Heterochromatin Controls γH2A Localization in Neurospora crassa

Abstract

In response to genotoxic stress, ATR and ATM kinases phosphorylate H2A in fungi and H2AX in animals on a C-terminal serine. The resulting modified histone, called γH2A, recruits chromatin-binding proteins that stabilize stalled replication forks or promote DNA double-strand-break repair. To identify genomic loci that might be prone to replication fork stalling or DNA breakage in Neurospora crassa, we performed chromatin immunoprecipitation (ChIP) of γH2A followed by next-generation sequencing (ChIP-seq). γH2A-containing nucleosomes are enriched in Neurospora heterochromatin domains. These domains are comprised of A·T-rich repetitive DNA sequences associated with histone H3 methylated at lysine-9 (H3K9me), the H3K9me-binding protein heterochromatin protein 1 (HP1), and DNA cytosine methylation. H3K9 methylation, catalyzed by DIM-5, is required for normal γH2A localization. In contrast, γH2A is not required for H3K9 methylation or DNA methylation. Normal γH2A localization also depends on HP1 and a histone deacetylase, HDA-1, but is independent of the DNA methyltransferase DIM-2. γH2A is globally induced in dim-5 mutants under normal growth conditions, suggesting that the DNA damage response is activated in these mutants in the absence of exogenous DNA damage. Together, these data suggest that heterochromatin formation is essential for normal DNA replication or repair.

INTRODUCTION

Heterochromatin is comprised of transcriptionally repressed, repetitive DNA sequences that remain condensed throughout the cell cycle (1). The condensed structure of heterochromatin and the repetitive nature of heterochromatic DNA sequences pose challenges to genome integrity during DNA replication and DNA repair. Repeated DNAs are hot spots for various types of genome instabilities because they can adopt non-B-form DNA structures that stall replication forks and because they are common sites of illegitimate recombination (2–6). Such events can lead to mutations, gross chromosomal rearrangements, or copy number variations often associated with human diseases (7, 8). Despite their deleterious potential, repetitive DNA sequences make up a significant fraction of the genome in many eukaryotes, including many filamentous fungi (9–12). These sequences likely persist in genomes because they perform essential functions in certain contexts. For example, the centromeres of most eukaryotes are flanked by large, repeat-rich heterochromatin domains (13), and centromeres of the filamentous fungus Neurospora crassa are completely heterochromatic (14, 15). Thus, cells limit deleterious effects of repetitive DNA sequences while preserving essential functions of heterochromatin domains.

Heterochromatin in mammals and N. crassa is enriched with specific molecular markers, including histone H3 lysine-9 methylation (H3K9me), heterochromatin protein 1 (HP1), and cytosine DNA methylation. In Neurospora, heterochromatin formation is initiated at A·T-rich repetitive DNA sequences by the H3K9 methyltransferase DIM-5 (defective in methylation 5), which exists in a multiprotein complex called DCDC (DIM-5/7/9–CUL4/DDB1 complex) (16–19). According to proposed nomenclature conventions (20), DIM-5 is sometimes also referred to as KMT1 (lysine [K] methyltransferase 1), based on its structural and functional homologies with the mammalian Suv39H1 (KMT1A) and Suv39H2 (KMT1B) enzymes. By methylating H3K9, DIM-5^KMT1 creates a binding site for multiple HP1-containing complexes, including the DIM-2 DNA methyltransferase complex (21), the HCHC histone deacetylase complex (HP1-chromodomain protein 2 [CDP-2]-histone deacetylase 1 [HDA1] complex) (22), and a complex containing a putative histone demethylase, DMM-1 (DNA methylation modulator 1) (23). The combined activities of these complexes are required for proper establishment and maintenance of heterochromatin domains in Neurospora. These domains include centromeres, subtelomeric regions, and hundreds of dispersed heterochromatin regions scattered throughout the genome (14, 15).

Proper heterochromatin formation appears to be important for normal genome stability in several organisms, but how specific heterochromatin components contribute to genome maintenance is not well understood (4). In the fission yeast Schizosaccharomyces pombe, replication fork stalling is observed in heterochromatin domains (24), and Clr4^KMT1 mutants, which lack H3K9me, exhibit high rates of illegitimate recombination within the repetitive ribosomal DNA (rDNA) locus (25). Cytological studies in Drosophila melanogaster revealed that H3K9me-deficient mutants exhibit spontaneous double-strand breaks (DSBs) in heterochromatin domains (4, 26–28). H3K9me may promote genome stability through recruitment of HP1, as HP1 homologs in Drosophila and mammals contribute to the DNA damage response pathway through both H3K9me-dependent and -independent mechanisms (29–38). Recent work in Neurospora indicates that H3K9 methylation is important for genome maintenance in this fungus as well. DCDC-deficient mutants are hypersensitive to the DNA-damaging agent methyl methanesulfonate (MMS) (16), suggesting that H3K9 methylation may be important for DNA replication or DNA repair. To investigate this possibility further, we examined γH2A—a well-characterized marker of genotoxic stress—in wild-type strains and heterochromatin-deficient mutants.

