Chl1 coordinates with H3K9 methyltransferase Clr4 to reduce the accumulation of RNA-DNA hybrids and maintain genome stability

Summary A genome-wide analysis in Schizosaccharomyces pombe indicated that double-deletion mutants of Chl1 and histone H3K9 methyltransferase complex factors are synthetically sick. Here, we show that loss of Chl1 increases the accumulation of RNA-DNA hybrids at pericentromeric dg and dh repeats in the absence of the H3K9 methyltransferase Clr4, which leads to genome instability, including more severe defects in chromosome segregation and increased chromatin accessibility. Localization of Chl1 at pericentromeric regions depends on a subunit of replication protein A (RPA), Ssb1. In wild-type (WT) cells, transcriptionally repressed heterochromatin prevents the formation of RNA-DNA hybrids. When Clr4 is deleted, dg and dh repeats are highly transcribed. Then Ssb1 associates with the displaced single-stranded DNA (ssDNA) and recruits Chl1 to resolve the RNA-DNA hybrids. Together, our data suggest that Chl1 coordinates with Clr4 to eliminate RNA-DNA hybrids, which contributes to the maintenance of genome integrity.


INTRODUCTION
Fission yeast pericentromeric heterochromatin contains highly repetitive dg and dh repeats that are characterized by condensed nucleosome arrays, transcriptionally inert regions, and histone H3K9 methylation (Grewal and Jia, 2007). This pericentromeric heterochromatin mediates sister chromatid cohesion by recruiting cohesin, a multiprotein complex, through Swi6 that recognizes methylated H3K9 specifically (Bernard et al., 2001). The association of sister chromatids ensures their segregation at anaphase. Accordingly, disruption of heterochromatin by deleting its regulators, such as histone H3K9 methyltransferase Clr4 and RNAi components, leads to chromosome segregation defects (Ekwall et al., 1996;Ellermeier et al., 2010). Moreover, depletion of H3K9 methylation by mutating Clr4 results in the release of transcriptional silence at heterochromatic regions (Ekwall et al., 1996).
In wild-type cells of fission yeast, noncoding RNAs transcribed from dg and dh repeats at heterochromatic regions can generate RNA-DNA hybrids (Nakama et al., 2012), but these regions are specifically transcribed at S phase and then repressed by the reestablishment of heterochromatin (Chen et al., 2008). A genome-wide mapping of RNA-DNA hybrids indicates that they are distributed at rDNAs, transposons, a subset of open reading frames (ORFs), and tRNAs (Chan et al., 2014;El Hage et al., 2014). Although RNA-DNA hybrid plays a positive role in DNA repair (Keskin et al., 2014;Ohle et al., 2016), it is also a threat to genome stability (Hamperl and Cimprich, 2014). One of the possible reasons is that the cotranscriptional RNA-DNA hybrid can form a three-stranded R-loop structure with the displaced single-stranded DNA (ssDNA). R-loops have physiological functions in transcriptional regulation, recombination, and DNA repair (Garcia-Muse and Aguilera, 2019; Niehrs and Luke, 2020;Santos-Pereira and Aguilera, 2015;Sun et al., 2013). Nonetheless, there is also considerable evidence showing that misregulation of R-loop causes hyperrecombination, transcription-replication conflicts, and DNA damage (Costantino and Koshland, 2018;Crossley et al., 2019;Hamperl et al., 2017;Skourti-Stathaki and Proudfoot, 2014), which may lead to neurological diseases, cancer, or other clinical disorders in humans. A range of factors have been found to process or resolve RNA-DNA hybrids, which protect against genome instability caused by the accumulation of RNA-DNA hybrids. On the basis of their biochemical activities, RNase H enzymes and helicases are two types of factors known to remove RNA-DNA hybrids. RNase H enzymes hydrolyze the RNA in RNA-DNA hybrid (Cerritelli and Crouch, 2009;Nowotny et al., 2005;Rychlik et al., 2010), whereas RNA/DNA helicases can unwind this structure (Kim et al., 1999;Song et al., 2017). Studies have shown that the rnh1D rnh201D In this study, we found that the fission yeast Chl1 prevents the accumulation of RNA-DNA hybrids generated from heterochromatic regions in the absence of Clr4. Disruption of heterochromatin in Clr4 mutant releases transcriptional silencing, which may increase the formation of RNA-DNA hybrids. Afterward, Ssb1, a subunit of replication protein A (RPA), binds to the displaced ssDNA, and then recruits Chl1 to these regions to resolve RNA-DNA hybrids, which protects genome integrity.

