DNA Replication Origin Function Is Promoted by H3K4 Di-methylation in Saccharomyces cerevisiae
Lindsay F. Rizzardi, Elizabeth S. Dorn, Brian D. Strahl, Jeanette Gowen Cook

Abstract

DNA replication is a highly regulated process that is initiated from replication origins, but the elements of chromatin structure that contribute to origin activity have not been fully elucidated. To identify histone post-translational modifications important for DNA replication, we initiated a genetic screen to identify interactions between genes encoding chromatin-modifying enzymes and those encoding proteins required for origin function in the budding yeast Saccharomyces cerevisiae. We found that enzymes required for histone H3K4 methylation, both the histone methyltransferase Set1 and the E3 ubiquitin ligase Bre1, are required for robust growth of several hypomorphic replication mutants, including cdc6-1. Consistent with a role for these enzymes in DNA replication, we found that both Set1 and Bre1 are required for efficient minichromosome maintenance. These phenotypes are recapitulated in yeast strains bearing mutations in the histone substrates (H3K4 and H2BK123). Set1 functions as part of the COMPASS complex to mono-, di-, and tri-methylate H3K4. By analyzing strains lacking specific COMPASS complex members or containing H2B mutations that differentially affect H3K4 methylation states, we determined that these replication defects were due to loss of H3K4 di-methylation. Furthermore, histone H3K4 di-methylation is enriched at chromosomal origins. These data suggest that H3K4 di-methylation is necessary and sufficient for normal origin function. We propose that histone H3K4 di-methylation functions in concert with other histone post-translational modifications to support robust genome duplication.

DNA replication initiates at discrete genomic loci termed “origins of replication.” Each eukaryotic chromosome is replicated from many individual origins to ensure complete and precise genome duplication during each cell division cycle. Individual origins vary both in the likelihood that they will initiate replication, or “fire,” in any given S phase and in the firing time within the S phase (Weinreich et al. 2004). Highly efficient origins fire in most cell cycles, whereas inefficient origins fire in only some cycles and are usually passively replicated by forks emanating from neighboring efficient origins. Although highly efficient origins that support initiation in most cell cycles have been identified in many genomes, the chromosomal determinants of origin location and function are still incompletely understood. Strikingly, while DNA sequence elements can be necessary, it is clear that sequence alone is insufficient to fully specify eukaryotic origin location and activity (Méchali 2010).

Like all DNA-templated processes, replication occurs on chromatin. Recent progress in the field has demonstrated that the chromatin structure surrounding origins plays an essential role in controlling origin activity. For example, the positioning of nucleosomes near origins can either stimulate or inhibit origin function (Simpson 1990; Crampton et al. 2008; Berbenetz et al. 2010; Eaton et al. 2010). The major protein components of chromatin, the histone proteins, can also be post-translationally modified by acetylation, methylation, phosphorylation, ubiquitination, and sumoylation (Kouzarides 2007). These modifications can alter DNA accessibility and serve as recognition sites for other proteins. Importantly, several individual histone modifications affect aspects of origin function. For example, acetylation of histones H3 and H4 accelerates the timing of origin firing within S phase and can increase origin efficiency (Aggarwal and Calvi 2004; Espinosa et al. 2010; Unnikrishnan et al. 2010). In addition, histone H3 lysine 36 mono-methylation (H3K36me1) by the Set2 methyltransferase has been implicated in the recruitment of the replication initiation protein, Cdc45 (Pryde et al. 2009). In metazoan genomes, PR-Set7-catalyzed histone H4 lysine 20 mono-methylation (H4K20me1) stimulates the loading of the core replicative helicase (Tardat et al. 2007, 2010).

It is clear that no single histone modification is absolutely required for origin function since loss of individual histone-modifying enzymes does not impact cell viability. This observation suggests that a combination of histone modifications facilitates efficient DNA replication in the form of a “histone code” similar to the combinations known to regulate transcription (Strahl and Allis 2000). While some elements of this code have been identified (e.g., H3 and H4 acetylation, H4K20 mono-methylation, and H3K36 mono-methylation), the complexity of DNA replication led us to hypothesize that additional histone modifications that impact origin activity remain to be discovered. We therefore sought to identify those histone modifications and chromatin modifiers that are integral to this process. We conducted a genetic screen to identify histone modifications that are required for the fitness of a hypomorphic replication yeast mutant, cdc6-1. This screen revealed a previously unidentified positive DNA replication role for histone H3 lysine 4 (H3K4) methylation by the COMPASS complex, and our subsequent analysis indicates that H3K4 di-methylation is the relevant modification for this activity. These findings contribute to elucidating the pattern of chromatin features that determines origin activity in eukaryotic genomes.

Materials and Methods

Yeast strains and growth conditions

The Saccharomyces cerevisiae strains used in this study are listed in Table 1, and any additional genotype information is available upon request. Construction of de novo gene deletion strains was performed by PCR-mediated disruption, and some double-mutant construction was performed by mating as indicated in Table 1.

View this table:
Table 1  Strains used in this study

Plasmids

All plasmids used in this study are listed in Table 2.