In response to DNA replication stress or DNA DSBs, the mammalian H2A variant H2A.X is phosphorylated by ATM or ATR kinase on a serine near the C terminus (serine-139). The resulting modified histone, referred to as γH2AX, acts to stabilize stalled replication forks and facilitate double-strand-break repair (39). Fungi lack an H2A.X variant, but the core H2A protein contains an H2A.X-like sequence near the C terminus (SQEL). Similar to the situation in mammals, this C-terminal serine is phosphorylated by ATM and ATR kinases (Tel1 and Mec1 in Saccharomyces cerevisiae) to form γH2A in response to genotoxic stress (40). γH2A then recruits numerous chromatin-binding proteins to regulate genome stability (41).

Because γH2A accumulates around stalled replication forks and DSBs, this modified histone is often used as a marker of genome instability (42–45). In budding and fission yeasts, genomewide analyses revealed that γH2A is enriched in heterochromatin domains during unperturbed vegetative growth (43, 45). Budding yeast (S. cerevisiae) lacks conserved features of heterochromatin, such as H3K9 methylation and HP1, but in S. pombe, γH2A was found to colocalize with H3K9me and HP1 (45). In the S. pombe Clr4^KMT1 mutant, γH2A enrichment was reduced in a heterochromatic region near the centromere and in a subtelomeric heterochromatin domain (45). In contrast, γH2A enrichment was unaffected in another heterochromatin domain, the silent mating-type locus. Thus, precisely how γH2A is directed to heterochromatin domains in either yeast species is not clear. In Drosophila, γH2A does not appear to be localized to heterochromatin domains in wild-type cells, but in H3K9-deficient mutants, high levels of γH2A are observed in heterochromatin domains, suggesting that heterochromatin-defective mutants suffer spontaneous DNA damage (26). In general, the functional and regulatory relationships between γH2A and heterochromatin are not well understood. We performed genomic, molecular, and cytological analyses of γH2A in Neurospora. We show that γH2A is a component of heterochromatin in Neurospora and that γH2A is significantly induced in a heterochromatin-defective mutant under normal growth conditions. These data suggest that ATM or ATR kinase is hyperactivated in the absence of normal heterochromatin. We propose that a repressive chromatin structure at repetitive, A·T-rich DNA sequences is important for normal genome stability in Neurospora.

MATERIALS AND METHODS

Strains and growth media.

All Neurospora strains used in this study are listed in Table S1 in the supplemental material. Strains were grown at 32°C in Vogel's minimal medium (VMM) with 1.5% sucrose (46). Liquid cultures were shaken at 150 rpm. Crosses were performed on modified synthetic cross medium (46). For plating assays, Neurospora conidia were plated on VMM with 2.0% sorbose, 0.5% fructose, and 0.5% glucose. Where relevant, plates included 200 μg/ml hygromycin or 400 μg/ml Basta (47). LB medium supplemented with either ampicillin (50 μg/ml) or kanamycin (50 μg/ml) was used to grow Escherichia coli DH5α and XL1-Blue (48). Plasmids used for transformation were isolated using Qiagen miniprep kits.

Construction of H3K9 mutants.

Primers used for site-directed mutagenesis are listed in Table S2 in the supplemental material. Primers were designed using previously described criteria (49). H3K9 mutations to glutamine or arginine were introduced by PCR, using plasmid pK9L as the template (50, 51). Plasmids containing Neurospora H3K9 mutations to glutamine or arginine were linearized with XbaI and introduced into Neurospora strain XStF9.1 by electroporation. This recipient strain contains the H3 coding sequence from Fusarium graminearum in place of the Neurospora hH3 gene, as well as a deletion of mus-51. These features ensure that transformation leads to replacement of the entire F. graminearum hH3 gene with the altered Neurospora H3 sequence. Homokaryotic H3 replacement strains were obtained by crossing primary transformants to the wild type. H3 replacement was confirmed by Southern blotting, followed by PCR and sequencing of the integrated DNA.

Molecular analyses.

Neurospora transformation (52), DNA isolation (53), protein isolation, histone isolation, and Western blotting (54) were performed as previously described. Southern blotting was performed as described previously (55), except that probe synthesis, hybridization, and detection were carried out using a North2South chemiluminescence hybridization and detection kit (Thermo). Chemiluminescent blots were imaged using a ProteinSimple FluorChem E imager. Primers used to generate probe templates are listed in Table S2 in the supplemental material. Chromatin immunoprecipitation (ChIP) was performed using 5-hour-old germinating conidia. Fifty-milliliter cultures containing 5 × 10⁶ conidia/ml were grown for 5 h, and conidia were harvested by centrifugation. Conidia were washed once in phosphate-buffered saline (PBS), and chemical cross-linking was performed by incubating conidia in PBS containing 1% formaldehyde at room temperature on a rotating platform for 30 min. The reaction was quenched with 125 mM glycine. Conidia were washed with PBS twice and resuspended in lysis buffer (50 mM HEPES, pH 7.5, 140 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% deoxycholate). For γH2A ChIP assays, the extraction buffer was supplemented with phosphatase inhibitor cocktail (Sigma). Chromatin was sheared by sonication, using an UltraSonic processor (duty cycle, 80; output, 3.5) (Heat System-Ultrasonics Inc.) to deliver 150 1-s pulses at 4°C. Lysates were centrifuged at 13,000 rpm for 5 min at 4°C. For γH2A-ChIP, 2 μl of anti-γH2A antibody (ab15083; Abcam) was used. For detection of H3K9 trimethylation (H3K9me3), 1 μl of antibody (Active Motif) was used. Protein A/G beads (20 μl) (sc-2003; Santa Cruz) were added to each sample. Following overnight incubation, beads were washed twice with 1 ml lysis buffer, once with lysis buffer containing 500 mM NaCl, once with 50 mM LiCl, and finally with TE (10 mM Tris-HCl, 1 mM EDTA). Bound chromatin was eluted in TES (50 mM Tris, pH 8.0, 10 mM EDTA, 1% SDS) at 65°C. Chromatin was de-cross-linked overnight at 65°C. The DNA was treated with RNase for 2 h at 50°C, treated with proteinase K for 2 h at 50°C, and extracted using phenol-chloroform. DNA pellets were washed with 70% ethanol and resuspended in TE buffer. Samples were then prepared for Illumina sequencing or subjected to analysis by quantitative real-time PCR.

qPCR and Illumina sequencing.