RESULTS
chl1D is synthetically sick with clr4D To confirm the negative genetic interaction between Chl1 and Clr4, we constructed single-or double-deletion mutants of Chl1 and Clr4 and compared the growth rate of wild type (WT), chl1D, clr4D, and chl1D clr4D cells. The results indicated that the growth curves of WT and chl1D were comparable, whereas the clr4D mutant grew slightly slower than WT. Moreover, as expected, a significant growth delay of chl1D clr4D cells was observed relative to WT and the single mutants ( Figure 1A). To determine the reason for the slow growth of double mutant, we stained the yeast cells with methylene blue and counted the number of blue cells. Methylene blue can be applied to identify the viability of cells, with live cells reducing the dye, so that only dead cells stain blue. Consistent with the growth curves, the percentage of dead cells of clr4D was slightly higher than those of WT and chl1D, whereas blue-stained chl1D clr4D cells were much more abundant than WT cells and the two single mutants ( Figure 1B). These data suggest that the growth defect of chl1D clr4D cells might be due to the high percentage of dead cells that cannot proliferate any further.
Heterochromatin is required for cohesion of sister chromatids. Disruption of heterochromatin by deleting genes that affect H3K9 methylation causes improper chromosome segregation during mitosis. Daughter cells without intact chromosomes lose viability. To elucidate whether the low viability of double mutants was due to chromosome missegregation, we grew WT, chl1D, clr4D, and chl1D clr4D cells harboring the nonessential minichromosome Ch16 on low-adenine medium and calculated the chromosome loss rates of the four strains (Niwa et al., 1986). The minichromosome Ch16 in WT and chl1D cells was faithfully segregated, whereas the loss rates increased to 2.2% and 4.6% per division in the clr4D and chl1D clr4D backgrounds, respectively ( Figure 1C). In addition to a high rate of chromosome loss, an elevated incidence of lagging chromosomes is also indicative of defects in chromosome segregation. Spindle microtubules and DNA were stained with an a-tubulin antibody and DAPI, respectively, and visualized by fluorescence microscopy ( Figure 1D). We counted 100 cells of each strain during anaphase and found that 1% and 2% of cells displayed lagging chromosomes in WT and the chl1D mutant, respectively, whereas 8% of clr4D cells and 22% of chl1D clr4D cells had lagging chromosomes ( Figure 1D). These findings suggest that deletion of Chl1 aggravated chromosome segregation defects in the clr4D mutant, which may have led to the low viability of the double-deletion mutant. Based on the enzymatic activity of ChlR, a human homolog of Chl1, disruption of heterochromatin might bring about a duplex structure that can be resolved by Chl1. Considering that heterochromatin is required for cohesin attachment, we propose that this chromatin Chl1 is involved in the repression of RNA-DNA duplexes generated from highly transcribed pericentromeric repeats Previous study showed that human ChlR has helicase activity on RNA-DNA duplexes (Hirota and Lahti, 2000), and we therefore tested whether fission yeast Chl1 can also unwind RNA-DNA hybrids in vitro.
When an RNA-DNA heteroduplex containing a biotin-labeled RNA oligonucleotide was used as substrate, Chl1 displaced the RNA from the hybrid, indicating that Chl1 of fission yeast has helicase activity on RNA-DNA hybrid ( Figure 2A). Depletion of Clr4 releases transcriptional silencing of pericentromeric dg and dh repeats (Volpe et al., 2002), implying that clr4D increases the chance of generating cotranscriptional RNA-DNA hybrids at these regions compared with WT cells. To detect the level of RNA-DNA hybrids at pericentromeric regions, we performed DNA-RNA immunoprecipitation (DRIP) using an S9.6 antibody specific for DNA-RNA hybrids. To our surprise, we did not observe the accumulation of RNA-DNA hybrids in WT and the two single mutants, whereas the levels of RNA-DNA hybrid in chl1D clr4D cells at dg and dh repeats were elevated. At the same time, there were no differences in Psm1 among these strains; Psm1 gene is transcribed at a low level in euchromatic region ( Figure 2B). The transcription of dg and dh repeats in chl1D clr4D slightly decreased as compared with that in clr4D ( Figure S1), suggesting that the enrichment of cotranscriptional RNA-DNA hybrids was not because of the increased transcription in chl1D clr4D, but due to the function of Chl1 in unwinding DNA-RNA hybrids generated from pericentromeric repeats in the absence of Clr4. RNA/DNA helicase and RNase H process RNA-DNA hybrids in different manners.
To confirm the accumulation of RNA-DNA hybrids, we overexpressed Rnh1 or Rnh201, two RNase H iScience Article enzymes of Schizosaccharomyces pombe, under the control of an inducible nmt1 promoter in chl1D clr4D cells, and examined the levels of RNA-DNA hybrid at dg and dh repeats. The DRIP experiment showed that overexpression of Rnh1 or Rnh201 can reduce RNA-DNA hybrids at these regions ( Figure 2B). These results suggest that Clr4 and Chl1 can suppress RNA-DNA hybrid formation by repressing transcription and resolving these duplexes, respectively.
The fact that Chl1 can eliminate the formation of RNA-DNA hybrid at pericentromeric regions in the absence of Clr4 prompted us to detect the localization of Chl1 at dg and dh repeats. We generated strains expressing Chl1-Flag in the WT and clr4D backgrounds and performed chromatin immunoprecipitation (ChIP) analysis with an antibody against the Flag tag. The ChIP signals of Chl1 were elevated in the absence of Clr4 within dg and dh repeats, suggesting that Chl1 can be recruited to these regions when transcription was released ( Figure 2C).
Next, we wanted to determine whether Chl1 can repress the accumulation of RNA-DNA hybrid in WT background. Pericentromeric repeats are specifically transcribed during S phase in fission yeast (Chen et al., 2008). If Chl1 resolves RNA-DNA hybrids generated from highly transcribed dg and dh repeats, an increase in the recruitment of Chl1 at these regions should be observed at S phase. To test this hypothesis, we treated WT cells and cells harboring the Flag-tagged Chl1 with hydroxyurea (HU), which can block the cell cycle in S phase by inhibiting ribonucleotide reductase, and then performed ChIP analysis with an anti-Flag antibody. As we expected, Chl1 signal was enhanced at dg and dh repeats ( Figure 2D). Consistent with this, deletion of Chl1 resulted in the accumulation of RNA-DNA hybrids at pericentromeric regions in iScience Article cells at S phase, whereas overexpression of Rnh1 or Rnh201 reduced their levels ( Figure 2E). Taken together, these data indicate that Chl1 plays a role in the suppression of RNA-DNA hybrids produced at pericentromeric regions. Nonetheless, as S phase is a short period during the entire cell cycle and the subsequent reestablishment of heterochromatin suppresses transcription, RNA-DNA hybrids cannot be detected at dg and dh repeats in asynchronous chl1D cells.

