Histone acetyltransferases (HATs) and histone deacetylases (HDACs) play important roles in cell regulation, including cell cycle progression, although their precise role in mitotic progression remains elusive. To address this issue, the effects of HDAC inhibition were examined upon a variety of mitotic mutants of the fission yeast Schizosaccharomyces pombe, which contains three HDACs that are sensitive to trichostatin A (TSA) and are similar to human HDACs. Here it is shown that HDACs are implicated in sister chromatid cohesion and separation. A mutant of the cohesin loader Mis4 (adherin) was hypersensitive to TSA and synthetically lethal with HDAC deletion mutations. TSA treatment of mis4 mutant cells decreased chromatin-bound cohesins in the chromosome arm region. By contrast, HDAC inhibitors and clr6 HDAC mutations rescued temperature sensitive (ts) phenotypes of the mutants of the ubiquitin ligase complex anaphase-promoting complex/cyclosome (APC/C), which display metaphase arrest. This suppression coincided with facilitated complex formation of APC/C. Moreover, our mass spectrometry analysis showed that an APC/C subunit, Cut23/APC8, is acetylated. HATs and HDACs might directly target adherin and APC/C to ensure proper chromosome segregation, and anti-tumour effects of HDAC inhibitors could be attributed to this deregulation.
Introduction
Two classes of enzymes control protein acetylation. Histone acetyltransferases (HATs) and histone deacetylases (HDACs) target lysine residues in the N-terminal tails of histones. Most HATs and HDACs have been found in transcriptional factor complexes and, therefore, have been considered to regulate transcription by modulating acetylation levels of the chromatin (Kouzarides, 1999; Kuo and Allis, 1998). Acetylation may be a major protein modification in cell regulations rivalling phosphorylation (Kouzarides, 2000). Acetylation plays a role in the cell cycle, based on studies identifying the biological function of HATs and HDACs. Budding yeast Esa1, a MYST (MOZ, Ybf/Sas3, Sas2, Tip60) family HAT, is essential for cell viability, and its mutation results in G2/M phase arrest (Clarke et al., 1999). Mammalian TAF250 and its budding yeast homologue Taf145, the largest subunits of the TFIID complex having HAT activity, are required for G1 phase progression (Suzuki-Yagawa et al., 1997; Walker et al., 1997). The importance of HDACs for cell cycle progression was suggested by studies on the effects of HDAC inhibitors on cell proliferation (Johnstone and Licht, 2003; Marks et al., 2001). Low concentrations of trichostatin A (TSA), a potent HDAC inhibitor, are able to arrest mammalian cells at the G1 and G2 phases (Noh and Lee, 2003; Yoshida et al., 1990). Consistently, recent analyses by siRNA and targeted disruption demonstrated that class I HDACs are required for cell proliferation in carcinoma cells and mouse embryos (Glaser et al., 2003; Lagger et al., 2002). However, the molecular mechanisms for cell cycle arrest induced by HDAC inhibition are not well understood.
HDACs form a large protein family and are categorized into three classes according to sequence and structural similarities to budding yeast HDACs; Class I and Class II HDACs are similar to Saccharomyces cerevisiae Rpd3p and Hda1p, respectively, whereas newly discovered Class III HDACs are the Sir2-like NAD-dependent HDACs thought to be involved in regulating cellular NAD levels and life span regulation. Fission yeast S. pombe has three Class I and II HDACs, Clr6, Clr3 and Hos2/Phd1/Hda1, that are sensitive to TSA and regulate histone acetylation levels in the genome, cooperating with Class III HDACs (Bjerling et al., 2002; Kim, Y. et al., 1998; Olsson et al., 1998; Wiren et al., 2005). Mutants of clr3 and clr6 were originally identified as defective in silencing of the mating type locus (Ekwall and Ruusala, 1994; Grewal et al., 1998; Thon et al., 1994). Currently, it has been established that HDACs have an important role in heterochromatin formation and maintenance. Culture of S. pombe cells for several rounds of doubling in the presence of TSA causes disruption of centromeric heterochromatin as well as an increased rate of chromosome loss (Ekwall et al., 1997; Olsson et al., 1999). A similar result was obtained for higher eukaryotic cells (Taddei et al., 2001). Furthermore, several lines of evidence suggest that HDACs are implicated in chromosome segregation. Clr6 is the sole essential HDAC for cell viability in fission yeast, and clr6-1 mutants show abnormal chromosome segregation with lagging chromosomes (Grewal et al., 1998; Nakayama et al., 2003; Silverstein et al., 2003). TSA-treated mammalian cells also show chromosome segregation defects and prolonged mitotic delay (Cimini et al., 2003; Taddei et al., 2001). Centromeric heterochromatin is considered to be important for centromere formation; therefore, mitotic defects caused by HDAC inhibition may be attributable to centromere dysfunction. However, it was recently shown that a human Class I HDAC, HDAC3 modulates Aurora B kinase function in mitosis by deacetylating and converting histone H3s into preferred substrates (Li et al., 2006), implying that HDACs might have a further role in mitotic progression, independent of transcriptional regulation and heterochromatin formation.
To understand the role of HDACs in the cell cycle, especially in mitotic progression, the inhibitory effect of HDACs was studied in fission yeast S. pombe, a well-established model organism for cell cycle analysis. Underlying molecular mechanisms were further investigated.
