Regulation of Swi6/HP1-dependent Heterochromatin Assembly by Cooperation of Components of the Mitogen-activated Protein Kinase Pathway and a Histone Deacetylase Clr6*

A study of gene silencing within the mating-type region of fission yeast defines two distinct pathways responsible for the establishment of heterochromatin assembly. One is RNA interference-dependent and acts on centromere-homologous repeats (cenH). The other is a stochastic Swi6 (the fission yeast HP1 homolog)-dependent mechanism that is not fully understood. Here we find that activating transcription factor (Atf1) and Pcr1, the fission yeast bZIP transcription factors homologous to human ATF-2, are crucial for proper histone deacetylation of both H3 and H4. This deacetylation is a prerequisite for subsequent H3 lysine 9 methylation and Swi6-dependent heterochromatin assembly across the rest of the silent mating-type (mat) region lacking the RNA interference-dependent cenH repeat. Moreover, Atf1 and Pcr1 can form complexes with both a histone deacetylase, Clr6, and Swi6, and clr6 mutations affected the H3/H4 acetylation patterns, similar to the atf1 and pcr1 deletion mutant phenotypes at the endogenous mat loci and at the ctt1+ promoter region surrounding ATF/CRE-binding site. These data suggest that Atf1 and Pcr1 participate in an early step essential for heterochromatin assembly at the mat locus and silencing of transcriptional targets of Atf1. Furthermore, a phosphorylation event catalyzed by the conserved mitogen-activated protein kinase pathway is important for regulation of heterochromatin silencing by Atf1 and Pcr1. These findings suggest a role for the mitogen-activated protein kinase pathway and histone deacetylase in Swi6-based heterochromatin assembly.

of the mating-type region provide insights for understanding the regulation of heterochromatin assembly in eukaryotes. Of particular interest, previous work (9) has defined sequential requirements for the establishment and maintenance of regional heterochromatic domains.
Heterochromatin assembly at the mating-type region containing the mat2 and mat3 silent donor loci and an 11-kb interval (K region) between them requires several cis-acting DNA sequences as well as trans-acting factors (8, 10 -13). Heterochromatin formation at the centromeres and within the silent mat2/3 interval requires many of the same silencing factors, including Clr3 and Clr6 (H3/H4-specific histone deacetylases), the Clr4-Rik1 complexes, and Swi6 (2, 14 -19). The DNA elements involved in silencing within the entire 20 kb of the mat2/3 silent mating-type interval include REII (20), the mat3-M element including putative ATF 1 /CREB-binding sites (21), and the 4.3-kb centromere-homologous repeat (cenH) sequence within the K region (22). The cenH region, which shares strong homology with the dg and dh centromeric elements, is a heterochromatin nucleation center and requires the RNA interference (RNAi) machinery similar to centromeric silencing (9). The cenH-mediated silencing via the RNAi machinery is required for initial formation of heterochromatin but not for its maintenance. In addition, it has been revealed that flanking sequences present in the K⌬ strain, which lacks the cenHcontaining K region, are capable of recruiting and maintaining H3 Lys-9 methylation only in the presence of Swi6. Moreover, heterochromatin formation at the mat2/3 region eventually occurs even without the RNAi-dependent cenH element, suggesting the existence of an additional RNAi-independent Swi6based mechanism for heterochromatin assembly (9). ATF/CREB family proteins are among the conserved effector molecules that are regulated by the stress-activated MAP kinase (MAPK) cascade. In fission yeast, Atf1 and Pcr1, members of ATF/CREB family, are phosphorylated by the stress-activated Wis1 (MAPK kinase) and Sty1/Spc1 (MAPK) protein kinases, and this phosphorylation induces transcriptional activation of target genes (23)(24)(25)(26)(27). Moreover, many lines of evidence suggest potential roles for ATF/CREB family proteins in chromatin remodeling and gene silencing (28 -30). Although deletion of two potential ATF/CRE-binding sites upstream of the mat3-M locus had a slight effect on heterochromatin silenc-* This work was supported in part by National Cancer Center Research Grant 0210110 (to Y. K. J.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ To whom correspondence should be addressed: National Cancer Center, 809 Madu-dong, Ilsan-gu, Goyang ing (21), there is still no clear evidence addressing the function of Atf1 and Pcr1 in heterochromatin silencing.
Here we describe how the ATF/CREB transcription factors regulate a Swi6-dependent heterochromatin assembly in fission yeast. We find that cooperation of the ATF/CREB transcription factors with common silencing factors including histone deacetylase Clr6 and Swi6 protein is important for histone deacetylation and H3 Lys-9 methylation via an additional RNAi-independent, Swi6-dependent mechanism that acts across the rest of the silent mat locus in the absence of cenH repeats.