View this table:
Table 2  Plasmids used in this study

Synthetic genetic array screen

Synthetic genetic array (SGA) analysis was carried out as previously described (Tong and Boone 2006). Briefly, 63 deletion strains (Table 3) were mated to the temperature-sensitive cdc6-1 strain (JCY332), and haploids carrying both mutations were isolated by growth on selective media. All of the deletion strains originated from the Yeast Knock Out library (Open Biosystems) except strains lacking SET1 or DOT1; the set1Δ strains were created de novo while the dot1Δ strain was previously published (Gardner et al. 2011). Additionally, SET1 and BRE1 deletions were recreated de novo in the cdc6-1 mutant in the BY4741 background (yLF058). All of the resulting double mutants were spotted in fivefold serial dilutions with an initial OD600 of 0.5 onto YPD, grown for 3 days at 32°, and growth was compared to that of the cdc6-1 single mutant. Double mutants displaying a synthetic growth phenotype were confirmed by analyzing three independent isolates. The fold change in growth is denoted by a score from 1 to 3 indicating an approximate 5- to 125-fold change compared to cdc6-1 alone. Negative values indicate growth defects, while positive values indicate enhanced growth or rescue. No genetic rescue was observed in any double-mutant strain.

View this table:
Table 3  Results from SGA screen for genetic interaction with cdc6-1

Minichromosome maintenance assays

Minichromosome (or plasmid) maintenance assays were performed as described previously (Tye 1999). Briefly, yeast strains containing YCplac33, YCplac111, or YCplac33+2XARS209 were grown to log phase in the appropriate selective media and 100–200 cells were plated on both selective and nonselective media to establish an initial percentage of plasmid-bearing cells. These cultures were also diluted to a concentration of 1 × 105 cells/ml in 5 ml of nonselective media and grown for 8–10 generations before once again plating on both selective and nonselective media. Precise generation numbers were calculated using the following formula: n = log(CF/CI)/log(2), where CF represents the final number of cells as measured by OD600 and CI represents the starting cell number of 105 cells/mL. After 2 days of growth, colonies were counted, and the plasmid loss rate (L) per generation (n) was calculated using the following formula: L = 1 − (%F/%I)(1/n), where %F is the final percentage of cells that retained the plasmid and %I is the initial percentage of cells that contain the plasmid.

Immunoblotting

Whole-cell extracts were prepared by extraction with trichloroacetic acid (TCA). Cell growth was halted by the addition of TCA to a final concentration of 5%, and the cell pellets were frozen at −80°. Pellets were resuspended in 200 μl TCA buffer (10 mM Tris–HCl, pH 8.0, 10% TCA, 25 mM NH4OAc, 1 mM EDTA) and broken by glass-bead lysis. Proteins were precipitated by centrifugation, resuspended in 100 μl resuspension buffer (0.1 M Tris–HCl, pH 11.0, 3% SDS), and boiled for 5 min. Samples were centrifuged, and the supernatant was quantified using the Dc Assay (BioRad). Equal concentrations of lysates were loaded onto 15% SDS-PAGE gels and transferred onto PVDF. The following antibodies were used: anti-H3 (1:10,000; ActiveMotif 39163), anti-H3K4me1 (1:2000; Millipore 07-436), anti-H3K4me2 (1:2000; Abcam 32356), anti-H3K4me3 (1:10,000; gift from M. Bedford), anti-H2B (1:5000; ActiveMotif 39237), and anti-LexA (1:5000; Millipore 06-719).

Chromatin immunoprecipitation

Yeast strains were cross-linked with 1% formaldehyde for 15 min at room temperature and quenched with 250 mM glycine for 5 min at room temperature. Forty OD600 units of cross-linked cells were harvested by centrifugation and washed thoroughly, and the pellets were stored at −80°. The cell pellets were resuspended in 300 mM FA-lysis buffer (50 mM HEPES-KOH, pH 7.5, 300 mM NaCl, 1 mM EDTA, 1% TritonX-100, 0.1% Na-deoxycholate) with protease inhibitors and broken by glass-bead lysis, and fixed chromatin was sheared by sonication using a Branson sonifier 250. Average DNA fragment lengths were 100–300 bp determined by gel analysis. After centrifugation and quantification via Bradford Assay (BioRad), 0.5 mg of soluble chromatin was incubated with 2 μl of antibody [anti-H3 (ActiveMotif), anti-H3K4me2 (Abcam), or anti-H3K4me3 (Millipore)] in 1.5-ml tubes overnight at 4° and immunoprecipitated with 10 μl Protein A Dynabeads (Invitrogen) for 1 h at 4°. The beads were washed sequentially with 300 mM FA-lysis buffer, twice with 500 mM FA-lysis buffer (50 mM HEPES-KOH, pH 7.5, 500 mM NaCl, 1 mM EDTA, 1% TritonX-100, 0.1% Na-deoxycholate), once with LiCl solution (10 mM Tris–HCl, pH 8.0, 250 mM LiCl, 0.5% NP-40, 0.5% Na-deoxycholate, 1 mM EDTA), and once with TE, pH 8.0. After washing, the chromatin was eluted from the beads in 200 μl elution buffer (0.1 M NaHCO3, 1% SDS) for 30 min at room temperature. The eluted material was treated with RNAse A and Proteinase K before de-cross-linking at 65° overnight. The DNA was purified using Genesee UPrep spin columns and eluted in 100 μl water. Immunoprecipitation (IP) samples and IP controls (set1Δ) were used undiluted while input samples were diluted 1:10. Samples were analyzed by quantitative PCR on the ABI 7900 HT (AppliedBiosystems) using SYBR Green master mix with Rox (Fermentas). Primer sequences are available upon request. Signals from the immunoprecipitates are reported as a percentage of the input and normalized to H3. Error bars represent the standard deviations of the average signals between experiments (n ≥ 3).