Results
TSA affects cell growth in a subset of fission yeast mitotic mutants
To investigate how HDACs are involved in mitotic progression, the effects of an HDAC inhibitor, TSA, were examined at the permissive temperature (26°C) on cell growth of S. pombe temperature-sensitive (ts) mutants defective in different aspects of mitosis (mutant strains examined are listed in Table 1). As previously shown, S. pombe Class I HDAC mutant Δhos2 showed hypersensitivity to TSA (Kim, Y. et al., 1998; Olsson et al., 1998) (Table 1 indicated by – –, and Fig. 1A shown in the presence of TSA at 36°C). A centromere/kinetochore mutant mis6-302 was also highly sensitive to TSA at 30°C (Fig. 1A). It is known that HDACs are involved in heterochromatin formation important for centromere function; therefore, the hypersensitivity of the mis6-302 mutant to TSA might be due to the synergistic effects of heterochromatin disruption and kinetochore defects. Here it was not further analysed. Unexpectedly, some of the mutants defective in the process of sister chromatid separation, e.g. sister chromatid cohesion and anaphase onset, showed marked responses to TSA, as shown in Table 1. So far there has been no established role for HDACs in sister chromatid separation. Hereinafter, the mechanisms underlying the effects of HDAC inhibition on these mutants are investigated in detail.
. | . | . | TSA sensitivity . | . | . | . | ||
---|---|---|---|---|---|---|---|---|
Function . | Homologues (human/S. cerevisiae) . | Mutant allele . | 6.25 μg/ml . | 12.5 μg/ml . | 25.0 μg/ml . | References . | ||
APC/C subunit | Apc1 | cut4-533 | ++ | ++ | ++ | (Yamashita et al., 1996) | ||
Apc3/Cdc27 | nuc2-663 | + | + | + | (Hirano et al., 1988) | |||
Apc4 | cut20-100 | + | + | + | (Yamashita et al., 1999) | |||
Apc6/Cdc16 | cut9-665 | + | + | + | (Samejima and Yanagida, 1994) | |||
Apc8/Cdc23 | cut23-194 | + | + | + | (Yamashita et al., 1999) | |||
Apc10/Doc1 | apc10-153 | ++ | ++ | ++ | (Matsumura et al., 2003) | |||
apc10-666 | ++ | ++ | ++ | This study | ||||
APC/C E2 | UbcH10/Ubc11 | ubcP4-140 | ++ | ++ | ++ | (Osaka et al., 1997) | ||
APC/C activator | Cdc20 | slp1-362 | – | – | – | (Matsumoto, 1997) | ||
Cdh1 | Δste9 | N | N | N | (Kitamura et al., 1998) | |||
Spindle checkpoint | Mad2 | Δmad2 | N | – | –– | (Toyoda et al., 2002) | ||
Proteasome subunit | S4/Yhs4 | mts2-1 | N | N | N | (Gordon et al., 1993) | ||
S14/Min1 | mts3-1 | N | N | N | (Gordon et al., 1996) | |||
Separase | hSeparase/Esp1 | cut1-206 | + | + | + | (Hirano et al., 1986) | ||
cut1-T693 | ++ | ++ | ++ | (Funabiki et al., 1996) | ||||
cut1-21 | + | + | + | (Funabiki et al., 1996) | ||||
Securin | hPTTG/Pds1 | cut2-364 | + | + | + | (Hirano et al., 1986) | ||
Cohesion | NIPBL/Scc2 | mis4-242 | –– | –– | –– | (Furuya et al., 1998) | ||
EFO1, 2/Ecol | eso1-H17 | N | N | N | (Tanaka et al., 2000) | |||
Cohesin subunit | Scc1 | rad21-K1 | N | – | – | (Tatebayashi et al., 1998) | ||
Scc3 | psc3-407 | N | N | N | (Yuasa et al., 2004) | |||
Smc1 | psm1-ST273 | N | N | N | This study | |||
Smc3 | psm3-ST426 | N | N | N | This study | |||
Condensin subunit | HCAP-H/Brn1 | cnd2-1 | N | N | N | (Aono et al., 2002) | ||
HCAP-E/Smc2 | cut14-208 | N | N | N | (Saka et al., 1994) | |||
HCAP-C/Smc4 | cut3-447 | N | N | N | (Saka et al., 1994) | |||
Kinetochore component | CENP-I/Ctf3 | mis6-302 | –– | –– | –– | (Saitoh et al., 1997) | ||
hMis12/Mtw1 | mis12-537 | N | N | N | (Goshima et al., 1999) | |||
Cenp-A/Cse4 | cnp1-1 | N | N | N | (Takahashi et al., 2000) | |||
HDACs | HDAC1, 2/Rpd3 | clr6-1 | N | N | N/– | (Grewal et al., 1998) | ||
HDAC1, 2/Rpd3 | nmt1-clr6HA | N | N | N | This study | |||
HDAC6,10?/Hda1 | Δclr3 | N | N | N | This study | |||
HDAC3?/Hos2 | Δhos2 | –– | –– | –– | This study |
. | . | . | TSA sensitivity . | . | . | . | ||
---|---|---|---|---|---|---|---|---|
Function . | Homologues (human/S. cerevisiae) . | Mutant allele . | 6.25 μg/ml . | 12.5 μg/ml . | 25.0 μg/ml . | References . | ||
APC/C subunit | Apc1 | cut4-533 | ++ | ++ | ++ | (Yamashita et al., 1996) | ||