Primers used for quantitative PCR (qPCR) are listed in Table S2 in the supplemental material. DNAs obtained from ChIP assays were diluted 1:50 in H₂O for input samples and 1:10 in H₂O for samples immunoprecipitated with H3K9me3, H3K4me2, or γH2A antibody. For PCR, iTaq Universal SYBR green Supermix (Bio-Rad) was mixed with specific primer pairs, and 1 μl of the diluted ChIP DNA was added. qPCR was performed using an iCycler IQ instrument (Bio-Rad). Statistical analyses were performed in Microsoft Excel. For Illumina sequencing, libraries were prepared using half of the total immunoprecipitated fraction following the instructions supplied with Illumina Tru-seq kits, except that genomic DNA adaptors were diluted 1:100 prior to ligation. Illumina sequencing was performed using an Illumina Hi-Seq 2000 genome analyzer at the University of Missouri DNA Core Laboratory.

Data analysis.

Sequence reads were mapped to the latest Neurospora genome annotation (version 12), available from the Neurospora genome database (11), by using bowtie2 (56). Read numbers were counted for 25-bp bins by using igvtools, and the read density was visualized using the Integrated Genome Viewer (IGV), available from the Broad Institute website (57, 58). We used IGV to normalize data to the total read number before plotting. To calculate normalized enrichment values, we created a custom feature annotation file containing genes, tRNAs, and DNA repeats. Repeated DNA sequences were identified by analyzing the Neurospora genome with RepeatScout (59). Repeat families were aligned to the Neurospora genome by using BLAT (60), and then coordinates were parsed using a custom perl script into a gene prediction format file that also contained coordinates for genes and tRNAs (downloaded from the Broad Institute genome database [11]). To calculate the normalized ChIP enrichment values (NLCS values) for each feature, we used EpiChIP software, which calculates enrichment values normalized for total read number and for length of the feature (61). Normalized H3K9me3, H3K4me3, and γH2A values for each feature were used to generate scatterplots and to calculate Pearson's correlation coefficients in Microsoft Excel. In addition, the NLCS values were used to plot the kernel density estimations for all features, genes, and repeats, using R (http://www.r-project.org). Where relevant, mapped reads were converted to bed format by use of bedtools software (62). Heterochromatin domains were classified as individual peaks by use of r-seg software (http://smithlabresearch.org/software/rseg/). The coordinates of H3K9me peaks are listed in Table S3 in the supplemental material.

Immunofluorescence.

For cytological analysis of γH2A, we adapted a method previously described for Aspergillus nidulans (63). Conidia were inoculated into VMM containing 1.5% sucrose and incubated at 32°C for 12 h on coverslips or in an 8-well μ-Slide (Ibidi). Cells were fixed for 30 min in a solution containing 3.5% formaldehyde, 5% dimethyl sulfoxide (DMSO), 25 mM EGTA, and 5 mM MgSO₄. Fixed cells were washed with PBS three times, followed by a 90-min incubation in a 50% egg white solution containing 50 mM piperazine-N,N′-bis(2-ethanesulfonic acid) (PIPES), pH 6.7, 25 mM EGTA, 5 mM MgSO₄, 1 mM dithiothreitol (DTT), and 1 mg/ml of lyticase (purified from Oerskovia xanthineolytica [64]; generously provided by Vincent Starai, University of Georgia). Cells were washed again with PBS and incubated overnight at 4°C in a PBS solution containing the primary antibody (1:200 dilution of the Abcam anti-γH2A antibody described above). Cells were washed three times with PBSA (PBS supplemented with 0.1% bovine serum albumin [BSA]) and incubated for 50 min at room temperature in PBS containing a 1:200 dilution of the secondary antibody [Alexa Fluor 488–goat anti-rabbit IgG(H+L); Life Technologies]. Cells were washed with PBSA three times prior to imaging. Microscopy was performed using a DeltaVision II microscope equipped with a Delta Vision standard filter set, which includes fluorescein isothiocyanate (FITC) and tetramethyl rhodamine isocyanate (TRITC) filters for green and red fluorescence acquisition, respectively. Quantitative analysis of fluorescence intensities was performed using Softworx 1.3 software (Applied Precision).

Nucleotide sequence accession numbers.

Sequence data from this study are available through the NCBI Short Read Archives (accession no. SRP042169).

RESULTS

γH2A is localized to heterochromatin domains in Neurospora crassa.