Results

Identification of histone modifiers that promote DNA replication

DNA replication origins in the budding yeast S. cerevisiae are defined by both sequence elements and local chromatin structure. Although DNA replication is essential for cell proliferation, the majority of histone modifications and chromatin-modifying enzymes are not individually required for yeast cell viability. This observation supports the model that a combination of histone modifications supports replication origin function. To identify new histone modifications that contribute to this combination, we conducted a genetic screen. We reasoned that individual chromatin elements that influence replication activity would be revealed as genetic suppressors or enhancers of cell growth in a strain bearing a hypomorphic mutation in an essential replication gene, CDC6.

The Cdc6 ATPase plays an essential role at origins in loading the replicative helicase complex composed of MCM2-7 (Bell and Dutta 2002; Takahashi et al. 2002). The cdc6-1 mutant harbors a G260D mutation in the catalytic domain resulting in failure to load MCMs at restrictive temperatures (Feng et al. 2000). Yeast cells harboring the cdc6-1 mutation produce a Cdc6 protein that functions normally at 29° and is nonfunctional at temperatures >34°, but retains partial function at intermediate temperatures between 30° and 33° (Feng et al. 2000). To identify suppressors or enhancers of cdc6-1, we deleted genes for most of the known histone modifiers, chromatin remodelers, and histone chaperones (63 total; see Table 3) in the cdc6-1 temperature-sensitive replication mutant strain. Double-mutant strains were tested for fitness at semipermissive temperatures and compared to the parent single-mutant strains (Tong and Boone 2006). The majority of double-mutant strains grew neither better nor worse than their respective parents under any of the tested growth conditions, and no null alleles improved the growth of the cdc6-1 mutant (Table 3). Although a role for the histone acetyltransferase (HAT) Gcn5 in DNA replication has been shown (Espinosa et al. 2010), it was not included in this screen because the null mutant strain has a slow-growth phenotype that would complicate interpretation of the double-mutant phenotype.

In contrast, 21 of the null alleles (including a positive control, tom1Δ) impaired growth in the cdc6-1 strain but had little effect in otherwise wild-type backgrounds. These genes represent a wide array of chromatin factors including HATs, histone deacetylases (HDACs), and histone methyltransferases (HMTs) (Table 3). Interestingly, many of these factors contributed either directly or indirectly to a single histone modification, H3K4 methylation, which is deposited by Set1, the catalytic subunit of the COMPASS complex (Figure 1A). H3K4 is mono-, di-, and tri-methylated by Set1 as part of the COMPASS complex. COMPASS consists of seven additional subunits that promote complex integrity (Swd1, Swd2, Swd3, Bre2), regulate catalytic activity (Sdc1 and Spp1), or have unknown function (Shg1) (Takahashi and Shilatifard 2010; Mersman et al. 2011; Takahashi et al. 2011). Unlike other COMPASS members, Swd2 also functions as part of the APT transcription termination complex, and its role in this complex is essential for cell viability (Soares and Buratowski 2012). For this reason, Swd2 was not included in our screen. The absence of Set1, Swd1, Swd3, Bre2, or Sdc1 impaired the growth of the cdc6-1 strain (Table 3).

Figure 1 

COMPASS and H3K4 are required for robust growth of the temperature-sensitive cdc6-1 replication mutant. (A) COMPASS complex model adapted from Takahashi et al. (2011). An asterisk denotes a complex member that genetically interacts with cdc6-1. (B) Wild-type (BY4741), cdc6-1 (yLF058), set1Δ (yLF062), and cdc6-1 set1Δ (yLF063) were transformed with an empty vector [pGLx2], a vector producing LexA-tagged wild-type Set1 [pGLx2-SET1], or catalytically dead Set1 [pGLx2-set1-H1017K] from a GAL1 promoter construct or a vector producing normal Cdc6 [pRS316-CDC6] from the CDC6 promoter as indicated. Fivefold serial dilutions were spotted onto SC-URA containing 1% galactose/2% raffinose and grown at the indicated temperatures for 4 days. (C) The cdc6-1 mutation was introduced into the H3-H4 “shuffle” strain (DY7803) transformed with HHT2 or hht2-K4R plasmids. Fivefold serial dilutions were spotted onto YPD and grown at the indicated temperatures for 3 days. (D) Fivefold serial dilutions of wild-type (YMS196), cdc6-1 (JCY332), cdc6-1 swd1Δ (yLF114), cdc6-1 bre2Δ (yLF120), bre2Δ (yLF060), or swd1Δ (yLF061) were spotted onto YPD and grown at the indicated temperatures for 3 days.

COMPASS activity and H3K4 methylation promote DNA replication

We confirmed the enhancer phenotype of the SET1 deletion strain by constructing a set1Δ allele de novo in the cdc6-1 parent strain. Growth of the cdc6-1 strain was only slightly impaired at 31° compared to wild-type or the cdc6-1 strain harboring wild-type CDC6 on a plasmid, but growth was substantially impaired when SET1 was deleted in this strain (Figure 1B). Expression of wild-type SET1, but not of the catalytically dead mutant set1-H1017K, rescued the synthetic growth defect of the set1Δ cdc6-1 strain (Figure 1B). Importantly, the synthetic growth defect of the set1Δ cdc6-1 strain was recapitulated in a cdc6-1 strain in which the only copy of histone H3 bears the K4R (unmethylatable) mutation (Figure 1C). These findings indicate that the catalytic activity of Set1 is important for robust growth of the cdc6-1 replication mutant.

