992). The HIRA histone chaperone has also been reported to be involved in nucleosome assembly and the establishment of repressive chromatin (Blackwell et al., 2004; Gunjan et al., 2005). In fission yeast, HIRAs are encoded by two genes, hip1+ and slm9+ (Blackwell et al., 2004; Kanoh and Russell, 2000); high levels of histone mRNAs are maintained throughout the cell cycle in cells lacking these genes (Anderson et al., 2009; Blackwell et al., 2004; Takayama and Takahashi, 2007). Recently, we reported that Ams2, a cell cycle-regulated GATA-type transcription factor (Chen et al., 2003a, 2003b), is also required for periodic expression of core histone genes at S phase (Takayama and Takahashi, 2007). Unlike HIRA, in the absence of Ams2, transcriptional activation of histone genes does not occur during S phase, leading to very low levels of histone transcription throughout the cell cycle. Ams2 protein appears to further promote centromeric localization of the centromere-specific histone H3 variant Cnp1 (the fission yeast CENP-A homolog) during S phase (Takayama et al., 2008). Ams2 protein accumulates on the nuclear chromatin when chromosomes are duplicated during S phase, whereas very little signal is detected during other phases of the cell cycle, including mid-to-late G2 and early M phase (Chen et al., 2003a; Takayama and Takahashi, 2007).
In this study, we explored the pathways underlying Ams2 protein oscillation during the cell cycle. Our results indicated that Ams2 is phosphorylated and degraded in an Hsk1 kinase- and ubiquitin-proteasome-dependent manner. Ams2 interacts with SCFPof3, which is essential for cell cycle-dependent degradation of Ams2. Furthermore, Hsk1-Dfp1 associates with and phosphorylates Ams2 at specific sites. We will discuss the physiological significance of Hsk1- and SCFPof3-mediated degradation of Ams2 for histone gene expression, cell viability, and specific chromatin structure at the centromere.
Our previous studies indicated that the ams2+ gene is expressed periodically with a peak at G1/S phase, and accordingly its protein is detectable only during G1, S, and early G2 (Chen et al., 2003a; Takayama and Takahashi, 2007). A number of proteins with this pattern of cell cycle-specific oscillation are regulated not only transcriptionally but also posttranslationally, i.e., by proteolysis (Baber-Furnari et al., 2000; Yamano et al., 2000). To examine Ams2 protein stability at different stages of the cell cycle, cell extracts were prepared from strains blocked at specific points in the cell cycle with temperature-sensitive (ts) cdc mutations or hydroxyurea (HU) treatment. To facilitate a protein half-life assay for Ams2, each strain contained the pREP41-Ams2 plasmid, which carried the ams2+ gene under the control of a thiamine-repressible nmt41 promoter. These cells were cultured in minimal medium in the absence of thiamine at the permissive temperature (derepressed, 26°C), followed by shift-up to the restrictive temperature (36°C). Transcription and translation of Ams2 were simultaneously attenuated by adding thiamine and the protein synthesis inhibitor cycloheximide (CHX), respectively. Flow cytometric analysis to determine the DNA content confirmed that these cells were arrested at specific cell cycle stages (see Figure S1C available online).
In asynchronous wild-type cell cultures consisting largely of G2 cells, Ams2 was extremely unstable, with a half-life of ~20 min (Figures 1A and 1B). In cdc10-129 G1-arrested cells, Ams2 protein was also degraded, although at a much slower rate than that in asynchronous cells. To investigate Ams2 stability at early S phase, cells were arrested with HU (Figures 1A and 1C) or by cdc22-C11 mutation (Figures S1A and S1B); as shown in these panels, Ams2 protein was substantially stabilized in early S phase cells. Ams2 in cdc2-33 cells, mostly G2 arrest (Novak and Mitchison, 1989), were degraded with similar kinetics to wild-type cells (Figures 1A and 1D), consistent with the data of asynchronous cultures. Ams2 was also unstable in cdc25-22 G2 mutants (Figures S1A and S1B). To examine Ams2 stability during M phase, nda3-KM311 cold-sensitive mutant, defective in β-tubulin, was used. Most nda3-KM311 cells were arrested at prometaphase after 14 hr incubation at the restrictive temperature, which was confirmed by visual inspection of their nuclear morphology (data not shown). As shown in Figure 1E, Ams2 protein in nda3-KM311 cells was unstable and degraded rapidly. Taken together, these results indicated that Ams2 protein is unstable during G2 and M phase, partially stabilized in G1 phase, and stable during S phase.