Chl1 is required for genome stability in clr4D cells
To determine whether accumulation of RNA-DNA hybrid is responsible for the slow growth of chl1D clr4D cells, we compared the growth rates of strains transformed with an empty vector or a vector harboring Rnh1 or Rnh201. The chl1D clr4D mutant overexpressing Rnh1 or Rnh201 had a growth rate comparable to that of clr4D ( Figure 3A), which were significantly faster than chl1D clr4D cells. These data suggest that the difference in growth rates between clr4D and chl1D clr4D resulted from RNA-DNA hybrid accumulation. In contrast, the minor slow-growth phenotype of clr4D was not relevant to RNA-DNA hybrids and might be due to failure in association with the cohesin complex. More importantly, overexpression of Rnh1 or iScience Article Rnh201 dramatically reduced the ratio of lagging chromosomes in chl1D clr4D cells ( Figure 3B), suggesting that chromosome segregation defects are caused by RNA-DNA hybrid accumulation. Conversely, a serial dilution plating assay showed that the rnh1D rnh201D chl1D clr4D quadruple-deletion strain displayed a severe growth defect relative to the other strains ( Figure 3C), implying that they had redundancy in the repression of RNA-DNA hybrids. We inferred that endogenous RNase H enzymes are not effective enough for processing all RNA-DNA hybrids produced from the pericentromeric regions in clr4D. The rest of unprocessed RNA-DNA hybrids can be unwound by Chl1. These results suggest that RNase H and helicase coordinately process RNA-DNA hybrids at specific chromatin regions.
In addition to cohesin attachment, chromatin compaction is also required for proper chromosome segregation. Studies have shown that R-loops associate with DNase I hyperaccessibility, chromatin decompaction, and lower nucleosome occupancy (Powell et al., 2013;Sanz et al., 2016). Because the cotranscriptional RNA-DNA hybrid and the displaced ssDNA can form an R-loop, we investigated whether RNA-DNA accumulation had effects on the chromatin structure at pericentromeric regions. We digested the chromatins from WT, chl1D, clr4D, and chl1D clr4D cells with different concentrations of micrococcal nuclease and performed Southern blotting with a probe that hybridized to dh repeats. The chromatin of double-deletion strain at the dh repeats was more easily digested by the enzyme than that of both WT and two single-deletion strains ( Figure 3D). This observation implies that the increased RNA-DNA hybrids interrupt the chromatin compaction, which leads to the defects in chromosome segregation. At same time, we found that chl1D clr4D cells failed to maintain the integrity of nuclei after recovery from genotoxic drug methanesulfonate (MMS) treatment. Four strains were grown in liquid medium containing 0.008% MMS for 6 h, washed, and further inoculated into medium without MMS for 16 h. During this process, cells from each step were fixed and stained with DAPI. After 16 h recovery from MMS treatment, 35% of chl1D clr4D cells displayed diffuse or fragmented nuclei, suggesting that deletion of both Chl1 and Clr4 resulted in failure of recovery from MMS-induced DNA lesions ( Figures 3E and 3F).
Chl1 of budding yeast associates with Ctf4 and Ctf7/Eco1 to play a role in sister-chromatid cohesion (Samora et al., 2016;Skibbens, 2004), but here, we did not detect direct interaction of Chl1 with Mcl1 or Eso1, the fission yeast homologs of Ctf4 and Ctf7/Eco1, by a yeast two-hybrid assay (data not shown). Meanwhile, the Chl1 single-deletion mutant did not exhibit severe chromosome segregation defects ( Figures 1C and 1D). These data suggest that fission yeast Chl1 does not function in sister chromatid cohesion but is implicated in maintaining a more condensed chromatin at pericentromeric regions by removing RNA-DNA hybrids in the absence of Clr4, which contributes to the genome integrity.

