Identification of a PRC2 Accessory Subunit Required for Subtelomeric H3K27 Methylation in Neurospora crassa

Polycomb repressive complex 2 (PRC2) catalyzes methylation of histone H3 at lysine 27 (H3K27) in genomic regions of most eukaryotes and is critical for maintenance of the associated transcriptional repression. However, the mechanisms that shape the distribution of H3K27 methylation, such as recruitment of PRC2 to chromatin and/or stimulation of PRC2 activity, are unclear. Here, using a forward genetic approach in the model organism Neurospora crassa, we identified two alleles of a gene, NCU04278, encoding an unknown PRC2 accessory subunit (PAS).

ABSTRACT

Polycomb repressive complex 2 (PRC2) catalyzes methylation of histone H3 at lysine 27 (H3K27) in genomic regions of most eukaryotes and is critical for maintenance of the associated transcriptional repression. However, the mechanisms that shape the distribution of H3K27 methylation, such as recruitment of PRC2 to chromatin and/or stimulation of PRC2 activity, are unclear. Here, using a forward genetic approach in the model organism Neurospora crassa, we identified two alleles of a gene, NCU04278, encoding an unknown PRC2 accessory subunit (PAS). Loss of PAS resulted in losses of H3K27 methylation concentrated near the chromosome ends and derepression of a subset of associated subtelomeric genes. Immunoprecipitation followed by mass spectrometry confirmed reciprocal interactions between PAS and known PRC2 subunits, and sequence similarity searches demonstrated that PAS is not unique to N. crassa. PAS homologs likely influence the distribution of H3K27 methylation and underlying gene repression in a variety of fungal lineages.

INTRODUCTION

Specification and maintenance of cellular fate in higher eukaryotes require the concerted effort of transcription factor networks and chromatin-modifying complexes. Polycomb repressive complex 2 (PRC2), which catalyzes methylation of lysine 27 on histone H3 (H3K27me), is one such chromatin-modifying complex essential for normal development in many organisms (1, 2). The three core components of PRC2, which are conserved in plants, animals, and fungi (3, 4), are EED, SUZ12, and the methyltransferase EZH2. Human PRC2 additionally associates with RBBP4 and RBBP7 (RBBP4/7), as well as other accessory proteins, which define the two human PRC2 subtypes: PRC2.1 and PRC2.2 (5). Analogous subtypes of PRC2 have also been described in Drosophila melanogaster (6–8) while in Arabidopsis thaliana, distinct PRC2 complexes are defined by different paralogs of the core PRC2 member, SUZ12 (9). Here, using a genetic approach, we provide evidence that a previously unidentified accessory subunit of PRC2 is responsible for functional specialization in the filamentous fungus Neurospora crassa.

N. crassa bears homologs of the human core PRC2 components EED (EED), SUZ12 (SUZ12), and EZH2 (SET-7), as well as RBBP4/7 (NPF) (10). While loss of either EED, SUZ12, or SET-7 completely abolishes H3K27 methylation, which normally covers ∼7% of the genome (10), loss of NPF affects H3K27 methylation in a region-specific manner (10). Considering that homologs of NPF are components of other chromatin-modifying complexes (11), it remained possible that the H3K27 methylation defect in Δnpf strains is not entirely attributable to PRC2 dysfunction. It is of obvious interest to identify all protein players involved in the establishment and maintenance of H3K27 methylation and to delineate their respective roles. Using a forward genetic selection for factors defective in Polycomb silencing in N. crassa, we report the isolation of mutant alleles of a previously uncharacterized gene (NCU04278) necessary for subtelomeric H3K27 methylation and silencing of associated genes. Immunoprecipitation followed by mass spectrometry (IP-MS) of NCU04278-interacting proteins demonstrated that NCU04278 is a PRC2 accessory subunit, and we therefore named it PAS. PAS homologs are present in lineages of both Sordariomycetes and Leotiomycetes, suggesting that PAS may play a crucial role in regulating PRC2 in a wide variety of fungal species.