As the first step in exploring the mechanism of Ams2 proteolysis, we examined the role of the proteasome pathway in this process. In fission yeast, the mts2+ and mts3+ gene products have been identified as subunits of the 26S proteasome (Seeger et al., 1996). We used temperature-sensitive mutations mts2-1 and mts3-1 to inactivate proteasome function. Mutant cells carrying pREP41-Ams2 were cultured at the restrictive temperature (36°C) for 3 hr and Ams2 synthesis was repressed by addition of thiamine and CHX as described above (Figure 2A). Whereas Ams2 underwent rapid proteolysis in wild-type cells (Figures 2A and 2B, open circles), degradation of Ams2 was markedly decreased in arrested mts2-1 (filled circles) and mts3-1 (data not shown) cells. These observations indicated that Ams2 proteolysis requires the proteasome-dependent pathway.
We next examined ubiquitylation of Ams2, again via mts2-1 and mts3-1 mutants. To visualize conjugated ubiquitin, these strains were transformed with pREP1-His6-ubiquitin or pREP1-HA-ubiquitin plasmid. After incubation for 3 hr at the restrictive temperature, His-tag-ubiquitylated proteins were purified and the resulting extracts were blotted with Ams2 antibody. As shown in Figure 2C and Figure S2, a characteristic smear pattern was seen in the His-ubiquitin lanes (lanes 1 and 3). Control cells expressing HA-ubiquitin did not show any bands (lanes 2 and 4), confirming the specificity of pull down. Note that, as expected, ubiquitylated Ams2 was more abundant in mts2-1 or mts3-1 than in wild-type cells (Figure 2C; Figure S2). Immunoblotting with ubiquitin antibody also showed similar results (data not shown). These results demonstrated that Ams2 is ubiquitylated in vivo.
We next examined the ubiquitin ligase responsible for Ams2 proteolysis. There are two ubiquitin ligases that regulate cell cycle progression, APC/C and SCF (Peters, 2006; Petroski and Deshaies, 2005). As reported previously (Takayama and Takahashi, 2007) and described above (see Figure 1), Ams2 protein is newly synthesized in G1/S phase, covalently modified in S phase, and degraded during the subsequent G2 phase. Because APC/C is likely to be inactive during G2 phase, we examined whether SCF is required for Ams2 degradation. Ams2 protein stability was measured with an HA-tagged Skp1 shut-off strain (Yamano et al., 2000). When Skp1 expression was repressed by adding thiamine, Ams2 protein was stable and detected until 120 min after CHX treatment (Figure 3A, +Thiamine). In contrast, Ams2 was degraded quickly in the absence of thiamine (−Thiamine), i.e., in the presence of Skp1. Note that the S phase cyclin Cig2, which is a substrate of SCF (Yamano et al., 2000), was also stabilized in the absence of Skp1. These results suggested that SCF is responsible for Ams2 proteolysis.
Fission yeast contains 18 ORFs encoding F-box proteins (Hermand, 2006). Three of these F box proteins, Pop1, Pop2, and Pof3, play crucial roles in cell cycle progression and coordination (Katayama et al., 2002; Toda et al., 1999). The half-life of Ams2 was examined with deletion mutants of each of these F box protein genes. The results indicated that Ams2 was stabilized in the pof3 deletion mutant, but not in pop1 or pop2 mutants (Figure 3B; Figure S3). Consistent with Ams2 stabilization in pof3 deletion mutant, immunoblotting with asynchronous cell extracts revealed that Ams2 protein was more abundant in pof3 mutant than in wild-type cells (Figure 3C). To confirm the accumulation of Ams2 protein in pof3 cells throughout the cell cycle, Ams2 protein was visualized with C-terminal GFP-tagged Ams2. The intensity of the Ams2-GFP signal was markedly stronger in pof3 mutant cells than in wild-type controls. Furthermore, the Ams2-GFP signal was clearly visible even during late G2 phase in pof3 mutant cells, whereas it was visible only during G1/S phase and early G2 phase in wild-type cells (Figure 3D; Chen et al., 2003a); the frequency of uninucleate cells (corresponding to G2 phase) displaying nuclear Ams2-GFP signals was increased to 63% in the mutant compared with only 30% in wild-type cells (Figure 3D). This result substantiated the suggestion that Pof3 is required for Ams2 proteolysis.
Finally, if Ams2 were a substrate for SCFPof3, Pof3 would bind Ams2. Immunoprecipitation was performed with nontagged wild-type or Pof3-13myc cells. Ams2, as well as Skp1, coimmunoprecipitated with myc antibody in the presence of Pof3-13myc but not in its absence (Figure 3E). Taken together, the results shown here support the suggestion that SCFPof3 is an E3 ubiquitin ligase responsible for cell cycle-dependent degradation of Ams2.
