Report

The COMPASS Subunit Spp1 Links Histone Methylation to Initiation of Meiotic Recombination

See allHide authors and affiliations

Science  11 Jan 2013:
Vol. 339, Issue 6116, pp. 215-218
DOI: 10.1126/science.1225739

Repair and Recombination

In most sexually reproducing organisms, the haploid gametes are produced by meiosis, a specialized cell-division during which recombination between the parental chromosomes ensures proper chromosome segregation. Double-strand breaks (DSB) generated in the DNA drive this recombination and are linked to recombination hotspots in the chromatin where histone H3 is methylated on lysine residue 4 (H3K4me) by the Set1C/COMPASS enzyme complex. Acquaviva et al. (p. 215, published online 15 November) show that the Mer2/Mei4/Rec114 complex of the DSB generating machinery is linked to the Set1C/COMPASS enzyme complex through the Spp1 subunit, which contains a PHD-finger capable of binding to H3K4me. The Spp1 subunit recruits the Mer2 protein to sites where recombination will occur, probably by binding to both the H3K4me-marked chromatin and other factors.

Abstract

During meiosis, combinatorial associations of genetic traits arise from homologous recombination between parental chromosomes. Histone H3 lysine 4 trimethylation marks meiotic recombination hotspots in yeast and mammals, but how this ubiquitous chromatin modification relates to the initiation of double-strand breaks (DSBs) dependent on Spo11 remains unknown. Here, we show that the tethering of a PHD-containing protein, Spp1 (a component of the COMPASS complex), to recombinationally cold regions is sufficient to induce DSB formation. Furthermore, we found that Spp1 physically interacts with Mer2, a key protein of the differentiated chromosomal axis required for DSB formation. Thus, by interacting with H3K4me3 and Mer2, Spp1 promotes recruitment of potential meiotic DSB sites to the chromosomal axis, allowing Spo11 cleavage at nearby nucleosome-depleted regions.

Meiotic recombination is initiated by the introduction of DNA double-strand breaks (DSBs) by Spo11, a meiosis-specific transesterase, which is highly conserved throughout evolution (1). To ensure at least one crossover per chromosome pair, a large number of DSBs are formed in each meiotic cell, and their occurrence is controlled at multiple levels, comprising both higher-order chromosome structure and local determinants (2). In Saccharomyces cerevisiae, ~150 DSBs are formed per meiosis, preferentially in promoter regions and with a highly variable frequency. DSB sites are largely independent of DNA sequence composition (3). The control of DSB formation also depends on a number of Spo11-accessory proteins that have been shown to form subcomplexes, but whose functions are only partially understood (4). In particular, the Mer2/Mei4/Rec114 complex plays a role in linking replication to subsequent DSB formation through Mer2 phosphorylation (57). This complex is also thought to tether potential DSB sites located on chromatin loops to the highly differentiated meiotic chromosome axis, leading to stepwise activation of Spo11 cleavage (810).

In S. cerevisiae, all H3K4 methylation is carried out by the COMPASS protein complex that includes Set1, the catalytic subunit, which acts as a scaffold for the other structural and regulatory components (11). In particular, the PHD-finger subunit Spp1 specifically regulates the H3K4me3 state (11, 12). Our previous studies revealed that the absence of Set1 severely reduces meiotic DSB levels (13, 14) and that the level of H3K4me3 is constitutively higher near DSB sites (14). However, beyond these correlations, the mechanistic link between H3K4 methylation and DSB formation has remained elusive (15).

To uncover functional connections, we first asked whether the mutation of each COMPASS subunit affected DSB frequencies at natural recombination hotspots (16). Similar to the deletion of SET1, the absence of each COMPASS subunit, except Shg1, reduced DSB frequencies at the BUD23 (Fig. 1A) and CYS3 hotspots (fig. S1A; strain genotypes in table S1). We observed the frequency reduction for Spp1, which is specifically required for H3K4me3 formation. The similar effect of set1Δ and spp1Δ mutants on DSB formation at hotspots also extends to DSB formation near the naturally cold PES4 locus (fig. S1B) (14). To check if the decrease in DSB formation at hotspots reflected the role of only the COMPASS complex on H3K4 methylation, we asked whether this decrease was recapitulated by mutation of the H3K4 residue. At BUD23, we observed equivalent DSB reduction in the set1Δ and H3K4R strains (fig. S1C). Parallel analysis of the set1∆ H3K4R double mutant revealed an identical effect, underscoring their epistatic relationship (fig. S1C). Together, these observations strongly support a role for the COMPASS complex and H3K4 trimethylation in the control of DSB formation and tie them to the same pathway.