Ssb1 is responsible for the recruitment of Chl1 to chromatin
To elucidate the regulation of Chl1, we performed affinity purification of Chl1-Flag and identified two subunits of RPA by mass spectrometry analysis, namely, Ssb1 and Ssb2 ( Figure S2). RPA is a heterotrimeric complex that binds to ssDNA, has versatile roles in DNA repair, replication, and recombination, and exhibits histone assembly ability. The fission yeast RPA consists of three subunits: Ssb1/Rpa1/Rad11, Ssb2/Rpa2, and Ssb3/Rpa3 (Ishiai et al., 1996). To confirm the interactions, we coexpressed Ssb1-Myc or Ssb2-Myc with Chl1-Flag at their native genomic loci and demonstrated that Chl1 coimmunoprecipitated with both Ssb1 and Ssb2 ( Figure 4A).
A proteome-wide screen showed that foci formation of Chl1 at induced DNA double-stranded break (DSB) occurred later than that of Ssb1 (Yu et al., 2013); therefore, we proposed that Chl1 was recruited by RPA subunit Ssb1 to perform its function. We introduced Ssb1-Flag into a strain containing the HO nuclease recognition site and HO gene integrated into the genome under the control of an inducible nmt1 promoter and then examined the localization of Ssb1 at the HO-induced DSB site in the presence and absence of Chl1. ChIP analysis showed that deletion of Chl1 did not affect Ssb1 accumulation at 0.2 kb, 2 kb, and 9 kb from the break site, suggesting that Ssb1 localization is independent of Chl1 ( Figure S3). Deletion of Ssb1 is lethal for fission yeast, but two temperature-sensitive mutants, ssb1 D223Y and ssb1-418 (Akai et al., 2011;Ono et al., 2003), reduced Ssb1 accumulation at these sites ( Figure 4B). Therefore, we checked whether these mutations affected the targeting of Chl1 at DNA break sites. ChIP assays demonstrated that Chl1 levels decreased at these three sites in the two Ssb1 mutants ( Figure 4C). This result indicates that Chl1 localization depends on Ssb1. Consistent with the enrichment of Chl1 at dg and dh repeats in clr4D cells ( Figure 2C), the accumulation of Ssb1 was also observed at these regions in the absence of Clr4, but not in chl1D cells, whereas overexpression of Rnh1 or Rnh201 repressed the localization of Ssb1 ( Figure 4D). iScience Article Because it has been reported that RPA complex binds to the ssDNA in the R-loops (Nguyen et al., 2017), our data suggest that the ssDNA in the three-stranded structure produced from transcription at dg and dh repeats is responsible for the recruitment of Ssb1. Collectively, we propose that Chl1 can be recruited to pericentromeric regions by Ssb1 to process RNA-DNA hybrids in clr4D cells, which protects genome integrity.

