The binding of Chp2’s chromodomain to methylated H3K9 is essential for Chp2’s role in heterochromatin assembly in fission yeast

The binding of heterochromatin protein 1 (HP1) to lysine 9–methylated histone H3 (H3K9me) is an essential step in heterochromatin assembly. Chp2, an HP1-family protein in the fission yeast Schizosaccharomyces pombe, is required for heterochromatic silencing. Chp2 recruits SHREC, a multifunctional protein complex containing the nucleosome remodeler Mit1 and the histone deacetylase Clr3. Although the targeting of SHREC to chromatin is thought to occur via two distinct modules regulated by the SHREC components Chp2 and Clr2, it is not clear how Chp2’s chromatin binding regulates SHREC function. Here, we show that H3K9me binding by Chp2’s chromodomain (CD) is essential for Chp2’s silencing function and for SHREC’s targeting to chromatin. Cells expressing a Chp2 mutant with defective H3K9me binding (Chp2-W199A) have a silencing defect, with a phenotype similar to that of chp2-null cells. Genetic analysis using a synthetic silencing system revealed that a Chp2 mutant and SHREC-component mutants had similar phenotypes, suggesting that Chp2’s function also affects SHREC’s chromatin binding. Size-exclusion chromatography of native protein complexes showed that Chp2-CD’s binding of H3K9me3 ensures Clr3’s chromatin binding, and suggested that SHREC’s chromatin binding is mediated by separable functional modules. Interestingly, we found that the stability of the Chp2 protein depended on the Clr3 protein’s histone deacetylase activity. Our findings demonstrate that Chp2’s H3K9me binding is critical for SHREC function and that the two modules within the SHREC complex are interdependent.


Introduction
Epigenetic changes affect genome function without an alteration of the DNA sequence, and they can be inherited by daughter cells and sometimes by offspring. Epigenetic changes in the genome are important in many diseases, including cancer, diabetes type II, and obesity [1,2].
were cleaved by BglII and introduced into the original chp2 + locus. To replace the wild-type chp2 + allele with the mutant chp2 allele, strains that lost the ura4 + gene through internal homologous recombination were isolated using a counter-selective medium containing 5-fluoroorotic acid (FOA). All strains used in this study are listed in S1 Table.

Recombinant protein production
The recombinant 6×His-tagged proteins used in isothermal titration calorimetry (ITC) analyses were expressed in E. coli BL21 (DE3) and purified by TALON Metal Affinity Resin (Clontech) according to the manufacturer's instructions. Recombinant proteins were further purified by anion-exchange chromatography (Source 15Q; GE Healthcare).

ITC
ITC was conducted using a MicroCal VP-ITC calorimeter (GE Healthcare) at 20˚C. Recombinant Chp2-CD proteins were dissolved in phosphate buffer (20 mM KPO 4 [pH 6.8] and 100 mM NaCl). A typical titration consisted of injecting 1-μl aliquots of ligand (H3K9me3; ARTKQTAR(Lys[Me]3)STGGKAPRY) into the protein sample at 3-min intervals to ensure that the titration peak returned to baseline. ITC data were analyzed using the Origin program.

Silencing assays and images of yeast colonies
Spot tests were performed as previously described [16] using the following procedure: midlog-phase cultures were adjusted to 3.2 × 10 6 cells/ml, serially diluted in five steps, and applied to plates in 5-μl drops. A Canon EOS 1100D with a Canon EF 50-mm lens or MP-E 65 mm f/2.8 1-5× Macro Lens objective was used to take pictures of the yeast colonies.

