Spt6 prevents transcription-coupled loss of posttranslationally modified histone H3

The tail of histone H3 is an ideal medium for storing epigenetic information because displacement of histone H3 is heavily restricted during transcription. To maintain the locus-specific modifications of histone H3, histone molecules should be retained locally at the original position through multiple rounds of transcription. Here, we found that fission yeast Spt6, a highly conserved RNA polymerase II-interacting histone H3–H4 chaperone, is essential for the maintenance of Lys-4 and Lys-9 methylation of histone H3 in euchromatin and heterochromatin, respectively. In euchromatin, loss of Lys-4 methylated histone H3 and deposition of newly synthesized Lys-56 acetylated histone H3 induced by Spt6 inactivation were coupled with transcription. While in heterochromatin, Spt6 prevents histone turnover and cryptic transcription in parallel with Clr3 histone deacetylase. We propose that Spt6 retains posttranslationally modified histone H3 during transcription to maintain epigenome integrity.

The tail of histone H3 is an ideal medium for storing epigenetic information because displacement of histone H3 is heavily restricted during transcription. To maintain the locus-specific modifications of histone H3, histone molecules should be retained locally at the original position through multiple rounds of transcription. Here, we found that fission yeast Spt6, a highly conserved RNA polymerase II-interacting histone H3-H4 chaperone, is essential for the maintenance of Lys-4 and Lys-9 methylation of histone H3 in euchromatin and heterochromatin, respectively. In euchromatin, loss of Lys-4 methylated histone H3 and deposition of newly synthesized Lys-56 acetylated histone H3 induced by Spt6 inactivation were coupled with transcription. While in heterochromatin, Spt6 prevents histone turnover and cryptic transcription in parallel with Clr3 histone deacetylase. We propose that Spt6 retains posttranslationally modified histone H3 during transcription to maintain epigenome integrity. P osttranslational modifications of histone molecules epigenetically control the state of the corresponding genomic loci in cis 1 . Transcription by RNA polymerase II (Pol II) can cause dissociation of pre-existing histone molecules in vitro 2,3 and in vivo [4][5][6][7][8][9] . Therefore, factors that oppose this dissociation by serving as molecular liaisons between histones and DNA should exist to ensure the integrity of the epigenome. Here, we show that Spt6 (Suppressor of Ty 6), a highly conserved histone H3-H4 chaperone that interacts with elongating Pol II [10][11][12][13][14] , is one such molecular liaison candidate. Spt6 has been shown to play important roles in nucleosome restoration in highly transcribed regions 11 , in genetic alterations at immunoglobulin loci 15,16 , and in methylation of Lys-4 and Lys-9 in histone H3 16,17 . However, the primary function of Spt6 in epigenetic regulation remains unclear.
We found that Spt6 prevents transcription-coupled nucleosome loss and loss of associated methylation of histone H3 at Lys-4 and Lys-9 in euchromatin and heterochromatin, respectively. The nucleosome loss caused by inactivation of Spt6 appears to be partially compensated for by Spt6-independent deposition of newly synthesized Lys-56 acetylated histone H3 molecules that do not carry locus-specific modifications. In other words, Spt6 prevents transcription-coupled histone turnover to maintain the posttranslational modifications of histone H3. We observed genetic interaction between Spt6 and Clr3 (cryptic loci regulator 3), a histone deacetylase that prevents histone turnover in heterochromatin 18 , indicating that Spt6 acts in parallel with Clr3 HDAC to prevent histone turnover and thereby maintain heterochromatin. These results strongly suggest that the primary function of Spt6 is to maintain epigenomic integrity by retaining posttranslationally modified histone H3 during transcription. grew slower than wild-type cells ( Supplementary Fig. 2a), whereas the complete deletion of spt6 (spt6D) caused a much more severe growth defect. Heterochromatin in fission yeast is marked by both Lys-9 methylation of histone H3 and the heterochromatin protein 1 (HP1) homolog Swi6 HP1 (switch 6), which recognizes Lys-9 methylated histone H3 20 . Insertion of ade6 in the pericentromere as a marker gene leads to passive construction of heterochromatin at the marker gene by spreading of heterochromatin marks from the native regions 21 . Chromatin immunoprecipitation (ChIP) followed by quantitative polymerase chain reaction (ChIP-qPCR) analysis for dimethylation of Lys-9 in histone H3 (K9me2) and Swi6 HP1 demonstrated that placement of these heterochromatin-specific marks in the pericentromere (ade6 and dg) and subtelomere (tlh) depends solely on the Suv39h family Lys-9-specific methyltransferase Clr4 Suv39h (cryptic loci regulator 4) 20 (Fig. 1b, c).