Fig. 1

GBD-Spp1 induces H3K4me3-independent meiotic DSBs at GAL2. (A) DSB formation in COMPASS mutants was analyzed by Southern blot at the BUD23-ARE1 region. The arrow indicates the DSB site. Quantification of the prominent DSB fragments (percentage of DSB fragment per total DNA) is indicated below the gel. (B) DSB analysis at GAL2 of GBD-SPP1 cells harvested at different times after transfer into SPM. (C) High-resolution analysis of DSBs at GAL2 16 hours after transfer to SPM was performed as described in (25). The five UASGAL2 sites are indicated by gray triangles. (D) Southern blot analysis of DNA extracted from GBD-SPP1 cells carrying the indicated mutations. Genomic DNAs were separated by pulsed-field gel electrophoresis and visualized with a chromosome XII left-end probe. GBD-SPO11 and WT were used as controls.

To elucidate the role of H3K4me3, Set1, and Spp1 in DSB formation, we next asked whether the tethering of Set1 and Spp1 proteins to UASGAL (UAS, upstream activating sequence) binding sites located in cold DSB regions was sufficient to stimulate DSB formation, as previously observed for the fusion of the Gal4 binding domain (GBD) with Spo11 (17). We generated and expressed in-frame fusions of the coding region with the GBD. With respect to H3K4 methylation and DSB formation, both GBD-Set1 and GBD-Spp1 fusions behaved the same as the wild type, indicating no adverse interference with COMPASS and the Spo11 machinery (fig. S2 and table S2). We examined DSB formation in the naturally cold GAL2 locus, which contains five UASGAL sequences in its promoter. Notably, the GBD-Set1 construct stimulated DSBs near the UASGAL2 sites (fig. S3, A and B). In terms of genetic requirements, the data reported in fig. S3, C to E, establish that the GBD-Set1–induced DSBs were not associated with a coincident increase in H3K4me3 at GAL2 but did depend on (i) the presence of Spo11, (ii) the integrity of the COMPASS, (iii) the histone methyl transferase activity of Set1, and (iv) the presence of the H3 lysine 4 residue.

Similar analyses revealed that GBD-Spp1 strongly stimulated DSB formation at GAL2. These DSBs appeared and disappeared upon repair with similar kinetics as natural DSBs (Fig. 1, B and C, see also Fig. 2A). Notably, DSBs targeted by GBD-Spp1 required neither the presence of Set1 or Bre2 nor the E3 ligase Bre1 that controls H3K4 trimethylation through the ubiquitinylation of H2B (18). In agreement with this observation, DSBs resulting from the tethering of Spp1 also appeared in the H3K4R mutant (Fig. 1D). Tethering Spp1 to the GAL1/GAL10 and GAL7 loci also efficiently induced DSBs (Fig. 2A), but, surprisingly, these breaks did not occur at the same location as those observed in the GBD-SPO11 (Fig. 2B). Whereas tethering of Spo11 introduced DSBs in the vicinity the UAS sites, tethering of Spp1 to the same sites produced cleavage in the 3′ end of the GAL10 coding sequence (Fig. 2B and fig. S4A). As for the majority of meiotic DSBs in S. cerevisiae, occurring in promoter regions, the unexpected location of these GBD-Spp1–induced DSBs coincides with the promoter of the well-characterized antisense GAL1 ucut (SUT013) noncoding RNA. This RNA is expressed in meiosis (19) under the control of the transcription factor Reb1 (20). This situation is also consistent with the high-resolution, genome-wide map of DSBs, which revealed a high enrichment of Spo11 binding in nucleosome-depleted regions adjacent to the Reb1 binding sites (3). As in GAL2, the DSBs generated at GAL10 required Spo11 and occurred in the absence of H3K4 methylation (fig. S4B). Together, these results demonstrated that the tethering of Spp1 efficiently stimulated Spo11-dependent DSBs independently of H3K4 methylation.