Phosphorylation of substrates is required for many F box proteins to recognize individual substrates (Patton et al., 1998). Moreover, our previous study showed that Ams2 protein migrates as two forms on gel electrophoresis, in which a faster migrating form appears first coincident with a peak of histone gene expression and is replaced by slower migrating forms, which disappear thereafter (Takayama and Takahashi, 2007). These observations strongly suggest that Ams2 is phosphorylated during S phase. To examine this point further, Ams2-HA was precipitated with HA antibody from HU-arrested cells, followed by λ-phosphatase (λ-PPase) treatment. We detected slower migrating forms of Ams2 (Figure 4A, lane 3), which were converted to a faster migrating form by λ-PPase treatment (lane 4). Importantly, a heat-inactivated phosphatase could not convert Ams2 banding patterns (lane 5), establishing that Ams2 is phosphorylated in vivo.
Visual inspection of the Ams2 amino acid sequence indicated the presence of two putative Cdc2 phosphorylation sites (Figure 4B, boxes and asterisks) with the consensus sequence S/T-P-X-K/R (where X is any amino acid). Interestingly, these two putative CDK sites are conserved in Ams2 homologs found in two other Schizosaccharomyces species (Schizosaccharomyces japonicus and Schizosaccharomyces octosporus, Figure S4). There are two additional conserved sites with the sequence S/T-P (Figure 4B, asterisks; Figure S4). We generated three types (M1, M2, and M3) of alanine-substituted mutants of Ams2 in corresponding threonine (T586) and serines (S492, S599, and S601) by site-directed mutagenesis. Expression of these mutants with episomal plasmids was able to rescue the slow growth of ams2 deletion mutant cells (data not shown), indicating that none of these alanine substitutions are loss-of-function mutations. We then examined the half-lives of Ams2 alanine mutants, either alone or in combination. As shown in Figure 4C, each single Ams2 mutant (M1, M2, or M3) or combinations of M1 and M2 (M1&M2) were still unstable. However, three types of M3 mutant (M1&M3, M2&M3, and M1&M2&M3) showed substantial stabilization (Figure 4C).
Given the stabilization of a specific set of Ams2 alanine mutants, the phosphorylation status of each mutant was examined by immunoprecipitation as described above (see Figure 4A). As shown in Figure 4D, stabilized forms of Ams2 mutant proteins (M2&M3), but not unstable forms (M2 or M3), were hypophosphorylated, although there still appeared to be some residual phosphorylation (lanes 13, 14, and 15). Thus, Ams2 is phosphorylated in multiple sites that are required for its degradation, which is consistent with the suggestion that Ams2 is a target of SCFPof3.
Having established that Ams2 is ubiquitylated and degraded by SCFPof3 in a phosphorylation-dependent manner, we examined which protein kinase(s) is responsible for Ams2 phosphorylation. As mentioned above, two of the phosphorylation sites (T586 and S601) match the consensus CDK site. However, our data showed that Ams2 is still unstable in the cdc2-33 mutant (see Figures 1A and 1D). This implies that protein kinase(s) other than Cdc2 is involved in Ams2 degradation. Accordingly, we focused on the Hsk1-Dfp1 (Cdc7/Dbf4) kinase, because its activity is high during S phase when Ams2 phosphorylation occurs (this study; Chen et al., 2003a; Takayama and Takahashi, 2007).
To address the involvement of Hsk1, we used a temperature-sensitive hsk1-89 mutant that cannot form colonies at 30°C–32°C, but can grow at 25°C or 37°C (Matsumoto et al., 2005). The in vitro kinase activity of Hsk1-89 mutant protein was significantly impaired even at permissive temperatures (Takeda et al., 2001). Wild-type and hsk1-89 cells were synchronized in S phase by HU treatment at 25°C or 37°C and immunoblotting with Ams2 antibody was performed. As shown in Figure 5A, unlike wild-type cells (Figure 5A, lanes 3 and 4), hsk1-89 mutant cells showed only a faster migrating band of Ams2 at either temperature (lanes 5 and 6). Identical results were also obtained in the absence of HU treatment (Figure S5A), indicating that hypophosphorylation of Ams2 in the hsk1-89 mutant is not attributable to HU-induced S phase arrest, but is instead due to the deterioration of Hsk1 function per se. Despite this apparent lack of Ams2 band-shift, however, careful immunoprecipitation, followed by phosphatase treatment, showed that Ams2 is still phosphorylated in this mutant (Figure 5B, lanes 4, 5, and 6). These observations suggested that Hsk1-89 mutant protein has some residual activity. Alternatively, another kinase may be involved in Ams2 phosphorylation.