The importance of fission yeast Spt6 in heterochromatin organization was previously studied with the spt6-1 mutant allele 17 . Cells carrying the spt6-1 mutant allele have been shown not to affect K9me2 in the pericentromeric repeats. However, we found that both the spt6-K20 and spt6D mutations caused a significant decrease in the levels of K9me2 and Swi6 HP1 (Fig. 1b, c). In spt6D cells, some K9me2 remained in the native regions (dg and tlh), whereas the modification was completely abrogated in the inserted marker gene (ade6) ( Supplementary Fig. 3a, showing K9me2 level per histone H3 molecule). Therefore, Spt6 is essential for the maintenance of K9me2, especially in the inserted marker gene. Lys-9 methylation of histone H3 is coupled with generation of small interfering RNA (siRNA) 22,23 . Consistent with the substantial decrease in the K9me2 level, siRNA corresponding to the pericentromeric repeats was undetectable in spt6 mutant cells ( Supplementary Fig. 4).
To examine the effect of spt6D on Pol II and histone H3 occupancy across the pericentromeric regions, we performed a ChIP analysis followed by high-throughput sequencing (ChIP-seq) of Pol II and histone H3. The Pol II occupancy in some pericentromeric cryptic regions was dramatically increased in spt6D cells, as was also shown in clr4D Suv39h cells 24 (Fig. 1d). Transcriptional activation in clr4D Suv39h cells did not lead to an overall decrease in histone H3 occupancy, except in several sensitive regions, as previously reported (Fig. 1d, red asterisks) 25 . It has been shown that the spt6-1 mutation does not cause a decrease in nucleosome occupancy in the pericentromere 17 . In spt6D cells, however, a significant decrease in histone H3 occupancy over the cryptic transcribed regions was observed ( Fig. 1d and Supplementary Fig. 3b). ChIP-qPCR and micrococcal nuclease treatment followed by qPCR (MNase-qPCR) analyses confirmed that nucleosome loss did occur in the cryptic transcribed regions ( Supplementary Fig. 5a-c). Even in a region where Pol II occupancy was very low ( Supplementary Fig. 3b, region dh-L), a 2-fold increase in Pol II occupancy was associated with a significant decrease in the levels of histone H3 occupancy ( Supplementary Fig. 3c) and K9me2 ( Supplementary Fig. 3a). Thus, in the absence of Spt6, nucleosome loss and a reduction in the level of K9me2 occur concomitantly in the cryptic transcribed regions in the pericentromere. Note that nucleosome loss did not appear to occur in some subregions within the cryptic transcribed regions in spt6D cells (Fig. 1d, blue asterisks). These subregions might be resistant to the cotranscriptional nucleosome loss caused by Spt6 inactivation.
Transcription-coupled nucleosome loss and compensatory deposition of Lys-56 acetylated histone H3 occur in spt6 mutant cells. Next, we studied the function of Spt6 in euchromatin. Partial inactivation of Spt6 causes nucleosome loss in actively transcribed genes 11,17 ; therefore, we performed ChIP-seq and MNase-qPCR analyses to examine the effect of complete inactivation of Spt6 on nucleosome occupancy. As expected, significant nucleosome loss over the actively transcribed act1 gene was observed in spt6D cells ( Fig. 2a and Supplementary Fig. 6). Genome-wide analysis of protein occupancy per gene revealed that the extent of the decrease in histone H3 occupancy in spt6D cells was greater in genes with higher Pol II occupancy (Fig. 2b), indicating that transcription induces nucleosome loss. In the naked regions from which pre-existing histone molecules have been dissociated, it is possible that nucleosomes could be reconstructed with other histone molecules. Newly synthesized histone H3, which does not carry ''locus-specific'' posttranslational modifications, is acetylated at Lys-56 (K56Ac) before deposition 9,26,27 . Therefore, this modification serves as a marker for newly deposited histone H3 28,29 . As shown in Figure 2c, Western blotting analysis of bulk histones in the chromatin fraction revealed that the amounts of histones H3 and H4 do not change dramatically, and that the level of Lys-56 acetylated histone H3 (K56Ac) was apparently increased in spt6 mutant cells (Fig. 2c).  These results strongly suggest that the cotranscriptional decrease in histone H3 occupancy is partially compensated for by Spt6independent deposition of Lys-56 acetylated histone H3.
