Synergistic regulation by H3K36 and H3K27 methylation defines the chromatin landscape to control virulence and secondary metabolism in a fungal pathogen

Distribution and phylogenetic analysis of SET domain proteins in fungi

To understand the diversity of SET domain proteins in fungal evolution, we performed domain-based functional annotations for all 23.64 million proteins from 1038 publicly available fungal genomes across nine phyla. This analysis identified 18,718 SET domain-containing proteins (PF00856) (Supplemental Table S1), averaging 18 per species but varying widely across phyla. A significant positive correlation (R = 0.52, P < 2.2 × 10⁻¹⁶) was found between total genes and SET domain abundance (Fig. 1A), suggesting a link between this family's evolution and genome expansion. We further performed a CLANS cluster analysis of these 18,718 proteins based on sequence similarity (Supplemental Fig. S1A), dividing them into three distinct clusters (Fig. 1B). Notably, nearly all known fungal HMTs with SET domains grouped into cluster 1, indicating evolutionary conservation (Fig. 1B,C; Supplemental Tables S2, S3). Given the critical role of HMTs in gene regulation, we focused on cluster 1 for further in-depth analysis. Phylogenetic analysis showed cluster 1 proteins formed distinct branches corresponding to different methylation types: Set1 (H3K4), Kmt1 (H3K9), Kmt6 (H3K27), and Kmt5 (H4K20). However, the SET2 family, responsible for H3K36 methylation, was divided into two subfamilies: Set2 and Ash1.

Figure 1.

Distribution and phylogeny of SET domain proteins across fungi kingdom. (A) Correlation between SET domain protein number and total gene count in fungal genomes. Pearson correlation coefficient: R = 0.52, P < 2.2 × 10⁻¹⁶. (B) Cluster analysis of 18,718 SET domain proteins from 1038 fungal species. Thirty groups within three clusters are shown by distinct colors and shapes. (C) Phylogenetic tree of cluster 1 SET domain proteins, highlighting histone methyltransferase (HMT) clades. Scale bar = 1. (D) Distribution of key HMT families (Set1, Kmt1, Kmt5, Kmt6) and subfamilies (Set2, Ash1) across seven fungal phyla. Average copy number per class is shown.

Open in new tabDownload PowerPointLink to figure

The distribution analysis revealed that these key HMTs are relatively conserved across fungi. However, the Kmt6 family and Ash1 subfamily appear to be less conserved (Fig. 1D), being absent in 417 and 382 species, respectively (Supplemental Fig. S1B). Approximately 36.8% of species lack Ash1 subfamily members, prompting an investigation into the evolutionary relationship between Set2 and Ash1. A phylogenetic tree showed both subfamilies forming distinct branches (Supplemental Fig. S1C), suggesting independent evolution and functional divergence. Additionally, we observed copy number variations for Set2 (≥2 copies in 108 of 1038 genomes) and Ash1 (≥2 copies in 35 of 1038 genomes) (Supplemental Fig. S1D,E). These results indicated that species-specific deletion or duplication events drive Set2 family variation. In summary, fungal SET domain protein numbers correlate with genome size, and fungal HMTs, particularly the Set2 family, exhibit extensive copy number variations. Next, we will use F. graminearum as a model to explore the biological functions of HMT candidates in cluster 1.

Cluster 1 SET domain proteins regulate histone methylation and pathogenicity in F. graminearum

F. graminearum harbors 19 SET domain proteins, with seven in cluster 1, 10 in cluster 2, and two in cluster 3 (Supplemental Table S4). To investigate the functions of these seven HMT candidates, we constructed deletion mutants for each gene and analyzed their roles in histone methylation and biological phenotypes. All mutants, except Δset3, exhibited significant hyphal growth defects on complete medium compared to wild-type PH-1 (Fig. 2A,B). Notably, Δash1 showed a particularly severe growth reduction, with only 5% of the wild-type relative growth rate. Next, we performed western blotting assays with site-specific antibodies for histone methylation to detect the impact of each mutant on histone methylation levels. Consistent with previous studies, Set1, Kmt1, Kmt6, and Kmt5 are responsible for H3K4, H3K9, H3K27, and H4K20 methylation, respectively, in fungi (Connolly et al. 2013; Liu et al. 2015; Bachleitner et al. 2021). Deletion of these genes resulted in complete loss or significant reduction of corresponding histone methylation levels in F. graminearum (Fig. 2C,D). Notably, H3K36me3 levels were significantly reduced in both Δset2 and Δash1 mutants, indicating that Set2 and Ash1 function as H3K36 methyltransferases in F. graminearum. In contrast, the Δset3 mutant shows no changes in tested histone methylation levels.

Figure 2.

Role of SET domain proteins in histone modification and pathogenicity in F. graminearum. (A) Colony morphologies of wild-type PH-1 and seven individual SET domain deletion mutants grown on complete medium (CM) for 3 days. (B) Colony diameters of strains in A. Data are mean ± standard deviation (s.d.) (n = 3). Different letters indicate statistically significant differences (ANOVA, Tukey's multiple comparisons test, P < 0.05). (C) Western blot analysis of PH-1 and deletion mutants using antibodies against H3K4me3, H3K9me3, H3K27me3, H3K36me3, and H4K20me3. Anti-H3 served as a loading control. (D) Relative histone methylation levels, normalized to H3, quantified by ImageJ software. PH-1 band intensity was set to 1. (E) Symptoms on flowering wheat heads inoculated with conidial suspensions of the indicated strains, photographed at 15 days postinoculation (dpi). (F) Symptoms on wheat seedling stems inoculated with mycelial plugs from the indicated strains, photographed at 5 dpi.