RESULTS

Isolation, mapping, and identification of a potential Polycomb group gene, pas (NCU04278).

We recently described a genetic scheme to select for mutants defective in Polycomb silencing in N. crassa (12). Briefly, a wild-type strain bearing two antibiotic resistance genes repressed by H3K27 methylation, hph and nat-1, was subjected to UV radiation, and mutants resistant to both hygromycin B and nourseothricin were isolated. One mutant recovered in this fashion, which we now designate pas^UV1, led to antibiotic resistance comparable to loss of the EZH2 homolog, SET-7, which is responsible for all known H3K27 methylation in N. crassa (Fig. 1A) (10). We mapped the causative mutation in the pas^UV1 strain, which was in the Oak Ridge genetic background, by crossing the strain to the highly polymorphic Mauriceville wild-type strain (13), pooling genomic DNA from antibiotic-resistant progeny, and scoring the percentage of Oak Ridge single nucleotide polymorphisms (SNPs) from whole-genome sequencing data (14). Oak Ridge SNPs were enriched on the right arm of linkage group (LG) V in a region that included a frameshift mutation in NCU04278 (designated M660fs; ATG → TTTG, mutation underlined) (Fig. 1B).

FIG 1.

Identification of NCU04278 as a potential component of the Polycomb silencing pathway. (A) Serial dilution spot test silencing assay for the indicated strains plated on the indicated media. All strains harbor P_NCU05173::hph and P_NCU07152::nat-1. (B) Whole-genome sequencing of pooled pas^UV1 mutant genomic DNA identified a region on the right arm of LG V, which is enriched for Oak Ridge single nucleotide polymorphisms (SNPs) and which contained an insertional frameshift in NCU04278 (protein structure is represented as a solid bar as it contains no predicted domains). Each point represents a running average (window size, 10 SNPs; step size, 1 SNP). (C) Serial dilution spot test silencing assay for the indicated strains. The pas^UV1 NCU04278^WT strain has a wild-type (WT) copy of NCU04278 at the his-3 locus. All strains harbor P_NCU05173::hph. (D) The same experiment as described in panel B but performed for pas^UV2, which contains a premature stop codon in NCU04278. (E) Schematic of the wild-type primary structure of NCU04278 and structural changes of NCU04278 in the two pas alleles. Predicted disordered regions of NCU04278 are displayed above. Light gray bars represent the wild-type sequence, and the dark gray bar represents the frame-shifted sequence.

To verify that the observed mutation in NCU04278 is the pertinent mutation in pas^UV1, we targeted a wild-type copy of NCU04278 to the his-3 locus in a pas^UV1 mutant strain. The ectopic copy of NCU04278 complemented the pas^UV1 mutant; i.e., it restored hygromycin B sensitivity (Fig. 1C). In addition, deletion of the wild-type allele of NCU04278 was sufficient to confer hygromycin B resistance (Fig. 1C). We subsequently isolated and mapped a second allele of NCU04278 (pas^UV2; designated K552*; AAA → TAA, mutation underlined) (Fig. 1D and E), further confirming the involvement of NCU04278 in the antibiotic-resistant phenotype. Strikingly, NCU04278 encodes a protein with no annotated domains and is predicted to be approximately 80% structurally disordered (Fig. 1E).

PAS is necessary for silencing subtelomeric H3K27-methylated genes.

Although our forward genetic selection was designed to identify novel components of the Polycomb repression pathway, in principle, mutations might confer resistance to hygromycin B and nourseothricin in some manner independent of derepression of the H3K27-methylated antibiotic resistance genes, such as by stimulating drug efflux (15) or by a global effect on transcription, leading to nonspecific derepression of hph and nat-1. To determine if loss of PAS has specific defects in Polycomb silencing, we performed mRNA sequencing (mRNA-seq) on Δpas and wild-type strains in biological replicates and compared the gene expression profiles with previously generated wild-type and Δset-7 data sets (16). We found that 33 genes were upregulated and that 12 were downregulated greater than 2-fold in Δpas strains compared to levels in wild-type strains (Fig. 2A and B). Although less than 9% of all genes are H3K27 methylated in a wild-type strain, 64% of the upregulated genes in Δpas strains were in this select group. Moreover, there was significant overlap between the upregulated genes in Δpas and Δset-7 strains (P = 6.0 × 10⁻²⁹) though loss of SET-7 appeared to derepress more H3K27-methylated genes (Fig. 2A); i.e., Δpas strains derepress a subset of SET-7 targets.