γH2A is a functionally conserved modification of H2A that acts to stabilize replication forks and facilitate double-strand-break repair in fungi and animals (43–45, 65). To determine if γH2A is enriched at specific genomic locations in the filamentous fungus Neurospora crassa, we performed ChIP of γH2A followed by high-throughput sequencing (ChIP-seq). We used a previously characterized anti-γH2A antibody that was specific for both yeast and Neurospora γH2A proteins (H2A proteins phosphorylated on serine-129 and serine-131, respectively) (66, 67). We also performed ChIP-seq experiments to detect two well-characterized chromatin modifications, H3K4me2 and H3K9me3, which are molecular markers of euchromatin and heterochromatin, respectively. As expected, H3K4me2 was enriched in active genes, while H3K9me3 was localized to A·T-rich, gene-poor heterochromatin domains (Fig. 1A). γH2A enrichment was highly correlated with heterochromatin domains identified by enriched H3K9 trimethylation (Fig. 1A and B; see Fig. S1 in the supplemental material). To validate the ChIP-seq results, we performed qPCR to examine γH2A at representative euchromatin (hH4 and cfp) and heterochromatin (8:A6 and 9:E1) domains. qPCR results were fully consistent with our ChIP-seq data, confirming that γH2A is a component of heterochromatin domains in Neurospora (Fig. 1C). To confirm that the γH2A antibody was specific, we performed ChIP experiments using a γH2A-deficient strain in which the single H2A gene was replaced with an H2A allele containing a serine-131-to-alanine substitution (66). Heterochromatic domains were not enriched in the hH2A^S131A strain when qPCR analyses were performed for representative heterochromatic and euchromatic regions (Fig. 1C). Similarly, heterochromatin domains were not enriched when the entire γH2A immunoprecipitate fraction was assayed by ChIP-seq (Fig. 1A and B; see Fig. S1), demonstrating that the γH2A antibody was specific for phosphorylated H2A serine-131.

FIG 1.

γH2A is a heterochromatin component in Neurospora. (A) ChIP-seq enrichment across linkage group VII is shown for H3K4me2, H3K9me3, and γH2A in the wild-type (WT) strain and for γH2A in the negative-control hH2A^S131A strain. The strain is indicated to the left of the histogram, and the antibody used to perform ChIP-seq is indicated on the right. The positions of genes (gray) and degenerate DNA repeats (red) are shown at the bottom of the plot. (B) ChIP-seq enrichment patterns for representative heterochromatin (8:A6 and 9:E1) and euchromatin (H4 and cfp) domains. Genes are shown in gray beneath each plot. The positions of PCR amplicons used to validate enrichment are indicated with a black line beneath each plot. (C) Quantitative real-time PCR analysis of the representative regions shown in panel B. The relative enrichment values for 8:A6, 9:E1, and cfp were determined by normalizing the indicated region to the hH4 gene. Enrichment values for the wild-type and hH2A^S131A strains were statistically different for the 8:A6 and 9:E1 regions (Student's t test; P < 2.7 × 10⁻⁶ and P < 3.8 × 10⁻⁸, respectively). (D) Scatterplots comparing enrichment of γH2A and H3K9me3 (left) or γH2A and H3K4me2 (right). The normalized ChIP enrichment value (NLCS) obtained for the indicated antibody was plotted for each genomic feature. Data for genes and tRNAs are shown in gray, and those for repeats are shown in red.

To validate that γH2A is localized with H3K9me3 at divergent repeated sequences throughout the genome, we used EpiChIP software (61) to calculate the NLCS values for all features in the genome. Features included genes, tRNAs, and repeated DNA sequences. We generated a scatterplot to compare the NLCS values for H3K9me3 and γH2A or to compare H3K4me2 and γH2A in genes and repeats. H3K9me3 and γH2A were similarly enriched in repeats, while exhibiting low levels of enrichment in genes (Fig. 1D) (Pearson's correlation coefficient = 0.80). In contrast, H3K4me2 was enriched in many of the genes and exhibited no correlation with γH2A (Fig. 1D) (Pearson's correlation coefficient = −0.16). We also calculated the kernel density estimations to examine the relative frequencies of enriched and unenriched features for H3K4me2, H3K9me3, and γH2A. Plotting the kernel densities revealed a bimodal distribution for all three modifications. In each case, one peak corresponds to features with background levels of enrichment, while the second peak represents features enriched with the modification. We also plotted the kernel densities for genes or repeats alone. For H3K4me2, genes were distributed in two peaks, corresponding to background and enriched features, while repeats were distributed in a single background peak (see Fig. S2A in the supplemental material). For both H3K9me3 and γH2A, genes were distributed in a single peak corresponding to background enrichment. In contrast, repeats were distributed in a single peak corresponding to enriched features (Fig. 2A and B). As expected, the kernel density estimation of γH2A enrichment in the hH2A^S131A strain produced a single peak corresponding to uniform background enrichment for all features across the genome (Fig. 2C). However, the enrichment value observed for repeats was slightly lower than that for genes due to a subtle bias against A·T-rich sequences observed in Illumina sequencing experiments (68, 69). Together, these data demonstrate that γH2A localizes to repeat-rich heterochromatin domains across the Neurospora genome.

FIG 2.