EXPERIMENTAL PROCEDURES
Media, Strains, and Plasmids-Fission yeast media and genetic methods have been described previously (31,32). The S. pombe strains used in this study are listed in Table I. Each gene deletion was carried out as described elsewhere (25,33). PCR-based hemagglutinin (HA)tagged and gene deletion strains were constructed as described elsewhere (34). To assay silencing in vivo, we used reporter strains, which contain ade6 ϩ gene inserted at the outer repeat of centromere 1 (Hu50), adjacent to mat3 (Hu51) and close to the telomere of minichromosomes (Hu60), SacI and BglII site of L region (AP136 and AP144, respectively), and ura4 ϩ gene placed close to mat3 (PG9). Strain AP152, whose cenH region was replaced with ade6 ϩ , was also used.
Silencing Test-Cell viability (plating efficiency) and colony color assays were performed to estimate the level of expression of the reporter gene placed at heterochromatic regions as described (20,35). 10-Fold dilutions of overnight cultures cells were plated on YES, EMM, and adenine-free (ϪAde) plates. The plates were incubated at 30°C for 3-4 days and were then photographed. For ade6-off to ade6-on transition assay, each red colony (ade6-off) was picked and transferred onto YE plates, and then colonies were counted after 3-4 days. Colonies with more than 50% white sector (ade6-on) and less than 50% red sector were scored as half-sectored colonies, and the rate was calculated by dividing the number of half-sectored colonies by the total red and half-sectored colonies. For transition assay of K⌬::ade6 ϩ strains in Fig. 5, each white colony with indicated vectors was picked and transferred onto low adenine EMM plates, and then colonies were counted after 3-4 days. The transition rate was calculated by dividing the number of sectored or red colonies by the total number of colonies. For each transition assay, more than three independent experiments were carried out, and more than a thousand colonies were counted.

Atf1 and Pcr1 Proteins Are General Silencing Regulators at the Silent Mating-type Loci and Their Deficiencies Cause Silencing Defects via Changes in Histone Acetylation Patterns-
Deletion of two putative ATF/CREB-binding motifs downstream of the cenH repeat caused a partial derepression of the endogenous Mc gene from mat3-M locus (21). To identify factors important for the stochastic heterochromatin assembly in an RNAi-independent, Swi6-dependent manner, we investigated the roles of bZIP transcription factors in heterochromatin silencing at the mating-type region. To test whether Atf1 is involved in heterochromatin silencing at centromeres and telomeres, in addition to the silent mat locus, we investigated the effect of atf1 deletion on silencing of reporter genes inserted within each heterochromatic domain. An atf1 deletion strain carrying the ade6 ϩ reporter gene inserted at the mat3-M locus displayed a loss of silencing of the reporter gene (Fig. 1A). In contrast, the atf1 deletion resulted in increased transcriptional repression at both centromeres and telomeres (Fig. 1A). Moreover, the absence of atf1 disrupted silencing of another reporter gene, ura4 ϩ , inserted at the silent mat3-M locus (data not shown). These data indicate that Atf1 is required for silencing RNA pol II-transcribed genes inserted into the silent mat3 region. In summary, this indicates that Atf1 acts as a positive regulator of mat3-M silencing but as a negative regulator of heterochromatin formation at other regions.
A previous report (25) demonstrated that Pcr1 binds to the ATF/CREB-binding sequences in vitro and that the presence of Pcr1 was critical for nuclear localization of Atf1 (23). Thus, we examined whether Pcr1, a binding partner of Atf1, is also required for mat3-M silencing. Similar to atf1 mutants, ⌬pcr1 showed reduced silencing, but the effect was weaker than that of ⌬atf1 or ⌬atf1⌬pcr1 (Fig. 1B). These results suggest that heterodimerization of Atf1 and Pcr1 is important for the mat3-M silencing but that to some extent Atf1 can compensate for silencing defects resulting from the absence of Pcr1.
To assess whether the position effect on ade6 ϩ was directly related to its transcriptional expression, we performed RT-PCR to measure levels of mRNA from wild-type ade6 ϩ inserted within the mat3-M locus and the corresponding minigene ade6-DN/N at the endogenous locus on wild type (WT), atf1, pcr1, and swi6 deletion derivatives (Fig. 1C). The relative expression of the ade6 ϩ marker can be determined by the ratio of the full-length ade6 ϩ to the ade6-DN/N products in a competitive RT-PCR. The ratio in the ⌬atf1 and ⌬pcr1 mutants was considerably higher than that in the isogenic parental strain (WT) but much lower than that in a swi6 deletion mutant. Thus, our data demonstrate that transcriptional derepression of ade6 ϩ at the inactive mat locus occurred via a disruption of silencing.