Real-time quantitative reverse transcription PCR (RT-qPCR)
S. pombe strains were grown in minimal medium (EMM) as shaken liquid cultures at 200 rpm, at 30˚C, until log phase (1 × 10 7 cells/ml). From each culture, 2 × 10 7 cells were harvested by centrifugation at 3,000 ×g for 3 min at 4˚C. Total yeast RNA was isolated with the RNeasy Mini Kit (Qiagen) according to the manufacturer's instructions. RNA quality and concentrations were determined using 1% agarose gel electrophoresis and the NanoDrop™ 1000 spectrophotometer (Thermo Scientific). cDNA was synthesized with the Maxima First Strand cDNA Synthesis Kit for RT-qPCR (Thermo Scientific). Experiments were done in biological triplicates, with technical duplicates for each biological sample. For RT-qPCR, we used an MJ Mini™ Thermal Cycler (Bio-Rad) or a MiniOpticon™ Real-Time PCR System (Bio-Rad) with SYBR Green technology and the 5 × HOT FIREPol 1 EvaGreen 1 qPCR Supermix (Solis Bio-Dyne). We used the following primers: ura4-Fw, 5'-CGTGGTCTCTTGCTTTTG-3', ura4-Rv, 5'-GTAGTCGCTTTGAAGGTTAGG-3'; act1-Fw, 5'-GGTTTCGCTGGA GATGATG-3', act1-Rv, 5'-ATACCACGCTTGCTTTGAG-3' [17]. Data were presented as the fold enrichment in ura4 + transcript levels relative to the wild-type strain. Briefly, the threshold-cycle (C T ) values were used to calculate Mean Normalized Expression (MNE) values, which were converted into fold-change values and normalized to gene expression in the wild-type strain [18]. Error bars indicated standard error of the mean (S.E.M.). at 120 rpm. Glycine was added to the cells to a final concentration of 125 mM while shaking gently for 5 min. Cells were collected by centrifugation at 3,000 ×g, for 5 min at 4˚C and subsequently washed three times with 40 ml ice-cold phosphate-buffered saline (PBS). Pelleted cells were resuspended in 400 μl ice-cold lysis buffer (50 mM HEPES-KOH [pH 7.5], 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% Na-deoxycholate [DOC], 1 mM phenylmethylsulfonyl fluoride [PMSF], and 1× protease inhibitor cocktail [Complete™ EDTA-free; Roche]). Cell suspensions were combined with 500 μl of ice-cold glass beads (0.5-mm diameter, BioSpec Products) in 2-ml skirted micro tubes (Sarstedt) and lysed four times for 20 sec with Fas-tPrep 1 FP120 Cell Disrupter (BIO 101/Savant). Cell lysates were collected in 2-ml DNA LoBind tubes (Eppendorf), resuspended, and transferred to 1.5-ml DNA LoBind tubes (Eppendorf) before sonication using a Bioruptor™ Next Gen (Diagenode) at 4˚C for 10 cycles, 30 sec on/60 sec off, with the power set to high. The samples were centrifuged at 16,000 ×g for 5 min at 4˚C. The supernatant, which contained the soluble chromatin, was transferred to 1.5-ml DNA LoBind tubes. The pellet was resuspended in 400 μl of ice-cold lysis buffer and sonicated again, and the supernatants were pooled after centrifugation. The pooled chromatin samples were centrifuged one more time at 16,000 ×g for 10 min at 4˚C, and the lysates (~800 μl) were placed in new 1.5-ml DNA LoBind tubes and kept on ice. DNA concentrations were measured with a NanoDrop™ instrument (Thermo Fisher Scientific).
In the ChIP assays, we used an anti-FLAG (F1804, Sigma-Aldrich) or anti-Myc (MA1-980, Thermo Fisher Scientific) antibody and Pierce ChIP-grade protein-A/G magnetic beads (Thermo Scientific/Pierce). Usually, each immunoprecipitation used 50-100 μg of lysate (50 μg of chromatin) and 20 μl of beads (50% slurry) bound to 2 μg of antibody, and total volumes were adjusted to 250 μl with the lysis buffer. Unbound magnetic beads were used for control immunoprecipitations. Input fractions (10%) were removed from the chromatin lysates prior to immunoprecipitation, adjusted to the total volumes of 250 μl with 50 mM Tris-HCl (pH 8.0), 10 mM EDTA, and 1% SDS, incubated overnight at 65˚C, cooled to room temperature, and treated with DNase-free RNase A (Thermo Fisher Scientific) for 1 h at 37˚C and Proteinase K (Thermo Fisher Scientific) for 3 h at 56˚C. The immunoprecipitated chromatin beads (bound fraction) and control (unbound, no antibody) beads were washed twice with lysis buffer, once with wash buffer I (10 mM Tris-HCl [pH 8.0], 250 mM LiCl, 0.5% NP-40, 0.5% DOC, and 1 mM EDTA), and once with TE buffer (10 mM Tris-HCl [pH 8.0] and 1 mM EDTA) containing 0.05% Tween 20. DNA was extracted twice using TES elution buffer (50 mM Tris-HCl [pH 8.0], 10 mM EDTA, and 1% SDS) for 30 min at 65˚C. The elution fractions were mixed, incubated overnight at 65˚C, and treated as described for input fractions. DNA was extracted from the solutions using the QIAquick PCR Purification Kit (Qiagen) or MiniElute PCR Purification Kit (Qiagen).