DISCUSSION
In comparison to RNase H enzymes that degrade RNAs in RNA-DNA hybrids, the physiological functions of helicases are less clear. In this study, we found that double deletion of the helicase Chl1 and histone H3K9 methyltransferase Clr4 reduced cell viability ( Figure 1B). Chl1 is involved in alleviating the accumulation of RNA-DNA hybrids and maintaining chromatin compaction at pericentromeric regions in clr4D cells ( Figures 2B and 3D), which are required for genome stability.
Centromeric R-loops drive the mitosis-specific ATR pathway to promote proper chromosome segregation (Kabeche et al., 2018).  iScience Article immunostaining with an S9.6 antibody showed that the percentage of RNA-DNA hybrids at heterochromatic regions is very low compared with that at euchromatic regions in S. pombe (Nakama et al., 2012). Consistent with this result, there was no significant enrichment of RNA-DNA hybrids at dg and dh repeats in asynchronized WT cells ( Figure 2B), which could be due to the fact that heterochromatin represses transcription in fission yeast. The roles of heterochromatin in cohesion and chromosome segregation have been extensively studied (Bernard et al., 2001;Gartenberg, 2009;McKinley and Cheeseman, 2016). Intact heterochromatin ensures proper sister chromatid cohesion and represses transcription at the heterochromatic region. Depletion of H3K9 methylation abolishes the attachment of cohesins, whereas Chl1 can partially suppress chromosome segregation defects by removing RNA-DNA hybrids that affect the chromatin state at dg and dh repeats and therefore increase cell viability. Moreover, the evidence that overexpression of two RNase H enzymes complements the growth defect of chl1D clr4D cells also confirms that the slow-growth phenotype results from RNA-DNA hybrid accumulation ( Figure 3A). Decreased ratios of lagging chromosomes in overexpressed strains indicate that chromosome segregation defect is one of the reasons for the low viability of chl1D clr4D cells ( Figure 3B). Thus, the less condensed pericentromeric heterochromatin structure brought about by RNA-DNA hybrids prevent faithful segregation of sister chromatids during mitosis. Considering that chromatin needs to be restored to its original state after DNA repair or replication (Czaja et al., 2012), more accessible chromatin at dg and dh repeats caused by double deletion of Chl1 and Clr4 may prevent DNA damage recovery, leading to genome instability and cell death ( Figures 3E and 3F). In contrast to Chl1 of budding yeast, single deletion of fission yeast Chl1 does not confer obvious defect in chromosome segregation ( Figures 1C and 1D), suggesting that its function is distinct from that in budding yeast.
Although our study focused on the function of Chl1 at dg and dh repeats of pericentromeric regions, evidence also suggested that Chl1 may function in DNA repair due to its recruitment to the induced DSBs by Ssb1( Figure 4C). However, because chl1D cells are insensitive to DNA damage reagents, we believe that Chl1 does not play a major role in DNA repair. A previous working model based on the colocalization of RPA with R-loops suggests that RPA binds the displaced ssDNA of R-loop and interacts with RNase H1, which can recognize and degrade RNAs in RNA-DNA hybrids (Nguyen et al., 2017). Furthermore, centromeric R-loop is shown to be covered with RPA (Kabeche et al., 2018). Our study demonstrates the recruitment of Ssb1 at highly transcribed pericentromeric regions partially depends on the accumulation of RNA-DNA hybrids ( Figure 4D). Hence, we propose here that RPA subunit Ssb1 associates with the displaced ssDNA of RNA-DNA hybrid and facilitates recognition of RNA-DNA hybrids by Chl1, which is able to remove this three-stranded structure by resolving RNA-DNA hybrids and alleviate chromosome segregation defects in the clr4D mutant ( Figure 4E). Certainly, repression of transcription by Clr4 is an alternative way to prevent the generation of RNA-DNA hybrids. At this time, the cohesin complex does not attach to heterochromatin through Swi6 without H3K9 methylation, which results in chromosome missegregation and cannot be rescued by Chl1. These results suggest that Chl1 and Clr4 work together to sustain genome integrity by reducing the formation of RNA-DNA hybrids at pericentromeric regions.