To test whether Ams2 degradation is dependent on Hsk1 function, Ams2 stability was examined by immunoblotting. Protein extracts were prepared form wild-type and hsk1-89 mutant cells cultured at 37°C, where hsk1-89 mutant cells kept growing with very little Hsk1 kinase activity. Whereas the slower migrating forms appeared in wild-type cells (Figure 5C, upper panel, arrows) that were mostly degraded within ~30 min, no corresponding slower forms were observed in hsk1-89 cells, and crucially Ams2 was significantly stabilized (lower panel). Taken together, these observations indicated that Ams2 is phosphorylated in an Hsk1-dependent manner during the cell cycle, which results in SCFPof3- and proteasome-mediated ubiquitylation and degradation.
The observations outlined above strongly suggest that Hsk1 kinase directly phosphorylates Ams2. To assess this possibility, we first examined physical interaction between Ams2 and Hsk1-Dfp1 kinase by yeast two-hybrid assay (Figure 5D; Figure S5B). Yeast cells expressing Hsk1 and Dfp1 showed strong β-galactosidase (β-gal) activity, consistent with their stable association. Notably, cells expressing Dfp1 and Ams2 also showed substantial β-gal activity, which indicated physical association of Ams2 with Dfp1. In contrast, Ams2 is unlikely to associate stably with Hsk1, because cells with Ams2 and Hsk1 did not show β-gal activity. These results suggest that Ams2 physically interacts with Hsk1 kinase through the Dfp1 subunit. Physical association of Ams2 with Hsk1-Dfp1 kinase was indeed confirmed by immunoprecipitation (Figure 5E); HA-tagged Ams2 was coimmunoprecipitated with FLAG-tagged Hsk1 and vice versa. It is of note that, because crosslinking by 1% formaldehyde treatment (Bailis et al., 2003) was required to detect coimmunoprecipitation, their interaction may be transient and/or unstable in vivo.
We next examined whether Hsk1 kinase phosphorylates Ams2 in vitro. For this purpose, the C-terminal fragment (aa 500–697) of wild-type and three types of mutant Ams2 (M2, M3, and M2&M3; see Figure 4B) were expressed and purified from bacteria. These Ams2 fragments were incubated with [γ-32P]ATP and Hsk1 immunopurified from fission yeast cells. As shown in the autoradiogram (Figure 5F, DDK), 32P was incorporated efficiently into the wild-type and M2 mutant Ams2 fragment (lanes 1 and 2, respectively), whereas the M3 and M2&M3 mutant fragments showed markedly reduced 32P incorporation (lanes 3 and 4, respectively). These results indicated that Hsk1 kinase phosphorylates Ams2 directly in vitro, and that at least one of the serine residues (S599 and S601) mutated in the M3 fragment is a major phosphorylation site.
We sought to establish the physiological significance of cell cycle-dependent oscillation of Ams2 levels. In wild-type cells, constitutive expression of Ams2 from the plasmid pREP41-Ams2 induced a modest delay in cell division (Figure 6A, top panel); the doubling time was 3.8 hr in cells expressing Ams2, whereas it was 2.8 hr when Ams2 was repressed. Further induction with a stronger promoter (pREP1-Ams2) resulted in complete growth inhibition (Figure 6A, second panel, and Figure 6B). Neither pREP41-Ams2 nor pREP1-Ams2 plasmid retarded cell growth when the promoter was turned off (Figure S6A). Consistent with Ams2 stabilization in the hsk1-89 mutant (Figure 5C), pREP41-Ams2 impeded cell division in this mutant background (Figure 6A, bottom). Therefore, Ams2 oscillation is important for proper cell proliferation and becomes particularly indispensable when Hsk1 activities are compromised.
As reported previously, the ams2+ gene is required for cell cycle-regulated expression of core histone genes (Takayama and Takahashi, 2007). Therefore, we performed northern blotting to investigate how Ams2 overexpression interferes with core histone gene expression. For this purpose, we used various cdc mutations or HU treatment to prepare the cells at individual stages of the cell cycle and Ams2 was mildly overexpressed from pREP41-Ams2 in these cells. As shown in Figure 6C and Figure S6B, all the core histone genes examined showed augmented expression by ectopic Ams2 (lanes 3 and 7). In the case of HU treatment, histone gene expression was already highly induced by HU alone (lane 6, OFF), and very little, if any, further induction, even lower than in the case of cdc10 or cdc2 mutants, was observed on overproduction of Ams2 (lane 5, ON), the reason of which remains currently unknown. In asynchronous cell population where most cells were in G2, constitutive expression of wild-type Ams2 as well as the M2&M3 mutant protein also increased the levels of histone expression (lanes 1 to 3 in Figure 6D). These results indicated that core histone genes are constitutively transcribed, irrespective of cell cycle stage by Ams2 overproduction. Surprisingly but consistent with the data presented earlier (Figure 5A), we found that the levels of histone mRNAs were elevated in hsk1-89 mutant cells even in the absence of an exogenous Ams2-expressing plasmid (lanes 7 to 9 in Figure 6D). This result further supports our conclusion that phosphorylation by Hsk1 kinase downregulates Ams2 activity by triggering SCF-dependent proteolysis, which subsequently attenuates the expression of histone genes. It would be noteworthy that, in contrast to constitutive expression from pREP plasmids, expression of the M2&M3 mutant Ams2 from its native promoter did not affect the levels of histone expression (Figure 6E). Because the protein levels of the mutant and the wild-type Ams2 were not significantly different when they are transcribed from the native promoter (data not shown), we hypothesized that the amount of Ams2 protein is regulated by not only proteolysis but also by transcriptional control. Consistent with this hypothesis, the mRNA levels of M2&M3 mutant ams2 decreased by 14% in comparison with that of wild-type ams2+ (bottom panel in Figure 6E). Native ams2+ transcription may be, thus, regulated by a negative feedback mechanism that maintains the appropriate expression levels of Ams2.