To amplify immunoprecipitated ura4 + , act1 + , and centromeric dg DNA sequences, realtime qPCR was performed in duplicates for each technical replicate using the same primers as for RT-qPCR. Relative DNA quantification was carried out at least for two biological replicates and technical duplicates for each biological replica. Real-time PCR data were analyzed by the percent input method [18]. Briefly, the C T values for input and bound (IP) fractions were averaged. The input DNA C T values for the ura4 + and act1 + genes were adjusted from 10% (starting input fraction) to the 100% equivalent by subtracting 3.32 C T (Log 2 10). The percent input was calculated by 100 x AE^(C T [adjusted Input]-C T [IP]), where AE is the amplification efficiency for a given pair of primers. We normalized the % input values for the ura4 + and act1 + genes relative to the wild-type strain; data were represented as the relative foldchange in % input for the corresponding gene, with error bars indicating the standard deviation (SD).

Native yeast-cell protein extracts
Yeast cells were grown at 30˚C to late log-phase (1.5 × 10 7 cells/ml) in 100 ml of rich-growth YEA medium shaken at 200 rpm. Cells were harvested by centrifugation at 3,000 ×g for 3 min at 4˚C, washed twice with 40 ml of ice-cold 1× stop buffer (150 mM NaCl, 50 mM NaF, 10 mM EDTA, and 1 mM NaN 3 ), and finally transferred into 3 × 2-ml micro tubes (Sarstedt) and centrifuged at 3,000 ×g for 5 min at 4˚C to remove the stop buffer. The pelleted cells were combined with 750 μl of ice-cold glass beads (0.5-mm diameter, BioSpec Products) along with 400 μl of ice-cold native lysis buffer (50 mM Tris-HCl [pH 8.0], 150 mM NaCl, 50 mM NaF, 10% glycerol, 2 mM EDTA, 10 mM EGTA, 0.5% Triton X-100, 0.5% IGEPAL 1 CA-630 [Sigma-Aldrich], 5 mM ß-glycerophosphate, 0.1 mM Na 3 VO 4 , 1 mM ß-mercaptoethanol, 2 mM dithiothreitol [DTT], 1 mM PMSF, and 2.5 × protease inhibitor cocktail [Complete™ EDTA-free; Roche]) and were lysed twice for 20 sec at 4˚C with a FastPrep 1 FP120 Cell Disrupter (BIO 101/Savant) on maximum power. An additional 400 μl of ice-cold native lysis buffer was added to the samples, and the crude native lysates were incubated for 15 min in a rotating wheel at 4˚C (for RNase A treatment, DNase-and protease-free RNase A [Thermo Scientific] was added to the native lysis buffer to a final concentration of 1 mg/ml). After incubation, the lysate was transferred to DNA LoBind tubes (Eppendorf), and the samples were centrifuged for 10 min at 16,000 ×g at 4˚C. The supernatants were carefully transferred to a glass beaker, mixed together, filtered with a 0.20-μm filter unit (Sarstedt), and kept on ice before loading into columns for size-exclusion chromatography. The protein concentration in the lysates, measured with a NanoDrop™ 1000 (Thermo Fisher Scientific), was typically 20-25 mg/ml.