FIG 2.

Loss of PAS upregulates a subset of SET-7-repressed genes near chromosome ends. (A) Venn diagram depicting genes that appear upregulated by mRNA-seq in both Δpas and Δset-7 strains (numbers in bold), only Δpas strains, or only Δset-7 strains, using a significance cutoff given by log₂(mutant expression/wild-type expression) > 1 and P < 0.05. The percentage of total upregulated genes that are H3K27 methylated for each gene set is indicated below the total gene count. Significance of overlapping gene sets was determined using a hypergeometric test. (B) The same experiment as described for panel A but for downregulated genes [log₂(mutant expression/wild-type expression) < −1]. (C) Box-and-whisker plot of the distances from the telomere of H3K27-methylated genes that appear upregulated by mRNA-seq in the indicated genotypes. Boxes represent interquartile range, horizontal lines represent the median, and whiskers represent minimum and maximum values (***, P < 0.001, two-tailed Mann-Whitney test). (D) RT-qPCR of H3K27-methylated genes that were replaced with antibiotic resistance genes (NCU07152 and NCU05173) and used for initial selection of mutants and H3K27-methylated genes that appeared upregulated in both Δpas and Δset-7 strains by mRNA-seq (NCU09633, NCU04991, NCU09604, NCU09306, NCU10070, and NCU10038). Each value was normalized to expression of NCU02840 (40) and is presented relative to that of the wild type. Filled bars represent the means from biological triplicates, and error bars show standard deviations (****, P < 0.0001; ***, P < 0.001; **, P < 0.01; *, P < 0.05; ns, not significant, for results compared to those for the wild type by a two-tailed, unpaired t test).

We previously demonstrated that there are at least two categories of H3K27 methylation in N. crassa: telomere dependent and telomere independent (17). Because loss of PAS affected only a subset of SET-7 targets, we considered the possibility that PAS is responsible for specifically silencing one category of H3K27-methylated genes, i.e., those that are telomere proximal or telomere distal. We therefore analyzed the genomic location of H3K27-methylated genes upregulated in Δpas and Δset-7 strains, respectively, and found that the genes derepressed by loss of PAS are closer to the chromosome ends than the genes derepressed by loss of SET-7 (P < 0.0005) (Fig. 2C). This suggested that PAS is responsible for silencing telomere-proximal H3K27-methylated genes.

We verified the results of our mRNA-seq experiments by performing reverse transcription followed by quantitative PCR (RT-qPCR) on RNA isolated from biological triplicates of wild-type, Δset-7, and Δpas strains (Fig. 2D). In addition to confirming the derepression of the native genes replaced by the antibiotic resistance genes in the selection strain (NCU07152 and NCU05173), we confirmed a significant increase in gene expression for five out of six H3K27-methylated genes that appeared upregulated, as measured by mRNA-seq in both Δpas and Δset-7 strains (Fig. 2D). We conclude that PAS represses a subset of SET-7 targets that are telomere proximal.

PAS is necessary for subtelomeric H3K27 methylation.