We next examined the effects of ectopically produced Ams2 on chromosome stability, because complete depletion of Ams2 causes severe chromosome instability. The rate of minichromosome loss was increased by more than 30-fold in cells constitutively producing Ams2 (Figure 6F). This result suggests that reduction of Ams2 level outside S phase is required for faithful chromosome transmission. Interestingly, it has been reported that hsk1-1312 thermosensitive mutation also caused severe chromosome instability (Snaith et al., 2000), which appears consistent with our finding that inactivation of Hsk1 stabilizes Ams2. We presumed that ectopic expression of Ams2 impairs the centromere chromatin structure, which was examined as described below.
The chromatin structure of the central centromere in fission yeast is distinct from that in other regions, by which micrococcal nuclease (MNase) digestion exhibits unstructured smeared patterns instead of regular nucleosomal ladders (Hayashi et al., 2004; Takahashi et al., 2000). We showed previously that this central centromere-specific chromatin structure is impaired in the absence of Ams2, converting to the ladder patterns (Chen et al., 2003a). Given the constitutive expression of histone genes and chromosome instability induced by Ams2 overproduction, we examined the chromatin structure of the centromere under these conditions. As shown in Figure 7A, Ams2 overproduction resulted in the appearance of nucleosomal ladders at the central centromere region (cnt) in parallel with those seen in ams2+ deletion mutants. Note that normal smeared patterns emerged when episomal Ams2 expression was repressed or an empty vector was introduced into the cells. Prolonged induction of Ams2 led to more pronounced ladder patterns (20 hr).
To determine how Ams2 overproduction affects nucleosome composition at the central centromere region, we performed chromatin immunoprecipitation (ChIP) assay with Cnp1, H3, and H4 antibodies. Unlike normal conditions where negligible levels of canonical H3 are found in cnt (see control cells without exogenous Ams2 expression [Vec and Ams2-OFF] in Figure 7B, IP:H3, cnt), Ams2-overproducing cells contained significant amounts of H3. Despite this, the amount of Cnp1 in this region was unaffected by Ams2 overproduction (IP:cnp1, cnt). In parallel with the increase in level of H3, incorporation of H4 was also enhanced substantially (IP:H4, cnt). This observation implied that ectopic Ams2 promotes additional incorporation of canonical H3/H4 nucleosomes into the cnt region. This effect appears not to be specific to the centromere, because we found that the act1+ gene, which was used as a control for euchromatin, exhibited similar increases in H3 and H4 loading (IP:H3, H4, act1). Therefore, it is possible that Ams2 overproduction enhances the amount of core histone binding to DNA throughout the entire genome. Nevertheless, we envisage that aberrant centromere chromatin structures are likely to be a reason, if not the sole reason, for chromosome instability resulting from Ams2 overproduction. It is worth noting that despite the abnormal structure of cnt chromatin, gene silencing at the centromere appeared to be less affected in Ams2-overproducing cells (Figure S7), the reason for which was not investigated further. The presence of Cnp1 may be sufficient for gene silencing in this region.
Taken together, we propose that cell cycle-dependent gene expression and proteolysis of Ams2, mediated by Hsk1 kinase and SCFPof3 ubiquitin ligase, play vital roles in the maintenance of genome integrity. Any perturbation in Ams2 levels would lead to chromosome instability and structural alterations in the centromere regions.