Set1 functions as the catalytic subunit of the COMPASS complex. We hypothesized that other members of this complex would display similar phenotypes when deleted in the cdc6-1 strain. Newly constructed deletions of BRE2, SDC1, SWD1, and SWD3 each impaired the growth of the cdc6-1 mutant at semipermissive temperatures, but not deletion of SPP1 or SHG1 (Figure 1D and Table 3). Bre2, Sdc1, Swd1, Swd2, and Swd3 (along with Set1) are the core structural components of the COMPASS complex required for full activity (Takahashi et al. 2011). These results further support the conclusion that COMPASS enzymatic activity and H3K4 methylation are important for proliferation when Cdc6 is crippled.

The poor growth of these double-mutant strains could be due to a general exacerbation of the replication defect caused by Cdc6 perturbation, or it could reflect a specific interaction between Cdc6 and H3K4 methylation. If Set1 and H3K4 methylation are generally important for efficient DNA replication, then we would expect similar proliferation defects from deleting SET1 in other replication mutant strains. To test this prediction, SET1 was deleted in two temperature-sensitive cdc7 mutants and one temperature-sensitive cdc45 mutant. These replication factors are required for origin firing at the G1/S transition, downstream of Cdc6 activity (Tercero et al. 2000; Labib 2010). Similar to the effect of deleting SET1 in the cdc6-1 strain, loss of SET1 in the cdc7-1, cdc7-4, and cdc45-27 mutants impaired growth at semipermissive temperatures (Figure 2A).

Figure 2 

H3K4 methylation is required for robust growth of multiple replication mutants. (A) Fivefold serial dilutions of the meiotic progeny from the cross of cdc7-1 (TSQ880), cdc7-4 (TSQ131), or cdc45-27 (TSQ694) with set1Δ (yLF062) were spotted onto YPD and grown at the indicated temperatures for 3 days. (B) Fivefold serial dilutions of wild-type (RUY121), ORC6-rxl CDC6-ΔNT (RUY028), swd1Δ (yLF051), and ORC6-rxl CDC6-ΔNT swd1Δ (yLF049) were spotted onto YP containing 2% dextrose (no re-replication) or galactose (re-replication induced) and grown for 2 days at 30°.

The poor growth of cdc6-1 set1Δ cells suggests that Set1 promotes replication; we thus predicted that loss of H3K4 methylation in a hypermorphic replication mutant would at least partially rescue the adverse phenotypes of that mutant. To test this hypothesis, we introduced the SET1 null allele into the hypermorphic replication mutant RUY028. This yeast strain harbors two mutations that deregulate replication origin licensing, resulting in re-replication, an aberrant phenomenon in which some origins fire more than once per cell cycle, leading to DNA damage and poor cell growth. The ORC6-rxl mutation prevents inhibitory phosphorylation of the Orc6 subunit of the origin recognition complex (ORC) by CDK, and the GAL1 pr-CDC6-ΔNT allele produces a hyperstable Cdc6 protein (Archambault et al. 2005). In this strain, re-replication is induced during growth on galactose, which induces transcription of the CDC6-ΔNT allele. As expected, loss of H3K4 methylation upon deletion of SWD1 or BRE1 partially rescues the poor growth of the re-replicating strain on galactose (Figure 2B). Together, these data provide strong evidence that H3K4 methylation promotes efficient DNA replication, most likely through regulation of origin activity.

A positive role for H3K4 methylation in replication suggests that this histone modification is enriched near origins of replication. To test this idea directly, we performed chromatin immunoprecipitation assays for both di- and tri-methylated H3K4 in asynchronous wild-type yeast cultures. Analysis of several genomic loci revealed that both H3K4 methylation states are enriched at both an early and a late-firing origin of replication (ARS315 and ARS822, respectively) relative to a nonorigin, telomere-proximal region (Figure 3).

Figure 3 

H3K4 methylation is present at replication origins. Chromatin immunoprecipitation experiments were performed on asynchronous wild-type (BY4741) or set1Δ (yLF062) strains grown at 30°. Immunoprecipitates using antibodies to total histone H3, di-methylated H3K4 (H3K4me2) in A and tri-methylated H3K4 (H3K4me3) in B were analyzed by quantitative PCR for chromosomal DNA fragments from a region near telomere VI-R and two replication origins, ARS315 and ARS822. Error bars represent the standard deviations of n ≥ 3 biological replicates. Significant enrichment of H3K4 methylation at origins compared to telomere VI-R was determined using the Student’s unpaired t-test (*P < 0.05).

The Rad6/Bre1 ubiquitin ligase complex promotes DNA replication

Methylation of histone H3K4 by the COMPASS complex requires prior mono-ubiquitination of histone H2B at lysine 123 (H2BK123) by the Rad6/Bre1 ubiquitin ligase complex (Lee et al. 2007; Nakanishi et al. 2009). Our original screen detected a strong synthetic growth defect when the bre1Δ library mutant was crossed with the cdc6-1 parent strain. De novo deletion of either BRE1 or RAD6 in the cdc6-1 mutant resulted in an ∼125-fold decrease in growth at semipermissive temperatures, validating this genetic interaction (Figures 4, A and B). Ectopic expression of wild-type BRE1 under control of its own transcriptional promoter fully rescued the exacerbated growth phenotype, but expression of a catalytically dead Bre1 (bre1-H665A) did not (Figure 4A). Moreover, BRE1 deletion also partially rescued the growth impairment of the re-replicating RUY028 strain similar to deletion of SET1 (Figure 4C). These data suggest that the Rad6/Bre1 ubiquitin ligase complex is also required for efficient DNA replication, likely through its involvement in promoting H3K4 methylation.