We next performed a transition assay to determine the effect of the ⌬atf1 and ⌬pcr1 mutations on the epigenetic inheritance of the repressed state of ade6 ϩ expression (Fig. 1D). In this system, ade6-off cells in the repressed state were distinguished by red coloration, whereas cells expressing ade6 ϩ , ade6-on cells were white. We found a striking increase in ade6-off to ade6-on conversion in ⌬atf1, ⌬pcr1, and ⌬atf1⌬pcr1 mutants, to levels that were significantly higher than that of the wild-type background (Hu51 strain) (Fig. 1D). This indicates that Atf1 and Pcr1 are required for stable epigenetic inheritance of the heterochromatic state at the mat locus.
To address whether Atf1 and Pcr1 are involved in silencing at another interval between mat1 and mat2, called the L region, we investigated the effect of ⌬atf1 or ⌬pcr1 null mutations on the silencing of ade6 ϩ reporter genes integrated at the BglII and SacI sites of the L region. Despite the fact that the L region has no apparent ATF/CREB-binding motifs, the ⌬atf1 and ⌬pcr1 deletion strains had uniformly white colonies on low adenine medium and higher cell viability on adenine-free selective medium relative to the parental strains (AP136 and AP144), demonstrating full derepression of silencing at the L region ( Fig. 2A). Furthermore, ChIP analysis showed that the absence of atf1 and pcr1 resulted in a considerable increase in histone H3/H4 acetylation and euchromatic-specific H3 Lys-4 methylation of the selected region (BglII-R) around the BglII site upstream of the mat2-P locus compared with that of wild type. The heterochromatin-specific H3 Lys-9 methylation was unlikely to be affected (Fig. 2B), perhaps because of the presence of other redundant nucleation mechanisms of heterochromatin assembly via the REII element (20) or cenH repeat (9,22). This indicates that the silencing defects observed in ⌬atf1 and ⌬pcr1 mutants were caused by the change in histone acetylation pattern and subsequent H3 methylation pattern. Furthermore, we found that atf1 and pcr1 deletions affect endogenous mat3-M expression in addition to altering the expression of the ade6 ϩ and ura4 ϩ reporter genes adjacent to the silent mat3-M locus (Fig. 2C). Thus, our results suggest that Atf1 and Pcr1 might act as general silencing factors of the inactive mat2/3 interval and nearby regions.
Phosphorylation Mediated by Sty1/Spc1 MAP Kinase Is Important for Atf1-dependent Regulation of Heterochromatin-Most surprisingly, the Wis1-Sty1/Spc1 kinase cascade is linked to the G 2 -M cell cycle control in optimal growth conditions (24,37), indicating that the role of the pathway is not restricted to stress-inducible responses. Atf1 and Pcr1 are required for heterochromatic silencing at the mat locus under optimal growth conditions, and we wondered whether a deficiency in other components of the Wis1-Sty1/Spc1 cascade could affect matingtype silencing. Most interestingly, we observed that unlike ⌬atf1 and ⌬pcr1 mutant strains, silencing at the mat region was comparable in wild-type, ⌬sty1, and ⌬wis1 mutant strains (Fig. 3A). Moreover, the ade6-off to ade6-on conversion in ⌬sty1 and ⌬wis1 mutants was significantly reduced relative to wildtype cells, indicating that sty1 and wis1 deletions enhanced stabilization of the epigenetic inheritance of the silent states (Fig. 3B). Furthermore, a loss of variegation of ade6 ϩ expression and the stabilized inheritance of the silenced state in the ⌬sty1 mutant strain were reversed in ⌬atf1-⌬sty1 double mutants (Fig. 3, A and B). This might be explained by previous work showing that the repressor activity of Atf1 protein is converted to activator activity via phosphorylation by Sty1/ Spc1 kinase, in the case of the UV-dependent catalase gene (ctt1 ϩ ) induction (38). Specifically, the absence of Sty1/Spc1 MAP kinase might strengthen the repressor activity of the Atf1 protein, leading to increased silencing at the mat region. Thus, we predict that as in general stress responses, Atf1 acts as a downstream effector of the Wis1-Sty1/Spc1 kinases for regulation of heterochromatin silencing.