The fission yeast GATA factor Ams2 is required for periodic expression of histone genes and is expressed only during G1/S phase. However, the mechanism of Ams2 protein oscillation and its physiological significance are not understood. In this study, we showed that Ams2 levels are regulated not only transcriptionally but also posttranslationally via the ubiquitin-proteasome pathway. We found that Ams2 is unstable during G2 and M phase, partially stabilized in G1 phase, and stable during S phase. We then explored the pathways responsible for the periodic fluctuations in Ams2 protein level. Our findings can be summarized as follows. (1) Ams2 binds SCFPof3 and is stabilized in pof3 or skp1 cells. (2) Ams2 is a cell cycle-dependent phosphoprotein and mutant hypophosphorylated Ams2 protein (M2&M3) is stabilized. (3) Hsk1-Dfp1 kinase, the activity of which is required for Ams2 phosphorylation and proteolysis, physically interacts with and phosphorylates Ams2 in vitro. (4) Ectopic expression of Ams2 leads to constitutive histone gene expression and alterations in nucleosome composition at the central centromere, which result in genome instability (Figure 7C).
To ensure the integrity of the genome, Ams2 must be degraded rapidly once S phase completes DNA replication. Surprisingly, proteolysis of Ams2 is triggered by Hsk1-dependent phosphorylation, which is widely viewed as an S phase-promoting kinase. Hsk1 has been reported to phosphorylate various proteins required for DNA replication, such as MCM, ORC, and primase (Francis et al., 2009; Lee et al., 2003), although the physiological significance of their phosphorylation remains obscure. To our knowledge, this is the first report showing that DDK-dependent phosphorylation regulates stability of a protein that is apparently not involved in DNA synthesis directly. Hsk1-DDK, therefore, may not only promote DNA replication but also destabilize proteins and play an important role in the integrity of histone organization via phosphorylation-mediated proteolysis of Ams2. It should be noted that earlier studies by Forsburg's group predicted late-S phase roles of Hsk1 other than initiation of DNA replication (Snaith et al., 2000). Our finding of the Hsk1 role in histone expression control may substantiate their prediction.
In vitro kinase assay clearly demonstrated that Ams2 is a substrate for Hsk1-DDK. M3 mutation (S599A and S601A) abolished Ams2 phosphorylation by DDK, so it is presumed that at least either S599 or S601 is phosphorylated by this kinase. Although the M2 mutation (T586A) was phosphorylated by DDK in vitro as efficiently as wild-type Ams2, we assume that T586 is also phosphorylated in vivo, because Ams2 was stabilized only when both the M2 and the M3 mutations were introduced into the cells, and single M3 mutant Ams2 still underwent phosphorylation-mediated SCF-dependent proteolysis (Figure 4C). Although the kinase responsible for Ams2 phosphorylation at T586 has not yet been identified, this unidentified kinase is likely to be activated by Hsk1-DDK, because hsk1-89 single mutation was sufficient for full stabilization of Ams2 (Figure 5C). Intriguingly, T586 of Ams2 is within the CDK consensus site (Figure 4B). Therefore, it is possible that Ams2 destruction is coordinately regulated by CDK and S phase promoting DDK. It is notable that S601 also constitutes the CDK phosphorylation site. One attractive possibility is that DDK acts as a priming kinase for CDK that subsequently phosphorylates T586 and S601.
F box Pof3 plays a critical role in the maintenance of genome integrity (Katayama et al., 2002). pof3 deletions induce spontaneous DNA damage and impinge on gene silencing at heterochromatin, including the centromere. A role for Pof3 in S phase progression was implied previously when we showed that SCFPof3 binds Mcl1, a Polα-associated protein (a homolog of budding yeast Ctf4 and human AND-1) (Mamnun et al., 2006; Williams and McIntosh, 2002). Mcl1 is required for appropriate S phase propagation, chromosome segregation, and importantly, centromere structures (Mamnun et al., 2006; Natsume et al., 2008; Williams and McIntosh, 2002), reminiscent of the defective phenotypes exhibited by misregulation of Ams2 levels or Pof3 deletions. Because upregulation of Ams2 level does not recapitulate all of the phenotypes of pof3 deletions, it is likely that Ams2 is not the sole target, but is one of multiple targets of SCFPof3. It is of note that in budding yeast the Pof3 homolog Dia2 is recently shown to associate with Ctf4 on the replication fork, by which this F box protein is intimately involved in the progression of DNA replication (Kile and Koepp, 2010; Mimura et al., 2009; Morohashi et al., 2009). Understanding how S phase progression and post-S phase events are coordinated in the cell via Hsk1- and SCFPof3-mediated pathways would be the next step of our study.
We have shown that both the constitutive presence and the absence of Ams2 led to malfunctioning of the central centromere and global cell division defects (Figures 6 and 7; Chen et al., 2003a; Takayama et al., 2008). Cell cycle oscillation of Ams2 levels, therefore, appears to be essential for maintenance of centromere integrity. Consistent with these suggestions, ChIP analysis and MNase digestion demonstrated that ectopic expression of Ams2 results in aberrant chromatin, which consists of a mosaic of H3/H4 and Cnp1/H4 nucleosomes, at the central centromere. Interestingly, it has been reported that mutations in hsk1+ or pof3+ gene severely impaired chromosome stability (Katayama et al., 2002; Snaith et al., 2000). Defects in Ams2 proteolysis might cause, at least in part, chromosome instability in these mutants.