To examine histone H3 occupancy with respect to transcription start sites (TSS) and termination sites (TTS), we analyzed four sets of genes transcribed at different levels (''Very high'', ''High'', ''Medium'', and ''Low'') ( Supplementary Fig. 7a, b, see Supplementary Methods for details). Median histone occupancy was slightly lower in the ''Very high'' and ''Low'' genes than in the ''High'' and ''Medium'' genes in wild-type cells ( Fig. 2d and Supplementary Fig. 7c). In spt6D cells, the decrease in histone H3 occupancy was indeed greater in genes with higher transcription levels ( Fig. 2d and Supplementary  Fig. 8). In contrast to previous studies in which Spt6 was not completely depleted 11 , in our study a significant decrease in histone H3 occupancy was observed even in the ''Low'' genes ( Fig. 2d and Supplementary Fig. 8). The weaker effect in the ''Low'' genes may reflect Spt6-independent deposition of K56Ac-marked histones that acts antagonistically to the cotranscriptional nucleosome loss. We noted that Pol II occupancy in genes with low expression levels was increased in spt6D cells (Fig. 2e). Consistently, the levels of both sense and antisense transcripts with low expression levels in wild-type cells were increased in spt6D cells ( Supplementary Fig. 9). Therefore, despite the observation that Spt6-independent deposition can partially compensate for the decrease in nucleosome occupancy, such quantitative compensation may be insufficient for the maintenance of the epigenetically repressed state. Histone H3 occupancy was decreased predominantly in the region corresponding to the first few nucleosomes (Fig. 2d and Supplementary Fig. 8, highlighted in light blue), suggesting that either nucleosome loss predominantly occurs in this region or that compensatory deposition rarely occurs here.
Spt6 prevents transcription-coupled loss of Lys-4 methylated histone H3. Newly deposited histone H3 molecules may not carry locus-specific posttranslational modifications. Therefore, the next question we sought to answer was whether or not inactivation of Spt6 facilitates cotranscriptional elimination of histone methylation in coding genes. We focused on four housekeeping genes in which Pol II occupancy was not dramatically altered by Spt6 inactivation (Fig. 3a). ChIP-qPCR analysis demonstrated that the extent of the decrease in histone H3 occupancy in spt6 mutant cells correlated with Pol II occupancy (Fig. 3a, b). In fission yeast, the Lys-4 residue of histone H3 is methylated solely by the Set1family methyltransferase Set1 30,31 . Western blotting analysis of bulk histones demonstrated the necessity of Set1 for di-and trimethylation at Lys-4 of histone H3 (K4me2 and K4me3, respectively) (Fig. 3c). In spt6 mutant cells, K4me2 and K4me3 could not be detected (Fig. 3c). ChIP-qPCR analyses consistently revealed that the level of K4me2 decreased significantly in spt6-K20 cells and was undetectable in spt6D cells ( Fig. 3d; Note: the y-axis is magnified). The significant decrease in the level of K4me2 suggests that either Spt6 mediates Set1-dependent placement of Lys-4 methylation in histone H3, or that Spt6 retains Lys-4 methylated histone H3 during transcription. If the former is true, partial inactivation of Spt6 should attenuate Set1 activity, leading to a uniform reduction in the level of K4me2 in each gene. Importantly, K4me2 levels in spt6-K20 cells were lower in genes with higher Pol II occupancy (Fig. 3a, d). In spt6D cells, the levels of K56Ac and acetylation of histone H3 at Lys-9 and Lys-14 (AcH3) were higher in genes with higher Pol II occupancy (Fig. 3a, e, f). The level of histone H4 acetylation (AcH4) was also altered in spt6D cells (Fig. 3g). In contrast, these histone acetylation marks were not altered in the heterochromatin-defective clr4D Suv39h cells (Fig. 3eg). These data suggest that inactivation of Spt6 causes cotranscriptional histone turnover, which in turn leads to impaired histone acetylation. Both the negative correlation between the levels of K4me2 and Pol II occupancy and the cotranscriptional histone turnover in spt6 mutant cells strongly suggest that transcription Spt6 acts in parallel with Clr3 HDAC to maintain heterochromatin in the pericentromere. Recent studies have shown that heterochromatic silencing in fission yeast is maintained in part by a Clr3 HDACmediated prevention of histone turnover 18,25,32,33 . In spt6 mutant cells, the majority of residual histone H3 in the pericentromere was not methylated at Lys-9 ( Supplementary Fig. 3a), suggesting that the loss of Lys-9 methylation occurs as a result of cotranscriptional histone turnover. Therefore, we examined the relationship between Spt6 and Clr3 HDAC genetically. ChIP-qPCR analysis showed that the levels of K56Ac in the pericentromere in both clr3D HDAC and clr4D Suv39h single mutant cells were higher than in wild-type cells (Fig. 4a), demonstrating that histone turnover is prevented by these factors 18 . The K56Ac level in the pericentromere was also increased in spt6-K20 single mutant cells (Fig. 4a).