The results presented above suggest that the kinase activity of Wis1 and Sty1/Spc1 is required for proper control of heterochromatin assembly by Atf1 and Pcr1. The Sty1/Spc1 kinase possesses conserved threonine 171 and tyrosine 173 residues that are phosphorylated by its cognate MEK Wis1 (23). Three mutants in which these conserved amino acids are modified, sty1AY (T171A), sty1TF (Y173F), and sty1AF (T171A, Y173F), displayed an elongated morphology phenotype that was identical to the phenotype of ⌬sty1 deletion cells (23). We used the sty1TF (originally designated spc1TF) mutant to elucidate the role of the phosphorylation state in silencing. To verify that phosphorylation of Sty1/Spc1 kinase by Wis1 is important for silencing of reporter genes inserted within the mat locus, we introduced the mat3-M::ade6 ϩ reporter into the Spc1/Sty1 phosphorylation mutant background (sty1TF) with genetic crosses. Most interestingly, silencing and maintenance of epigenetic inheritance of the silenced state at the mat3 locus were comparable in the ⌬sty1 deletion and the phosphorylation mutant strains (Fig. 3, A and B). These analyses suggest that indeed phosphorylation signaling via the Wis1-Sty1/Spc1 kinase cascade is important for regulation of heterochromatin by Atf1 and Pcr1.
general silencing factors acting on the entire silent mat2/3 region. Moreover, deletion of the mat region containing the potential ATF/CREB-binding sequences caused a partial derepression of endogenous Mc expression at the mat3-M locus (21). Thus, these findings raised the possibility that Atf1 and Pcr1 are associated with an additional RNAi-independent Swi6based mechanism for heterochromatin assembly. To test whether Atf1 and Pcr1 are involved in an alternative Swi6based silencing in the absence of the RNAi-dependent cenH element, we used a K⌬::ade6 ϩ reporter strain, in which a part of the K region containing the cenH was replaced by an ade6 ϩ marker (Fig. 4A). The K⌬::ade6 ϩ strain predominantly exhibited the silent state (designated as ade6-off), showing red coloration (Fig. 4B). Moreover, we confirmed that the deletion of an RNAi component named rdr1 ϩ (RNA-dependent RNA polymerase) in K⌬::ade6 ϩ ade6-off cells did not influence the silencing phenotype (Fig. 4B). This supports the model of a Swi6-dependent mechanism for the residual silencing capability in the K⌬::ade6 ϩ reporter strain lacking the RNAi-depend-  and ⌬pcr1 (YKJ178). B, comparison of H3/H4 acetylation and H3 Lys-4 and Lys-9 methylation levels in atf1 and pcr1 deletion derivatives of the AP144 strain. The bar located downstream of BglII site (A, indicated as BglII-R in B) indicates location of the primer set used in PCR. The levels of histone acetylation and euchromatin-specific H3 Lys-4 methylation were significantly increased, suggesting that atf1 and pcr1 deletions affect the acetylation patterns and subsequently caused the change of methylation patterns at the silent mat loci. The relative fold enrichment shown below each lane was calculated by dividing the ratio of the selected loci/leu1 PCR products in the ChIP sample with that in the input sample. Ac, acetylation; K, lysine; Me, methylation. C, effects of ⌬atf1 and ⌬pcr1 deletions on expression of the endogenous mat3-M gene. Cells were grown in nitrogen-rich (ϩ) and starved (Ϫ) media. Total RNA was prepared from the cultures and subjected to RT-PCR. Competitive PCR for the expression of the Mc and the act1 ϩ genes was carried out. Strains used are as follows: WT (AP137), ⌬atf1 (YKJ135), ⌬pcr1 (HS631), and ⌬swi6 (YKJ129).

FIG. 3. Antagonistic effects of Sty1/Spc1 and Wis1 protein kinases on mat3-M silencing and epigenetic inheritance of the silent states correlate with phosphorylation.
A, silencing phenotype of the Spc1/Sty1 phosphorylation mutant strain was identical to those of ⌬sty1 and ⌬wis1 deletion mutants in a wild-type background as confirmed by cell viability and colony color assays. Comparisons of colony growth ability (top) and color (bottom) on selective medium (ϪAde) or low adenine medium are shown. Colonies are as follows: WT (Hu51); ⌬sty1 (YKJ109); ⌬wis1 (YKJ103); sty1TF (YKJ121); and ⌬atf1⌬sty1 (HS42). B, both the absence of sty1 and wis1 kinases and mutations in phosphorylation sites of Sty1/Spc1 resulted in increased stability of epigenetic inheritance. The increased stability was reversed by combination with an ⌬atf1 null allele. Transition assays were carried out as described in Fig. 1D. ent cenH region. Thus, we investigated the effects of atf1 and pcr1 deletions on the silencing phenotype of K⌬::ade6 ϩ ade6-off cells. Most surprisingly, the silencing defects at the mat2/3 region were comparable in the ⌬atf1, ⌬pcr1, and ⌬swi6 mutants (Fig. 4B). Moreover, our ChIP analysis revealed that the deficiencies of atf1, pcr1, or swi6 completely abolished heterochromatin-specific H3 Lys-9 methylation, concomitant with a significant increase of euchromatin-specific H3 Lys-4 methylation at the selected chromosomal loci across the rest of mat2/3 region lacking the cenH element (Fig. 4, C and D). This dem-onstrates that in K⌬::ade6 ϩ cells lacking the cenH repeat, the presence of H3 Lys-9 methylation strictly depends on Atf1 and Pcr1, in a manner similar to its dependence on Swi6. Furthermore, ChIP analysis using