Considering that the loss of PAS led to the derepression of H3K27-methylated genes near the chromosome ends, we wondered if there was also a concomitant loss of H3K27 methylation at subtelomeric regions. To address this possibility, we performed chromatin immunoprecipitation of di- and trimethylated H3K27 (H3K27me2/3) followed by sequencing (ChIP-seq) on two Δpas siblings and compared the results with the distribution of H3K27me2/3 in a wild-type strain (18) (Fig. 3A). We found that the general distribution of H3K27me2/3 in Δpas strains was comparable to that of wild-type strains, except for clear losses near the chromosome ends, including the genes NCU05173 and NCU07152 that were replaced with antibiotic resistance cassettes (Fig. 3A). A Western blot for H3K27me3 confirmed that Δpas strains had reduced global H3K27me3 (Fig. 3B). Detailed analysis of the H3K27me2/3 ChIP-seq results revealed that 284 genes in Δpas strains had at least 2-fold reductions in H3K27me2/3 compared to levels of the wild-type strains while 34 genes showed at least 2-fold-greater H3K27me2/3 levels than wild-type strains (Fig. 3C). The set of genes that lost H3K27me2/3 in Δpas strains were significantly closer to telomeres than all genes marked by H3K27me2/3 in the wild type (P < 0.0001) (Fig. 3D). To confirm the H3K27me2/3 ChIP-seq results, we performed H3K27me2/3 ChIP followed by qPCR (ChIP-qPCR) on wild-type, Δset-7, and Δpas strains in biological triplicate. We confirmed the findings for three regions that appeared to have no change in H3K27me2/3 (NCU05086, NCU08085, and NCU08251), one region with an apparent gain in H3K27me2/3 (NCU07801), and three regions with apparent losses in H3K27me2/3 (NCU05173, NCU07152, and telomere IIIR) (Fig. 3E). Thus, PAS is critical for normal subtelomeric H3K27 methylation.

FIG 3.

Loss of PAS results in global reduction of H3K27 methylation. (A) H3K27me2/3 ChIP-seq tracks from merged biological replicates of wild-type, Δpas, and Δnpf strains displayed for all seven linkage groups (LGs) of N. crassa. LG I is split at the right end of its centromere into IL and IR. The y axis units are reads per kilobase per million. Scale bar, 500 kb. (B) Western blot showing H3K27me3 and total histone H3 (hH3) in the indicated strains. Biological replicates are shown. The same lysate was analyzed on separate gels, and hH3 was used as a sample processing control. The number displayed below each genotype is the average ratio of the mutant H3K27me3 level (normalized to the hH3 level) to the wild-type H3K27me3 level (normalized to the hH3 level). (C) Scatter plot showing the correlation of H3K27me2/3 levels at all genes (dots) in wild-type and Δpas strains based on merged biological replicates of ChIP-seq data. Dots representing genes are colored based on their expression in the Δpas strain compared to that in the wild type. Red, log₂(Δpas strain expression/wild-type expression) > 1; gray, −1 < log₂(Δpas strain expression/wild type expression) < 1; blue, log₂(Δpas strain expression/wild-type expression) < −1. Representative genes that gained (NCU07801) or lost (NCU07152) H3K27me2/3 in the Δpas strain are indicated. (D) Box-and-whisker plot of the distances from the telomere of genes that lose H3K27me2/3 in the indicated genotypes. Boxes represent interquartile range, horizontal lines represent the median, and whiskers represent minimum and maximum values (****, P < 0.0001, two-tailed Mann-Whitney test). (E) H3K27me2/3 ChIP-qPCR to validate ChIP-seq data at eight regions: NCU05086, NCU08085, and NCU08251 (retained H3K27me2/3); NCU07801 (gained H3K27me2/3 in the Δpas strain); NCU05173, NCU07152, and telomere IIIR (lost H3K27me2/3 in the Δpas strain); hH4 (negative control). Filled bars represent the mean for biological triplicates, and error bars show standard deviations (****, P < 0.0001; ***, P < 0.001; *, P < 0.05; ns not significant, for results relative to those of the wild type by a two-tailed unpaired t test).

Strains lacking PAS have H3K27 methylation defects similar to Δnpf strains.