Enrichment of γH2A depends on DIM-5. (A to D) The distributions (kernel density estimations; densities) of normalized read counts (NLCS values) are shown for all genomic features (genes and repeats; black), genes only (gray), and repeats only (red). For the wild type, two peaks are observed for both H3K9me3 and γH2A, corresponding to features with background signals (peak B; genes) and features enriched in the modification (peak E; repeats). For the hH2A^S131A and Δdim-5 strains, all features exhibited background levels of γH2A enrichment. (E) ChIP-seq data for H3K9me3 and γH2A are shown for wild-type, Δdim-5, and hH2A^S131A strains at two representative regions. The strain is indicated to the left of the histogram, and the antibody used to perform ChIP-seq is indicated on the right. Genes are shown in gray beneath each plot. The positions of PCR amplicons used to validate the ChIP-seq experiment are indicated with black lines beneath the plot. (F) γH2A ChIP samples were subjected to qPCR to analyze enrichment at each of the amplicons shown in panel E. Relative enrichment was determined by normalizing the enrichment value at each region to the euchromatic flank adjacent to each peak (region 90-1 or 230-1).

H3K9 methylation is required for normal γH2A localization.

H3K9 methylation by DIM-5^KMT1 is an early step in heterochromatin formation (50, 70). We were interested to learn if γH2A localization depends on DIM-5^KMT1 or its enzymatic product, H3K9me3. We performed γH2A ChIP-seq experiments with a dim-5 deletion strain and found that enrichment of γH2A was significantly reduced at all heterochromatin domains (see Fig. S2B in the supplemental material). Consistent with this apparent loss of enrichment, plotting the kernel density for γH2A enrichment in the dim-5 strain revealed a single peak corresponding to uniform background enrichment across the genome for all features. We performed qPCR to validate the ChIP-seq data for representative heterochromatin domains on linkage groups (LG) II and V (Fig. 2E). We used primers to test enrichment adjacent to each domain (regions 90-1 and 230-1), at the edge of each domain (regions 90-2 and 230-2), and in the center of each domain (regions 90-3 and 230-3). qPCR data were fully consistent with the ChIP-seq data. γH2A enrichment was significantly reduced at both domains tested (peak 90 and peak 230) (see Table S3); however, a low level of γH2A in the center of the heterochromatin region remained (Fig. 2F) (Student's t test; P < 0.008 for region 90-2, P < 0.008 for region 90-3, P < 0.0002 for region 230-2, and P < 0.007 for region 230-3). Together, these data suggest that γH2A localization is altered in the dim-5 mutant strain. We next asked if γH2A enrichment depends on H3K9me3. It was previously reported that H3K9 is an essential residue in Neurospora, based on the observation that H3K9L mutants were not recovered from crosses in which the mutant hH3 allele was integrated at an ectopic locus (51). We constructed H3K9-to-arginine (R) and H3K9-to-glutamine (Q) replacement alleles, which mimic unacetylated and acetylated lysine, respectively. Homokaryons were obtained for both alleles, demonstrating that H3K9 is not essential (see Fig. S3 in the supplemental material). As expected, both H3K9 substitution strains lacked DNA methylation (see Fig. S3), indicating that heterochromatin formation was defective. Both strains also exhibited severe growth defects, similar to dim-5 strains (see Fig. S4). We next performed ChIP-seq experiments with these H3K9me3-deficient strains. Like the case in dim-5 strains, γH2A was no longer enriched in repetitive sequences in these mutants (Fig. 3A and B). Loss of enrichment was evident across the entire genome (see Fig. S2C). Data for three representative heterochromatin regions are shown in Fig. 3B. We confirmed these data by performing ChIP-qPCR to assay γH2A enrichment at the 8:A6 heterochromatin region (Fig. 3C) (Student's t test; P < 4 × 10⁻⁵ for the hH3^K9Q mutant and P < 1.5 × 10⁻⁵ for the hH3^K9R mutant). These data show that heterochromatin formation directed by DIM-5 is required for normal localization of γH2A.

FIG 3.

H3K9me3 is required for γH2A enrichment. (A) The distributions (kernel density estimations; densities) of normalized read counts (NLCS) are shown for all genomic features (genes and repeats), genes only, and repeats only for strains harboring H3K9 substitution alleles. In both strains, all features displayed equivalent background levels of γH2A enrichment. (B) ChIP-seq data for γH2A are shown for the wild-type and Δdim-5 strains and for two strains harboring H3K9 substitution alleles at two representative regions (peak 90 and peak 230). A scale bar at the top of each graph indicates the size of the region plotted. Genes are shown in gray beneath each plot. The position of the 8:A6 PCR amplicon used to validate the ChIP-seq experiment is indicated with a black line beneath the plot. (C) γH2A ChIP samples were subjected to qPCR to analyze enrichment at the 8:A6 region. Relative enrichment was determined by normalizing the enrichment value at 8:A6 to that at the hH4 gene.

Normal γH2A localization depends on HP1.