Open in new tabDownload PowerPointLink to figure

Mutant pathogenicity was evaluated by point-inoculating conidial suspensions into flowering wheat heads and placing mycelial plugs on wheat seedling stems. In wheat heads, wild-type PH-1 inoculation resulted in scab symptoms developing on inoculated spikelets and rapidly spreading to neighboring spikelets. Fifteen days postinoculation, wild-type PH-1 caused severe, typical scab symptoms (Fig. 2E). In contrast, all mutants, except Δset3, showed significantly reduced virulence. The Δset1, Δkmt6, and Δash1 mutants were nonpathogenic, whereas the Δset2 mutant exhibited scab symptoms restricted to the inoculated spikelets without rachis spread (Fig. 2E). The Δkmt1 and Δkmt5 mutants infected the inoculated spikelets and spread to neighboring spikelets, but their virulence was significantly reduced compared to the wild type. Additionally, these mutants demonstrated attenuated virulence on wheat seedlings (Fig. 2F). Collectively, six of seven SET domain proteins in cluster 1 regulate histone methylation, growth, and pathogenicity in F. graminearum. Given that the roles of Set1, Kmt1, Kmt6, and Kmt5 are already characterized, we will specifically focus on Set2 and Ash1 to investigate their distinct or overlapping contributions in regulating H3K36 methylation, gene expression, and biological functions.

Distinct roles of Set2 and Ash1 in H3K36 methylation, development, virulence, and stress response in F. graminearum

To determine if F. graminearum Set2 and Ash1 possess H3K36 methyltransferase activity, we heterologously expressed and purified them in Escherichia coli for in vitro enzyme assays. A recombinant H3K36 peptide, devoid of pre-existing histone modifications, served as the substrate, with S-adenosylmethionine (SAM) as the methyl donor. Western blotting and LC–MS/MS analyses evaluated the reaction products’ methylation status. Western blotting revealed specific H3K36me2 and H3K36me3 signals when probed with corresponding antibodies (Fig. 3A). LC–MS/MS analysis further showed mass shifts of 28.03 Da and 42.05 Da at the H3K36 residue, consistent with di- and trimethylation (Supplemental Fig. S2; Supplemental Table S5). These findings confirm that both Set2 and Ash1 possess in vitro H3K36 methyltransferase activity. We then assessed the in vivo H3K36 methyltransferase activity of Set2 and Ash1 and explored their roles in various biological processes. Localization analyses with GFP-tagged Set2 and Ash1 demonstrated nuclear localization in mycelia, confirmed by colocalization with H1-mCherry (Supplemental Fig. S3). We then evaluated relative H3K36me2 and H3K36me3 levels via western blotting, using specific antibodies, in single (Δset2, Δash1), double (Δset2/ash1), and H3K36A (H3K36 substituted with alanine) mutants. As shown in Figure 3, B and C, SET2 mutation significantly reduced global H3K36me3 levels to ∼10% of wild-type, whereas ∼80% H3K36me3 was retained in Δash1. In both single mutants, global H3K36me2 levels decreased to ∼45% of PH-1. Importantly, H3K36me2 and H3K36me3 levels were completely abolished in both the Δset2/ash1 and H3K36A mutants (Fig. 3B,C). These findings suggest that Set2 and Ash1 redundantly contribute to full H3K36me2/3 activity in F. graminearum, with Set2 playing a more dominant role in regulating H3K36me3.

Figure 3.

Set2 and Ash1 regulate development, virulence, and stress responses in F. graminearum. (A) Western blot analysis of Set2 and Ash1 in vitro HMT activity, probed with anti-H3K36me2 and anti-H3K36me3 antibodies. (B) H3K36me2 and H3K36me3 levels in PH-1, deletion mutants, and complemented strains. Anti-H3 served as a loading control. (C) Quantification of H3K36me2 and H3K36me3 signal intensities using ImageJ software. PH-1 intensity was set to 100%. (n.d.) not detectable. Different letters indicate significant differences among all groups shown. (D) Colony diameters of PH-1 and mutant strains after 3 days on PDA and CM. Different letters indicate statistically significant differences among all tested groups. Representative colony morphologies are shown in E. (F) Conidial production of strains grown in carboxymethyl cellulose (CMC) liquid medium for 5 days. Different letters indicate statistically significant differences among strains. (G) Virulence of PH-1 and mutants on wheat heads after point-inoculated with conidia, photographed at 15 dpi. (H) Deoxynivalenol (DON) production by PH-1 and mutants after 7 days in trichothecene biosynthesis-inducing (TBI) medium. Different letters indicate statistically significant differences among strains. (I) DON toxisome formation in PH-1, Δset2, and Δash1 visualized by Tri1-GFP after 36 or 48 h in TBI medium. Scale bar, 5 µm. (J) Heat map showing sensitivity of PH-1 and mutants to various abiotic stresses. Growth inhibition of PH-1 under each condition was normalized to 1.0. Data are mean ± standard deviation (n = 3). For panels C, D, F, and H, data are presented as mean ± standard deviation (s.d.) from three independent experiments. Statistical analysis was performed using ANOVA with Tukey's multiple comparisons test (P < 0.05).