Gel filtration chromatography
Native protein extracts were fractionated by fast protein liquid chromatography (FPLC). Briefly, 1 ml of pre-cleared native protein extract was loaded onto a HiLoad 16/600 Superdex 200 prep-grade column (GE Healthcare) attached to an Ä KTAxpress FPLC system (GE Healthcare) and equilibrated with running buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 0.05% Tween 20, and 2 mM DTT). The column was also calibrated with a Gel Filtration HMW Calibration Kit (GE Healthcare) at running conditions. The proteins were fractionated at 4˚C with a flow rate of 1 ml/min. After discarding the first 30-ml flow-through, 2-ml fractions were collected, and the proteins were precipitated according to the Peterson trichloroacetic acid (TCA)-DOC protocol with some modifications. Briefly, a 1/10 volume of 0.15% DOC solution was added to the fraction sample and mixed by inverting the tube several times, and the sample was incubated on ice for 15 min. Next, 1/10 of the original sample volume of chilled 72% TCA was added, and the sample was mixed by inverting the tube several times, and then incubated on ice for 15 min. The samples were centrifuged at 16,000 ×g for 15 min at 4˚C and the supernatant carefully discarded. Pre-chilled (-20˚C) acetone containing 10 mM ß-mercaptoethanol was added to the samples and the tubes were vortexed for 30 sec. The samples were left overnight at -20˚C, then centrifuged at 16,000 ×g for 15 min at 4˚C. The supernatants were discarded, and the pellets were air-dried at room temperature for 5-10 min and solubilized with 30 μl Urea-SDS buffer (50 mM Tris-HCl [pH 8.0], 8 M urea, 1% SDS, 10 mM ß-mercaptoethanol, and 10 mM DTT).

Chp2's H3K9me binding is essential for its function in heterochromatic silencing
To investigate the role of H3K9me binding in Chp2's function, we created a strain that expressed a mutated Chp2 protein containing W199A (Chp2-W199A) from the endogenous Chp2-CD's H3K9me binding and heterochromatin assembly in fission yeast chp2 + locus (Fig 1A; see Materials and methods). Since the level of endogenous Chp2 expression is relatively low [10], we introduced multiple FLAG tags simultaneously into the Chp2 Nterminus to allow us to readily detect Chp2 (F-Chp2-W199A) (S1A Fig). As a control, we also created a strain expressing the wild-type Chp2 protein with the same FLAG tags from the endogenous promoter (F-Chp2) (S1A Fig). Next, we examined the impact of Chp2's H3K9me binding on its function in heterochromatic silencing. The F-chp2-W199A allele or the control F-chp2 was combined with the mat3-M::ura4 + reporter gene (Fig 2A), and the silencing states were evaluated by a spotting assay in which cells were cultured, serially diluted, and spotted onto either non-selective plates (N/S), plates to select for ura4 + -expressing cells (-Ura), or counter-selective plates containing 5-fluoroorotic acid (FOA) (Fig 2B) [21]. The ura4 + gene inserted into the mating-type region was tightly repressed in wild-type cells, which grew poorly on-Ura plates but grew well on FOA plates (Fig 2B, wt). A lack of Chp2 derepressed the reporter gene, as evidenced by good cell growth on-Ura plates and no growth on plates containing FOA (Fig 2B, chp2Δ); this agrees with previously published data [10,12]. RT-qPCR also confirmed the derepression state of mat3-M::ura4 + (Fig 2B, right). We confirmed that adding the FLAG tags to Chp2's N-terminus did not disturb its function, since the silencing state of the mat3-M::ura4 + reporter was comparable to that in wild-type cells (Fig 2B, F-chp2).
Cells expressing F-Chp2-W199A showed defective silencing in the spotting assay and in RT-qPCR (Fig 2B), although RT-qPCR revealed slightly milder derepression in F-chp2-W199A than in chp2Δ cells, with slight variations between independent isolates (Fig 2B, last three rows). There were two possible explanations for this partial derepression: either silencing was alleviated at a similarly low level in all cells, or the culture contained cells with variegated derepression states [16,22]. To distinguish between these possibilities, we used a mat3-M::ade6 + allele, in which the ade6 + gene was inserted into the same position as the ura4 + reporter (Fig 2A). When wild-type cells repressing mat3-M::ade6 + were plated on a medium containing limited amounts of adenine, the cells formed red colonies due to the accumulation of an intermediate in the adenine-synthesis pathway (Fig 2C, top left) [23]. A control strain expressing the F-Chp2 protein also formed red colonies on the plates, like the wild-type strain (Fig 2C,  bottom left), while the chp2Δ and F-chp2-W199A strains formed two types of coloniessmaller pink and larger white colonies (Fig 2C, right panels). The existence of two cell populations in these mutant strains indicated that there was indeed an epigenetic shift between the repressed and derepressed chromatin states at the mating-type region. However, repressed colonies were light pink rather than red indicating that the ade6 + -reporter gene was not fully repressed, suggesting that both chp2 mutations caused a partial derepression in all cells and full derepression in a certain percentage of the cells.