histone acetylation-specific antibodies showed that deficiencies of atf1, pcr1, and swi6 resulted in a significant increase of histone H3/H4 acetylation at all positions tested relative to the parental wild-type strain (Fig. 4E). This suggests that Atf1 and Pcr1 act in early steps of heterochromatin assembly and are required for the establishment of heterochromatin-specific histone modification patterns such as FIG. 4. Deficiencies in atf1 ؉ , pcr1 ؉ , and swi6 ؉ affect heterochromatin-specific histone modification patterns and subsequent silencing across the rest of the silent mat region lacking the cenH repeat. A, schematic representation of K⌬:: ade6 ϩ , in which the cenH region was replaced with ade6 ϩ . The bars marked by numbers or names indicate location of primer sets used for PCR amplification. B, three genes, atf1 ϩ , pcr1 ϩ , and swi6 ϩ , are required to maintain silencing, whereas the RNA-dependent RNA polymerase gene, rdr1 ϩ , is dispensable for maintenance of heterochromatin silencing in the K⌬:: ade6 ϩ ade6-off strain. Comparison of each colony growth and color on selective (ϪAde) or low adenine medium is shown. Strains used are as follows: WT (AP152), ⌬atf1 (HS3001), ⌬pcr1 (HS3051), ⌬swi6 (HS3072), and ⌬rdr1 (HS3091). C, levels of H3 Lys-9 methylation in ⌬atf1, ⌬pcr1, ⌬swi6, and ⌬rdr1 derivatives from Fig. 3B were determined by ChIP analysis. DNA extracted from ChIP or input was amplified by competitive PCR by using the primer sets shown in A. An endogenous leu1 fragment served as an internal control. Relative fold enrichment shown below each lane was calculated by dividing the ratio of the selected loci/leu1 PCR products in the ChIP sample with that in the input sample. The absence of atf1 and pcr1 reduced H3 Lys-9 methylation to a level comparable with swi6. D, levels of H3 Lys-4 methylation in the strains used in C were determined by ChIP analysis. The absence of atf1 and pcr1 resulted in a significant increase of euchromatin-specific H3 Lys-4 methylation at the mat loci indicated as ade6-R. PCR was performed using the primer set of ade6-R and leu1. E, levels of H3/H4 acetylation in the same strains used in C and D were determined by ChIP analysis. The absence of atf1 and pcr1 resulted in a significant increase of histone H3/H4 acetylation at all positions tested at the mat loci indicated as ade6-R. PCR was performed using primer sets of ade6-R and leu1. Relative fold enrichment in each graph was calculated by dividing the ratio of the ade6-R locus/leu1 PCR products in the ChIP sample with that in the input sample. Ac, acetylation; K, lysine; Me, methylation. deacetylation at H3 Lys-9 and H3 Lys-14. Taken together, these findings suggest that flanking sequences present in the K⌬::ade6 ϩ strain are capable of establishing heterochromatinspecific histone modification patterns such as H3/H4 deacetylation and H3 Lys-9 methylation only in the presence of Atf1, Pcr1, and Swi6.
The Swi6 and Clr4 proteins were previously known as limiting chromatin components that convert the epigenetic imprint from the expressed state to the silenced state when overexpressed, emphasizing their dosage-critical roles in heterochromatin formation (39). To evaluate the establishment of a heterochromatin-like structure in the K⌬::ade6 ϩ ade6-on (white color colony), we quantified the ade6-on to ade6-off conversion following overexpression of Swi6 and Clr4. We observed increases in ade6-on to ade6-off conversion by overexpressed Swi6 and Clr4 (Fig. 5A, left panel), consistent with previous work (9). To delineate the role of Atf1 and Pcr1 in a Swi6-based silencing mechanism in the absence of the RNAi-dependent cenH repeat, we investigated whether atf1 deletion affects the ade6-on to ade6-off conversion in K⌬::ade6 ϩ ade6-on when Swi6 or Clr4 is overexpressed. Most surprisingly, the absence of atf1 abolished the dosage-dependent effect of Swi6 and Clr4 on the ade6-on to ade6-off conversion (Fig. 5A, right panel). More importantly, only re-introduction of a wild-type atf1 ϩ copy recovered the ade6-on to ade6-off conversion to a level similar to the isogenic parental strain in K⌬::ade6 ϩ ade6-on cells bearing a null allele of atf1 (Fig. 5A). This demonstrates that the residual capacity for ade6-on to ade6-off conversion found in K⌬::ade6 ϩ ade6-on cells was completely dependent on the presence of Atf1 (Fig. 5A). This suggests that Atf1 and Pcr1 are indispensable for the initiation of heterochromatin formation via an alternative Swi6-based mechanism in the absence of the RNAi-dependent cenH element. We next measured the levels of H3 Lys-9 methylation in Swi6-or Clr4-overexpressing derivatives of K⌬::ade6 ϩ ade6-on cells using a ChIP assay. Consistent with the silencing phenotypes (Fig. 5A), ChIP analysis showed that overexpression of Swi6 or Clr4 recovered H3 Lys-9 methylation in K⌬::ade6 ϩ ade6-on cells with an atf1 ϩ wild-type copy but not in an ⌬atf1null background (Fig. 5B). The overexpressed protein levels induced in thiamine-depleted media were confirmed by Western blot analysis using anti-HA antibodies (Fig. 5C). Taken together, these results suggest that Atf1 and Pcr1 are key regulators of Swi6-dependent heterochromatin formation and silencing at the silent mat region in the absence of the RNAidependent cenH element.