Fig. 2

GBD-Spp1 induces DSBs distal to the UASGAL7-GAL10. (A) Kinetics of DSB formation at GAL10 induced by GBD-Spp1 in SAE2 and sae2∆ cells. (B) Analysis of DSB formation in the GAL1-GAL7 region (GAL7 probe). Asterisks indicate the position of the DSBs induced by GBD-Spp1.

A simple interpretation of the above results is that Spp1 recruits a component of the DSB machinery. To search for interaction partners, we used the full-length Spp1 protein to perform yeast two-hybrid screening (16). We readily identified Set1 and Mer2 as prominent interactors of Spp1 (table S3). We mapped the Spp1-interacting region of Mer2 to the central part of the protein (residues 105 to 172) (fig. S5A). We validated the Mer2-Spp1 interaction by glutathione S-transferase–Mer2 pull-down experiments using various in vitro–translated Spp1 polypeptides. We found that the 131–amino acid C-terminal region of Spp1 was required for interaction with Mer2 (fig. S5B). Finally, we used tagged versions of both proteins to examine the in vivo interaction of Spp1 and Mer2 during meiosis. The C-terminally tagged SPP1-HA3 strain showed a level of H3K4me3 near that of the wild type and a slight reduction (35%) of DSB formation at the BUD23 hotspot, but, nevertheless, wild-type (WT) spore viability, indicating efficient DSB formation throughout the genome (fig. S6 and table S2). Spp1-HA3 was efficiently coimmunoprecipitated with Mer2-myc9 during meiosis, with a peak at 4 hours (Fig. 3A). Deoxyribonuclease and phosphatase treatments of the Myc immunoprecipitated proteins from the double-tag strain did not reduce the recovery of Spp1-HA (fig. S5, B and C), suggesting an interaction of Spp1 with the nonphosphorylated form of Mer2 (5, 7, 9). A schematic representation of the Spp1-Set1 and Spp1-Mer2 interaction domains is illustrated in fig. S5D.

Fig. 3

Spp1 interacts with Mer2. (A) (Top) DNA replication monitored by fluorescence-activated cell sorting. (Bottom) Protein extracts (INPUT) and myc-immunoprecipitated proteins (IP myc) from a Spp1-HA3 Mer2-myc9 strain were analyzed by Western blot with an antibody to myc (anti-myc) (upper gel) and reprobed with an anti-HA antibody (lower gel) (16). Arrows indicate the positions of Mer2-myc9 (the asterisk denotes the phosphorylated form) and Spp1-HA3. A strain expressing only Spp1-HA3 was used as negative control. IgG, immunoglobulin G. (B) GBD, GBD-SET1, and GBD-SPP1 strains expressing Mer2-myc9 were harvested after 0 and 4 hours in SPM. Binding of the GBD-fusion proteins and Mer2-myc9 at the GAL2, GAL7, and GAL10 UASs was analyzed by chromatin immunoprecipitation (ChIP) with anti-GBD and anti-myc antibodies. Polymerase chain reaction (PCR) primers specific for each GALUAS and TELXV-L (control) were used to amplify immunoprecipitated DNAs (16). (C) Chromatin binding of Mer2-myc9 in the GBD-SPP1 strain was analyzed by ChIP–quantitative PCR after 4 hours in SPM along the genomic region comprising GAL7, GAL10, and GAL1 at positions indicated by two convergent arrowheads. The asterisk indicates the Gal1 ucut promoter. Error bars represent SDs from three independent experiments.

The essential role of Mer2 in DSB formation and its interaction with Spp1 prompted us to determine whether Mer2 was present at the Spp1-tethered DSB sites. At the control CTR86 and SED4 regions known to associate with Mer2 in WT cells (9), we found that Mer2 was enriched in the strains expressing GBD and GBD-Spp1 (fig. S7A), suggesting that the expression of these GBD fusion proteins did not alter chromatin occupancy of Mer2. Four hours after transfer to the sporulation media (SPM), Mer2 was strongly enriched at the UASGAL sites in the strain expressing GBD-Spp1, but not in the strains expressing GBD alone (Fig. 3B). Importantly, Mer2-myc was also enriched in the vicinity of the Gal1 ucut promoter, where the major DSB site is detected (Fig. 3C). This enrichment appears to be specific, as its level is greater than the one expected to spread for the UAS sites.