To confirm that the two mutant strains had similar phenotypes, we restreaked pink and white colonies and counted how many pink versus white colonies were generated in the offspring (Fig 2D and Table 1). A restreaked pink colony formed by the chp2Δ strain generated 78% pink and 22% white new colonies; the chp2-W199A strain formed 79% pink and 21% white new colonies. This very similar frequency of switching indicates that the two mutations had very similar effects on mating-type silencing.
To determine whether the chp2-W199A mutant's phenotype was due to a reduced binding of Chp2-W199A to the mating-type region, we conducted ChIP assays with an anti-FLAG antibody and primers against the act1 + (control) and ura4 + genes. The strain expressing wildtype F-Chp2 protein was strongly enriched at the mat3-M::ura4 + region (Fig 2E, F-chp2), whereas three different isolates of the strains expressing F-Chp2-W199A were only slightly or moderately enriched in F-Chp2-W199A bound to the mat3-M::ura4 + gene. The enrichment of F-Chp2-W199A bound to the act1 + control gene was slightly higher than that of the wild-type  (Fig 2E; compare F-chp2 to F-chp2-W199A). Taken together, these results indicated that Chp2's ability to bind H3K9me is crucial for its silencing function.

Chp2's H3K9me binding is necessary for targeted silencing
Having established that Chp2's molecular function is to bind chromatin via H3K9me, we next focused on Chp2's function in the SHREC complex. We used a minimal silencing system in which an engineered Clr4 protein lacking the CD and fused to a Gal4 DNA-binding domain (GBD-Clr4ΔCD) silences a reporter gene placed downstream of a triple Gal4-binding site (3×gbs) integrated at a euchromatic location (Fig 3A) [24]. In fission yeast, several distinct pathways cooperate to assemble repressive heterochromatin, making it difficult to define the exact role of trans-acting factors based on a reporter gene inserted into the mating-type region. However, a minimal silencing assay allows us to define factors that are specifically required for Clr4-mediated heterochromatin assembly. Previous studies using this system revealed distinct roles of the RNAi pathway and demonstrated that three SHREC components-Clr2, Clr3, and Chp2-are necessary for targeted silencing [22,24] (Fig 3B).
https://doi.org/10.1371/journal.pone.0201101.g002 Chp2-CD's H3K9me binding and heterochromatin assembly in fission yeast necessary for targeted silencing. We also confirmed that Clr3's HDAC activity was crucial for ectopic silencing; in a strain with a clr3-735 loss-of-function mutation [25], this silencing was clearly absent (Fig 3B; compare rows 9 and 10). Importantly, the strain expressing the F-Chp2-W199A protein also failed to silence the reporter gene; it had a derepression state comparable to that of the SHREC-component mutants (Fig 3B, 13th row). These results confirmed that Chp2-CD's H3K9me binding is important in targeted silencing, and indicated a functional link between Chp2 and SHREC's targeting to chromatin. Chp2-CD's H3K9me binding and heterochromatin assembly in fission yeast

Chp2's H3K9me binding affects Clr3's chromatin targeting
SHREC was previously shown to contain two functional modules and to target chromatin through separable recruitment mechanisms [12,14]. To test whether Chp2's H3K9me binding affects SHREC's chromatin targeting, we combined yeast strains expressing myc-tagged Clr3 (Clr3-myc), V5-tagged Clr2 (V5-Clr2) [7,22], or F-Chp2 with each of the SHREC-component mutant alleles. The protein complexes in native whole-cell lysates of these strains were separated by gel-filtration chromatography (see Materials and methods). Proteins in the collected fractions were resolved by SDS-PAGE and analyzed by western blotting. Clr3-myc has an expected molecular mass of 92.4 kDa, and structural analysis suggests that Clr3 forms a dimer [14]. Clr3-myc in the native lysate was detected in fractions 11-24 with at least two peaks: Most of the Clr3-myc was eluted in fractions 16-19, corresponding tõ 500 kDa, whereas a minor Clr3-myc population was detected near the void fractions 11-14 (>600 kDa) (Fig 4A, wt). Although the molecular mass and stoichiometry of the native SHREC complex is not clear, the faster-eluted peak probably represents Clr3-myc in the SHREC complex. It should be noted that bulk H3 was detected in fractions near the void ( Fig 4D); thus, it is also possible that the faster-eluted peak corresponds to Clr3-myc associated with chromatin through other SHREC components.