Atf1 and Pcr1 Can Form Complexes with Clr6 and Swi6 -For silencing at the mat2/3 region, Atf1 and Pcr1 proteins might cooperate with common silencing factors. In particular, we reasoned that Clr3, Clr6, and Swi6 are potential candidates because atf1 deletion affected the histone deacetylation and methylation patterns responsible for establishment and maintenance of heterochromatin ( Fig. 2B and Fig. 4, C-E). When cells were treated with tricostatin A, a histone deacetylase inhibitor, the basal level of ctt1 ϩ expression in the ⌬sty1 deletion background was equivalent to that of ⌬atf1 mutant cells. 2 To find out whether Atf1 and Pcr1 interact with common silencing factors such as Clr3, Clr6, and Swi6, we performed GST-pull down assays. Protein extracts from cells expressing each protein were used for GST-pull downs followed by Western blotting. Most interestingly, GST-Atf1 could form complexes with Clr6-HA, but interactions with Clr3-HA were minimal (Fig. 6A, left top panel). GST-Swi6 binds to Atf1-HA and Pcr1-HA (Fig. 6A, left bottom). Expression of the GST-tagged proteins was confirmed by Western blot by using an anti-GST antibody (Fig. 6A, right). Thus, these data suggest that complex formation between Atf1, Pcr1, Clr6, and Swi6 is important for heterochromatin assembly at the mat2/3 region.
FIG. 5. Atf1 and Pcr1 are critical for silencing and H3 Lys-9 methylation via a Swi6-dependent mechanism across the silent mat region in the absence of the RNAi-dependent cenH element. A, overexpression of Swi6 and Clr4 caused increases in the ade6-on to ade6-off conversion in wild type but not in ⌬atf1 mutants. After introduction of pREP41 plasmid-derived overexpression cassettes for HAtagged Swi6 and HA-tagged Clr4 or Atf1-HA proteins (or vector only) into K⌬:: ade6 ϩ ade6-on atf1 ϩ (AP152) or K⌬:: ade6 ϩ ade6-on ⌬atf1 (HS3011) cells, white (transformant) colonies were selected and replated onto adenine-limiting minimal medium without thiamine. After a 5-day incubation, red (ade6-off) colonies were counted, and the frequency of red to total colonies was calculated. The ade6-on to ade6-off conversion in Swi6-or Clr4-overexpressing strains increased 6 -15-fold relative to that of the wild type strains carrying vector only (left). The absence of atf1 abolished the effect of Swi6-or Clr4-overexpression on ade6-on to ade6-off conversion (right). B, ChIP analysis of H3 Lys-9 methylation in Swi6-or Clr4-overexpressing derivatives of K⌬:: ade6 ϩ ade6-on cells with or without a wild type atf1 ϩ copy. Consistent with the results shown in the ade6-on to ade6-off conversion (A), Swi6-or Clr4overexpression is capable of recovering the lost H3 Lys-9 methylation against an atf1 ϩ background but not a ⌬atf1-null background. C, appropriate expression of Swi6 and Clr4 in yeast cells was confirmed by Western blot (WB) analysis. Crude extracts of the strains overexpressing HA-tagged Swi6 and HA-Clr4 were subjected to immunoblot using an anti-HA antibody.

