Abstract
Mating-type switching (MTS) in fission yeast Schizosaccharomyces pombe is a highly regulated gene conversion event. In the process, heterochromatic donors of genetic information are selected based on the P or M cell type and on the use of two recombination enhancers, SRE2 promoting use of mat2-P and SRE3 promoting use of mat3-M. Recently, we found that the histone H3K4 methyltransferase complex Set1C participates in donor selection, raising the question of how a complex best known for its effects in euchromatin controls recombination in heterochromatin. Here, we report that the histone H2BK119 ubiquitin ligase complex HULC functions with Set1C in MTS, as mutants in the shf1, brl1, brl2 and rad6 genes showed defects similar to Set1C mutants and belonged to the same epistasis group as set1Δ. Moreover, using H3K4R and H2BK119R histone mutants and a Set1-Y897A catalytic mutant indicated that ubiquitylation of histone H2BK119 by HULC and methylation of histone H3K4 by Set1C are functionally coupled in MTS. Cell-type biases in mutants further showed that the regulation might be by inhibiting use of the SRE3 enhancer in M cells, in favor of SRE2.
Consistently, imbalanced switching in the mutants was traced to compromised association of the directionality factor Swi6 with the recombination enhancers in M cells. Based on their known effects at other chromosomal locations, we speculate that HULC and Set1C might control nucleosome mobility and strand invasion near the SRE elements. In addition, we uncovered distinct effects of HULC and Set1C on histone H3K9 methylation and gene silencing, consistent with additional functions in the heterochromatic domain.
Author Summary Mating-type switching in the fission yeast Schizosaccharomyces pombe occurs by gene conversion using a donor, mat2 or mat3 located in a heterochromatin region. Multiple studies have shown that donor selection is critically affected by heterochromatic factors. Here, we document the role of euchromatic factors, the histone H2BK119 ubiquitin ligase complex HULC and the H3K4 methyltransferase complex Set1C, in donor selection. Mutational analysis indicated that HULC and Set1C inhibit donor choice at the mat3 cis-acting recombination enhancer SRE3 in M cells, through concerted histone modifications by the two complexes. Linking this effect with heterochromatin, mutants in each complex, shf1Δ and set1Δ strains, exhibited decreased association of the HP1 heterochromatic factor Swi6 with the SRE2 and SRE3 recombination enhancers at the mat2 and mat3 silent loci in M cells. These joint effects by the two complexes on MTS were observed even though other aspects, the methylation state of histone H3K9 and effects on gene silencing differed between the shf1Δ and set1Δ strains. The results provide insight into the regulation of recombination by chromatin structure.
Introduction
Homothallic strains (h90) of the fission yeast Schizosaccharomyces pombe switch between two mating types, P and M. This process is known as mating-type switching (MTS) and it takes place at the mat locus on chromosome 2. The mat locus comprises three cassettes, mat1, mat2 and mat3 (Fig 1A). The active mat1 cassette expresses either P or M mating-type specific genes and determines the mating-type of a haploid cell [1]. The genes in the mat2 and mat3 cassettes, P (mat2-P) and M (mat3-M), respectively, are invariant, and are silenced by heterochromatin. Each cassette is flanked by short homology boxes called H1 and H2. MTS is a result of gene conversion of the mat1 cassette by either the mat2-P or mat3-M donor cassette using the homology boxes [2].
A cell-type specific regulation takes place at the donor-selection step, where cells usually select the mating-type donor cassette opposite to the allele present at mat1, thus, P cells (mat1-P) preferentially choose mat3-M as a donor, while mat1-M, M cells (mat1-M) preferentially choose mat2-P (Fig 1B) [3]. This donor selection requires the mating-type switching factors Swi2 and Swi5, the heterochromatin protein 1 (HP1) homologue Swi6, and cis-acting DNA elements [4-6]. The cis-acting DNA elements are called Swi2-dependent Recombination Enhancers, SRE2 and SRE3, and are located next to the H1 box of mat2-P and mat3-M, respectively [4, 5]. The Swi2 protein localizes to these elements according to cell type: in P cells, Swi2 localizes only to SRE3; in M cells, Swi2 localizes to both SRE2 and SRE3 [4, 5]. From yeast two-hybrid assays, Swi2 has been shown to interact with Swi5 and with the recombination factor Rad51; therefore, it has been suggested that the Swi2-Swi5 complex promotes homologous recombination at the SRE2- or SRE3-adjacent cassette [7]. Furthermore, the Swi6 protein spreads at the mat locus between the boundary elements IR-L and IR-R and it contributes to form heterochromatin over the entire region [8, 9]. The association of Swi6 with the region differs between P and M cells, with greater enrichment of Swi6 in M cells than P cells [8]. As Swi2 interacts with Swi6 in vitro, the localization pattern of Swi2 is believed to be connected to its interaction with Swi6 [4, 7, 10].