Open in new tabDownload PowerPointLink to figure

We next assessed fungal development, pathogenicity, and stress tolerance in these deletion mutants and their complemented strains. All mutants exhibited significant growth defects on both tested media compared to PH-1, with Δash1 displaying the most severe reduction (only 5% of the growth rate of PH-1). Notably, Δset2/ash1 partially restored its growth rate compared to Δash1 (Fig. 3D,E). Additionally, conidiation was significantly reduced across all mutants (Fig. 3F). Pathogenicity was evaluated by point-inoculating conidial suspensions onto flowering wheat heads (Fig. 3G). Fifteen days postinoculation, scab symptoms caused by Δset2 and Δset2/ash1 were restricted to inoculated spikelets, whereas Δash1 was entirely nonpathogenic. In contrast, PH-1 and the complemented mutants induced severe scab symptoms across the wheat heads (Fig. 3G). Deoxynivalenol (DON), a critical virulence factor, is essential for fungal spread within a spike. Given the mutants’ restricted spread or complete loss of pathogenicity, we quantified DON production. Cultivation in trichothecene biosynthesis induction (TBI) medium for 3 days revealed that all mutants produced significantly lower DON amounts compared to PH-1 and complemented strains (Fig. 3H). To further confirm, we investigated DON-toxisome formation using GFP-tagged Tri1 (Tri1-GFP) as a marker in TBI medium. As shown in Figure 3I, no toxisomes were detected in either Δset2 or Δash1, whereas they were clearly visible in PH-1. The H3K36A mutant mirrored Δset2/ash1 phenotypes, including defects in growth, conidiation, virulence, and DON biosynthesis.

Subsequently, we evaluated the sensitivity of all mutants to various abiotic stresses (Fig. 3J; Supplemental Fig. S4). Compared to PH-1, Δash1 and Δset2 exhibited increased sensitivity to osmotic stress agents KCl and NaCl and to the cell membrane stress agent sodium dodecyl sulfate (SDS). However, these single deletion mutants showed significantly increased tolerance to oxidative stress induced by hydrogen peroxide (H₂O₂) and vitamin K₃. Moreover, Δset2 displayed heightened sensitivity to the fungicide tebuconazole but increased tolerance to the ionic stress agent CaCl₂. In contrast, Δash1 exhibited an opposite pattern of sensitivity to tebuconazole and CaCl₂ relative to Δset2. Both Δash1/set2 and the H3K36A mutant showed increased sensitivity to all tested stress agents. Taken together, these findings indicate that Set2 and Ash1 have both overlapping and distinct functions in regulating fungal development, virulence, secondary metabolism, and response to diverse abiotic stresses.

Distinct contributions of Ash1 and Set2 to H3K36me3

To investigate the roles of Ash1 and Set2 in H3K36me3 and transcriptional regulation, we performed chromatin immunoprecipitation sequencing (ChIP-seq) to profile H3K36me3 distribution in PH-1, Δset2, and Δash1 strains. In PH-1, we identified 8390 significant H3K36me3 peaks. In contrast, Δset2 had 2366 peaks, whereas Δash1 showed 4610 peaks. Over 85% of peaks in both mutants overlapped with PH-1 (Fig. 4A), indicating substantial yet distinct contributions from the two enzymes. Δset2 showed increased H3K36me3 enrichment at transcriptional start sites (TSSs) but reduced methylation across gene bodies. Conversely, Δash1 displayed diminished H3K36me3 at TSSs, with other genomic regions largely unaffected. These observations suggest that Set2 primarily deposits H3K36me3 within gene bodies, whereas Ash1 preferentially targets TSS regions (Fig. 4B,C).

Figure 4.

Ash1 and Set2 deposit H3K36me3 in distinct genomic regions of F. graminearum. (A) Venn diagram of H3K36me3 peaks in PH-1, Δset2, and Δash1 strains. (B) Heat maps showing genome-wide H3K36me3 signal intensity across all genes within 2.0-kb genomic regions in PH-1, Δset2, and Δash1. TSS: Transcription start site; TTS: transcription termination site. (C) Metagene plots showing H3K36me3 distribution across target transcription units and adjacent intergenic regions in PH-1, Δset2, and Δash1. (D) Integrative Genomics Viewer (IGV) tracks of H3K36me3 ChIP-seq signals across four chromosomes in PH-1, Δset2, and Δash1. Y-axis: reads per million (RPM), normalized to total mapped reads. mRNA abundance profiles in PH-1 shown as reference.

Open in new tabDownload PowerPointLink to figure

Integrative Genomics Viewer (IGV) (Thorvaldsdóttir et al. 2013) tracks showed H3K36me3 in PH-1 is broadly distributed across all four chromosomes, with notable absence in centromeric regions (Fig. 4D). Δset2 and Δash1 mutants displayed distinct, mutually exclusive H3K36me3 patterns, collectively recapitulating the wild-type H3K36me3 landscape. Specifically, H3K36me3 was mainly enriched in telomeric and centromere-proximal regions in Δset2, whereas Δash1 showed H3K36me3 enrichment in euchromatic regions. To explore the functional significance of these patterns, we examined the relationship between H3K36me3 distribution and mRNA expression in PH-1 (Fig. 4D). Ash1-mediated H3K36me3 was enriched in genes with low or no expression, whereas Set2-mediated H3K36me3 was found in highly expressed genes. These findings suggest that Ash1 and Set2 contribute differently to H3K36me3 deposition, with Ash1 primarily in facultative heterochromatin regions and Set2 in transcriptionally active euchromatin.