HP1 binds methylated H3K9 and functions as a molecular scaffold to recruit additional chromatin-modifying enzymes to heterochromatin domains (21). Neurospora HP1 forms distinct complexes with the histone deacetylase HDA1 and the DNA methyltransferase (MTase) DIM-2 (22, 54). To determine if HP1, HDA1, or DIM-2 is required for proper control of γH2A localization, we performed ChIP-qPCR experiments using deletion strains that lack these proteins. As expected based on prior work, enrichment of H3K9me3 was abolished at the 8:A6 region in the hpo and hda-1 strains but not in the dim-2 strain. Similarly, γH2A was lost from the 8:A6 region in the hpo and hda-1 strains but not in the dim-2 strain (Fig. 4A) (Student's t test; P < 0.0001 for the hpo strain and P < 1.0 × 10⁻⁷ for the hda-1 strain). We also performed qPCR to examine γH2A enrichment at the same representative regions on LGII and LGV as those examined above (peaks 90 and 230). For both regions, γH2A enrichment was lost at the edge of the heterochromatin domain in the hpo and hda-1 strains (Student's t test; P < 0.001 for region 90-2 in the hpo strain, P <0.001 for region 90-2 in the hda-1 strain, P < 0.002 for region 230-2 in the hda-1 strain, and P < 0.003 for region 230-2 in the hpo strain). The dim-2 strain exhibited a small decrease in γH2A enrichment at the edge of the LGV domain (region 230-2), but the observed difference was not statistically significant. The dim-2 strain also failed to exhibit significant differences in the level of γH2A enrichment in the centers of these heterochromatin domains (Fig. 4B and C). Enrichment was slightly reduced in the centers of the LGII and LGV regions in the hda-1 strain (Fig. 4C) (Student's t test; P < 0.013 for region 90-3 and P < 0.025 for region 230-3). Enrichment appeared to be slightly reduced in the middle of the LGV domain in the hpo strains, but this difference was not statistically significant (Fig. 4C) (Student's t test; P < 0.17 for region 230-3). These data suggest that the HCHC complex contributes to normal localization of γH2A. However, the loss of enrichment in the HCHC-deficient strains was less severe than that in the dim-5 mutants at the regions tested. This suggests that another H3K9me-binding protein may be important for normal γH2A localization. In contrast, DNA methylation is not required for γH2A enrichment at heterochromatin domains.

FIG 4.

γH2A enrichment is reduced in the hpo and hda-1 strains. ChIP was performed for the indicated strains, using antibodies to H3K9me3 (left) and γH2A (right). Relative enrichment levels at the 8:A6 region (A) and at sites within and adjacent to peak 90 (B) and peak 230 (C) were determined by qPCR. Enrichment at the 8:A6 region is shown relative to that at the euchromatic hH4 gene. Enrichment values for peaks 90 and 230 are shown relative to the euchromatin region adjacent to each peak (PCR amplicons 90-1 and 230-1, respectively).

γH2A is not required for H3K9 methylation or DNA methylation.

Since γH2A is enriched in heterochromatin domains, we tested the possibility that γH2A regulates H3K9me3 or DNA methylation. qPCR experiments revealed similar enrichments of H3K9me3 at the 8:A6 region in both wild-type and hH2A^S131A strains (Fig. 3C). To confirm this, we performed ChIP-seq experiments to examine H3K9me3 in the hH2A^S131A strain. These analyses revealed that H3K9 methylation patterns were qualitatively similar in the wild type and the hH2A^S131A strain (Fig. 5A). To next determine if γH2A was required for proper control of DNA methylation, genomic DNAs from the wild-type and hH2A^S131A strains were digested with the methylation-sensitive enzyme BfuCI and the methylation-insensitive isoschizomer DpnII. The digested DNAs were analyzed by Southern blotting to examine DNA methylation at known methylated regions (8:A6, 8:G3, and centromere VII) (71). The membrane was also probed for the unmethylated H3 gene to confirm that the DNAs had been digested completely. Wild-type patterns of DNA methylation were observed for all four loci (Fig. 5B). These data show that γH2A is not required for H3K9 trimethylation or cytosine DNA methylation.

FIG 5.

Heterochromatin formation is independent of γH2A. (A) ChIP-seq enrichment across linkage group VII is shown for H3K9me3 in the wild-type strain and the hH2A^S131A strain. The strain is indicated to the left of the histogram. Genes are shown in gray, and degenerate DNA repeats are shown in black at the bottom of the plot. (B) Genomic DNAs from the wild-type, Δdim-5, and hH2A^S131A strains were digested with methylation-sensitive BfuCI (lanes B) and methylation-insensitive DpnII (lanes D). Cytosine methylation levels were analyzed by visualizing digested DNA with ethidium bromide (EtBr) and by probing Southern blots with the indicated methylated (8:A6, 8:G3, and centromere VII [CenVII]) and unmethylated (hH3) regions.

γH2A is induced in the dim-5 strain.

We asked if the loss of γH2A enrichment observed in the dim-5 strain corresponded to a global reduction in γH2A levels. We isolated total histones from the wild-type, dim-5, and hH2A^S131A strains. As a control, we also isolated histones from the same strains following exposure to the DNA-damaging agent MMS, and we performed Western blotting to examine the level of γH2A in each strain. To ensure equal loading, gels were stained with Coomassie brilliant blue, and we performed Western blotting using antibodies that recognize H3K4me2, which was previously demonstrated to be unchanged in the dim-5 background (54). In the wild type, γH2A levels are low in minimal medium, but exposure to MMS leads to induction of γH2A, as observed previously (66). In contrast, analysis of the dim-5 strain revealed that γH2A levels were significantly elevated in minimal medium (Student's t test; P = 0.02). γH2A levels remained high following exposure to MMS (Fig. 6). No signal was observed for the hH2A^S131A strain.

FIG 6.

γH2A levels are elevated in the Δdim-5 strain. (A) Histones were isolated from the indicated strains grown in VMM or VMM plus 0.015% MMS and subjected to Western blot analyses (WB) of γH2A. As a loading control, Western blotting was performed using antibodies to H3K4me2, and gels were stained with Coomassie brilliant blue (CBB). (B) The levels of γH2A for each strain were determined by densitometry and normalized to those of H3K4me2 for each strain grown in VMM and VMM plus MMS.