Swi6 localization at the mat locus is controlled by several histone modifications [11]. An essential histone modification for Swi6 localization is di-or tri-methylation of histone H3 at lysine 9 (H3K9me2 and -me3, respectively) catalyzed by the methyltransferase Clr4, the homologue of human SUV39H1 and SUV39H2 [12-14]. H3K9me2 and -me3 are detected in constitutive heterochromatin such as mat, centromeres and telomeres, and in facultative heterochromatin such as meiotic genes [15, 16]. Constitutive heterochromatic regions contain repetitive DNA sequences that nucleate H3K9 methylation through RNA interference (RNAi) [15]. At the mat locus, RNAi is triggered by the transcription of cenH, homologous to centromeric repeats [17, 18]. dsRNAs originating from cenH are cleaved by the ribonuclease Dcr1, and the siRNA products are bound by the RNAi-induced transcriptional silencing complex (RITS) [19, 20]. RITS loaded with siRNAs recruits the Clr4-Rik1-Cul4 complex (CLRC) to methylate H3K9 [21]. At the mat locus, there is an additional silencing pathway involving the CREB-like transcription factor, Atf1-Pcr1 [22]. The Atf1-Pcr1 dimer binds to consensus sequences within the mat locus that exist at the REIII silencer and ∼1.4 k bp away from REIII, close to cenH (Fig 1A) [23]. Other histone-modifying enzymes such as histone deacetylases (HDACs) are also required for donor selection and heterochromatin establishment. These include the NAD+-dependent histone deacetylase Sir2 [24] and the Snf/Hdac-containing repressor complex (SHREC), Clr1, Clr2 and Clr3 [12, 25]. Clr4, Clr3 and another HDAC, Clr6, interact with Atf1, therefore Atf1-Pcr1 has been suggested to recruit these histone modifying enzymes to the mat locus [22, 26, 27].
We recently conducted a genetic screen for factors that affect mating-type switching and we identified the six genes encoding the H3K4 mono-, di-and tri-methyltransferase complex, Set1C/COMPASS, as well as a component of the H2BK119 monoubiquitin ligase HULC [3, 28]. In S. pombe, no other enzyme catalyzes these reactions. Ubiquitylation of H2B (H2Bub at K119 in fission yeast, at K123 in budding yeast, and at K120 in human) is required for the tri-methylation of H3K4 by Set1/COMPASS family from yeast to human [29]. Therefore, we hypothesized that H2Bub and H3K4me might work together to regulate MTS. Interestingly, the histone modifications by HULC and Set1C are generally observed at active genes, in euchromatin, and have been proposed to antagonize heterochromatin formation at the mat locus [8, 30, 31]. However, HULC and Set1C have also been reported to have a positive effect on gene silencing in heterochromatin [32-34]. HULC was proposed to promote loading of heterochromatin factors in centromeric repeats by facilitating a wave of transcription that fuels the RNAi machinery during S-phase [32]. In the case of Set1C, the deletion of the catalytic subunit Set1 causes a slight derepression of centromeric and telomeric reporter genes, and transcription of the cenH element in the mat region [33, 34]. In addition, Set1C also functions in the repression of the stress-response gene ste11 [35, 36] and Tf2 retrotransposons [34]. Interestingly, Set1 localizes with Atf1 binding sites at centromeres, where it contributes to the heterochromatin assembly with the Clr3 HDAC [37].
In this study, we performed genetic and molecular analyses of HULC, Set1C, and histone mutants to better understand the role of the HULC and Set1C complexes in mating-type switching. Our results support the view that, like Set1C [38], HULC affects donor choice by reducing the effectiveness of the SRE3 enhancer in M cells and that H2B ubiquitylation by HULC and H3K4 methylation by Set1C operates in the same pathway for this function. Both modifying enzymes regulate Swi6 enrichment positively at the SRE2 and SRE3 elements. On the other hand, the two enzymes differentially affect heterochromatin formation and heterochromatic silencing at the mat locus. When combined with dcr1Δ, the set1Δ mutation caused a derepression of a reporter gene at SRE3 while shf1Δ did not. Thus, beyond increasing our understanding of donor selection through HULC and Set1C, our findings also shed light on gene silencing mechanisms by these complexes.