In order to determine whether Spt6 acts in the Clr3 HDAC -dependent prevention of histone turnover, we combined the spt6-K20 mutation with clr3D HDAC . In the spt6-K20 clr3D HDAC double mutant cells, the K56Ac level in the pericentromere was higher than that in each of the single mutant cells (Fig. 4a), suggesting that Spt6 and Clr3 HDAC have redundant mechanisms to prevent histone turnover. The increase in the K56Ac level was associated with an increase in the acetylation level of the N-terminal tail of both histone H3 (AcH3) and histone H4 (AcH4) (Fig. 4b, c). In addition, the levels of K9me2 and Swi6 HP1 were further decreased in the double mutant cells to levels comparable to those in clr4D Suv39h cells (Fig. 4d, e). Consistently, Pol II occupancy, which was higher in each of the single mutant cells, was increased further in the double mutant cells (Fig. 4f). The additive effect observed here suggests that Spt6 acts in parallel with Clr3 HDAC to prevent histone turnover and the   In wild-type cells, Pol II transcribes the nucleosome templates, keeping histones H3 and H4 in its wake. The histone H3-H4 chaperone Spt6 interacts directly with the phosphorylated C-terminal domain of elongating Pol II. Spt6 retains posttranslationally modified histone H3 during transcription so that the state of the chromatin is maintained. In the absence of Spt6 activity (spt6D), modified histones are displaced during transcription. Spt6independent deposition of K56Ac-marked histones partially compensates for the loss of nucleosome occupancy. However, as the newly deposited histones do not carry appropriate locus-specific modifications, epigenetic control of the transcribed region is impaired.

Discussion
Our data indicate that when Pol II lacking a functional Spt6 transcribes through the nucleosome template, nucleosome loss occurs in the wake of Pol II and eventually leads to Spt6-independent deposition of histones without locus-specific modifications. Therefore, we propose that during transcription, the histone H3-H4 chaperone Spt6 retains histone H3 to keep the rate of transcription-coupled histone turnover to a minimum in order to maintain locus-specific posttranslational modifications (Fig. 5). Alternatively, the transcription-coupled loss of histone modifications could result from an error in the recruitment of histone modification enzymes to the transcribed regions that is potentially mediated by Spt6. However, transcription-coupled nucleosome loss and increased acetylation of histone H3 at Lys-56 in the spt6 mutant cells support the idea that elimination of locus-specific histone modifications occurs as a result of histone turnover. The precise mechanism of Spt6-dependent maintenance of histone modifications should be studied at the molecular level in the future.
The essentiality of Spt6 for the maintenance of histone positioning and associated modifications suggests that intracellular mechanisms exist that inactivate Spt6 in cis to reset the chromatin state or create accessible chromatin 34 . As almost all of the genome is transcribed to some extent 35 , inactivation of Spt6 could cause a catastrophic breakdown in epigenomic integrity. Furthermore, Spt6 breakdown in human cells could be involved in the epigenetic alterations frequently observed in cancer cells 36 .

Methods
Genetic manipulations. A fission yeast wild-type strain was mutagenized with 254nm ultraviolet light to obtain mutants defective in pericentromeric silencing. The corresponding mutation was identified by whole-genome sequencing.
Chromatin immunoprecipitation. Exponentially growing cells were subjected to chromatin immunoprecipitation (ChIP). For ChIP-qPCR analysis of histone modifications, the immunoprecipitation efficiency of K4me2, K9me2, K56Ac, AcH3 and AcH4 at each locus was first normalized to that of the C-terminal region of histone H3 or histone H4 (for AcH4) in order to evaluate the level of modification per histone molecule. The modification levels were then compared to those in the wildtype control (y-axis: relative to wild-type) or the internal control regions indicated. For ChIP-qPCR analysis of K9me2 and Swi6 in Figures 1 and 4, the immunopreciptation efficiency at each locus was compared to those in the act1 region. For ChIP-seq analysis, approximately 10 million single-end 49-base reads were mapped onto the 972 reference genome. Pileup data generated by MACS 37 were analyzed using the R-statistical environment (http://www.R-project.org). The Rscripts used in this study are available upon request.
Full Methods and any associated references are available as Supplementary Information.