Set2 and Ash1 differentially regulate H3K36me3-marked gene expression

To elucidate the regulatory functions of Set2- and Ash1-mediated H3K36me3, we integrated ChIP-seq with RNA-seq to correlate H3K36me3 enrichment and gene expression. We first identified genomic regions with significantly reduced H3K36me3 occupancy in Δset2 and Δash1 mutants. In Δset2, 5025 H3K36me3 peaks associated with 5680 genes were significantly diminished, whereas Δash1 showed 3810 peaks linked to 3813 genes with reduced H3K36me3 (Fig. 5A). Notably, in Δset2, 60.3% of reduced H3K36me3 signals were confined to gene bodies, whereas in Δash1, reductions occurred in both promoters (42.7%) and gene bodies (49.6%) (Supplemental Fig. S5A). We then performed RNA-seq on PH-1, Δset2, Δash1, and Δset2/ash1 under the same ChIP-seq conditions. Principal component analysis (PCA) revealed distinct transcriptomic profiles for Δset2 and Δash1 compared to PH-1, with the double mutant clustering closer to Δset2 (Fig. 5B). Compared to PH-1, Δset2 exhibited 1002 differentially expressed genes (DEGs), with 604 upregulated and 398 downregulated (Supplemental Fig. S5B). In contrast, Δash1 exhibited a more pronounced transcriptional effect, with 3317 DEGs (1893 upregulated and 1424 downregulated (Supplemental Fig. S5C). At the genomic scale, H3K36me3-marked genes in PH-1 did not differ significantly in expression from nonmarked genes (P = 0.062) (Fig. 5C). However, in the Δset2 and Δash1, H3K36me3-marked genes were downregulated in Δset2 but upregulated in Δash1, suggesting opposing regulatory roles for Set2 and Ash1 in gene expression (Fig. 5C). Beyond its role in gene regulation, Set2-mediated H3K36me3 suppresses cryptic transcription in other eukaryotes (Carrozza et al. 2005; Kim et al. 2016; DiFiore et al. 2020; Lee et al. 2021). To determine whether this function is conserved in F. graminearum, we analyzed read distributions across ORF regions to detect aberrant transcription initiation within gene bodies. Compared to PH-1, cryptic transcription emerged at 67 genes in Δset2, all marked by Set2-dependent H3K36me3 across their gene bodies (Supplemental Fig. S6; Supplemental Table S6). These findings indicate Set2-mediated H3K36me3 contributes to both transcriptional regulation and cryptic transcription suppression in F. graminearum.

Figure 5.

Distinct transcriptional roles of Set2 and Ash1 via H3K36me3 in F. graminearum. (A) Heat maps of Ash1- and Set2-dependent H3K36me3 peaks across all genes ±3.0 kb. (B) Principal component analysis (PCA) of transcriptomes for PH-1 and deletion mutants. (C) Violin plots of expression for genes with/without H3K36me3. Different letters indicate statistically significant differences among all groups shown. (D) Venn diagram: Δash1 upregulated versus Ash1 targets (top); Δset2 downregulated versus Set2 targets (bottom). RF: Representation factor. (E,F) Gene Ontology (GO) enrichment of overlapping sets in D. (G) IGV tracks of H3K36me3 and RNA expression of LEU1 in PH-1, Δset2, and Δash1. (H) Colony morphology of strains on FGA (fructose gelatin agar) medium with/without 0.3 mM leucine after 3 days. (I) Mycelial dry weight of strain of incubation in liquid FGM with/without 0.3 mM leucine after 3 days (25°C, 180 rpm). Different letters indicate statistically significant differences within each treatment condition. (J) IGV tracks of H3K36me3 and RNA expression of STF in PH-1, Δset2, and Δash1. (K) Perithecium formation in mating cultures at 14 days postfertilization. Scale bar, 0.5 cm. (L) IGV tracks of H3K36me3 and RNA expression of NRPS7 in PH-1, Δset2, and Δash1. (M) Ultraperformance liquid chromatography-mass spectrometry (UPLC-MS) ion chromatograms of fusaristatin A in PH-1 (blue) and Δash1 (red). Peaks: m/z 659.43 ([M+H]⁺) and 681.41 ([M+Na]⁺). (N) IGV tracks of H3K36me3 and RNA expression of CHSA in PH-1, Δset2, and Δash1. (O) Colony morphologies on CM with Congo red (CR, 0.2 g/L) or calcofluor white (CFW) (100 mg/L) at 3 days. (P) Inhibition rates under indicated stress conditions. Different letters indicate statistically significant differences within each treatment condition. Data in I and P represent the mean ± s.d. of three independent experiments. For panels C, I, and P, statistical significance (P < 0.05) was determined by ANOVA followed by Tukey's multiple comparisons test.

Open in new tabDownload PowerPointLink to figure

To further elucidate the regulatory interplay of Set2- and Ash1-mediated H3K36me3 and their impact on gene expression, we investigated the relationship between H3K36me3-enriched genes and DEGs in Δash1 and Δset2. Our results showed a substantial overlap, with 46.2% (184/398) of downregulated genes in Δset2 linked to Set2-dependent H3K36me3, and 46.1% (872/1893) of upregulated genes in Δash1 linked to Ash1-mediated H3K36me3 (Fig. 5D). These overlapping genes were defined as direct targets of Set2 or Ash1. Gene Ontology (GO) analysis revealed that Set2-regulated genes were enriched in pathways related to metal ion transport, putrescine and β-alanine biosynthesis, and RNA polymerase II transcription (Fig. 5E). Ash1-regulated genes were enriched in proteolysis, siRNA processing, and secondary metabolite biosynthesis (Fig. 5F).