Clr6 Is a Major Histone Deacetylase Essential for Regulation of Heterochromatin Assembly and Gene Silencing by Atf1 and
Pcr1-To explain the biological significance of this physical interaction between Atf1, Pcr1, and Clr6, the effect of clr6 mutations on histone acetylation and methylation patterns was examined. As shown in Fig. 6B, under semi-permissive growth conditions (30°C), mutations in clr6 ϩ essential for cell growth resulted in an increased level of histone acetylation at all positions tested, a phenotype similar to that of ⌬atf1 and ⌬pcr1 mutant cells (see Fig. 4). In addition, the clr6 mutations showed a moderate but consistent decrease in heterochromatin-specific H3 Lys-9 methylation, concomitant with a moderate increase of euchromatin-specific H3 Lys-4 methylation at the selected chromosomal mat loci (ade6-R). This functional interaction between Atf1 and Clr6 was also supported by the findings that both ⌬clr3-clr6-1 and ⌬atf1-⌬clr3 double mutants showed synergistic effects on the silencing and histone acetylation patterns at the mating-type region (data not shown; see Ref. 17).
To investigate whether the functional interaction between Atf1 and Clr6 is required for euchromatic gene silencing, we performed ChIP analysis for localization of Atf1 and histone acetylation patterns at the stress-inducible ctt1 ϩ promoter region in ⌬atf1, ⌬pcr1, clr6-1, and ⌬clr3 mutants. ChIP analysis revealed that Atf1-HA was highly enriched at the promoter region encompassing the ATF/CREB-binding site (Fig. 6C, left). Moreover, the ChIP assay using histone acetylation-specific antibodies showed that, similar to ⌬atf1 and ⌬pcr1 deletion mutants, the clr6 mutation resulted in a moderate but consistent increase (about 2-fold) in histone H3/H4 acetylation at the ctt1 ϩ promoter region (Fig. 6C, right). In contrast, the clr3 deletion mutant showed little effect on histone acetylation (Fig.   FIG. 6. Cooperation between Atf1, Pcr1, Clr6, and Swi6 is important for heterochromatin assembly and gene silencing. A, GST-Swi6 can form complexes with Atf1-HA and Pcr1-HA and GST-Atf1 interacts with Clr6-HA but little with Clr3-HA. The total protein extracts of HS1001 (expressing HA-tagged Clr6 at the endogenous genomic locus) and HS1005 (expressing HA-tagged Clr3) cells overexpressing GST or GST-Atf1 were incubated with GST-Sepharose beads for a GST-pull down assay, and the bound proteins were subjected to Western blot analysis using anti-HA antibodies. In addition, the total extracts of HS801 (expressing Atf1-HA at the endogenous genomic locus) and HS821 (expressing Pcr1-HA) cells transformed with GST-or GST-Swi6-overexpressing plasmid were subjected to GST-pull down assay. B, the clr6 mutation affects histone acetylation and methylation patterns at the selected genomic locus (ade6-R) within the mating-type region. Levels of H3/H4 acetylation and H3 Lys-4 and Lys-9 methylation were determined by ChIP analysis. PCR was performed using primer sets for ade6-R and leu1. Relative fold enrichment below each lane was calculated by dividing the ratio of the ade6-R locus/leu1 PCR products in the ChIP sample with that in the input sample. Strains used are as follows: WT (AP152) and clr6-1 (HS7010). Ac, acetylation; K, lysine; Me, methylation. C, Atf1-HA was enriched at the promoter region of ctt1 ϩ surrounding the ATF/CREB-binding site, and histone acetylation patterns at the same region were affected by the clr6 mutation but not in the clr3 mutant. A physical map of the ctt1 ϩ genomic locus is shown (top). The PCR fragment indicated as Ϫ0.4 includes the ATF/CREB-binding site and the TATA box. In vivo association of Atf1-HA with the chromosomal region surrounding the ctt1 ϩ genomic locus in a strain expressing HA-tagged Atf1 was determined by ChIP. Levels of H3 Lys-9-Ac and H4 Lys-16-Ac at the selected locus (Ϫ0.4) in clr6-1 K⌬:: ade6 ϩ strain grown at 30°C (semi-permissive temperature) were determined by ChIP analysis. Strains used are as follows: Atf1-HA (HS801), WT (AP152), ⌬atf1 (HS3001), ⌬pcr1 (HS3051), clr6-1 (HS7010), and ⌬clr3 (HS1051). 6C, right). Thus, these findings suggest that Clr6 is a key histone deacetylase that regulates heterochromatin assembly and gene silencing by the Atf1-Pcr1 heterodimer. DISCUSSION Most interestingly, we showed that disruption of Atf1 and Pcr1 resulted in a marked decrease of gene silencing at the silent mat region, but increased repression at centromeres and telomeres, suggesting that this silencing is locus-specific (Fig.  1). The phenotype could be explained by a competition model based on the functional interaction of Atf1 with common silencing factors such as Clr6 and Swi6 (Fig. 6). This phenotype is reminiscent of the relocalization of Sir3 from telomeres to rDNA loci and the subsequent increase of rDNA silencing in Sir4-null mutant cells of S. cerevisiae (40). Likewise, in the ⌬atf1 mutant cells, general silencing factors such as Clr6, Clr4, and Swi6 normally associated with Atf1 might be released from specific genomic loci including the mat loci and potential Atf1repressive promoter regions (41), and then accumulate in other regions of heterochromatin such as centromeres and telomeres, leading to their increased silencing.