Loss of subtelomeric H3K27 methylation has been previously observed in strains lacking the PRC2 component NPF (10). For this reason, we repeated H3K27me2/3 ChIP-seq on Δnpf siblings to compare the result with the distribution of H3K27me2/3 observed in Δpas strains (Fig. 3A). We found that while the distribution of H3K27 methylation in each strain was unique, the overall changes in the two strains were similar (Fig. 4A). Indeed, comparison of H3K27me2/3 levels over all genes using Spearman’s correlation coefficient showed that Δnpf and Δpas strains are more similar to each other than they are to wild-type strains (Fig. 4B). Interestingly, unlike Δnpf strains (10), strains lacking PAS do not exhibit a linear growth defect (Fig. 4C). Therefore, the phenotypic consequences of losing NPF or PAS are not equivalent.

FIG 4.

H3K27 methylation defects in Δpas and Δnpf strains are similar. (A) H3K27me2/3 ChIP-seq of merged biological replicates of wild-type, Δpas, and Δnpf strains for select genomic regions. The y axis units are reads per kilobase per million. (B) Spearman correlation matrix comparing H3K27me2/3 levels of all genes in ChIP-seq biological replicates of wild-type, Δpas, and Δnpf strains. Spearman correlation coefficients are displayed, and associated boxes are color coded according to the legend. (C) Linear growth rates measured by race tubes (41) are shown for two biological replicates of wild-type (N3752 and N3753), Δset-7 (N4718 and N4730), Δpas (N7999 and N8002), and Δnpf (N8003 and N8004) strains. Lines represent the means from three technical replicates (circles). (D) Histogram of wild-type gene expression levels of all genes and genes that specifically gain H3K27me2/3 in Δpas strains.

Relationship between gene expression changes and H3K27 methylation levels in Δpas strains.

To test for a correlation between the gene expression changes in Δpas strains and the losses or gains of H3K27 methylation in a Δpas strain, we examined the intersection of these two data sets (see the supplemental material). We found that while the loss of H3K27 methylation in Δpas strains was not sufficient for gene activation, as with loss of H3K27 methylation in Δset-7 strains, all but one (NCU08790) of the 21 H3K27-methylated genes upregulated in Δpas strains lost H3K27 methylation. In contrast, there was no overlap between the genes that gained H3K27 methylation in Δpas strains and the genes that were downregulated in Δpas strains. Analysis of the genes that gained H3K27me2/3 in Δpas strains revealed that the majority are barely expressed in wild-type strains (Fig. 4D). In summary, the loss of H3K27 methylation in Δpas strains is associated with, but not sufficient for, increased gene transcription, and the gain of H3K27 methylation in Δpas strains is mostly associated with genes that are normally transcribed at low levels.

PAS is an accessory subunit of PRC2.

To determine if PAS acts in a protein complex, we immunopurified PAS fused to three copies of a FLAG tag (PAS-3×FLAG) from N. crassa lysate and identified its copurifying proteins by mass spectrometry. In addition to PAS itself, we identified the four known components of PRC2: SET-7, SUZ12, EED, and NPF (Fig. 5A) (10). We also immunopurified SUZ12-3×FLAG and a negative control (3×FLAG–EPR-1) (12) and analyzed their copurifying proteins by mass spectrometry. In addition, we reexamined previously collected mass spectrometry data from a 3×FLAG-EED purification (10). The purifications of the known PRC2 components, SUZ12 and EED, yielded PAS as well as the other known members of PRC2 (Fig. 5A). The negative-control (3×FLAG–EPR-1) purification did not detect any peptides of SET-7, SUZ12, EED, or PAS (Fig. 5A). Besides the proteins listed in Fig. 5A, the only other proteins with peptides detected in the PAS, SUZ12, and EED purifications but absent in the control (EPR-1) were NCU01249 (importin α), NCU02407 (dihydrolipoyl dehydrogenase), and NCU06482 (pyruvate dehydrogenase). Thus, it appears that PRC2 in N. crassa can form a five-member complex and that PAS is an accessory component. For this reason, we have designated NCU04278 a PRC2 accessory subunit (PAS) (Fig. 5B).

FIG 5.