To validate these roles, we selected two candidate genes from each Set2- and Ash1-regulated sets and constructed corresponding deletion mutants for phenotypic assessment. For Set2, we selected LEU1 (FGSG_09589), encoding isopropyl malate isomerase, and STF (FGSG_00404), a sexual-specific transcription factor. Both genes showed reduced H3K36me3 in their coding regions and decreased transcript levels in Δset2 (Fig. 5G,J). Phenotypic analysis revealed Δleu1 exhibited significantly impaired growth on solid and liquid FGA medium without leucine supplementation (Fig. 5H,I), and Δstf exhibited defective perithecium formation (Fig. 5K). Notably, Δset2 exhibited partial phenotypes resembling Δleu1 and Δstf, suggesting Set2 regulates leucine biosynthesis and sexual development. For Ash1, we selected NRPS7, involved in fusaristatin A biosynthesis, and CHSA (FGSG_02354), encoding chitinase. Both genes showed reduced H3K36me3 at their promoter and/or coding regions, with elevated expression in Δash1 (Fig. 5L,N). As expected, fusaristatin A production increased in Δash1 compared to PH-1 (Fig. 5M). In addition, CHSA overexpression led to increased sensitivity to cell wall stress, paralleling Δash1 (Fig. 5O,P). Collectively, these results indicate that Set2 and Ash1 play distinct roles in the transcriptional landscape of H3K36me3-marked genes in F. graminearum, with Set2 associated with actively transcribed genes and Ash1 enriched at loci with lower transcriptional activity.

Ash1 activity modulates H3K27me3 accumulation

Previous studies have shown that H3K27me3 is deposited by Kmt6, the catalytic subunit of PRC2, marking facultative heterochromatin and mediating transcriptional repression in F. graminearum (Connolly et al. 2013; Tang et al. 2021). Notably, Ash1-catalyzed H3K36me3 is also enriched in facultative heterochromatin and implicated in transcriptional repression, partially overlapping in genomic distribution with H3K27me3 (Fig. 6A). In Δash1, we observed a substantial global reduction in H3K27me3, whereas no decrease was detected in Δset2 (Fig. 6B). ChIP-seq further confirmed this, revealing significantly reduced H3K27me3 occupancy in Δash1 compared to PH-1 (Fig. 6C). These observations suggest functional cross-talk between Ash1-mediated H3K36me3 and H3K27me3 in gene expression regulation.

Figure 6.

Ash1 modulates H3K27me3 deposition and target gene transcription. (A) IGV tracks of H3K36me3 in Δset2 and H3K27me3 in PH-1 across all four chromosomes. (B) Immunoblot analysis of H3K27me3 levels in PH-1 and deletion mutants. Relative intensities were quantified using ImageJ; anti-H3 as a loading control. (C) Metagene plots and heat maps of average H3K27me3 distribution in PH-1 and Δash1 around H3K27me3-marked genes (±2.0 kb). (D) Venn diagrams of overlap between H3K27me3-marked genes and Ash1-dependent H3K36me3-marked genes. (E) GO enrichment analysis of overlapping genes from D (n = 1092). (F,G) Heat maps of loci with increased (F) or decreased (G) H3K27me3 in Δash1, with corresponding transcript changes. (H,I) IGV tracks of H3K36me3/H3K27me3 and expression for MFS1, LPP1 (H), and P450A, P450B (I) in PH-1 and Δash1.

Open in new tabDownload PowerPointLink to figure

Approximately 70% (1092/1566) of H3K27me3-enriched genes were also co-occupied by Ash1-mediated H3K36me3 (Fig. 6D). GO analysis revealed these comarked genes were predominantly enriched in transport and secondary metabolism (Fig. 6E). To assess whether Ash1 influences H3K27me3 deposition, we analyzed H3K27me3 ChIP-seq profiles in Δash1. Loss of ASH1 led to increased H3K27me3 occupancy at 63 genes, with 17 having reduced transcript levels (Fig. 6F). Conversely, 77 genes showed decreased H3K27me3 enrichment, with 42 displaying transcriptional upregulation (Fig. 6G). For instance, MFS1 (FGSG_03725) and LPP1 (FGSG_03120), which gained H3K27me3 in Δash1, showed reduced expression, indicating Ash1-dependent H3K36me3 may promote repression by facilitating H3K27me3 accumulation (Fig. 6H). In contrast, two cytochrome P450 genes that lost both H3K36me3 and H3K27me3 in Δash1 showed increased expression, indicating transcriptional derepression (Fig. 6I). Together, these findings suggest that Ash1-mediated H3K36me3 exerts dual roles in modulating H3K27me3 dynamics, enabling both repression and activation of comarked genes in F. graminearum.

H3K36me3 coordinates with H3K27me3 in secondary metabolism regulation

Secondary metabolite (SM) gene clusters, including polyketide synthases (PKSs), nonribosomal peptide synthases (NRPSs), terpene synthases (TPSs), and cytochrome P450s, are predominantly located in telomeric and pericentromeric regions of the F. graminearum genome. Previous studies show that Kmt6 represses SM gene expression through H3K27me3 deposition (Tang et al. 2021; Atanasoff-Kardjalieff and Studt 2022). Loss of H3K27me3 in Δkmt6 derepressed genes involved in SM biosynthesis. Given H3K36me3 enrichment in telomeric and pericentromeric regions and Ash1's role in modulating H3K27me3 deposition, we investigated whether H3K36me3 contributes to SM regulation in F. graminearum.