Although upregulation of Ams2 leads to constitutive expression of canonical histones, it is unlikely that abnormal histone expression itself impairs the chromatin structure at the central centromere; although deletion of hip1+ and slm9+ genes results in elevated histone expression outside S phase in fission yeast (Blackwell et al., 2004; Takayama and Takahashi, 2007), deletion of these HIRA genes does not impair the chromatin structure at the central centromere (Blackwell et al., 2004). Instead, HIRA deletion derepresses expression of marker genes embedded in pericentromeric heterochromatin, which is normally silenced in the wild-type cell (Greenall et al., 2006). Therefore, we postulated that ectopically produced Ams2 promotes not only expression of canonical histones but also the deposition of canonical nucleosomes into the central centromere in cell cycle phases other than S phase. As reported previously (Takayama and Takahashi, 2007), Ams2 promotes the deposition of Cnp1/H4 nucleosomes at the centromere core in S phase. Although it remains to be clarified as to how the Cnp1/H4 nucleosomes are specifically incorporated by Ams2, presumably ectopic Ams2 exhibits promiscuous substrate specificity, and therefore may impose incorporation of canonical histones into the central centromere.
Although obvious Ams2 homologs have not been identified in either budding yeast or metazoans, this is probably ascribable to the high rate of drift of Ams2 sequences during evolution, as even among Schizosaccharomyces species (S. pombe, S. japonicus, and S. octosporus), Ams2-related proteins have diverged very rapidly (28% identity, Figure S4). In all organisms examined to date, canonical histone genes are transcribed in a cell cycle-dependent manner, peaking at G1/S phase (Marzluff et al., 2008; Osley, 1991). Our results established that cell cycle-dependent expression of histone genes is crucial for proper cell division and genome integrity. To achieve this oscillatory pattern, the fission yeast has evolved a regulatory network, which consists of the transcription factor Ams2, Hsk1-DDK, and SCFPof3-mediated ubiquitin-proteasome pathways. Because DDK is a universal S phase kinase, we envisage that the regulatory pathways described here may be functionally conserved through evolution.
The yeast strains used in this study are listed in Table S1. The general techniques and media used for manipulation of fission yeast were described previously (Mamnun et al., 2006; Moreno et al., 1991; Takayama et al., 2008). Experiments were performed at 33°C unless otherwise stated. Micrococcal nuclease digestion assay and ChIP assay were performed as described in Takahashi et al. (2000) and Takayama et al. (2008).
Wild-type, cdc10-129, cdc2-33, cdc25-22, and mts2-1 cells carrying pREP41-Ams2 (wild-type) or alanine mutant plasmids were grown in the absence of thiamine overnight at 26°C, followed by temperature shift up to 36°C for 3.5 hr. For HU treatment, wild-type cells carrying pREP41-Ams2 plasmid were grown in a similar manner and HU (12 mM) was added upon temperature shift-up. Wild-type and nda3-KM311 cells carrying pREP41-Ams2 plasmid were grown overnight at 33°C, followed by temperature shift down to 20°C for 14 hr. Then, 2 μM thiamine and 100 μg/ml CHX (time 0) were added to all these cultures. For experiments shown in Figures Figures3B3B and and5C,5C, only 100 μg/ml CHX were added (time 0) to wild-type and pop1, pop2, pof3, and hsk1-89 cells. Cells were harvested at indicated time points and cell extracts were prepared as described previously (Mamnun et al., 2006; Takayama and Takahashi, 2007) and immunoblotting with Ams2 antibody was performed.
ImageJ software (v.1.37, NIH) was used to quantify Ams2 protein levels. α-tubulin (detected by TAT-1 antibodies, kindly provided by Dr. Keith Gull, Oxford University, UK) was used as a loading control.
Ubiquitylation analysis was performed as described by Takeda and Yanagida (2005). His6-tagged ubiquitin (His-Ub) or HA-tagged ubiquitin (HA-Ub) under the nmt1 promoter was overproduced in a wild-type strain or a proteasome-defective mts2-1 or mts3-1 mutant. These cells were precultured at 26°C in the minimal EMM2 liquid media containing 2 μM thiamine, resuspended in EMM2 without thiamine, and grown overnight, and the cultures were shifted to 36°C for 3 hr. For preparation of cell lysates, the cells were washed with buffer I (10 mM Tris-HCl [pH 7.5], 100 mM Na phosphate, 0.1% NP40, 10 mM Imidazole, 6 M Guanidine-HCl) and broken by vortexing with glass beads, and the lysates were cleared by centrifugation at 13,000 × g for 10 min. Cell lysates were incubated for 4 hr at room temperature with the TALON beads (BD bioscience Clontech, Palo Alto, CA), and washed twice with buffer I and 4 times with buffer II (10 mM Tris-HCl [pH 7.5], 100 mM Na phosphate, 0.1% NP40, 10 mM Imidazole, 1 mM PMSF). Ams2 protein pulled down with the beads was detected by immunoblotting with Ams2 antibody.