PAS is a PRC2 accessory subunit with homologs in other fungi. (A) Exclusive unique peptide counts from immunoprecipitation followed by mass spectrometry (IP-MS) experiments. Each column represents the results of a single IP-MS experiment with the indicated baits. The 3×FLAG-EED data set was previously published (10). The 3×FLAG–EPR-1 sample is a negative control. (B) Diagram of putative PRC2 complex containing PAS. (Adapted from reference 30 with permission of the publisher.) (C) Phylogenetic tree highlighting representative fungal species in Sordariomycetes and Leotiomycetes that have predicted PAS homologs. Length, in amino acids (AA), of predicted PAS homologs in representative species and results of the Basic Local Alignment Search Tool (BLAST), using Neurospora crassa PAS as the query, for each of the indicated species, are also displayed. Colored bars indicate an aligned region. Gray lines are for visual reference. Amino acid positions in PAS are indicated below the alignments. Alignment scores are color coded according to the legend.

PAS homologs are present in a variety of fungal lineages.

In an effort to determine if homologs of PAS exist outside N. crassa, we performed an iterative sequence similarity search (19). We detected homologs of PAS in lineages of Sordariomycetes and Leotiomycetes (Fig. 5C). The majority of these fungal species have H3K27 methylation either experimentally validated or inferred from their genomic sequences (20), supporting the notion that the detected PAS homologs associate with PRC2. Sequence alignments of the detected PAS homologs from representative species compared to the sequence of N. crassa PAS revealed that significant sequence similarity between the proteins is restricted to the C terminus of N. crassa PAS (Fig. 5C). In addition, the total predicted length of PAS homologs is variable (Fig. 5C), further highlighting the divergence of these homologs. We conclude that while apparent PAS homologs are in lineages of Sordariomycetes and Leotiomycetes, they exhibit considerable sequence divergence.

DISCUSSION

Ever since the identification of Polycomb response elements, sequences that reliably induce H3K27 methylation in Drosophila melanogaster (21), researchers have searched for cis- and trans-acting factors that direct PRC2 activity in other organisms. In plants, this line of inquiry has been somewhat fruitful, with the identification of sequence-specific transcription factors that directly recruit PRC2 (22–24). In mammals, unmethylated CpG islands can recruit PRC2 (25–27), but the mechanism of this recruitment is unclear. Recently, noncore subunits of mammalian PRC2.1 and PRC2.2 were demonstrated to be required to target PRC2 (28, 29), but how these structurally diverse components target specific genomic sites is unknown. Defining the full complement of cis- and trans-acting factors that recruit PRC2 and understanding their mechanism of action remain an outstanding challenge.

Utilizing a relatively simple eukaryote, the fungus N. crassa, we performed a forward genetic selection to identify novel factors implicated in Polycomb repression. This yielded an unknown accessory component of N. crassa PRC2, PAS (NCU04278), that is necessary for subtelomeric H3K27 methylation and associated gene silencing. Interestingly, the H3K27 methylation defect observed in Δpas strains bears a striking resemblance to that of strains lacking the PRC2 component NPF (RBBP4/7 in mammals; p55 in Drosophila) (Fig. 3A) although their respective defects are distinct (Fig. 4A). Homologs of NPF are known to associate with other protein complexes (11); therefore, the observed H3K27 methylation defects in Δnpf strains could, in principle, be due to non-PRC2 activities of NPF. However, the finding that PAS results in H3K27 methylation defects similar to those of Δnpf strains but appears to interact only with PRC2 components suggests that the H3K27 methylation defects in Δnpf strains are largely due to the dysfunction of PRC2. Considering that the in vivo role of mammalian NPF homologs (RBBP4/7) in PRC2 function cannot be tested since they are essential for proliferation (30), studies from more amenable organisms, such as N. crassa, are extremely valuable.