Results
The histone H2B ubiquitylation complex HULC is involved in MTS
Recently, we determined that a subunit of HULC, the brl2+ gene product, is required for efficient MTS by screening the gene deletion library, Bioneer version 5 [38]. In an independent screen for MTS defects with version 2 of the same library, we identified another subunit of HULC, the shf1+ gene product. Therefore, we systematically investigated the switching phenotypes of mutants in each of the HULC subunits, the shf1Δ, rad6Δ, brl1Δ and brl2Δ mutants. MTS defects can be detected by measuring mating-type ratios in saturated liquid cell cultures because efficient switching results in an equal proportion of each cell type. Mutants in a few factors such as Swi2 mutants display different mating-type ratios in independent cultures while other mutations uniformly bias cell populations towards P or M [5, 38]. Thus, for each HULC mutant, four independent strains were analyzed to evaluate clonal variation by multiplex PCR. The wild-type h90 strain, PG4045, had nearly equal proportions of P and M cells, as expected, but all HULC mutants displayed a biased mating-type ratio which was around 33% P cells in all cultures examined (Figs 2A and S1A). The deletion mutants were also examined by iodine staining of colonies, a classical assay for MTS efficiency. Colonies of the control PG4045 strain grown on EMM plate stained darkly with iodine, indicative of efficient MTS, conjugation and sporulation during colony growth. The HULC mutants were less stained (Figs 2B and S1B) consistent with the observed cell-type bias. Double deletion of shf1 and, respectively, rad6, brl1 or brl2 did not show any additive effect (Figs 2C and S1C). We concluded that HULC is involved in MTS.
The Rad18/Rad6 pathway does not influence MTS
In addition to its interaction with the E3 ubiquitin ligase Brl, the E2 Rad6 enzyme also operates with the E3 Rad18 in a conserved DNA damage tolerance pathway by mono-ubiquitylating PCNA [39]. We tested whether the Rad6/Rad18 pathway is required for MTS. As expected, both the rad6 and rad18 deletion mutant strains were sensitive to UV exposure, similar to a rad51 deletion mutant strain (Fig 2D). On the other hand, HULC mutants, respectively shf1Δ, brl1Δ and brl2Δ, did not show UV damage sensitivity (Fig 2D). Conversely, multiplex PCR and iodine staining assays did not detect any switching defect in the rad18Δ mutant, unlike for the shf1Δ, brl1Δ and brl2Δ mutants (Figs 2A, B, S1A and S1B). These results indicate that Rad6 functions in different complexes for MTS and DNA damage tolerance.
HULC regulates donor selection at SRE3
Depending on which step is affected, mutants deficient in MTS can be separated into three classes, Class Ia, Class Ib and Class II, by Southern blots [6]. Class Ia mutants lack an imprint at mat1; the imprint is required for MTS as it is converted to a single-ended double-strand break just past the H1 homology box during DNA replication, allowing invasion of mat2-P or mat3-M by the H1 sequence (Fig 1) [2, 3]. The imprint also creates a fragile site that causes breakage during DNA preparation and it is thus visible on Southern blots as a DNA double-strand break. Class Ib mutants have the imprint but fail to use it properly. Class II mutants are deficient at a later step, in the resolution of the gene conversion, which causes characteristic mat2-mat3 cassette duplications inserted at mat1 that are not found in Class I mutants. The shf1 deletion was assigned to Class Ib in which a functional imprint is detected at the mat1 cassette (S2A Fig) similar to the previously examined brl2 deletion [38]. The Class Ib group includes mutations that impair donor selection. These can be identified and further classified according to their effects in h09 cells that have swapped mat2-M mat3-P donor cassettes [40]. At least four groups of h09 Class Ib mutants can be distinguished: Group 1a displaying a strong bias towards P cells; Group 1b with a mild bias towards P cell; Group 2 with clonal variations; and Group 3 with a bias towards M cells [38]. Deletions removing HULC subunits each caused a mild bias towards P cells in the h09 background (∼48 % P cells instead of ∼19 % in the presence of functional HULC), placing them in Group 1b (Figs 3A and S2B).
The SRE elements are necessary for donor selection [4, 5]. We investigated the importance of SRE2 and SRE3 in the presence or absence of the shf1 gene using SRE deletion mutants (Fig 3B). In an SRE3Δ strain, the shf1 deletion did not affect the mating-type ratio (∼ 83% P cells in shf1Δ instead of ∼84 % in shf1+ cells) (Figs 3B and S2C). In an SRE2Δ strain, the shf1 deletion slightly affected the mating-type ratio towards M cells (∼ 12% P cells in shf1Δ instead of ∼19 % in shf1+ cells) (Figs 3B and S2C). Taken together with the biases caused by HULC mutations in h90 (towards M cells, using SRE3) and h09 strains (towards P cells, using SRE3), these results suggest that, in M cells, HULC normally prevents the use of SRE3, rather than facilitating the use of SRE2.