Apc3/Cdc27 | nuc2-663 | + | + | + | (Hirano et al., 1988) | |||
Apc4 | cut20-100 | + | + | + | (Yamashita et al., 1999) | |||
Apc6/Cdc16 | cut9-665 | + | + | + | (Samejima and Yanagida, 1994) | |||
Apc8/Cdc23 | cut23-194 | + | + | + | (Yamashita et al., 1999) | |||
Apc10/Doc1 | apc10-153 | ++ | ++ | ++ | (Matsumura et al., 2003) | |||
apc10-666 | ++ | ++ | ++ | This study | ||||
APC/C E2 | UbcH10/Ubc11 | ubcP4-140 | ++ | ++ | ++ | (Osaka et al., 1997) | ||
APC/C activator | Cdc20 | slp1-362 | – | – | – | (Matsumoto, 1997) | ||
Cdh1 | Δste9 | N | N | N | (Kitamura et al., 1998) | |||
Spindle checkpoint | Mad2 | Δmad2 | N | – | –– | (Toyoda et al., 2002) | ||
Proteasome subunit | S4/Yhs4 | mts2-1 | N | N | N | (Gordon et al., 1993) | ||
S14/Min1 | mts3-1 | N | N | N | (Gordon et al., 1996) | |||
Separase | hSeparase/Esp1 | cut1-206 | + | + | + | (Hirano et al., 1986) | ||
cut1-T693 | ++ | ++ | ++ | (Funabiki et al., 1996) | ||||
cut1-21 | + | + | + | (Funabiki et al., 1996) | ||||
Securin | hPTTG/Pds1 | cut2-364 | + | + | + | (Hirano et al., 1986) | ||
Cohesion | NIPBL/Scc2 | mis4-242 | –– | –– | –– | (Furuya et al., 1998) | ||
EFO1, 2/Ecol | eso1-H17 | N | N | N | (Tanaka et al., 2000) | |||
Cohesin subunit | Scc1 | rad21-K1 | N | – | – | (Tatebayashi et al., 1998) | ||
Scc3 | psc3-407 | N | N | N | (Yuasa et al., 2004) | |||
Smc1 | psm1-ST273 | N | N | N | This study | |||
Smc3 | psm3-ST426 | N | N | N | This study | |||
Condensin subunit | HCAP-H/Brn1 | cnd2-1 | N | N | N | (Aono et al., 2002) | ||
HCAP-E/Smc2 | cut14-208 | N | N | N | (Saka et al., 1994) | |||
HCAP-C/Smc4 | cut3-447 | N | N | N | (Saka et al., 1994) | |||
Kinetochore component | CENP-I/Ctf3 | mis6-302 | –– | –– | –– | (Saitoh et al., 1997) | ||
hMis12/Mtw1 | mis12-537 | N | N | N | (Goshima et al., 1999) | |||
Cenp-A/Cse4 | cnp1-1 | N | N | N | (Takahashi et al., 2000) | |||
HDACs | HDAC1, 2/Rpd3 | clr6-1 | N | N | N/– | (Grewal et al., 1998) | ||
HDAC1, 2/Rpd3 | nmt1-clr6HA | N | N | N | This study | |||
HDAC6,10?/Hda1 | Δclr3 | N | N | N | This study | |||
HDAC3?/Hos2 | Δhos2 | –– | –– | –– | This study |
++, the growth is strongly improved; +, the growth is slightly improved; N, the growth is not affected by TSA; ––, the growth is strongly inhibited; –, the growth is slightly inhibited. The indicated temperature-sensitive mutants and null mutants grown exponentially at 26°C were diluted and spotted with wild-type cells on YPD plates containing 6.25-25.0 μg/ml TSA. Those plates were incubated at 26, 30, 33 and 36°C. Effects of TSA on the growth of those mutants were determined by comparing with the growth of the mutants in the absence of TSA, and with wild type. Names of their homologous proteins in human and S. cerevisiae and the references for each strain are also shown
The cohesion loader mutant mis4-242 is hypersensitive to HDAC inhibition
Two mutants defective in sister chromatid cohesion, a cohesin subunit mutant rad21-K1 (Tatebayashi et al., 1998) and the adherin mutant mis4-242 (Furuya et al., 1998), were sensitive to TSA, as shown in Table 1 and Fig. 1B. Whereas rad21-K1 was only slightly sensitive to TSA, mis4-242 was extremely hypersensitive even at the permissive temperature, 26°C. Conversely, mutants of other cohesin subunits, the cohesin establishment factor Eso1 and condensin subunits all failed to show any hypersensitivity to TSA (Table 1 and Fig. 1B). Mis4 is similar to budding yeast Scc2 and human NIPBL, which is the causal gene of Cornelia de Lange syndrome, and fly NippedB, which is implicated in gene expression for development (Dorsett, 2007). In the presence of TSA at the permissive temperature (26°C), mis4-242 frequently revealed missegregation with unequal daughter nuclei (Fig. 1C, upper panel; indicated by arrowheads). The phenotype was different from that of mis4-242 at the restrictive temperature (36°C) in the absence of TSA, which showed condensed chromosomes indicative of a mitotic arrest due to the spindle checkpoint activation (Fig. 1C, lower panel; indicated by arrows) (Toyoda et al., 2002).