Our research group has previously identified telomere repeats, (TTAGGG)_n, as effective inducers of H3K27 methylation at chromosome ends in N. crassa (17). This subtelomeric H3K27 methylation is largely lost in Δpas, Δnpf, and Δtert (telomerase reverse transcriptase) strains alike (Fig. 3A) (17). However, the mechanism connecting these cis- and trans-acting factors is unclear. In mammals, the accessory subunits of PRC2 are known to affect both the catalytic activity and the recruitment of PRC2 to chromatin (28–30). Conceivably, PAS and NPF may work together to stimulate PRC2 activity on nonideal subtelomeric targets and play no role in recruitment. It is also possible, however, that they recruit PRC2 specifically to subtelomeres without influencing catalysis. Both activity and recruitment models can equally account for the observed losses and gains of H3K27 methylation in Δpas strains.

The location of the mutations in the identified pas^UV alleles and the sequence alignments of predicted PAS homologs demonstrate that the C terminus of PAS is a critical, conserved region. This invariant C-terminal region may be responsible for assembly of PAS into PRC2, whereas the divergent N termini may dictate chromatin targeting or regulation of PRC2 activity that is species specific. Although N. crassa PAS is responsible for subtelomeric H3K27 methylation, homologs of PAS in other species may be responsible for H3K27 methylation present at diverse genomic regions. Future work in other fungal species should elucidate the conservation of function, or lack thereof, of PAS-containing PRC2 complexes.

MATERIALS AND METHODS

Strains, media, and growth conditions.

All N. crassa strains used in this study are listed in Table S1 in the supplemental material. Preparation of media and growth conditions for experiments was carried out as previously described (12).

Selection for mutants defective in H3K27 methylation-mediated silencing.

Conidia from strain N6279 were mutagenized with UV light and challenged with hygromycin B and nourseothricin to select for antibiotic-resistant colonies as previously described (12). Primary mutants were rendered homokaryotic by crossing to strain N3756.

Whole-genome sequencing, mapping, and identification of pas^UV1 and pas^UV2 alleles.

Whole-genome sequencing and mapping were performed as previously described (12). Briefly, antibiotic-resistant, homokaryotic mutants were crossed to a Mauriceville wild-type strain (13), and antibiotic-resistant progeny were pooled for whole-genome sequencing. Mapping of the critical mutations was performed as previously described (12, 14). FreeBayes and VCFtools were used to identify novel genetic variants present in our pooled mutant genomic DNA (31, 32).

RNA isolation, RT-qPCR, and mRNA-seq.

Extraction of total RNA from germinated conidia was performed as previously described (12) and used either for cDNA synthesis and subsequent qPCR (see Table S5 in the supplemental material for primers) (12) or for mRNA-seq library preparation (16). Mapping and analysis of gene expression levels were performed as previously described (12).

ChIP, ChIP-qPCR, and ChIP-seq.

H3K27me2/3 chromatin immunoprecipitation (ChIP) using anti-H3K27me2/3 antibody (39536; Active Motif) was performed as previously described (12), and the isolated DNA was used for qPCR (see Table S6 in the supplemental material for primers) or prepared for sequencing (12). Mapping, visualization, and analysis of H3K27me2/3 ChIP sequencing (ChIP-seq) reads were performed as previously described (12). Spearman’s correlation coefficient analysis was performed using deepTools2 (33) on the Galaxy public server (34).

Western blotting.

N. crassa tissue lysates were prepared as previously described (12) and used for Western analysis. Anti-H3K27me3 (9733; Cell Signaling Technology) and anti-histone H3 (hH3) (ab1791; Abcam) primary antibodies were used with IRDye 680RD goat anti-rabbit secondary antibody (926-68071; LI-COR). Images were acquired with an Odyssey Fc imaging system (LI-COR) and analyzed with Image Studio software (LI-COR).

Immunoprecipitation followed by mass spectrometry (IP-MS).