Histone residues modified by HULC and Set1C are required for MTS
We noticed that the genetic analyses in Fig 3 showed similar trends for the MTS defects detected in HULC mutants and the defects previously reported for set1 deletion mutants [38]. The ubiquitylation of H2BK119 by HULC is essential for H3K4me3 by Set1C in S. pombe [41], therefore we performed an epistasis analysis to test the relationship between HULC and Set1C in MTS. The switching defect of the h90 shf1Δ set1Δ double mutant was quite similar to the switching defect of the h90 set1Δ strain (∼42% P cells in shf1Δ set1Δ mutant versus ∼43% P cells in set1Δ mutant in the multiplex PCR assay), with set1Δ slightly suppressing the cell-type bias of the shf1Δ strain (∼33% P cells in shf1Δ mutant) (Figs 4A and S3A). This placed the two mutations in the same epistasis group, suggesting coordinated action of HULC and Set1C.
Next, we investigated the importance of the histone H2BK119 and H3K4 residues that are modified by HULC and Set1C, respectively, in MTS. The proportion of P cells in the H2BK119R mutant (∼34% P cells) was not only similar to the shf1Δ deletion mutant, but also to the double mutant shf1Δ H2BK119R (Figs 4B and S3B). On the other hand, the proportion of P cells in the H3K4R mutant was ∼27% P cells, less than in the set1Δ deletion mutant (∼44% P cells), but similar to the H3K4R set1Δ double mutant (∼30% P cells) (Figs 4B and S3C). We also investigated the requirement for the catalytic activity of Set1 in MTS. The SET domain has a highly conserved tyrosine residue which is suggested to be a catalytic residue from the crystal structure (S3D Fig) [42]. Replacement of tyrosine Y1054 with alanine in S. cerevisiae Set1 causes loss of H3K4 methylation [43]. We created a Set1-Y897A mutant in S. pombe, corresponding to Set1-Y1054A in S. cerevisiae (S3D Fig). The mating-type ratios in the set1-Y897A mutant analyzed by multiplex PCR were similar to the ratios in the set1Δ mutant (Figs 4D and S3E). We verified that the Set1-Y897A protein tagged with 9×V5 epitope (9×V5-Set1-Y897A) was present in the cells (S3F Fig). The swd1+ gene encodes a subunit of the Set1C complex [41]. An swd1Δ single mutant and swd1Δ Set1-Y897A double mutant also showed similar ratios to the set1Δ mutant (Figs 4D and S3E). These observations point to both H2BK119 ubiquitylation by HULC and H3K4 methylation by Set1C playing an important role in the directionality of MTS.
Several lines of evidence have indicated that the RNA polymerase II-associated factor 1 complex Paf1C [44-46], functionally conserved from yeast to mammals [47, 48], can recruit HULC and Set1C [49] and in fission yeast Paf1C prevents heterochromatin propagation across the IR-L boundary of the mat locus, among other effects [49]. We constructed strains lacking the Paf1C components Leo1 and Paf1, respectively, to test the requirement for Paf1C in MTS. The leo1Δ and paf1Δ mutants did not show a switching defect in the multiplex PCR assay (Figs 4E and S3G). Thus, the effects of HULC and Set1C in MTS occur independently of Paf1C.
Set1 and Shf1 are involved in the mating-type-specific localization of Swi6 at SRE2 and SRE3, but they have different effects on histone H3K9 methylation
The results presented so far allow updating a previous model [38] by now proposing that both HULC and Set1C reduce use of the SRE3 recombination enhancer in M cells by modifying histone H2BK119 and H3K4. Given that differential Swi6 enrichment in the mating-type regions of M and P cells is a determinant of donor choice [4, 5, 40], we next investigated the effects of HULC and Set1C on Swi6 occupancy. To this end, we performed ChIP-qPCR with Flag-tag antibody for 3×Flag-Swi6 using heterothallic strains with a fixed mating type, P or M (Fig 5A). In a wild-type background, Swi6 showed an approximately three-fold higher enrichment at both SRE2 and SRE3 in M cells compared to P cells (Fig 5B). No significant difference between P and M cells was observed at the K region, a location between mat2-P and cenH (Fig 5B). In both the shf1Δ and set1Δ backgrounds, the high, M-specific, Swi6 occupancy at SRE2 and SRE3 was decreased (Fig 5B). It thus appears likely that Shf1 and Set1, and by extension HULC and Set1C, control donor choice at least in part by ensuring high Swi6 occupancy at recombination enhancers in M cells. We note however that the double shf1Δ swi6Δ deletion mutant showed a statistically significant slightly lower (in h90) or higher (in h09) proportion of P cells compared with the single swi6Δ mutant (S4 Fig), suggesting some effects of HULC in MTS are not through Swi6.