Figure 4 

H2BK123 mono-ubiquitination promotes robust growth of the temperature-sensitive cdc6-1 replication mutant. (A) Wild-type (BY4741), bre1Δ (yLF151), cdc6-1 (yLF058), and cdc6-1 bre1Δ (yLF150) were transformed with either an empty vector [pRS315], a vector expressing BRE1 [pRS315-9xMyc-BRE1], or bre1-H665A [pRS315-9xMyc-bre1-H665A] from the native BRE1 promoter. Fivefold serial dilutions were spotted onto SCD-LEU and grown for 3 days at the indicated temperatures. (B) Fivefold serial dilutions of wild-type (YMS196), cdc6-1 (JCY332), rad6Δ (yLF154), and cdc6-1 rad6Δ (yLF117) were spotted onto YPD and grown for 3 days at the indicated temperatures. As previously reported, the rad6Δ strain is cold sensitive at temperatures <30° (McDonough et al. 1995). (C) Wild-type (RUY121), ORC6-rxl CDC6-ΔNT (RUY028), bre1Δ (yLF052), and ORC6-rxl CDC6-ΔNT bre1Δ (yLF050) were spotted onto YP containing 2% dextrose or 1% galactose (re-replication induced) and grown for 3 days at 30°.

H3K4 methylation is required for efficient minichromosome maintenance

To assess DNA replication more specifically than general cell proliferation, we used a plasmid loss assay that measures the ability of cells to maintain a minichromosome bearing a centromere and a single replication origin (Maine et al. 1984). Growth in nonselective medium for several cell divisions allows cells that failed to initiate replication of the minichromosome to survive, and these cells are then counted as colonies on nonselective medium. Strains lacking either SET1 or BRE1 displayed significantly higher plasmid loss rates per cell division than wild-type strains did (Figure 5A, P < 0.05). Similar to the phenotype resulting from SET1 deletion, cells expressing the mutant histone H3K4R displayed a significant increase in plasmid loss rate (Figure 5B).

Figure 5 

H3K4 methylation is required for efficient origin-dependent mini chromosome maintenance. (A) Plasmid loss rates of a single-origin-bearing plasmid [YCpLac33] were measured in a wild-type strain (BY4741) and in strains lacking SET1 (yLF089), BRE1 (yLF151), or DOT1 (YNL037). Loss rates are reported per cell division. (B) Plasmid loss rates of a single-origin-bearing plasmid [YCpLac111] were measured in the histone “shuffle” strain (DY7803) transformed with plasmids expressing wild-type HHT2, hht2-K4R, or hht2-K79R. (C) Plasmid loss rates of plasmids bearing either a single origin [YCpLac33] or three origins [YCpLac33 + 2X ARS209] were measured. For all experiments, the average loss rates were obtained from at least three independent transformants, and the error bars indicate standard deviations. Statistics were performed using the Student’s unpaired t-test (*P < 0.05).

Notably, H2B mono-ubiquitination by Rad6/Bre1 is a prerequisite not only for H3K4 methylation, but also for H3K79 methylation (Nakanishi et al. 2009). Neither deletion of DOT1, the histone H3K79 methyltransferase, nor mutation of H3K79 to arginine had any effect on plasmid maintenance (Figure 5, A and B). Moreover, loss of Dot1 did not impair the growth of the cdc6-1 strain (Table 3). Efficient growth and plasmid maintenance by these strains indicate that the elevated plasmid loss rate of the BRE1 null strain is likely due to the subsequent loss of H3K4 methylation rather than loss of H3K79 methylation. Furthermore, these results demonstrate that inefficient minichromosome maintenance is not a universal phenotype of strains lacking histone-modifying enzymes or expressing mutant histones.

If these minichromosome maintenance defects are specific to perturbations of origin function and not chromosome segregation or expression of the selectable marker, then adding multiple origins to the test plasmid should rescue the elevated loss rates of the set1Δ and bre1Δ strains. Additional origins multiply the chances for a successful origin initiation event on the plasmid, and success at any one origin allows replication and transmission to both daughter cells. This property has been used by others to validate origin-specific phenotypes (Hogan and Koshland 1992). We modified the single ARS1-containing plasmid by adding two copies of ARS209. As before, the plasmid harboring only one origin was lost more frequently from set1Δ and bre1Δ strains than from wild-type strains, but this effect was reversed with the addition of multiple origins (Figure 5C). This result supports the conclusion that H2BK123 mono-ubiquitination by Bre1 and the consequent H3K4 methylation by Set1 promote DNA replication origin function in yeast.

H3K4 di-methylation is required for efficient origin function

The Set1/COMPASS complex is responsible for mono-, di-, and tri-methylation of H3K4. These methylation states are differentially distributed over gene bodies (Pokholok et al. 2005), and several studies have shown that the different methylation states of H3K4 and H3K36 can have different functions (Taverna et al. 2006; Pinskaya et al. 2009; Pryde et al. 2009). The individual H3K4 methylation states depend on other histone residues and are influenced by the presence or absence of individual COMPASS subunits (Chandrasekharan et al. 2010; Mersman et al. 2011). To gain insight into which H3K4 methylation states were required for efficient DNA replication, we measured plasmid loss rates in strains lacking different COMPASS complex members. Like the set1Δ strain (Figure 5A), both swd1Δ and bre2Δ strains displayed significantly elevated plasmid loss compared to their isogenic wild-type counterparts (Figure 6A). These two proteins are both required for Set1 stability (Takahashi et al. 2011), and in their absence no H3K4 methylation is detectable (Figure 6B). Unlike Swd1 and Bre2, Spp1 is a COMPASS subunit that is required for H3K4 tri-methylation, but not mono- or di-methylation (Takahashi et al. 2009) (Figure 6B); loss of SPP1 had no effect on plasmid maintenance (Figure 6A). This result is consistent with our earlier observation that Spp1 loss did not affect proliferation of the cdc6-1 strain (Table 3). Taken together, these data suggest that H3K4 di-methylation is sufficient for proper origin function and that H3K4 tri-methylation is dispensable.