Strain N7807 (overexpressing NCU04278::10×Gly::3×FLAG) or strain N4666 (endogenous suz12::10×Gly::3×FLAG) was grown for 7 to 10 days in a 250-ml flask containing 50 ml of Vogel’s minimal medium containing 1.5% sucrose and 1.5% agar. Approximately 1 × 10⁹ conidia were collected from each culture and filtered through sterile cheesecloth. Filtered conidia were used to inoculate 1 liter of Vogel’s minimal medium containing 1.5% sucrose and then grown for 16 h with shaking (150 rpm) at 30°C. Tissue was collected by filtration using a Buchner funnel and washed with water. Dry tissue (∼10 g) was ground using a 6870 Freezer/Mill Cryogenic Grinder (SPEXSamplePrep). Ground tissue was added to 40 ml of extraction buffer (EB) (50 mM HEPES [pH 7.5], 150 mM NaCl, 10 mM EDTA, 10% glycerol, 0.02% NP-40, 1× Halt protease inhibitor cocktail [78438; Thermo Scientific]) to achieve a final volume of 50 ml and then suspended by rotation at 4°C for 1.5 h. Insoluble material was pelleted with one 10-min centrifugation step at 2,000 rpm and two consecutive 10-min centrifugation steps at 8,000 rpm. Soluble material was precleared with 250 μl of equilibrated protein A-agarose (15918014; Invitrogen) with rotation for 1 h at 4°C. The protein A-agarose was pelleted by centrifugation at 2,000 rpm, and the supernatant was collected and then incubated with 400 μl of equilibrated anti-FLAG M2 affinity gel (A2220; Sigma-Aldrich) overnight, with rotation, at 4°C. Resin was pelleted by centrifugation at 1,000 rpm, and the supernatant was removed. Resin was washed with EB, rotated at 4°C for 10 min, and pelleted with spinning at 1,000 rpm five consecutive times. All liquid was removed after the final spin. Protein was eluted twice from the resin by incubation with 300 μl of 500 μg/ml 3×Flag peptide (A6001; APExBIO) with rotation at 4°C for 20 min and one final wash with 300 μl of EB. Eluate was precipitated with trichloroacetic acid (10% final concentration) on ice for 1 h, pelleted by centrifugation at 14,000 rpm, and washed three times with ice-cold acetone. The pellet was air dried by placing it in a heat block at 100°C for 30 s. Samples were sent to and processed by the University of California Davis Proteomics Core Facility for mass spectrometry and subsequent analysis.

Bioinformatic analysis of PAS.

Structurally disordered regions of N. crassa PAS (GenBank accession number Q1K790) were predicted using MobiDB (35). Homologs of PAS were detected using a JACKHMMER iterative search (19), which converged after three consecutive iterations. Regions of significant alignment between detected PAS homologs from representative species and N. crassa PAS were analyzed using the BLAST web server (36). Phylogenetic relationships between fungal species were based on previous work (37).

Replacement of NCU04278 with trpC::nat-1.

The 5′ and 3′ flanks of NCU04278 were PCR amplified from wild-type genomic DNA with the primer pair 6405 and 6406 (5′) and the pair 6409 and 6410 (3′) (see Table S3 in supplemental material). The 5′ and 3′ flanks were separately PCR stitched to plasmid 3237 (source of trpC::nat-1) using the primer pair 6401 and 4883 and the pair 4882 and 6351, respectively. These two split-marker PCR products were cotransformed into strain N2930, and NCU04278 gene replacements were selected on nourseothricin-containing medium.

Generation of an N. crassa strain expressing PAS-3×FLAG.

NCU04278 was PCR amplified from wild-type genomic DNA with primers 6397 and 6398 (Table S4) and cloned into plasmid 2401 (38) using XbaI and PacI restriction sites to create plasmid 3340 (Table S2). Plasmid 3340 was subjected to Sanger sequencing using primers 6397, 6407, 6426, 6427, and 6715 to verify that the sequence matched that of the wild type. Plasmid 3340 was linearized with NdeI and targeted to his-3 in N6762, as previously described (39). A his-3⁺ primary transformant was then crossed to N7742 to generate N7807.

Data availability.

Whole-genome sequencing data from pas^UV1 and pas^UV2 mapping experiments are available from the NCBI Sequence Read Archive under accession number PRJNA559544. The results of our mRNA-seq and ChIP-seq experiments are available on the NCBI Gene Expression Omnibus (GEO) database under accession number GSE140787. Results of the mass spectrometry experiments are included in the supplemental material.