Swi6 recognizes di- and trimethylation of histone H3K9 [14, 50]. We thus examined the di-and tri-methylation levels of H3K9 in shf1Δ and set1Δ strains using the same chromatin fixed samples as in Fig 5B and antibodies specific for di- and tri-methylation of H3K9 (H3K9me2 and H3K9me3, respectively) (Fig 5C and D). Comparing P and M cells, we observed that H3K9me2 enrichments in wild-type strains were similar at the three locations examined and did not vary with cell type, with the possible exception of SRE2 for which H3K9me2 was slightly lower in M cells (Fig 5C). In shf1Δ strains, H3K9me2 was significantly reduced in both cell types with the similar tendency of slightly lower enrichments in M cells (Fig 5C). The opposite trend was seen for set1Δ cells where H3K9me2 was unchanged or, in the case of M cells, increased (Fig 5C).
For H3K9me3, M cells showed somewhat higher enrichments than P cells in the wild-type background (Fig 5D). This difference between P and M cells was attenuated in the shf1Δ background, but globally the levels of H3K9me3 enrichments remained unchanged in shf1Δ cells (Fig 5D). As observed for H3K9me2 (Fig 5C), H3K9me3 levels were increased by set1Δ, at all locations and in both cell types (Fig 5D). These ChIP-qPCR analyses suggest that the enrichment levels of H3K9me2 and H3K9me3 are regulated by Shf1 and Set1, however these regulations work in different manners and correlations between H3K9 methylation and Swi6 enrichment levels were not systematically observed.
Shf1 and Set1 have differential effects on silencing and MTS
Two main pathways of heterochromatin formation have been well documented at the mat locus, one is RNAi nucleating heterochromatin at the cenH region [15, 17, 18], and the other is an RNAi-independent pathway involving the Atf1-Pcr1 complex [22, 26, 51, 52]. Both H2Bub and H3K4me are believed to control transcription by regulating RNA polymerase II activity in S. pombe [31, 34, 53]. This suggests that HULC and Set1C might participate in silencing the mat locus by facilitating RNAi which uses transcription products from cenH to nucleate heterochromatin. Alternatively, the fact that H2Bub and H3K4me mediate gene silencing at the ste11 gene locus [35, 36], together with the fact that Set1 localizes to Atf1 binding sites at centromeres and at the ste11 gene locus [37], suggests that HULC and Set1C might co-operate with Atf1-Pcr1 also in the mating-type region. To investigate whether Shf1 or Set1 participate in the RNAi or Atf1-Pcr1 pathway of heterochromatin formation, we created double mutants, respectively shf1Δ dcr1Δ, shf1Δ pcr1Δ, set1Δ dcr1Δ and set1Δ pcr1Δ in a strain with a ura4+ reporter gene inserted in the SRE3 region (PG1899 with (EcoRV)::ura4+) (Fig 6A). As previously reported [22], dcr1Δ and pcr1Δ single deletion mutants with the (EcoRV)::ura4+ reporter were resistant to 5-FOA, which is toxic to cells that express ura4+, while the dcr1Δ pcr1Δ double deletion caused sensitivity, which was the same level as the swi6Δ strain (Fig 6B). The shf1Δ single mutant was resistant to 5-FOA, and the shf1Δ dcr1Δ and shf1Δ pcr1Δ double mutants remained resistant. In contrast, the set1Δ and set1Δ pcr1Δ strains were resistant to 5-FOA, but the set1Δ dcr1Δ strain was clearly sensitive, its growth on 5-FOA was as severely affected as growth of the dcr1Δ pcr1Δ strain (Fig 6B). This places set1Δ, but not shf1Δ, in the same epistasis group as pcr1Δ for the effects of these mutations on silencing in the mating-type region.
We examined the switching phenotype of the strains shown in Fig 6B by multiplex PCR (Figs 6C and S5). The PG1899 strain was slightly biased toward P cells (∼59% P cells) in comparison with the normal h90 configuration (∼50% P cells). This switching phenotype probably comes from the insertion of the ura4+ gene slightly affecting SRE3 due to its proximity (see Fig 6A). However, populations of a swi6Δ derivative of PG1899 showed mating-type biases similar to the h90 swi6Δ mutant, hence we analyzed all mutant strains derived from PG1899 for MTS. The single dcr1Δ and pcr1Δ mutants showed mating-type ratios similar to PG1899. The double dcr1Δ pcr1Δ mutant was biased towards M cells (∼45% P cells). The double shf1Δ dcr1Δ and shf1Δ pcr1Δ mutants had cell-type ratios similar to the shf1Δ single deletion mutant. In the case of set1Δ, the set1Δ pcr1Δ deletion strain had a cell-type ratio similar to the set1Δ strain, but the set1Δ dcr1Δ double deletion strain showed a larger bias toward M cells than the set1Δ strain, the largest bias of all strains examined. Thus, both gene silencing and multiplex PCR assays converge to show that Set1 functions in parallel to Dcr1 at the silent mat locus.