Clr3-myc was distributed similarly in the wild-type strain and strains lacking a SHREC component-with the exception of the clr1Δ strain, in which the signal intensity decreased dramatically in fractions 11-14 (Fig 4A, clr1Δ). This finding is consistent with reports that Clr1 is important for the integrity of the SHREC complex [12,14]. The Clr3-myc levels also decreased slightly in the chp2Δ strain in fractions 11-14 (Fig 4A, chp2Δ). Importantly, a similar reduction was also observed in the chp2-W199A strain (Fig 4A, chp2-W199A). These results support the idea that the faster-eluted peak includes chromatin-associated Clr3-myc, and suggest that Clr3's chromatin targeting depends in part on Chp2's H3K9me binding.
Clr2 possesses an MDB-like domain, and the Clr1-Clr2 complex binds RNA and DNA through this domain [14]. Although this activity of Clr2 is thought to tether SHREC to chromatin, it might also help SHREC interact with cellular RNAs. To investigate whether SHREC contains an RNA component, or whether SHREC's association with chromatin is dependent on RNA, native lysates were treated with RNaseA prior to chromatographic separation (S3 Fig). However, the distribution patterns of Clr3-myc, V5-Clr2 and F-Chp2 upon RNase A treatment did not change compared to the untreated control, indicating that RNA is not critical for the integrity of the SHREC complex (S3 Fig). We used the same chromatographic approach to investigate the elution profile of F-Chp2 (Fig 4C). F-Chp2 was primarily detected in fractions 11-13 and 17-23. The faster-eluted peak overlapped with the H3 peak and the faster-eluted peaks of Clr3-myc and V5-Clr2 (Fig 4D), and thus most likely corresponded to chromatin-associated F-Chp2. Indeed, the Chp2-W199A mutant protein was not detected in fractions 11-14 (Fig 4D, chp2-W199A). Again, the result confirmed that Chp2-CD's H3K9me binding was critical for its own association with chromatin. While the estimated size of the slower-eluted peak was clearly larger than the molecular mass calculated for F-Chp2 (47.5 kDa), a previous study using recombinant proteins showed that Chp2 forms a stable dimer that is detected in fractions corresponding to a molecular mass of 300-350 kDa, presumably due to its elongated shape [10]. Based on this previous observation, we concluded that the slower-eluted peak corresponds to the free F-Chp2 dimer. The F-Chp2 distribution was not affected by the lack of Mit1 or Clr2, suggesting that Chp2 targets chromatin independently of Mit1 and Clr2.
To further confirm the relationship between Chp2's H3K9me binding and Clr3's chromatin targeting, we conducted ChIP assays with primers against centromeric dg repeats (Fig 5A), where Clr3's preferential binding was previously observed [13]. The strain expressing Clr3-myc was enriched at the dg locus, whereas the enrichment was partially decreased by the lack of Chp2 or by the Chp2-W199A mutation, but not by Clr2 depletion (Fig 5B), which is well consistent with the results obtained by gel-filtration chromatography (Fig 4A). Using the same primer sets, Chp2's centromeric dg binding was also examined. As observed for the mat3-M::ura4 + region (Fig 2E), F-Chp2's chromatin association was compromised by Chp2-W199A mutation (Fig 5C). Although F-Chp2's distribution was not noticeably affected in gel-filtration chromatography (Fig 4C), the enrichment was moderately decreased by the lack of Clr1 or Mit1 (Fig 5C, clr1Δ and mit1Δ), suggesting that the SHREC's remodeling module may facilitate Chp2's H3K9me binding. Interestingly, F-Chp2's chromatin association was increased in clr2Δ cells (Fig 5C, clr2Δ). This result may suggest that Chp2 and Clr2 competes with each other for binding to H3K9me-enriched heterochromatic regions.