Next, it was examined whether inhibition of any specific HDAC among the three TSA-sensitive HDACs in S. pombe causes the hypersensitivity of the mis4-242 mutant to TSA. Three double mutants were constructed and examined for TSA sensitivities; mis4-242 clr6-1, mis4-242 Δclr3 and mis4-242 Δhos2. mis4-242 showed a certain additive effect on temperature sensitivity with Δclr3 or Δhos2 at 30°C, whereas single mutants Δclr3 and Δhos2 did not show any temperature sensitivity at all (Fig. 1D, upper panels). The clr6-1 single mutant was temperature-sensitive, and the mis4-242 clr6-1 double mutant showed a less additive effect, if any, on temperature sensitivity for colony formation growth than the other double mutants (Fig. 1D, lower panels).
TSA treatment decreases Mis4 protein levels and the amount of Rad21 associated with the chromosome arm region
To understand why mis4-242 is strongly affected by TSA, the behaviour of Mis4 protein was analysed for up to 12 hours after TSA treatment at 26°C. The cell extracts of wild type and mis4-242 were immunoblotted with antibodies against Rad21, Mis4, Cut9 and Cdc2. Expectedly, the levels of acetylated histones (AcH3 and AcH4) in the wild-type cells increased within 1 hour after TSA addition (Fig. 2A). In conjunction with the hypersensitivity of mis4-242 to TSA, Mis4 protein levels were reduced in the wild-type cells and the mutant cells after TSA treatment (Fig. 2A). The amount of Mis4 protein in the mutant cells was already much lower than in wild-type cells before treatment, and it further decreased at 4 hours and became undetectable after 8 hours. In mis4-242 mutant cells, the level of Rad21 was similar to that in wild-type cells but less phosphorylated (more abundant lower bands) (Tomonaga et al., 2000), probably due to reduced chromatin loading. Curiously, the Rad21 level increased and peaked at around 4 hours after the addition of TSA in mis4-242 mutants. The levels of Cut9, an anaphase-promoting complex/cyclosome (APC/C) subunit, however, were not significantly affected in the presence of TSA, whereas the levels of Cdc2 (detected by PSTAIR antibody) slightly increased after the addition of TSA. mRNA levels of Mis4 were not decreased by TSA treatment, either in wild-type or mis4-242 mutant cells (supplementary material Fig. S1). These results imply that HDAC inhibition might cause destabilization of Mis4 protein, which might account for the hypersensitivity of mis4-242.
Mis4 (adherin) regulates the loading of the cohesin complex onto chromosomes. To investigate whether HDAC inhibition affects this cohesin loading function of Mis4, chromatin immunoprecipitation (ChIP) of Rad21 was performed before (–) and after (+) TSA treatment, using strains containing chromosomally integrated HA-tagged rad21+ gene. An APC/C subunit mutant cut9-665 was also used to arrest cells in mitosis. TSA (12.5 μg/ml) or buffer alone was added to cultures of these strains growing exponentially at 26°C, before the temperature was shifted to 36°C. After a 4 hour incubation, the cells were fixed with 1% formaldehyde. The cell extracts were then subjected to immunoprecipitation with anti-HA antibody after genomic DNA was sheared. In the cells treated with TSA, considerable reduction in the level of chromatin-bound Rad21HA was observed in the chromosome regions lys1 and c83 derived from non-centromeric regions of chromosome I and II, respectively (Fig. 2B). However, Rad21HA associated with chromatin in the cnt1 and dg regions located in the centromere did not significantly change after TSA treatment (Fig. 2B). A similar change in chromatin association of Rad21HA as a result of TSA treatment was also observed in cut9-665 mutants. Importantly, the total protein level of Rad21HA in the cell extracts remained unchanged (Fig. 2B, bottom panels).
The manner of the decrease was studied more precisely. No tag wild-type and rad21HA cells were grown at 26°C, followed by the addition of 12.5 μg/ml TSA to the cultures. Samples were taken 1, 2 and 4 hours after the addition of TSA, and PCR amplification was carried out with the seven primer sets indicated in Fig. 2C. The levels of Rad21 bound to the arm regions (lys1, c83, 08c and ARS2004) all decreased to control levels within 1 hour after TSA treatment (Fig. 2C, lower panels). By contrast, Rad21 bound to the centromeric region (cnt1, dg and imr1) slightly oscillated upon TSA treatment but returned to normal levels after 4 hours (Fig. 2C, upper panels). Thus, HDAC inhibition might affect the Mis4 function of cohesin loading specifically in the chromosome arm region.
Rescue of APC/C mutants by HDAC inhibitors
In contrast to the known anti-proliferation effects on mammalian cells and the potent growth inhibition to mis4-242, TSA treatment greatly enhanced cell growth of the ts mutants of APC/C at the restrictive temperature (Table 1 and Fig. 3A). Every APC/C subunit mutant examined (cut4-533, cut9-665, nuc2-663, cut23-194, cut20-100, apc10-153 and apc10-666) was rescued by TSA treatment. The suppressive effects on cut4 and apc10 mutants were particularly strong, as these mutants formed colonies nearly indistinguishable from wild type in the presence of TSA (12.5 μg/ml) at the restrictive temperature, 36°C (Fig. 3A). Cut4/Apc1 is the largest subunit of APC/C and is proposed to bridge two subcomplexes within APC/C, whereas Doc1/Apc10 was suggested to have a role in recognizing APC/C substrates or increasing processivity of polyubiquitination by limiting substrate dissociation (Passmore et al., 2003; Thornton et al., 2006).