Heat map analysis revealed that primary metabolism gene expression remained largely unaffected in Δash1, Δset2, and Δset2/ash1, consistent with patterns in Δkmt6 (Fig. 7A). In contrast, SM genes exhibited greater transcriptional variability, particularly in Δash1, where 35.8% (245/683) of SM genes were differentially expressed. A similar trend was observed in Δkmt6, which showed differential expression of 59.5% (407/683) SM genes (Fig. 7B). Notably, 130 SM genes were upregulated in both Δash1 and Δkmt6 (Fig. 7C). Furthermore, 85% of SM genes upregulated in Δash1 were restored to wild-type expression in Δset2/ash1, suggesting Set2 also participates in their regulation (Fig. 7B,C). Further analysis identified 62 key SM genes, including PKSs, NRPSs, TPSs, and cytochrome P450s, marked by H3K36me3 in PH-1 and showing significant changes in the single or double mutants (Fig. 7D). Among these, 34 genes were comarked by H3K36me3 and H3K27me3 and showed elevated expression in Δash1 (Fig. 7D). ChIP-seq showed a substantial reduction in both H3K36me3 and H3K27me3 at the promoters and gene bodies of derepressed genes in Δash1, with H3K36me3 levels nearly abolished at promoters (Fig. 7E).

Figure 7.

Ash1 regulates secondary metabolism via H3K36me3-H3K27me3 interplay. (A,B) Heat maps of transcript changes in primary metabolism (A, n = 2590) and secondary metabolism (SM) genes (B, n = 683) across mutants. (C) Overlap of upregulated SM genes in Δash1 and Δkmt6. (D) Heat maps of PKS, NRPS, and P450 expression changes in mutants. H3K36me3 and H3K27me3 enrichment in PH-1 is marked (√, enriched; ×, not enriched). (E) Metagene plots of average H3K36me3 and H3K27me3 in PH-1 and Δash1 over upregulated gene from D (±2.0 kb). (F) IGV tracks of H3K27me3 and H3K36me3 across cluster C62. Bar graph shows transcriptional changes of genes within the cluster. (G,H) ChIP-qPCR analysis of H3K36me3 and H3K27me3 enrichment at the promoter (P1), 5′ (F1), and 3′ (F2) region of AUR1 (G) and PKS12 (H) in PH-1 and Δash1. Different letters indicate significant differences between PH-1 and Δash1 at each indicated region. (I) Relative expression of AUR1 and PKS12. Expression levels in PH-1 were set to 1, with ACTIN as the internal normalization control. Different letters indicate statistically significant differences among all groups. Error bars, s.d. (n = 3). Student's t-test (P < 0.05). (J,K) Δash1 shows pigment overproduction on CM plates and in liquid cultures. (L) Colony morphologies and diameters of strains grown on CM for 3 days. (M) Suppressor sectors (red arrows). Colony morphologies and diameters of strains grown on CM for 3 days. Data are mean ± s.d. (n = 3). (N) Spontaneous suppressor mutations in CTK1 and SET2. (fs) frameshift. (O) RT-qPCR of representative SM genes in PH-1, Δash1, Δset2/ash1, and Δctk1/ash1 strains. ACTIN served as the normalization control. Different letters indicate statistically significant differences among the strains for each gene. Data in G, H, I, L, M, and O are presented as mean ± s.d. from three independent experiments. Statistical significance (P < 0.05) was determined as follows: Student's t-test was used for pairwise comparisons in G and H, whereas ANOVA followed by Tukey's multiple comparisons test was used for multiple group comparisons in I and O. Error bars, s.d. (n = 3).

Open in new tabDownload PowerPointLink to figure

To further elucidate Ash1's effects on SM gene regulation, we analyzed the aurofusarin biosynthetic gene cluster as an example (Fig. 7F). All 13 genes in this cluster were upregulated in Δash1, coinciding with reductions of H3K36me3 and H3K27me3 at their promoters (Fig. 7F). This trend was further validated by ChIP-qPCR and RT-qPCR for two representative genes—AUR1, encoding the transcription factor, and PKS12, encoding polyketide synthase. Both genes exhibited decreased promoter enrichment of H3K36me3 and H3K27me3, with increased transcript levels (Fig. 7G–I). Notably, H3K36me3 enrichment across the gene bodies of several cluster genes, including AUR4, PKS12, AUR8, and AUR9, was not altered in Δash1 (Fig. 7F). Elevated expression of the aurofusarin biosynthesis cluster in Δash1 led to increased pigment accumulation on CM plates and in liquid cultures (Fig. 7J,K). Additionally, AUR1 overexpression under a constitutive tubulin promoter (∼25-fold increase) enhanced pigment production but reduced vegetative growth by ∼70% compared to PH-1 (Fig. 7J,L). Subsequent experiments revealed that purified aurofusarin significantly suppressed mycelial growth, conidiation, and conidial germination in PH-1 (Supplemental Fig. S7). These suggest that the growth defects in Δash1 are partially due to increased aurofusarin production. To test this, we generated a Δash1/aur1 double mutant, which partially rescued the growth defects in Δash1 (Fig. 7L). These findings show that Ash1-mediated H3K36me3 and Kmt6-mediated H3K27me3 cooperatively repress SM gene expression via promoter occupancy and that growth defects in Δash1 are partially due to overproduction of toxic SMs, such as aurofusarin.