The Chp2 protein's stability depends on Clr3's HDAC activity
In the course of experiments detecting Chp2 proteins, we found that Chp2 protein levels were partially decreased in a strain lacking Clr3 (S1B Fig). To test whether Chp2's stability depends on Clr3's presence or its enzymatic activity, we examined the effect of trichostatin A (TSA) treatment (Fig 6). Clr3 is a class II histone deacetylase that is inhibited by TSA [7,26]. A yeast strain expressing Clr3-myc, V5-Clr2, and F-Chp2, was treated with increasing concentrations of TSA during the growth period, resulting in a gradual reduction in F-Chp2 (Fig 6; F-Chp2). Interestingly, TSA treatment also gradually decreased and degraded Clr3-myc (Fig 6; Clr3-myc). Histone H3 was slightly reduced in a TSA-treated wild-type strain and a strain lacking Clr3, which is consistent with previous reports [11]. Although the ß-actin level was also somewhat reduced, which could indicate that TSA treatment decreased the protein levels in general, there was no significant change in the V5-Clr2 level (Fig 6). These results suggested that the Chp2 protein's stability depends strongly on Clr3's enzymatic HDAC activity.

Chp2's ability to bind H3K9me is crucial for its silencing function
ITC revealed that Chp2-CD bound the H3K9me3 peptide ( Fig 1C) with an affinity similar to a previously reported range [10]. Chp2 might bind H3K9me-containing nucleosomes in vivo more strongly than might be expected from our ITC experiment, since Swi6's affinity for chromatin increases if it forms an oligomer or is phosphorylated [27,28]. Our genetic approaches demonstrated that Chp2's H3K9me-binding activity is essential for Chp2's function in heterochromatin assembly (Figs 2 and 3). Interestingly, both chp2Δ and chp2-W199A mutants had an unstable silencing phenotype, with the expression of reporter genes in the mating-type region Chp2-CD's H3K9me binding and heterochromatin assembly in fission yeast switching on and off (Fig 2B). However, the off state was incomplete, with some degree of derepression, and the cells produced pink colonies instead of the red colonies seen for the wild-type strain when grown on plates with low adenine levels ( Fig 2B). There are other mutations that produce a similar switch between different epigenetic states of the mating-type Chp2-CD's H3K9me binding and heterochromatin assembly in fission yeast region; this observation agrees with the self-propagating model of heterochromatin formation [16,22,[29][30][31].

Biochemical fractionation supports a two-module model for targeting SHREC to chromatin
Biochemical analysis of three SHREC components allowed us to determine which of these components are important for SHREC's integrity and chromatin targeting. All three SHREC components analyzed in this study were cofractionated in the same fractions as histone H3, which was eluted in the fractions with higher molecular mass (Figs 4 and 5). These findings are consistent with the previously demonstrated association between all of the SHREC components and chromatin [7,13]. In addition, the failure of a SHREC component to cofractionate with histone H3 indicated a loss of association to chromatin. Our data showed that Clr3-myc was lost from high molecular-weight fractions in a clr1Δ strain, indicating that Clr1 is necessary for Clr3's association with chromatin, the stability of the SHREC complex, or both. This finding is consistent with reports that Clr1 is the platform upon which the other SHREC components assemble [12,14].