Yeast cell (YAP61-1 and mts3-1, Table S1) extracts were prepared in RIPA lysis buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS) according to the standard method described previously (Moreno et al., 1991). Soluble proteins (22 mg) were incubated with anti-myc antibody-coupled affinity resin (Covance), followed by SDS-PAGE and immunoblotting. For coimmunoprecipitation of Ams2 and Hsk1, we modified a previously published method (Bailis et al., 2003). Cells were crosslinked with 1% formaldehyde for 1 hr on ice then broken by glass beads in HB buffer (25 mM Tris-HCl [pH 7.5], 15 mM EGTA, 15 mM MgCl2, 300 mM NaCl, 60 mM β-glycerophosphate, 15 mM p-nitrophenylphosphate, 1 mM DTT, 0.1% NP40, 0.5 mM Na3VO4, 0.1 mM NaF, 1 mM PMSF, and 1 × Complete Mini EDTA-free; Roche Diagnostics). Extracts were then treated with DNaseI (Takara, 700U) and subjected to immunoprecipitation.
Cell lysates were prepared in HB buffer broken by vortexing with glass beads. The cleared lysates (1 mg of protein) were incubated with HA antibodies (12CA5; Roche) for 2 hr at 4°C followed by addition of Dynabeads anti-mouse IgG (Dynal Biotech). After incubation, the beads were washed 4 times with PBS and incubated for 30 min at 30°C in the presence of 200 units of λ-phosphatase (New England Biolabs) or with heat-denatured phosphatase. The beads were then washed with PBS, and samples were run on a 6% SDS-polyacrylamide gel, followed by immunoblotting with appropriate antibodies.
Immunoprecipitation of Dfp1/Hsk1 (DDK) was performed essentially as described previously (Brown and Kelly, 1998). Yeast cell (SP4024, Table S1) lysates were prepared with glass beads in lysis buffer (50 mM HEPES [pH 7.8], 10% glycerol, 10 mM sodium β-glycerophosphate, 50 mM NaF, 1 mM Na3VO4, 1 mM PMSF, mammalian protease inhibitor cocktail; Sigma). NaCl (250 mM) and 0.05% Tween-20 were added and incubated for 5 min after lysis. The extracts were spun to clarify and then incubated with anti-FLAG M2 affinity gel (A2220; Sigma) for 2 hr. After washing three times with lysis buffer plus salt and Tween-20, the beads were eluted with 3 × FLAG peptide (Sigma). In vitro DDK assay was carried out as described previously (Kakusho et al., 2008).
Fluorescence microscopy was performed with a Zeiss Axioplan microscope, a chilled video charge-coupled device camera (C4742-95; Hamamatsu Photonics), and the Volocity 3.1 software (Improvision Inc.) and processed with Adobe Photoshop version 7.0.1.
The minichromosome loss assay was performed as described previously (Takahashi et al., 1994). The cells containing Ch10-CN2 minichromosome were precultured in thiamine-containing EMM2 liquid medium without adenine. Cells were then plated on thiamine-lacking EMM2 plates supplemented with leucine (7 mg/L) and adenine (2.5 mg/L). The numbers of total colonies and red colonies (Ade-) were counted after 4 day incubation at 33°C.
We would like to thank Robin Allshire, Grant Brown, Keith Gull, Osami Niwa, Mitsuhiro Yanagida, Hisao Masai, Pierre Hentges, and Antony M. Carr for providing the reagents used in this study. We also thank Kojiro Ishii and our laboratory members for helpful discussions and support. Special thanks are due to Kohta Takahashi, who contributed to this work in the initial phase of the study. Y.T. was supported by a postdoctoral fellowship from the Japan Society for the Promotion of Science. Y.M.M. was supported by postdoctoral fellowships from the FWF (J-2398) and EMBO (ALTF53-2004). This work was supported by Cancer Research UK (T.T.), Grant-in-aid for Young Scientists (B) from the Japan Society for the Promotion of Science (S.S. and Y.T.), the Ishibashi Foundation for the Promotion of Science (Y.T.), and Marie Curie Cancer Care and the Association for International Cancer Research (H.Y. and M.T.).
Published: March 15, 2010
Supplemental Information includes Supplemental Experimental Procedures, seven figures, and one table and can be found with this article online at doi:10.1016/j.devcel.2009.12.024.