Discussion
Mating-type switching is governed by chromatin conformation. Thus, the H3K9 methyltransferase CLRC and several HDACs have well documented roles on the formation of heterochromatin required for the directionality of switching. Our genetic screens identified euchromatic factors, the H2B ubiquitin ligase HULC and the H3K4 methyltransferase Set1C, as additional MTS factors (Fig 2) [38]. A potentially important feature is that these complexes engage in a cross-talk where H2B ubiquitylation upregulates H3K4 methylation by Set1C [53]. Here, we obtained evidence that HULC and Set1C function in a common pathway for MTS to inhibit the preferential use of the SRE3 enhancer in M cells, plausibly through this crosstalk. Notably, the effects of the two complexes determined by mutational analyses were not quite identical and distinct contributions to gene silencing were detected for Set1 and Shf1 that are likely to reflect some divergent functions in heterochromatin formation.
Strong biases towards the M cell-type were observed in both HULC and Set1C mutants denoting increased use of the SRE3 enhancer in these mutants (Figs 2 and 3) [28]. An epistasis analysis assigned the two complexes to the same pathway (Fig 4), even though a slightly more pronounced bias towards M cells was observed in shf1Δ cells than in set1Δ or shf1Δ set1Δ cells (Fig 4A). It has been reported that Set1 is still expressed in cells that lack HULC and ubiquitylation of H2BK119 in vivo [31, 34, 53] and H3K4 di-methylation is slightly detected in H2BK119R strain [53]. In S. cerevisiae, H3K4 mono-methylation is still detected in rad6 deletion cells [54]. Therefore, we speculate that the residual mono- and/or di-methylation of H3K4 in cells lacking H2Bub might increase the bias towards M cells, either directly through effects on recombination or indirectly. Our study also revealed that while the H2BK119R, shf1Δ and combined H2BK119R shf1Δ mutations had the same effect on MTS (Fig 4B), the switching defects in the H3K4R and H3K4R set1Δ mutants were much stronger than in the single set1Δ strain (Fig 4C). Transient acetylation of H3K4 has been proposed to facilitate the association of Swi6 with histone H3K9me2 through a chromodomain switch where Swi6 replaces H3K9me2-bound Chp1 or Clr4 at centromeres [55]. The same mechanism facilitating Swi6 association at SRE elements at the mat locus might account for the pronounced effects of H3K4R on MTS where Swi6 association is paramount.
The differential association of Swi6 with the mating-type region is a distinguishing factor between P and M cells: Swi6 is present at a low level over the region in P cells, coinciding with Swi2 specifically at SRE3, but at a high level in M cells, coinciding with Swi2 at both SRE2 and SRE3 [4]. These associations are believed to favor the use of SRE2 over SRE3 in M cells [5, 56]. We found here that Shf1 (and to a lesser extent Set1) is required for the high Swi6 occupancy at SRE2 and SRE3 in M cells. In the shf1Δ and set1Δ mutants, Swi6 occupancy remained abnormally low at both enhancers in M cells, similar to what is normally seen in P cells (Fig 5). This profile could account for SRE3 being preferred over SRE2 in both cell types in the mutants, where all cells would in essence behave as P cells. The association of Swi6 with the enhancers was most strongly reduced in the shf1Δ mutant (Fig 5), where donor choice was also most impaired (Fig 4). These effects would place HULC and Set1C upstream of the enhanced Swi6 association with recombination enhancers in M cells (Fig 6D), without excluding other points of action.
Our experiments are consistent with and support an active repression by HULC and Set1C at SRE3, relevant to the central question of how SRE2 outcompetes SRE3 in M cells when Swi2 is present at both enhancers. SRE2 might be inherently more efficient than SRE3 under the conditions, or recombination might be actively repressed at SRE3. Here, shf1Δ and set1Δ mutants showed a bias towards M cells in h90 cells (Figs 2 and 4) and towards P cells in h09 cells (Fig 3A), consistent with the two factors repressing use of SRE3. Moreover, the ∼80% bias towards P in SRE3Δ cells remained unchanged in the shf1Δ mutant, indicating SRE2 is functional in the absence of Shf1. In contrast, in the SRE2Δ strain, SRE3 was increasingly used in shf1Δ cells, indicating Shf1 inhibits use of SRE3 (Fig 3B). Thus, our genetic analyses suggest that HULC inhibits selection of SRE3 - and thereby mat3-M donor selection -in h90 cells, similar to Set1C (Fig 6D) [28].