As observed, Δash1 exhibited phenotypic instability, producing fast-growing sectors after 2 weeks of incubation. Whole-genome resequencing of two suppressor mutants identified frameshift mutations in CTK1 (FGSG_16718) and SET2 (FGSG_05558) (Fig. 7M,N). Previous studies show that Ctk1, a cyclin-dependent kinase, phosphorylates the C-terminal domain (CTD) of RNA polymerase II (RNAPII) at serine 2 (Ahn et al. 2004; Röther and Sträßer 2007), facilitating Set2 recruitment and H3K36me3 deposition during transcription elongation (Youdell et al. 2008; Dronamraju and Strahl 2014). We hypothesize that frameshift mutations in Ctk1 and Set2 impair transcription elongation, particularly in SM genes, which partially alleviates the growth defects in Δash1. Consistent with this, transcript levels of representative SM genes were reduced in both Δash1/ctk1 and Δash1/set2 compared to Δash1 (Fig. 7O). Moreover, the growth rates of these double mutants were faster than Δash1 and comparable to the suppressor strains (Fig. 7M).

In conclusion, these findings highlight the cooperative roles of H3K36me3 and H3K27me3 in regulating SM gene expression. In PH-1, SM gene promoter regions are co-occupied by Ash1-catalyzed H3K36me3 and PRC2-catalyzed H3K27me3, maintaining transcriptional repression. Deletion of ASH1 results in a near-complete loss of H3K36me3 at SM gene promoters, with a marked reduction in H3K27me3 enrichment. This epigenetic disruption relieves repression, leading to SM gene activation. Set2 catalyzes H3K36me3 in the SM gene coding regions, playing a key role in elongation. This process is facilitated by Ctk1 kinase, which phosphorylates Ser2 residues in RNAPII's CTD, transitioning RNAPII into a highly phosphorylated state. Phosphorylated Ser2 enables Set2 recruitment to coding regions, where it catalyzes H3K36me3, ensuring efficient transcription elongation. Importantly, H3K36me3 levels in SM gene coding regions remain robust in Δash1, highlighting its critical role in sustaining transcription elongation of SM genes (Fig. 8).

Figure 8.

Model of H3K36me3-H3K27me3 interplay in regulating secondary metabolism gene expression. In PH-1, SM genes are repressed by Ash1-catalyzed H3K36me3 (purple) and PRC2-medicated H3K27me3 (red) at promoters (left). ASH1 deletion causes loss of H3K36me3 and reduced H3K27me3, releasing repression and activating transcription (right). During elongation, kinase Ctk1 phosphorylates the Ser2 residue of the C-terminal domain (CTD) of RNA polymerase II (Pol II), facilitating Set2 recruitment. Set2 deposits H3K36me3 along gene bodies to sustain elongation of activated SM genes. Green arrow, activation; red arrow, repression. Figure created with BioRender (https://www.biorender.com).

Open in new tabDownload PowerPointLink to figure

Identification of SET domain proteins in fungi

To systematically identify SET domain proteins across fungi, a total of 23.64 million protein sequences were retrieved from 1038 fungal genomes spanning nine phyla, all obtained from the NCBI database. Domain annotations were performed using Pfam (PF00856), SMART (SM000317), and InterPro (IPR001214). Proteins containing at least one SET domain were retained, resulting in the identification of 18,718 SET domain proteins. To investigate subfamily relationships, the identified sequences were clustered with CLANS v2.0 (https://github.com/inbalpaz/CLANS), a Python-based tool for sequence-similarity network analysis, executed in non-GUI mode with the “-nogui” flag.

Phylogenetic analysis of SET domain proteins in fungi

Phylogenetic reconstruction was performed for cluster 1, which contained 6640 sequences grouped into 11 groups. Multiple sequence alignments were generated with MAFFT v7.475 (Katoh and Standley 2013), and poorly aligned regions were removed using trimAl v1.4.1 (Capella-Gutiérrez et al. 2009) with the “gappyout” setting. A maximum-likelihood tree was inferred with FastTree v2.1 (Poon et al. 2010). The reference species tree for the 1038 fungal genomes was obtained from a previous study (Li et al. 2021), which established phylogenetic relationships using 290 single-copy BUSCO genes. To assess the distribution of major SET domain subfamilies, the average copy number of SET domain proteins was calculated across 42 fungal classes spanning seven phyla (each represented by more than one species). The groups examined included group03-H3K9, group04-H3K4, group06-Set2 (H3K36), group15-Ash1 (H3K36), group07-H4K20, and group14-H3K27.

Strain construction

The wild-type F. graminearum strain PH-1 (NRRL 31084) served as the parental strain for transformants generated in this study. All gene deletions, complementations, and H3K36A point mutation were generated using the double-joint PCR method (Supplemental Fig. S9A–C), and the resulting DNA fragments were introduced into protoplasts via polyethylene glycol (PEG)-mediated transformation as described previously (Yun et al. 2014). Transformants were selected on hygromycin (100 µg/mL) or neomycin (100 µg/mL), depending on the resistance cassette used, and verified by PCR or sequencing (Supplemental Fig. S9D–G). Primers used in this study are listed in Supplemental Table S8. Indel profiles of Δset2, Δash1, Δset2/ash1, complemented strains, and the H3K36A mutant were confirmed by whole-genome sequencing (WGS) (Supplemental Fig. S9H; Supplemental Table S9). For gene deletions, the ORF was replaced with either the hygromycin B phosphotransferase gene (HPH) or the neomycin resistance gene (NEO). Complementation strains of SET2 and ASH1 were generated by in situ replacement with GFP-tagged ORFs fused to flanking regions and the NEO cassette. The PH-1::Tri1-GFP strain was constructed using the same strategy. Double mutants (Δset2/ash1, Δash1/aur1, and Δash1/ctk1) were obtained by replacing the target ORF with the NEO cassette in the Δash1 background. The H3K36A allele was generated by substituting lysine with alanine at position 36 of HISTONE3 (FGSG_04290) and confirmed by sequencing. Overexpression of AUR1 and CHSA was driven by either the native TUB promoter (FGSG_09530) or the A. nidulans GPDA promoter (Frandsen et al. 2008), fused with the NEO cassette, and introduced into the corresponding deletion strains.