Limitations of the study
One of the limitations of this study is that the current evidence is insufficient to determine whether accumulation of RNA-DNA hybrids at heterochromatin is the only reason for genome instability. Chl1 localized at DNA double-stranded break after Ssb1, so its role in RNA-DNA hybrids processing might also have a small contribution to genome integrity outside of heterochromatin. Moreover, Chl1 also has helicase activity on DNA-DNA hybrids. Our study did not address the question of whether the DNA helicase activity of Chl1 is involved in the maintenance of chromatin compaction and cell survival. In addition to Clr4, deletions of several components in CLRC complex and factors implicated in RNAi pathway lead to the loss of histone H3K9 methylation and increased transcription at heterochromatin as well. More experiments are required to reveal whether Chl1 can repress the RNA-DNA hybrids that generate from highly transcribed heterochromatin in the mutants of these genes.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

ACKNOWLEDGMENTS
We thank Yan Zhang for experimental assistance. We thank Li-Lin Du and Daochun Kong for yeast strains. We thank Zhizhong Gong, Li-Lin Du, and Huiqiang Lou for advice on experiments. This work was supported by grants from the National Natural Science Foundation of China (31671296).

DECLARATION OF INTERESTS
The authors declare no competing interests. Costantino, L., and Koshland, D. (2018

Lead contact
Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Yu Wang (yw2250@cau.edu.cn).

Materials availability
All newly created strains generated in this study are available upon request.

Data and code availability
All data reported in this paper will be shared by the lead contact upon request. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Yeast strains and cell cultures
Yeast strains harboring deleted or epitope-tagged genes were constructed using a PCR-based module method (Bahler et al., 1998;Wach et al., 1994). Genetic crosses were used to construct double mutants. Cell used in this study were cultured in YEA at 30 C unless otherwise noted.

HO induction
HO induction was performed as described previously. Cells were first cultured in EMM-Leu supplemented with thiamine at 30 C and then shifted to thiamine-free EMM-Leu for 22 h to induce DSB. For HO induction of temperature-sensitive strains, Cells were shifted to 37 C for 3.5 h after incubated at 30 C for 22 h, and then used for ChIP assay.

Growth curves
The overnight cultures were diluted to an OD 600 of 0.1 as the initial value of the growth curve. Cells were further grown for 12 h with agitation and OD 600 was measured hourly.

Dilution analysis
Logarithmic phase cells were normalized to an OD 600 of 0.5. Ten-fold serial dilutions were spotted onto YEA plates and YEA with MMS (Sigma Aldrich). Cells were grown at 30 C for 3-4 days.

HU synchronization assay
Exponentially growing cells were synchronized in S phase by being treated with 15 mM HU (Sigma) for 4.5 h. Samples were released by washing twice in hydroxyurea-free media and then re-inoculated into medium lacking HU for further grown. Ten OD units of cells were collected every 30 min after release from HU block. At the same time, 1 ml of sample was collected, fixed with 70% ethanol and stained with calcofluor. Cell cycle progression was monitored by microscopically counting septated cells.