How might an inhibition of recombination by HULC/Set1C take place at SRE3? A relevant effect of HULC/Set1C could be by controlling nucleosome occupancy or positioning. In S. cerevisiae, nucleosome occupancy is decreased genome-wide in H2BK123, rad6Δ, and lge1Δ mutants [57, 58]. In S. pombe, H2Bub decreases chromatin remodeling by RSC at the ste11 promoter to repress transcription [35, 36]. In this case, the effect of H2Bub is through H3K4me and histone deacetylation, supporting the idea that HULC, Set1C, and HDACs might work in concert to position nucleosomes at the mat locus as well. Nucleosome positioning at SRE3 might mask the enhancer or prevent strand invasion. An intriguing alternative that our results do not exclude is that nucleosome positioning or modification by HULC/Set1C might facilitate recombination near SRE2 in M cells. Positive effects on recombination repair and on recombination-dependent bypass of DNA lesions have been reported for the RNF20/RNF40 mammalian homolog of Bre1 [59, 60] and for the S. cerevisiae counterpart of HULC [61]. Nucleosome depletion by associated remodelers, rather than nucleosome stabilization, often appears instrumental during repair, highlighting the context dependency of the effects of the modifying and remodeling complexes [60, 62]. It will be important in the future to understand how remodeling complexes might contribute to MTS, taking into account the fact that Swi6 also interacts with many remodelers [63].
Finally, we uncovered effects of HULC and Set1C on the heterochromatic structure of the mating-type region that bring light to how these complexes might affect Swi6 association and MTS. In the case of shf1Δ, we observed reduced H3K9me2 at the three locations tested (Fig 5C). HULC has been suggested to associate with the RNAi machinery in centromeric region when repetitive sequences are transcribed during S-phase [32], by analogy cenH in the mating-type region might be an entry point for HULC. In the case of set1Δ, we observed a synthetic silencing defect when combining the set1Δ and dcr1Δ mutations (Fig. 6B), showing that Set1 and Dcr1 participate in parallel pathways of heterochromatin formation. The effect is clearly relevant to MTS as the double mutant showed a strong bias towards M cells (Fig 6C). Similarly, Dcr1 and the Clr3 HDAC operate in parallel to recruit Clr4 to the mat locus, Dcr1 through RNAi at cenH and Clr3 through the Atf1-Pcr1 binding sites [27]. Here, we placed set1Δ in the same epistasis group as pcr1Δ for the effects of these mutations on silencing (Fig 6B), suggesting that the Atf1-Pcr1 binding sites at the mat locus constitute entry points not just for Clr3 but also for Set1C. Independently, a genome-wide study found that Set1 cooperates with Clr3 to repress transcription at other Atf1-Pcr1 binding sites [37] and an important effect of Clr3 in heterochromatin is to suppress histone turnover [51, 52, 64]. Thus, taken altogether, HULC and Set1C might be recruited in several ways to the mat locus where they would cooperate with Clr3 to regulate aspects of heterochromatin formation and nucleosome occupancy important for donor selection. To further analyze this mechanism, we will need to understand how the regulation would be exerted in a cell-type-specific manner, as well as spatially, whether the regulation has to occur locally at the enhancers, or whether global effects on nucleosome mobility that would differ in P and M cells might lead to the observed biases in enhancer use.
Material and Methods
Yeast strains, strains constructions and strain manipulations
S. pombe strains used in this study are described in S1 Table. Standard techniques were used to cultivate, sporulate, cross and genetically manipulate S. pombe [65]. Strains were generated by transformations or genetic crosses. The H3K4R mutant strain (TM504: h90 hht1-H3K4R hht2-H3K4R leu1-32 his3-D1) was generated from EM20 (h90 leu1-32 his3-D1 ade6-M375) by CRISPR/Cas9-mediated gene editing. EM20 was transformed by a gRNA expression plasmid (pEM59), a Cas9 expression plasmid and HR donor templates. To mutate the hht1 and hht2 genes, the common target sequence, 5’-TCTACCGGTGGTAAGGCACC-3’, was inserted into the gRNA scaffold portion in pEM59. To select cells in which Cas9 was active, the ade6-M375 mutation was edited in the same transformation. The gRNA targeting ade6-M375, 5’ CCTGCCAAACAAATTGATTG was also expressed from pEM59. The HR donor templates for mutagenesis, purchased from Integrated DNA Technologies, were amplified using primer sets, hht1-H3K4R: (5’-CTGCAGTACGCTTGCGTTTC-3’ and 5’-GGGACGATAACGATGAGGCTTC-3’), hht2-H3K4R (5’-GGGAAGCCGAAATCGCAATC-3’ and 5’-CCAGGACGATAACGATGAGGCTTC-3’) and ade6+ (5’-GTGGTCAATTGGGCCGTAT-3’ and 5’-CGTGCACTTCTTAGACAGTTCA-3’) by PCR. Cells were grown on low-adenine plates and white colonies (ade6+) were selected. TM863 and TM896 were derived from TM504. TM501 was also generated by Cas9-mediated gene editing, with the gRNA target sequence (5’-TACTTATGATTACAAGTTTC-3’) and the HR donor template amplified by primer set (5’-GGGAAATATCGCGCGTTTC-3’ and 5’-CTAGTTTAAATAGCCACGACATGT-3’). All mutants selected for further analysis were confirmed by PCR and sequence analysis.