Furthermore, in liquid cultures at the restrictive temperature, TSA nearly fully rescued the cellular phenotypes of cut4-533, and the proliferation rates were almost identical between wild type and the mutant at 36°C in the presence of TSA (Fig. 3B). In the absence of TSA, the cell proliferation of the mutant was halted at 36°C after 4 hours, with the cells showing metaphase arrest phenotypes (Fig. 3B, right top panel). Aberrant mitotic arrest phenotypes of cut4-533 at 36°C were hardly observed in the presence of TSA (Fig. 3B, right bottom panel). Thus, TSA appears to suppress the metaphase arrest phenotypes specific to APC/C mutants.
To verify that TSA-induced suppression is due to the inhibition of the HDAC activity rather than an unknown effect of TSA, other drugs also known to inhibit HDACs were used. A hydroxamic acid-containing peptide CHAP31, and cyclic peptides Trapoxin B (TPXB) and FK228, inhibit HDAC activity in vitro and possess anti-tumour activity (Furumai et al., 2001; Furumai et al., 2002; Kijima et al., 1993; Komatsu et al., 2001). Wild type and the cut4-533 mutant were grown at 26°C, then shifted to 36°C in the presence of serial dilutions of these drugs. In wild type, the effects of TPXB and FK228 were stronger than that of TSA in regard to the inhibition of cell growth, whereas CHAP31 was not inhibitory at all in the range of the drug concentrations used (Fig. 3C, left panel). In cut4-533 mutants, CHAP31, TPXB and TSA all suppressed the growth defects, with CHAP31 suppressing to the same level as TSA, confirming that the suppression effects were due to inhibition of HDACs by these drugs rather than an unknown effect of TSA (Fig. 3C, right panel). The suppression effect of FK228 was not observed at the concentrations examined, probably due to its strong toxicity.
Mutants of proteins involved in APC/C-dependent ubiquitination and proteolysis were also examined for TSA sensitivity. In agreement with the above results, the growth defect of ubcP4-140 (Osaka et al., 1997), a mutant in the fission yeast orthologue of human UbcH10, the E2 enzyme required for the ubiquitination activity of APC/C, was also suppressed by TSA (Fig. 3D, upper panels). No significant suppression, however, was observed for mts2-1 and mts3-1, ts mutants of 26S proteasome subunits, suggesting that HDAC inhibition upregulates APC/C activity rather than affects proteasome functions (Fig. 3D, lower panels). A deletion mutant of Mad2, a spindle checkpoint component known to bind the APC/C activator Slp1/Cdc20 and thus inhibit APC/C (Hwang et al., 1998; Kim, S. et al., 1998), showed hypersensitivity to TSA, which might be due to premature anaphase onset with APC/C activated by TSA treatment. A mutant of Ste9, the G1 phase-specific APC/C activator required for G1 arrest upon nitrogen starvation (Kitamura et al., 1998; Matsumoto, 1997; Yamaguchi et al., 1997), was not affected by TSA (Fig. 3D, upper panels). However, slp1-362, a mutant of the mitotic APC/C activator was slightly sensitive to TSA. Slp1 was suggested to have a role in mitotic entry after DNA damage checkpoint in addition to APC/C activation (Matsumoto, 1997). Thus, it is possible that TSA might affect this function of Slp1 before cells enter mitosis.
Suppression of APC/C mutants by Clr6 HDAC mutation
To determine which HDAC gene is implicated in the suppression of APC/C mutants, double or triple mutants with HDAC mutation or deletion were constructed. In contrast to effects seen with mis4-242, the double mutant cut4-533 clr6-1 showed significantly better growth than cut4-533 single mutants at 36°C, whereas Δclr3 and/or Δhos2 did not produce a significant suppressive effect on cut4-533 (Fig. 4A). Another clr6 mutant, nmt1-clr6HA, in which the chromosomal clr6+ gene is replaced by the HA-tagged clr6+ gene under the inducible nmt1 promoter leading to repressed expression on YPD plates, almost completely rescued cut4-533 at 36°C. The clr6-1 mutation was also highly effective in suppressing the ts phenotypes of the two apc10 mutants and the cut9-665 mutant (Fig. 4B). In cytological analyses of the liquid cultures, the frequent mitotic phenotypes observed in cut4-533 at 36°C were not observed in the double mutants cut4-533 clr6-1 and cut4-533 nmt1-clr6HA (Fig. 4C). By contrast, the double mutants cut4-533 Δclr3 and cut4-533 Δhos2 displayed the highly frequent aberrant arrest phenotypes seen in cut4-533 single mutants. Therefore, it is plausible that the reduction of the Clr6 activity is responsible for the suppression of APC/C mutants by TSA treatment.
TSA treatment and clr6 mutation enhance the assembly of 20S APC/C
These results raised a question as to how TSA could suppress the phenotype of APC/C mutants. HDACs are known to be involved in repression of gene transcription, and, accordingly, HDAC inhibition could affect protein expression. To examine the possibility that HDAC inhibition suppresses APC/C mutants via increased expression of the mutant proteins, the levels of various APC/C subunits were examined in wild type, cut4-533 and cut9-665. Cell extracts were prepared from cells cultured at 26°C in the presence of TSA for 12 hours. As shown in Fig. 5A, no significant increase in the cellular concentration of Cut4 and Cut9 mutant proteins was observed in the presence of TSA. However, after TSA addition, due to unknown protein modifications the upper bands of Cut4 (Yamashita et al., 1996) migrated more slowly and the level of Cut9 protein appeared to slightly increase. Overexpression of wild-type Cut9 did not suppress the ts phenotypes in cut4-533 (Y.K., unpublished); therefore, it is unlikely that the increase of the Cut9 protein is the chief reason for the suppression of APC/C mutants by TSA treatment.