Figure 6 

H3K4 di-methylation promotes efficient mini-chromosome maintenance. (A) Plasmid loss rates of the single-origin-bearing plasmid [YCpLac33] were measured for wild-type (BY4741), swd1Δ (yLF061), bre2Δ (yLF060), and spp1Δ (yLF153). (B) Immunoblot analysis of whole-cell extracts (from strains shown in A) using antibodies to total H3, H3K4me1, H3K4me2, and H3K4me3. (C) Plasmid loss rates of the single-origin-bearing plasmid [YCpLac111] were measured for the H2B “shuffle” strain (FY406) transformed with plasmids expressing HTB1 [pZS145], htb1-K123R [pZS146], htb1-R119D [pZS473], or htb1-R119A [pZS145-R119A]. (D) Immunoblot analysis of whole-cell extracts (from strains shown in C) using antibodies specific for total H2B, H3, and H3K4 methylation as in B. All plasmid loss data represent the mean and standard deviation of at least three independent transformants. Statistics were performed using the Student’s unpaired t-test (*P < 0.05).

To examine the role of H3K4 di-methylation without perturbing the COMPASS complex, we took advantage of H2B mutants that were previously shown to differentially affect H3K4 methylation (Chandrasekharan et al. 2010). Mutation of H2BK123 to arginine abolished mono-ubiquitination of this residue and consequently eliminated both H3K4 di- and tri-methylation (Figure 6D). Importantly, similar to loss of BRE1, this mutation induced a significantly higher plasmid loss rate than wild type (Figure 6C). This result, in combination with the failure of the Bre1 catalytically dead mutant to complement the growth defect of the bre1Δ cdc6-1 mutant, strongly argues that the H2BK123 ubiquitination function of Bre1 is required for full origin function. Additionally, it supports the importance of H3K4 di-methylation in this process.

A previous study reported that mutational alteration of H2B arginine 119 to alanine does not prevent H2BK123 mono-ubiquitination, but does affect the degree of H3K4 methylation (Chandrasekharan et al. 2010). As previously reported, changing H2BR119 to alanine (H2BR119A) reduces the chromatin association of Spp1 and therefore H3K4 tri-methylation, whereas changing H2B119 to aspartic acid (H2BR119D) results in loss of H3K4 di- and tri-methylation to the same extent as eliminating H2BK123 mono-ubiquitination (Chandrasekharan et al. 2010). Our analysis of these histone H2B mutants revealed that strains expressing htb1-R119D displayed elevated plasmid loss rates similar to those of the htb1-K123R strain (Figure 6C). Mutation of R119 to alanine abolished H3K4 tri-methylation as expected, while having only a moderate effect on H3K4 di-methylation (Figure 6D). This H2B mutant showed a plasmid loss rate similar to the wild-type HTB1 strain. The strict correlation between the ability to produce H3K4 di-methylation and normal plasmid maintenance underscores the importance of H3K4 di-methylation, but not necessarily H3K4 tri-methylation, for origin function (Figure 6C).

Our initial observation that loss of SET1 in a cdc6-1 mutant strain resulted in a proliferation defect, coupled with the requirement for Set1 for proper minichromosome maintenance, clearly indicates that H3K4 methylation is necessary for robust DNA replication. To determine if H3K4 di-methylation is sufficient for DNA replication, we transformed the cdc6-1 set1Δ strain with plasmids directing the production of wild-type Set1, catalytically dead Set1 (Set1H1017K), or a Set1 protein lacking the RRM domain. The latter mutant can only di-methylate H3K4 (Schlichter and Cairns 2005). Each of these Set1 proteins was fused to a LexA tag for detection on immunoblots (Figure 7B) and expressed from the GAL1 promoter on galactose-containing medium. Immunoblot analysis of the three H3K4 methylation states confirmed that the fusions generated the expected methylation states (Figure 7B). As before, production of active Set1 suppressed the growth phenotype caused by deleting SET1 in the cdc6-1 strain (Figure 7A). Importantly, production of Set1ΔRRM also fully rescued this proliferation defect without the ability to produce H3K4 tri-methylation (Figure 7B). These data demonstrate that H3K4 di-methylation is both necessary and sufficient for robust growth of the cdc6-1 mutant and suggest that this histone modification plays an important positive role in origin function.

Figure 7 

H3K4 di-methylation is sufficient for robust growth of the cdc6-1 mutant. (A) Wild-type (BY4741), cdc6-1 (yLF058), set1Δ (yLF062), and cdc6-1 set1Δ (yLF063) were transformed with an empty vector [pGLx2], a vector producing LexA-tagged Set1 [pGLx2-SET1], or Set1 lacking the RRM domain [pGLx2-set1-ΔRRM] from the GAL1 promoter. Fivefold serial dilutions were spotted onto SC-URA containing 1% galactose and grown at the indicated temperatures for 5 days. (B) Immunoblot analysis of whole-cell extracts from either wild-type (BY4741) or set1Δ (yLF062) strains transformed with empty vector [pGLx2] or vectors producing LexA-tagged normal (“WT”) Set1 [pGLx2-SET1], catalytically dead (“CD”) Set1 [pGLx2-set1-H1017K], or Set1 lacking the RRM domain (“RRM”) [pGLx2-set1-ΔRRM] from the GAL1 promoter. Blots were probed with antibodies specific for LexA, total H3, H3K4me3, H3K4me2, and H3K4me1. A single asterisk represents a likely degradation product unique to the positive control construct; a double asterisk represents a nonspecific band detected by the H3K4me1 antibody.