Iodine staining
Cells were streaked on MSA plates, grown at 26°C for 3∼4 days and colonies were exposed to iodine vapors.
Multiplex PCR
Isolated colonies of the strains of interest were propagated in 2 ml YE5S cultures at 30°C, to saturation. 500 μl of each culture were harvested in a 1.5 ml microcentrifuge tube, followed by DNA extraction with a Dr. GenTLE (from yeast) high recovery kit (Takara Bio). Quantification of genomic DNA concentration was performed using the Quantifluor® ONE dsDNA Dye System (Promega). Multiplex PCR was performed as previously described [28].
UV damage sensitivity
Serial dilutions of exponentially growing cell cultures were plated on complete medium (YES) and subjected to UV irradiation by exposure to a germicidal lamp (254 nm; 100 J/m2). UV intensities were measured with a UV Radiometer (TOPCON UVR-2). Plates were incubated at 30°C for 3∼4 days.
Chromatin Immunoprecipitation (ChIP)
ChIP was performed as in Kimura et al., 2008 with a few modifications [66]. EMM2 medium (50 mL) containing 0.1 g/l each of leucine, adenine, histidine, uracil and arginine was used for cell culture. The cultures were propagated to 1.0 × 107 cells/ml at 30°C, and then shifted to 18°C for 2 h. 5.0 × 108 cells were cross-linked with 1% formaldehyde for 15 min at 25°C and then incubated in 125 mM glycine for 5 min. Cross-linked cell lysates were solubilized by a multi-beads shocker (Yasui Kikai) at 4°C, 15 cycle of 1 min on and 1 min off, and sonicated using a Bioruptor UCD-200 (Diagenode) at 2 cycle of 10 min each with alternating pulses of 40 sec on and 30 sec off at high level. The sheared samples were centrifuged at 20,000 g for 10 min at 4°C. The supernatants were incubated with 30 μl Dynabeads Protein A (Thermo Fisher) preloaded with 1.2 μl anti-FLAG M2 antibody (Sigma-Aldrich) for 6 h at 4°C. The beads were washed sequentially with wash buffer 1 (50 mM Hepes-KOH [pH7.5], 1 mM EDTA, 0.1 % Sodium deoxycholate, 0.5 M NaCl), wash buffer 2 (10 mM Tris-HCl [pH8.0], 0.25 M LiCl, 1 mM EDTA, 0.5% NP40, 0.5% SDS) and TE (twice), and materials coprecipitated with the beads were eluted with elution buffer (50 mM Tris-HCl [pH 7.6], 10 mM EDTA and 1% SDS) for 20 min at 65°C. The eluates were incubated at 65°C overnight to reverse cross-links and were then treated with 10 μg/ml RNase A for 1 hr at 37°C, followed by 20 μg/ml proteinase K for 3 hr at 50°C. DNA was purified with a MonoFas DNA purification kit I (GL Sciences). Quantitative PCR was performed with SYBR Premix Ex Taq II (TaKaRa Bio) or TB Green Premix DimerEraser (Takara Bio) on a Mx3000P qPCR system (Agilent). Primer sequences are in S2 Table B. For ChIP of H3K9me2 and -me3, 20 μg of H3K9me2 antibody or 20 μg of H3K9me3 antibody [67] were preloaded to 40 μl Dynabeads M-280 Sheep anti-Mouse IgG (Thermo Fisher).
Funding
We acknowledge the following grant support: Grants-in-Aid for Scientific Research (A) (JP18H03985) to HI, for Scientific Research (B) (JP18H02371) to HT, and for Scientific Research on Innovative Areas (JP18H05527) to HK from the Japan Society for the Promotion of Science (JSPS); grant R35 GM127029 to JEH from NIH; the World Research Hub Initiative (WHRI) program at the Tokyo Institute of Technology to JEH; grant R167-A11089 to GT from the Danish Cancer Research Society; Otsuka Toshimi Scholarship Foundation to AEC.
Author contributions
A.E.C. and T.M. conducted experiments. T.M. and H.I. were responsible for conceptualization and project design. T.M., H.T., T.H., H.K., G.T. and H.I. supervised the study. H.K. provided materials. A.E.C., T.M., J.E.H, G.T and H.I. wrote the manuscript. T.M., J.E.H, G.T and H.I. were responsible for data analysis and funding acquisition.
Conflict of interest
The authors have no conflicts of interest to declare.
Acknowledgments
We are grateful to members of the Iwasaki Laboratory for discussion. We thank to Jason Tanny for providing us with the htb1 mutant strain and Jun-ichi Nakayama for providing us with the Flag-Swi6 strain. We thank the Biomaterials Analysis Division, Open Facility Center, Tokyo Institute of Technology for sequence analysis.