Examination of the genetic requirements for DSB formation indicated that the cleavage site induced by the tethering of Spp1 (i) was independent of Set1, (ii) required Mer2, and (iii) did not result from Spp1 overexpression (fig. S7B). The previously described Spp1 zinc finger–like domain (SZF) was included in the Mer2-interacting region (21). We deleted the SZF canonical CXXC motif in GBD-Spp1 (GBD-Spp1∆CXXC) and tested whether deleting this evolutionarily conserved motif impairs DSB formation and Mer2 binding at GAL10 (16). DSB formation and Mer2 binding were both strongly reduced at GAL10, providing a strong correlation between the ability of GBD-Spp1 to recruit Mer2 at GAL10 and DSB formation (fig. S8). This finding outlines the importance of this interaction for DSB formation.

What is the role of H3K4 methylation in the control of DSB formation? We found that the amount of Spp1 was strongly reduced both in the absence of Set1 and in COMPASS mutants affecting the stability of Set1, but either remained normal in the catalytically inactive set1G951S mutant or was slightly reduced in the H3K4R, bre2∆, and sdc1∆ mutants (fig. S9, A and B). Therefore, decreased levels of Spp1 might only partially explain the DSB formation defect of mutants that are compromised in H3K4 methylation. Our interpretation is that the function of Spp1 in DSB formation depends on H3K4 methylation through an interaction of its PHD-finger with H3K4me2/3 (21). We therefore deleted the PHD domain of Spp1 (16) and tested whether the presence of the PHD domain was important for H3K4me3 and DSB formation. We detected a level of H3K4me3 near that of the wild type in the spp1∆PHD mutant (Fig. 4A), whereas DSB formation was clearly reduced at the BUD23 and CYS3 hotspots (Fig. 4B). We conclude that the Spp1 PHD domain itself plays a specific role in DSB formation.

Fig. 4

The PHD-finger domain of Spp1 regulates meiotic DSB formation. (A) Analysis of H3K4me3 levels (16). (B) DSB formation at the BUD23 and CYS3 hotspots. (C) Extension of the chromatin loop-axis model. The tethering of a H3K4me-rich region to the chromosomal axis via the Spp1-Mer2 interaction allows Spo11 cleavage at a proximal nucleosome-depleted region (NDR). The red circles labeled with “m” indicate di- and trimethylated H3K4.

Our work led us to propose an enriched chromatin loop-axis model (Fig. 4C) for the regulation of DSB formation that addresses how the meiotic DSB sites are mechanistically selected. We propose that the interaction between the COMPASS subunit, Spp1, and Mer2 brings potential meiotic DSB sites to the chromosome axis for further downstream events that will ultimately lead to Spo11-dependent DSB formation at axis-proximal regions that are depleted of nucleosomes. In mammals, H3K4 methylation has been reported to be enriched at recombination hotspots where the meiosis-specific PRDM9 H3K4 methyltransferase is known to act (2224). How PRDM9 connects to the mammalian DSB machinery is not known, but, as in yeast, it may be the consequence of a direct interaction with a protein of the DSB machinery. Our results indicate that H3K4me3 is required for the function of Spp1, probably through its recognition by the PHD domain within Spp1, and this requirement can be bypassed by tethering Spp1 to the DNA locus. The broadly localized H3K4me3 modification has the virtue of permitting the initiation of recombination at numerous places of the genome, a molecular strategy that ensures a large diversity of recombinant haplotypes to be transmitted by the gametes. In conclusion, this model offers a clue of how chromosome structure and DSB regulation are interrelated, and it attributes a pivotal role to Spp1 in the recruitment of components acting at meiotic DSB sites to the chromosomal axis.

Supplementary Materials

www.sciencemag.org/cgi/content/full/science.1225739/DC1

Materials and Methods

Figs. S1 to S9

Tables S1 to S5

References (2631)

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
  2. Acknowledgments: We thank F. Klein and V. Borde for materials, J. B. Boulé for his help, and S. Kowalczykowski for reading of the manuscript. Work in the V.G. and A.N. laboratories is supported by the “Ligue contre le Cancer” (Equipe Labelisées). Work in the laboratory of B.D. was supported by the Swiss National Science Foundation. L.A. was supported by grants from the Fondation ARC and L.Sz. by the European Union and the Hungarian Research Fund.
View Abstract

Navigate This Article