These data appeared to be inconsistent with our ChIP-seq data, which suggested an apparent loss of enrichment in the dim-5, H3^K9R, and H3^K9Q strains (Fig. 2 and 3). However, overall increased γH2A levels in euchromatin would also result in the loss of relative γH2A enrichment observed in H3K9me3-deficient strains. To test this possibility, we performed immunofluorescence assays to visualize γH2A localization in the wild-type and dim-5 strains. Both strains contained an H1-RFP reporter construct to enable visualization of nuclei. In wild-type cells, the observed fluorescence was similar to that for the hH2A^S131A strain, consistent with the low levels of γH2A observed in Western blot experiments (Fig. 7A; see Fig. S4 in the supplemental material). In contrast, nuclei in dim-5 cells were intensely stained with the γH2A antibody (Fig. 7A). γH2A appeared to be enriched throughout dim-5 nuclei, although some nuclei displayed nonuniform γH2A staining (Fig. 7B). Immunofluorescence assays revealed that γH2A accumulated in wild-type nuclei following exposure to the DNA-damaging agent MMS (see Fig. S5), demonstrating that this method is able to detect elevated levels of γH2A in wild-type cells. We quantified green and red fluorescence in wild-type and dim-5 nuclei grown in minimal medium (Fig. 7C). These data confirmed that dim-5 strains exhibit a significant induction of γH2A under normal growth conditions, consistent with defective DNA replication or repair in these strains. We next asked if γH2A misregulation is responsible for the drug sensitivity of dim-5 strains. We tested colony survival of wild-type, dim-5, hH2A^S131A, and dim-5; hH2A^S131A strains on MMS and the topoisomerase inhibitor camptothecin. Double-mutant strains were more sensitive than either single mutant, suggesting that DIM-5 regulates additional factors, along with γH2A, to promote proper DNA replication or repair (see Fig. S6).

FIG 7.

γH2A levels are elevated throughout Δdim-5 nuclei. (A) Immunofluorescence of γH2A is shown for wild-type and Δdim-5 cells. Both strains expressed an H1-dTomato fusion protein to allow visualization of nuclei (H1-RFP). Bars, 10 μm. (B) Immunofluorescence data for three individual nuclei from each strain. Bars, 1 μm. (C) Red and green fluorescence was quantified for three representative nuclei by using the SoftWorx Explorer “line profile” tool. The plots show fluorescence intensities measured along a 4-μm-line drawn across the center of the nucleus.

DISCUSSION

γH2A associates with heterochromatin in Neurospora.

We found that γH2A is a component of heterochromatin domains in wild-type Neurospora cells. Similar results were observed in S. cerevisiae and S. pombe (43, 45, 72), suggesting that γH2A is a conserved heterochromatin component in fungi, or at least in the ascomycetes. These data raise the possibility that γH2A is enriched in heterochromatin in other eukaryotes as well. Like the case in other organisms that have been studied, Neurospora γH2A is generated by ATM and ATR kinases in response to replication stress or DNA damage (66, 67). The MRN complex (Mre11, Rad50, and Nbs1) orchestrates ATM recruitment and activation at double-strand breaks (73, 74), while ATR is activated by accumulation of single-stranded DNA (ssDNA) (75). Because ssDNA can accumulate during various repair processes, ATR is able to respond to a variety of different types of DNA damage and genotoxic stresses. For example, ssDNA is generated during end resection of a double-strand break and at stalled replication forks, where the replicative helicase is thought to become uncoupled from DNA polymerase (76). It is possible that the heterochromatin structure leads to stalled replication forks in the wild type, thereby activating ATR kinase. Studies from S. cerevisiae show that γH2A overlaps regions that stall replication forks, including heterochromatin domains, and that recruitment of γH2A to heterochromatin depends on Sir3 (43, 72, 77). In S. pombe, heterochromatin-associated γH2A depends on the checkpoint kinase RAD3 (a homolog of ATR) (45). Certain DNA-binding proteins can inhibit replication fork progression (5). Thus, the tightly packaged nucleosomes found in heterochromatin may act as a natural impediment to replication forks. Indeed, analysis of replication intermediates by two-dimensional gel electrophoresis revealed stalled replication forks in heterochromatin domains of wild-type S. pombe cells (24). However, it remains possible that H2A kinases, such as ATR and ATM, are recruited directly to heterochromatin domains. Further studies are needed to determine the mechanisms responsible for γH2A deposition in heterochromatin in wild-type cells.

In yeasts and animals, γH2A functions as a signal to recruit proteins that stabilize stalled replication forks or regulate DNA repair (41, 78, 79). For example, S. pombe Brc1 binds to γH2A via its BRCT domains to enable recovery from replication stress (78). Neurospora heterochromatin domains are comprised of AT-rich repetitive sequences that can adopt non-B-form DNA structures, such as cruciforms or hairpins. Such structures are known to stall replication forks in bacterial, yeast, and mammalian cells (80). It is therefore possible that γH2A functions to recruit proteins to stabilize replication forks that encounter natural sequence impediments found in heterochromatin domains. On the other hand, repeated DNA sequences can also provide substrates for illegitimate recombination, leading to chromosome rearrangements. Work with Drosophila revealed distinct DSB repair mechanisms in heterochromatin and euchromatin (32). The Neurospora dim-5 mutant exhibits elevated rates of illegitimate recombination between heterochromatin-associated transgenes arranged in tandem (81). Thus, γH2A may recruit proteins to regulate the type of DNA repair that occurs in heterochromatin. The mammalian γH2A-binding protein PTIP binds 53BP1 to suppress homologous recombination (HR) and promote nonhomologous end joining (NHEJ) (82, 83), consistent with this possibility. The Neurospora genome encodes 10 BRCT domain-containing proteins (11). Future studies are required to determine if any of these proteins are localized to heterochromatin and, if so, what functions they might perform at these genomic regions.