In fission yeast, APC/C is known to sediment as a 20S particle, which was disrupted in cut4-533, cut9-665 and nuc2-663 mutant cells (Yamada et al., 1997; Yamashita et al., 1996). To examine the possibility that formation of the 20S APC/C complex might be affected by HDAC inhibition, sucrose gradient centrifugation was performed on the extracts from cut4-533 mutants cultured in the absence or presence of TSA. In wild type, the subunits of APC/C, Cut4, Cut9 and Nuc2, enriched in the fractions around 20S (Fig. 5B, WT –TSA). In cut4-533 mutant cells, the levels of APC/C subunits that sedimented around 20S were much lower even at the permissive temperature 26°C in the absence of TSA; however, their levels at 20S greatly increased in the presence of TSA (Fig, 5B, cut4-533 –TSA and +TSA). Recovery of the 20S complex occurred within 2 hours after TSA addition in cut4-533 mutants and was also observed in cut9-665 mutants (supplementary material Fig. S2A,B) (Y.K., unpublished). Thus, the 20S complex can re-form in APC/C mutant cells if TSA is added into cultures, suggesting that suppression of APC/C mutants by TSA treatment might be due to the restoration of APC/C complex formation.
To examine whether the clr6-1 mutation also affects the APC/C complex formation in a manner similar to the addition of TSA, cell extracts of the cut4-533 clr6-1 double mutant were analysed by sucrose gradient centrifugation along with three other strains, wild type, single clr6-1 and single cut4-533 mutant cells (Fig. 5C). Cnd2, a component of the condensin complex, was used as a sedimentation marker for the position of 13S (Sutani and Yanagida, 1997). Expectedly, the levels of Cut4, Cut9 and Nuc2 in the 20S fractions were found to greatly increase in cut4-533clr6-1 double mutant cells cultured at both 26 and 36°C, compared with those in single cut4-533 mutant extracts (Fig. 5C, left panels). Thus, Clr6 plays a main role in restoring the level of the 20S APC/C complex by TSA treatment. In the whole cell extracts, a slight increase in the levels of non-modified Cut4 mutant protein (indicated by the arrows) and the Cut9 protein was detected in the double mutants at both 26 and 36°C, compared to the single cut4 mutants (Fig. 5C, right panels). APC/C subunit proteins might therefore be stabilized when the 20S complex is properly assembled.
It was also asked whether the wild-type APC/C complex is affected by TSA. Cell extracts prepared from wild-type cells cultured in the presence or absence of TSA were examined using sucrose gradient centrifugation. As shown in Fig. 5D, the 20S peak of the APC/C subunits was observed in both extracts, and the Cut4 band was more intense in the presence of TSA. To confirm that these APC/C subunits actually form the complex, immunoprecipitation was performed using the strain containing chromosomally integrated apc2+ gene C-terminally tagged with 3HA (apc2HA). As shown in Fig. 5E, more Cut4, Cut9 and Nuc2 proteins were indeed co-precipitated with Apc2HA in the presence of TSA. Thus, HDAC inhibition appears to enhance APC/C assembly in wild-type cells as well as in APC/C mutant cells.
Cut23, a subunit of APC/C, is acetylated
The above results hinted that the ability of APC/C subunits to assemble into the complex might be regulated by acetylation-deacetylation cycle of APC/C subunits. To examine this possibility, 6His-tagged APC/C subunits (Cut4, Apc2, Nuc2, Cut9, Cut23, Apc10, Apc11 and Hcn1) were overexpressed in wild-type cells grown in the presence or absence of TSA, then purified by nickel beads under denaturing conditions and immunoblotted by anti-acetyl-lysine antibodies. No change in acetyl-lysine levels was observed in the presence of TSA for any of the proteins analysed (Y.K. and A.M., unpublished). The antibodies used in this experiment were raised against acetylated histones or bovine serum albumins so that they might not be able to detect acetylated APC/C subunits. Therefore, mass spectrometry was used to analyse individual APC/C subunits. 3HA6His-tagged APC/C subunits were overexpressed in wild-type cells grown in the presence of TSA and purified by nickel beads under denaturing conditions. Cut23/APC8 was found to be acetylated. The purified Cut23 protein was separated by SDS-PAGE and stained with Coomassie Blue (Fig. 6A), and the band corresponding to the Cut23 subunit was cut out and digested by trypsin. By liquid chromatography and tandem mass spectrometry (LC-MS/MS) analyses (Ohta et al., 2002), four putatively acetylated peptides were identified in Cut23 protein (Fig. 6B). Among the four acetylation sites in Cut23, only the K461 residue is conserved from yeast to human. Cut23 contains six putative destruction box sequences (RxxL), which are known as substrate recognition motifs for APC/C, and one of them (465RALL) is close to the conserved acetylated sequence (Fig. 6C). Three of the four putative acetylation residues, including K461, were located within the tetratricopeptide repeat (TPR) domain of Cut23 involved in protein-protein interaction (Fig. 6C). To investigate the functions of these acetylation sites, strains that contain HA-tagged Cut23 with all four sites mutated to arginine or glutamine (cut23K4RHA and cut23K4QHA, respectively) were constructed, as well as a strain carrying HA-tagged wild-type Cut23 (cut23HA). C-terminal HA-tagging alone partially affects the function of Cut23 protein, as the ts growth defect of cut4-533 was increased when it was combined with cut23HA (Fig. 6D, –TSA). Notably, the growth of cut4-533cut23K4RHA and cut4-533cut23K4QHA was even slower than cut4-533cut23HA at 26 and 30°C, implying that the acetylation sites have some role in Cut23 functions. However, temperature sensitivity of all double mutants was suppressed by TSA treatment or clr6-1 mutation (Fig. 6D) (Y.K., unpublished). Therefore, the four acetylation sites in the Cut23 protein are not essential for suppression of APC/C mutants by HDAC inhibition, at least not on their own. As yet unidentified acetylation sites on Cut23 and/or other APC/C subunits might therefore be important for the formation of APC/C.