Clr2 was detected in chromatin-enriched fractions, and this elution pattern was not changed by deleting any other SHREC components, indicating that Clr2 can target chromatin independently of other SHREC components (Fig 4B). Biochemical fractionation also revealed V5-Clr2 peaks in fractions 16-19 and 27-29. The slower-eluted peak appeared to correspond to free V5-Clr2, whereas the faster-eluted peak corresponded to an estimated molecular mass of about 500 kDa, probably representing a subcomplex containing V5-Clr2. The decreased V5-Clr2 signal in clr1Δ or clr3Δ cells would seem to indicate that this subcomplex contains Clr1 and Clr3 in addition to V5-Clr2. However, Clr3-myc was also detected in fractions [16][17][18][19], and this elution profile did not change noticeably in clr1Δ or clr2Δ cells (Fig 4A). Thus, it seems likely that the subcomplex consists of V5-Clr2 and Clr1, whereas Clr3 might indirectly affect the stability of this subcomplex. The reduced association of Clr3-myc with high molecular weight fractions in the strains carrying point mutations in clr2, but not in the strains completely lacking Clr2 (S2A Fig) was surprising. It might indicate a compensatory function of Chp2 in the targeting of Clr3 (SHREC) when Clr2 is lacking. A more severe phenotype of a point mutation as compared to a complete deletion is not unprecedented [32].
In chp2Δ cells or cells carrying the chp2-W199A mutation, the Clr3-myc signal diminished substantially in high molecular-weight fractions, indicating that either Clr3's association with chromatin or SHREC's stability (or both) depend partially on Chp2's H3K9me binding (Fig 4). As with the silencing assays (Figs 2 and 3), a chp2 deletion or chp2-W199A mutation had a very similar effect on Clr3-myc's elution profile (Figs 4 and 5), supporting the conclusion that Chp2's function relies on its ability to bind H3K9me via its CD.
Another interesting observation was that the Chp2 protein level decreased dramatically in clr3Δ cells, indicating that Chp2 was destabilized in the absence of Clr3 (S1B Fig). Further investigation demonstrated that treating yeast cells with TSA, which inhibits class I and class II HDAC activity, led to the degradation of Chp2 and of Clr3 itself, strongly suggesting that Clr3's HDAC activity was involved in Chp2's targeting and stability and in the stability of Clr3 itself (Fig 6).
In conclusion, data presented here along with previously published work suggest a two-step mechanism for SHREC's association with chromatin, in which Clr2 binds chromatin to form a platform for recruiting Clr1 and Clr3, while Chp2 may recruit Mit1 and additionally bind and stabilize the SHREC complex at the site (Figs 4 and 5) [14]. This mechanism means that if Chp2 or Clr2 are missing, an incomplete, partially functional complex will form. This scenario agrees with an idea proposed by Job et al., that the SHREC complex functions as two separable modules, one a remodeling module containing Mit1 and Chp2, and the other an HDAC module comprised of Clr2 and Clr3 and held together by Clr1 [14]. The two are targeted independently to chromatin: the remodeling module is targeted via Chp2 and the HDAC module via Clr2, which has been shown to bind RNA and DNA [14]. However, it is well established that all of the components of the SHREC complex are required for heterochromatin formation; this is evidenced also by the minimal silencing system described in this study, in which the reporter gene was not silenced if any of the SHREC proteins were absent (Fig 3) [12,13]. Moreover, Clr3's HDAC activity greatly affected Chp2's targeting or stability, demonstrating the functional dependency of the two modules within the SHREC complex (Fig 6 and S1B Fig). Supporting information S1 Fig. Western blots of Chp2 proteins. (A) Whole-cell lysates prepared from the SPYB106 (wt), SPYB148 (chp2Δ), SPM2238 (F-chp2), and SPM2291 (F-chp2-W199A) strains were separated by SDS-PAGE and analyzed by western blotting with antibodies against the FLAG tag (top) or against Chp2 (middle). After western blotting, the membrane was stained by Amido black to assess the amount of loaded proteins (bottom). Asterisk: nonspecific protein band; arrowhead: endogenous Chp2 band. (B) Whole-cell lysates prepared from the PJ78 (wt), SPAH101 (clr3Δ #1), and SPAH102 (clr3Δ #2) strains were separated by SDS-PAGE and analyzed by western blotting with antibodies against Chp2 (top). After western blotting, the membrane was stained by Amido black to assess the amount of loaded proteins (bottom). (TIF)