While γH2A is best known for its response to replication stress and DNA damage, it is possible that this modified histone performs other functions at heterochromatin domains. Recent work with S. cerevisiae suggested that γH2A is important for long-range interactions between silent mating-type loci (77), raising the possibility that γH2A contributes to three-dimensional organization of the nucleus. On the other hand, γH2A may play a role in transcriptional silencing. The mammalian γH2A-binding protein MDC1 (mediator of DNA damage checkpoint 1) appears to be important for silencing sex chromosomes during male meiosis (84). In S. cerevisiae, the homolog of Brc1, Esc4/Rtt107, was implicated in silencing because it binds to the silencing protein SIR3 (85). In S. pombe, the γH2A-binding protein Brc1 is required for normal transcriptional silencing of a reporter gene embedded in pericentromeric heterochromatin (86). It is likely, however, that this protein contributes to silencing independently of γH2A, as γH2A itself is not required for silencing or for normal heterochromatin formation in S. pombe (45). We found that γH2A is not required for normal H3K9me3 or DNA methylation in Neurospora, demonstrating that heterochromatin formation does not depend on phosphorylation of H2A serine-131. In particular, we found that DNA methylation at the 8:A6 region is not affected in the Neurospora H2A^S131A mutant. This result suggests that γH2A is not required for transcriptional silencing, because this locus frequently loses DNA methylation in mutants that affect transcriptional silencing (15, 22, 54).

Heterochromatin formation is required for normal genome stability in Neurospora.

The DCDC complex is required for growth on the DNA-damaging agent MMS (16), suggesting that this complex is required for normal DNA replication or repair. In this study, we found that Δdim-5 strains have high levels of γH2A during normal replicative growth. These data suggest that Δdim-5 strains suffer spontaneous DNA damage during replication, perhaps due to frequent replication fork collapse. γH2A in S. cerevisiae and S. pombe heterochromatin-deficient mutants has been analyzed by ChIP. S. cerevisiae sir3 mutants show reduced enrichment of γH2A in heterochromatin (72), and for S. pombe, it was reported that γH2A enrichment in centromeres and subtelomeric regions depends on Clr4, a homolog of DIM-5^KMT1 (45, 87). Based on the results obtained with S. pombe, it was suggested that H3K9 methylation by Clr4 is required for γH2A deposition in heterochromatin. Similarly, we observed reduced enrichment of γH2A at heterochromatin domains in the Δdim-5 strain. Taken together, our results suggest that heterochromatin-deficient Neurospora mutants do not exhibit reduced γH2A in heterochromatin but, rather, display elevated levels of γH2A throughout the genome. This interpretation is fully consistent with our global analyses of γH2A by Western blotting and immunofluorescence assays, as well as with our ChIP-seq analyses. It is possible that S. pombe heterochromatin mutants display a similar induction of γH2A. The reported loss of heterochromatin-associated γH2A in Clr4 mutants was determined by normalizing γH2A enrichment in heterochromatin domains to a euchromatic control locus. Thus, in S. pombe, loss of enrichment may result from increased γH2A in euchromatin, as observed in our experiments. This would be consistent with recent work showing that heterochromatin components are required for proper genome integrity of centromeres in the absence of fork stability components (88).

In Drosophila mutants lacking the DIM-5 homolog Su(var)3-9, γH2A is induced in foci that colocalize with heterochromatin (26, 27). Our results show that γH2A is dramatically induced in the dim-5 strain and appears to be found throughout the nucleus. Given that γH2A is induced by stalled replication forks and by DSBs, these data suggest that DIM-5 is required for DNA replication or repair in Neurospora. This interpretation is supported by previous work demonstrating that DIM-5-deficient cells are hypersensitive to the DNA-damaging agent MMS (16). It is possible that DIM-5 is a global regulator of genome stability. However, γH2A is deposited in extremely large domains around DSBs (0.5 to 2 Mb in human cells) (42). In yeast, γH2A can spread to the undamaged domains close to the site of damage both in cis and in trans, presumably because ATR modifies serine-129 of H2A in close proximity (89). Our results may indicate that the Δdim-5 strain accumulates DSBs in heterochromatin domains, leading to massive spread of γH2A into euchromatin. Such a model is supported by recent analyses of Drosophila mutants defective for H3K9 methylation (4, 26, 27). These Drosophila mutants exhibit spontaneous DNA damage, including double-strand breaks, and suffer frequent genome rearrangements. These defects are presumably due to defective replication of heterochromatin-associated repeat sequences. Future studies are required to determine how heterochromatin might function to preserve genome integrity. The Neurospora genome includes abundant heterochromatin domains that share important molecular features with higher eukaryotes. Given these similarities with other eukaryotes, future work with Neurospora is likely to lead to important insights regarding the relationships between heterochromatin and genome maintenance.