HDAC inhibition facilitates sister chromatid separation
At the onset of anaphase Cut2/Securin is polyubiquitinated for destruction by APC/C allowing activation of Cut1/Separase, which in turn cleaves cohesin subunits. If HDAC inhibition causes the facilitation of anaphase according to our hypothesis, it is expected that cut1 and cut2 mutants defective in the activation of Cut1 and sister chromatid separation (Funabiki et al., 1996; Kumada et al., 1998) would also be suppressed by TSA. Consistently, the ts phenotypes of separase mutants, cut1-206, cut1-T693 and cut1-21, and the securin mutant, cut2-364, were all partially rescued in the presence of TSA (Fig. 7A). Mitotic phenotypes of cut1-206 in the liquid culture at the restrictive temperature were significantly suppressed by the addition of 12.5 μg/ml TSA (K.N., unpublished).
The spindle checkpoint and the cAMP/PKA pathway are known to negatively regulate APC/C (Kim, S. et al., 1998; Yamashita et al., 1996). We then examined whether suppression of APC/C mutants by TSA requires the presence of cyclic AMP-dependent protein kinase Pka1 and/or spindle checkpoint protein Mad2. Suppression took place in the double mutants cut9 Δpka1 and cut9 Δmad2 in the presence of TSA (Fig. 7B,C), indicating that neither PKA or Mad2 pathways were essential for the suppression.
Discussion
HDAC inhibitors are a promising class of anticancer agents (Bolden et al., 2006; Johnstone and Licht, 2003; Yoshida et al., 2003). However, our understanding of how HDAC inhibitors act within cells and why different cell types respond in different ways is limited. We employed the fission yeast S. pombe as a model, because this organism is sensitive to HDAC inhibitors and contains a set of HDAC genes similar to those of mammalian organisms. In the present study, we focused on the mitotic functions of three TSA-sensitive HDACs in fission yeast. Our results suggest that Clr6 negatively regulates APC/C independently of the PKA pathway and Mad2. By contrast, Mis4, the cohesin loader, is positively controlled by HDACs. HDAC inhibitors thus reduce the level of Mis4 and facilitate the exit from mitosis via the assembly of APC/C complex, leading to the direction of sister chromatid separation in dividing cells (Fig. 7D).
Given the known anti-proliferation effects of HDAC inhibitors on tumour cells, our result in S. pombe is surprising, as APC/C mutants could produce colonies by TSA treatment, clr6-1 ts mutation or repression of Clr6 expression. Three features are prominent for this suppression of APC/C mutants by HDAC inhibition. The first is the universality of suppression: all the APC/C mutants examined were suppressible by the drugs used. In addition, ubcP4-140, the mutant of APC/C-specific E2, was also suppressed. The effect of TSA and clr6-1 hence is confined to ubiquitination, as 26S proteasome subunit mutants are not suppressed. The second is that Clr6, rather than Hos2 and Clr3, appears to be the target of HDAC inhibition for the suppression. The third is that HDAC inhibition apparently affects the level of assembled APC/C rather than the expression of individual APC/C subunits. Restoration of the 20S APC/C complex seems to correspond to the recovery of APC/C activity. Similar correlation between the ts suppression and the 20S APC/C assembly was reported in the suppression of APC/C mutant cut4-533 via the inactivation of the cAMP/PKA pathway (Yamashita et al., 1996). However, the suppression of APC/C mutants by TSA or clr6 mutation required neither Pka1 or Mad2, thus the effect of HDAC Clr6 on APC/C seems to occur through a mechanism independent of the cAMP/PKA pathway and the spindle checkpoint. In any case, the use of HDAC inhibitors does not necessarily cause the arrest of the cell proliferation in S. pombe. In certain mutants described in this study, HDAC inhibitors appear to facilitate the exit from mitosis and so normal cell division.
The additive lethality of the cohesin loader mutant mis4-242 by HDAC inhibition is interpreted as a result of the diminished levels of Mis4 under HDAC inhibition and so the stability of Mis4 may be under the positive regulation of HDACs. TSA treatment results in strong growth inhibition in mis4-242 but not in the mutants of cohesin subunits and the establishment factor Eso1. Mis4 or its partner molecule are possibly regulated by direct acetylation for the stability of Mis4. Curiously, TSA treatment decreases the chromatin-bound cohesin subunit Rad21, especially in the non-centromeric regions. The level of centromeric Rad21 only slightly decreases 2 hours after TSA treatment and recovers to a normal level by 4 hours. Mis4 or budding yeast adherin Scc2p are required for cohesin loading on both centromere and arm regions (Ciosk et al., 2000; Tomonaga et al., 2000) and localizes to the entire chromatin region in a punctate manner (Furuya et al., 1998). However, the chromatin localization of Mis4-GFP was not affected by TSA treatment (Y.K., unpublished). We have no obvious explanation for this arm-specific cohesin unloading promoted by HDAC inhibition. Thus, HDAC inhibition seems to affect only a part of the Mis4 functions.