Discussion

This study documents a novel role for H3K4 di-methylation in DNA replication origin function. Yeast strains with hypomorphic mutations in multiple replication genes are highly dependent on H3K4 di-methylation for robust growth. In these replication mutants (cdc6-1, cdc7-1, cdc7-4, cdc45-27) under semipermissive conditions, replication activity is reduced to the minimum needed for normal growth, and any further reduction in active origins caused by loss of H3K4 di-methylation results in severely impaired proliferation. Additionally, normal propagation of a minichromosome containing a single origin requires H3K4 di-methylation even in an otherwise wild-type strain. Every mutation that prevents H3K4 di-methylation, including loss of the prerequisite histone H2BK123 mono-ubiquitination, causes a similar replication phenotype. These data provide clear evidence that H3K4 di-methylation is important for full origin function.

The Set1 histone methyltransferase and the Bre1 E3 ubiquitin ligase have well-established roles in regulating gene expression (Shukla et al. 2006; Mutiu et al. 2007). Nonetheless, several lines of evidence suggest that the phenotypes that we have detected are due to reduced replication activity as opposed to altered gene expression. First, many chromatin-modifying enzymes showed no synthetic growth phenotype when combined with the cdc6-1 mutation. Some of these non-interacting genes include those with much more profound effects on patterns of gene expression than Set1, suggesting that replication phenotypes are not a general outcome of perturbed gene expression control. Interestingly, even the Rpd3 histone deacetylase that regulates origin firing time within S phase (Knott et al. 2009) did not genetically interact with cdc6-1 (Table 3), implying that the synthetic growth phenotypes observed here are relatively specific for origin function and not origin firing time. Second, a genome-wide analysis identified only 55 transcripts that changed significantly in a set1Δ strain compared to a wild-type strain, and none of those genes are predicted to directly affect origin activity (Lenstra et al. 2011). Third, the mini-chromosome maintenance phenotypes associated with loss of H3K4 di-methylation were largely suppressed by the addition of extra origins to the test plasmid, indicating that the phenotypes are closely tied to origin function and not to other biological parameters. Fourth, we directly detected H3K4 di- and tri-methylation at two yeast origins. Moreover, our analysis of a published genome-wide H3K4 tri-methylation data set identified peaks of methylation distinct from nearby transcription-associated peaks (Radman-Livaja et al. 2010 and our unpublished observations). Finally, our observation that loss of H3K4 methylation exacerbates poor growth of a replication hypomorphic strain but suppresses the poor growth of an origin-firing hypermorphic strain indicates that the replication phenotypes reported here are most likely direct positive effects of H3K4 di-methylation at origins. Taken together, these data provide strong evidence that the role of H3K4 di-methylation in origin function is direct and separate from any indirect transcriptional effects.

DNA replication is an essential process for proliferation, yet neither Set1 nor Bre1 are essential gene products. Interestingly, very few null mutations in yeast chromatin-modifying enzymes show significant growth defects despite their importance for several essential processes, such as transcription, replication, and repair. The function of histone modifications in transcription control has been described as a “histone code” in which combinations of post-translational modifications promote or repress transcription at a given locus (Strahl and Allis 2000; Oliver and Denu 2010). In this model, no single histone modification generates an active or inactive promoter, and thus the effects of mutations that alter local chromatin structure are cumulative. We propose that the same concept applies to chromatin structure at DNA replication origins. Consistent with the idea that, like promoters, origins can accommodate elimination of a single positive element of chromatin structure, the kinetics of S-phase progression are normal in a set1Δ strain grown under standard conditions (data not shown). Also analogous to transcriptional control at promoters, we propose that histone H3K4 di-methylation is an important element of origin chromatin structure, but its absence alone does not severely inhibit origin function. In fact, origin activity is robust enough that even substantial reductions in expression of MCM proteins or deletion of many origins from a yeast chromosome causes no growth defects (Dershowitz et al. 2007; Ge et al. 2007; Blow et al. 2011). Nevertheless, the effect of H3K4 di-methylation loss over several cell cycles or in combination with other replication perturbations causes a significant loss of replication fitness.

How does H3K4 di-methylation promote DNA replication? We propose two models by which H3K4 di-methylation could facilitate DNA replication (Figure 8). The first is a recruitment model whereby a replication factor directly interacts with di-methylated H3K4 to associate with replication origins. This possibility is supported by precedent since it has been suggested that mono-methylation of H3K36 can recruit the replication initiation factor Cdc45 (Pryde et al. 2009). In addition, a member of the origin recognition complex, Orc1, contains a bromo-adjacent homology domain that mediates nucleosome binding in budding yeast and specifically binds to di-methylated H4K20 at origins in human cells (Müller et al. 2010; Kuo et al. 2012). Such histone-binding domains in core replication factors or as-yet-unidentified bridging proteins could link H3K4 di-methylation to recruitment of the replication machinery at origins.