The findings presented here suggest that phosphorylation by the Wis1-Sty1/Spc1 kinase cascade is important for regulation of heterochromatin by Atf1 and Pcr1. How phosphorylation can influence Atf1-dependent heterochromatin silencing might be explained by an activator-repressor model for Atf1 activity, as proposed by Degols and Russell (38). This hypothesis suggests that Atf1 has an intrinsic repressor activity, which is converted to an activator upon phosphorylation by the Spc1/StyI kinase. This model was supported by the previous findings that Atf1 negatively regulates ctt1 ϩ expression in the absence of Spc1/ Sty1 kinase activity (38), and Spc1/Sty1 directly regulates Atf1 activity through phosphorylation (24,26). In addition, the model is supported by reports that the budding yeast Hog1 kinase converts a repressor complex containing the ATF/CREB repressor Sko1 (Acr1) into an activator in response to osmotic stress (42). What is the biological relevance of interconversion between an activator activity and a repressor activity of Atf1? Swi6, one of the common silencing factors, physically interacts with several proteins involved in mating-type switching (10), suggesting that Swi6 provides an interface between heterochromatin silencing and mating-type interconversion by establishing a chromatin structure favorable to both processes (10). Similarly, we propose that regulation of Atf1 activity by the MAP kinase cascade may be required to establish a dynamic chromatin structure favorable to both silencing and switching processes at the mating-type region.
As indicated previously (7,9), the precise mechanism of the Swi6-dependent heterochromatin assembly at the silent mat regions in the absence of RNAi-dependent cenH is still not clear. Our present findings demonstrate that Atf1 and Pcr1 regulate Swi6-dependent silencing at the mat region independently of cenH-mediated silencing. In our current model, Atf1 and Pcr1 bind specifically to putative ATF/CREB-binding sites upstream of mat3-M (21), as a result of their sequence-specific binding properties (25,30,33); subsequently, they recruit Clr6 histone deacetylase and Swi6 to the mat locus. This hypothesis is consistent with our findings that Atf1 and Pcr1 can form complexes with Clr6 and Swi6 (Fig. 6) and that histone acetylation patterns found in clr6-1 mutant cells were similar to those of ⌬atf1 and ⌬pcr1 deletion mutant cells (Figs. 4E and 5B). In the absence of the main nucleation site of the cenH element, the establishment of histone deacetylation patterns at both H3 and H4 is initiated by Atf1 and Pcr1. These events nucleate heterochromatin by subsequently targeting H3 Lys-9 methylation and creating Swi6-binding sites around the ATF/ CREB-binding sequence. Subsequently, the spread of hetero-chromatin across the whole mat2/3 region occurs in a Swi6dependent, self-propagating manner (6,9). These Swi6-based activities initiated by Atf1 and Pcr1 might function in an inefficient and highly stochastic manner, as proposed previously (9), because the dosage-critical role of Swi6 and Clr4 in silencing is strictly dependent on the presence of Atf1 (Fig. 5, A  and B).
This proposed mechanism is reminiscent of mammalian gene silencing by the retinoblastoma-SUV39H1-HP1 complex (43), mating-type, and telomeric silencing by the budding yeast Rap1-Sir protein complexes (44,45) and centromeric heterochromatin assembly by the CENP-B-Swi6 complex (46). A tumor suppressor protein retinoblastoma recruits histone-modifying factors and HP1 protein to the cyclin E promoter, leading to H3 Lys-9 methylation and gene silencing (43). In addition, it has been proposed that both Sir3 and Sir4 can directly and independently bind to Rap1 at mating-type silencers and telomeres, suggesting that Rap1-mediated recruitment of Sir proteins operates through multiple cooperative interactions (44,45).
Collectively, our findings suggest that Atf1 and Pcr1 are key regulators that are involved in an early step for nucleation of a Swi6-dependent, RNAi-independent heterochromatin assembly at the mating-type region through cooperation with the histone deacetylase Clr6 in the fission yeast. Moreover, cooperation between Atf1-Pcr1 heterodimers and Clr6 is also required for euchromatic gene silencing of transcriptional targets of Atf1, such as the stress-inducible ctt1 ϩ . Although the conserved role of ATF-2, the mammalian counterpart to Atf1, in heterochromatin silencing remains to be demonstrated experimentally, our data provide novel insights into the roles of members of the ATF/CREB family in heterochromatin formation and gene silencing in higher eukaryotes.