TSA treatment of S. pombe cells leads to disruption of heterochromatin, which in turn causes lagging chromosomes and increased rates of chromosome loss (Ekwall et al., 1997). However, in our experimental conditions described (6.25∼25.0 μg/ml TSA, for 12 hours), no significant chromosome segregation phenotype was observed in wild-type cells, implying that the centromeric heterochromatin was not strongly affected. Consistently, the levels of Rad21 bound to the centromeric heterochromatic region (dg), recruited by Swi6/HP1 (Bernard et al., 2001; Nonaka et al., 2002), were not significantly changed after TSA treatment (Fig. 2B,C). Strong effects of HDAC inhibition are detected only in the mutant strains. The converse is seen in APC/C and mis4 mutants, where sister chromatid separation is facilitated. For APC/C mutants, the drug promotes apparently normal sister chromatid separation and viable cell division; however, for Mis4 mutant cells, sister chromatid separation is premature and consequently becomes lethal. Thus, in S. pombe, HDAC inhibitors can be inhibitory or enhancing for cell proliferation under different genetic backgrounds, whereas the drug promotes sister chromatid separation via different HDACS. This variation of effects might be applicable for human cells and should be taken into consideration when the effects of HDAC inhibitors on different types of tumour are assessed.
Materials and Methods
Strains, media and TSA sensitivity test
All the mutant strains used have the genetic background of h– leu1-32. The strain apc2HA::LEU2 was constructed by transforming wild-type cells with pYC11-ΔNapc2-3HAHis6 plasmid, which contains LEU2 gene as a selection marker and the C-terminus of fission yeast apc2+ gene (from an internal NheI restriction site) followed by the sequences coding 3HAHis6 tag at its 3′ end. The strains nmt1-clr6HA, Δclr3 and Δhos2 were constructed by a PCR-based method slightly modified from Bähler et al. (Bähler et al., 1998). apc10-666 was isolated as a ts mutant showing metaphase arrest from the ts library described in Yamashita et al. (Yamashita et al., 1999). Strains psm1-ST273 and psm3-ST426 were isolated through screening mutants resistant to overexpression of Cut2 central fragments (Nagao and Yanagida, 2006). To examine TSA sensitivity, cells growing exponentially at 26°C were diluted with EMM2 medium and spotted at a series of concentrations: 5×104, 1×104, 2×103, 4×102, 80 and 16 cells per spot, on YPD plates containing TSA (6.25-25 μg/ml).
Antibodies
Anti-acetylated histone H3 and H4 antibodies (upstate #06-599 and #06-598, respectively), and anti-HA antibody (12CA5, Roche) were used for immunoblotting and immunoprecipitation. Polyclonal antibodies specific for Cut4, Cut9, Nuc2, Cnd2, Mis4 and Rad21 were used as previously described (Aono et al., 2002; Tomonaga et al., 2000; Yamashita et al., 1996). Anti-PSTAIR and TAT-1 monoclonal antibodies are gifts from Dr Y. Nagahama and Dr K. Gull, respectively.
Sucrose gradient centrifugation and ChIP
Sucrose gradient centrifugation was performed as previously described (Yamashita et al., 1996). ChIP of Rad21HA was performed following the previously described procedure (Tomonaga et al., 2000). The primer sets for cnt1, dg and lys1 are the same as those previously used (Saitoh et al., 1997; Tomonaga et al., 2000). The primers used for c83, 08c and ARS2004 were newly synthesized to amplify the non-coding region on the arm of chromosome II corresponding to the cosmid SPBC83, 30244-30440, the coding region of the SPBC28F2.08c gene and the non-coding region in the ARS2004 on the arm of chromosome II, respectively.
Purification of Cut23 protein and mass spectrometry analysis
To purify Cut23 protein from fission yeast cells, Cut23-3HAHis6 was overexpressed by the pREP1 vector in the presence or absence of 12.5 μg/ml TSA. The 2×109 cells were collected and disrupted by glass beads in 1 ml NG8 buffer (10 mM Tris-HCl at pH 8.0, containing 100 mM Na-phosphate, 50 mM NaCl, 6 M guanidine hydrochloride and 0.1% NP-40). The extracts were centrifuged at 14 000 rpm for 15 minutes, and 100 μl of Ni-NTA beads (QIAGEN) were added to the supernatant. After incubation for 2 hours at room temperature, the beads were washed twice with NG8 buffer, twice with NG7 (pH 7.0) and then twice with buffer N8 (as NG8 but without guanidine hydrochloride). The beads were boiled in SDS sample buffer after the addition of 100 mM EDTA. Mass spectrometry analysis using in-gel digested peptides was performed as described in Ohta et al. (Ohta et al., 2002).
Acknowledgements
We thank Dr A. Klar for clr6-1 strain, Drs Y. Nagahama and K. Gull for the antibodies and Drs H. Yamano, C. Goding, A. McAinsh, M. Trickey and S. McClelland for reading the manuscript and discussion. This study was supported by the CREST Research Project of the Japan Science and Technology Corporation (JST), the COE Research Unit of the Ministry of Education, Culture, Sports, Science and Technology, and Marie Curie Cancer Care.