Figure 8 

Two models by which H3K4 di-methylation could promote efficient DNA replication. The recruitment model suggests either direct or indirect recruitment of replication factors by H3K4 di-methylation, while the accessibility model suggests recruitment of a HAT that acetylates nucleosomes near the origin, resulting in an open chromatin state that allows replication factors to access and bind the origin.

The second model suggests that H3K4 di-methylation near origins recruits another chromatin modifier such as a HAT. Acetylation of nearby nucleosomes resulting in increased origin accessibility would allow more efficient association of replication factors (Figure 8). Acetylation is already known to affect the efficiency and timing of origin firing (Vogelauer et al. 2002; Goren et al. 2008). Additionally, several HAT complexes contain subunits that specifically recognize H3K4 methylation. The HAT complex SAGA contains the catalytic subunit Gcn5 as well as two proteins, Chd1 and Spt8, that genetically interact with cdc6-1 (Table 3). Chd1 is a chromatin remodeler whose mammalian counterpart is capable of binding H3K4 methylation (Sims et al. 2005) while Spt8 impacts transcription by directly recruiting TATA-binding protein (Sermwittayawong and Tan 2006). SAGA also contains the Sgf29 subunit that has recently been reported to bind di- and tri-methylated H3K4 to facilitate SAGA recruitment to some promoters (Bian et al. 2011). SAGA is responsible for acetylation of multiple H3 residues as well as H4K8 and H2BK11 and K16 (Grant et al. 1999). Two other HAT complexes were also identified in our screen, NuA3 and NuA4; both contain subunits, Yng1 and Yng2, respectively, that bind H3K4 methylation (Martin et al. 2006a,b; Taverna et al. 2006; Shi et al. 2007). Further examination of these and other H3K4 methylation readers will shed light on the role of H3K4 methylation in the context of origin-specific chromatin.

Prior research has identified other histone modifications that impact origin function. Mono-methylation of H3K36 by Set2 facilitates recruitment of the replication initiation factor Cdc45 in yeast (Pryde et al. 2009). In addition to methylation, H3K4 can also be acetylated by both Gcn5 and Rtt109 (Guillemette et al. 2011). At promoters, this acetylation is typically found just upstream from the peak of H3K4 tri-methylation. Whether H3K4 acetylation is also found at origins is currently unknown. Multi-acetylated histones H3 and H4 have been shown by several groups to impact the ability of origins to fire (Unnikrishnan et al. 2010) and the timing of origin firing in yeast, Drosophila, and mammalian cells (Vogelauer et al. 2002; Hiratani et al. 2009; Schwaiger et al. 2009). Additionally, the HATs Hat1 and Gcn5 physically interact with the replication machinery (Suter et al. 2007; Espinosa et al. 2010). In mammalian cells, Gcn5 acetylates Cdc6 at several lysine residues, and this acetylation is required for subsequent phosphorylation by cyclin/CDKs (Paolinelli et al. 2009). In vertebrates, the H4-specific HAT, Hbo1, binds (Iizuka and Stillman 1999; Burke et al. 2001) and acetylates several origin-binding proteins and histone H4 (Iizuka et al. 2006; Miotto and Struhl 2010), and this acetyltranferase activity is required for efficient origin licensing. In metazoans, methylation of H4K20 by PR-Set7 (a.k.a. Set8) is cell cycle regulated (Tardat et al. 2010; Beck et al. 2012) and is required for proper S-phase initiation and prevention of re-replication (Wu and Rice 2011). In S. cerevisiae, histone H3 is phosphorylated at threonine 45 in a cell cycle-dependent manner by the Cdc7-Dbf4 kinase. Mutating this residue to alanine causes sensitivity to hydroxyurea and camptothecin, indicating its importance in proper DNA replication (Baker et al. 2010).

Because all organisms except S. cerevisiae lack common nucleotide sequence motifs at origins, discovering a chromatin signature that promotes origin function is essential to understanding the location and activity of replication zones in higher eukaryotes. Given the complexity of metazoan genomes, it may be that several different chromatin signatures specify origins in different chromosomal domains. In fact, many histone modifications have been found at human replication origins, including H3K4me3, H3K56me, H4K20me1, H4K20me2, and multiple acetylations on H3 and H4 (Rampakakis et al. 2009; Miotto and Struhl 2010; Tardat et al. 2012; Yu et al. 2012). Our discovery that H3K4 di-methylation is able to promote DNA replication adds to the growing number of histone modifications that could function together to confer origin function at discrete genomic loci. Further work will determine which suite of histone modifications is common to all origins or is required only at specific genomic loci to successfully promote DNA replication.

Acknowledgments

We thank Zu-Wen Sun for strains and H2B plasmids; Charlie Boone for the cdc45-27, cdc7-1, and cdc7-4 mutant strains; Alisha Schlichter and Brad Cairns for Set1 mutant plasmids; Doug Koshland for plasmids containing the tandem ARS209 insert; Ali Shilatifard for wild-type and catalytically dead Bre1 plasmids; Fred Cross for RUY121 and RUY028 re-replication strains; Mark Bedford for the H3K4 tri-methyl antibody; and members of the Cook and Strahl labs for many helpful discussions during the course of this work. This research was supported by National Institutes of Health (NIH) grants T32 GM07092-34 (L.F.R.), T32 CA071341-15 (E.S.D.), and NIH R01 GM68088 (B.D.S.) and by funding from the American Heart Association (10PRE3740013 to L.F.R. and BGIA 0865097E to J.G.C.).

Footnotes

  • Communicating editor: J. C. Schimenti

  • Received June 8, 2012.
  • Accepted July 18, 2012.

Literature Cited

View Abstract