Pathogenicity and DON production assays

To evaluate the pathogenicity on wheat heads, 10 µL of conidial suspension (1 × 10⁵ conidia/mL) were injected into the central spikelet of individual flowering wheat heads. Each strain was tested on 15 biological replicates. The number of infected spikelets per head was recorded 14 days postinoculation (dpi). For wheat seedling stem assays, fresh wheat stems were inoculated with mycelial plugs, and lesion development was documented at 5 dpi. For deoxynivalenol quantification, strains were cultured in TBI liquid medium at 28°C in the dark with shaking at 150 rpm for 7 days. DON was extracted and quantified using a Deoxynivalenol (DON) ELISA kit (WS2024-011, Wise Science & Technology Development, Jiangsu, China) according to the manufacturer's instructions. DON assays were conducted with three independent biological replicates.

In vitro histone methyltransferase assay

In vitro HMT assays were performed as described previously (Zhang et al. 2024a). Each 200-µL reaction contained 10 µg of synthetic recombinant histone H3 peptide (PSTGGVKKPHRYKPGT), 2 µg S-adenosyl-L-methionine (Solarbio), and ∼100 µg of purified MBP-Set2, MBP-Ash1, or MBP control protein in HMT buffer (300 mM NaCl, 50 mM Tris-HCl [pH 7.5], 5% glycerol, 1 mM DTT). Reactions were incubated at 30°C for 20 h and terminated by adding SDS-PAGE loading buffer, followed by boiling at 95°C for 5 min. Reaction products were analyzed by western blot using anti-H3K36me3 (Abcam, #ab9050) and anti-H3K36me2 (Abcam, #ab9049,) antibodies, and by ultraperformance liquid chromatography-mass spectrometry (UPLC-MS) on a Waters UPLC system (Waters Corp.). Raw MS data were processed with PEAKS Studio X+ (Bioinformatics Solutions Inc.) and searched against the F. graminearum PH-1 protein database from UniProt. Search parameters were: parent mass tolerance, 10 ppm; fragment mass tolerance, 0.02 Da; maximum missed cleavages, 2; variable modifications, oxidation (M), mono-, di-, and trimethylation (K), acetylation (K, protein N-term); false discovery rate (FDR) < 1%; minimum peptide score, −10lgP ≥ 10. Label-free quantification was performed based on extracted-ion chromatogram peak areas. Peptide-spectrum matches and posttranslational modifications were validated using PEAKS’ built-in FDR estimation and manual inspection. Proteins identified with ≥2 unique peptides were considered confidently identified.

Chromatin immunoprecipitation and ChIP-seq analysis

ChIP assays were performed on PH-1, Δash1, and Δset2 strains using fresh mycelia grown in CM medium at 25°C for 24 h, following previously described protocols (Tang et al. 2021). Briefly, 0.6 g of mycelia was washed twice with cold PBS, crosslinked with 1% formaldehyde at 25°C for 10 min, and quenched with 125 mM glycine. Samples were lysed in buffer containing 50 mM Tris-HCl (pH 8.0), 10 mM EDTA, 1% SDS, and protease inhibitor cocktail. Chromatin was extracted on ice and sonicated to 200- to 500-bp fragments. A 20-µL aliquot of sonicated chromatin was reserved as input, and 100 µL were immunoprecipitated with 10 µg of anti-H3K36me3 (Abcam, #ab9050), anti-H3K27me3 (Active Motif, #39155), or anti-H4K20me3 (Active Motif, #91107) antibodies at 4°C overnight. The following day, 30 µL of protein A magnetic beads were added, and samples were incubated for 3 h. Beads were sequentially washed once with buffer I (20 mM Tris-HCl [pH 8.1], 50 mM NaCl, 2 mM EDTA, 1% Triton X-100, 0.1% SDS), twice with buffer II (10 mM Tris-HCl [pH 8.1], 250 mM LiCl, 1 mM EDTA, 1% NP-40, 1% sodium deoxycholate), and twice with TE buffer (10 mM Tris-HCl [pH 7.5], 1 mM EDTA). DNA was eluted in 300 µL elution buffer (100 mM NaHCO₃, 1% SDS), treated with RNase A (8 µg/mL) at 65°C for 6 h, and digested with Proteinase K (345 µg/mL) overnight at 45°C. Immunoprecipitated DNA was used to construct sequencing libraries with the NEXTFLEX ChIP-seq Library Prep kit for Illumina (Bioo Scientific, #NOVA-5143-02) and sequenced on an Illumina NovaSeq 6000 platform using a paired-end 150-bp (PE150) strategy. Low-quality reads were filtered using Trimmomatic v0.36 (Bolger et al. 2014), and clean reads were mapped to the F. graminearum reference genome using BWA v0.7.15 (Li and Durbin 2009). PCR duplicates were removed with SAMtools v1.3.1 (Li et al. 2009). Peak calling was performed with MACS2 v2.1.1 (bandwidth, 300 bp; model fold, 5–50; Q-value ≤ 0.05) (Zhang et al. 2008). ChIP-seq results were validated by ChIP-qPCR using primers listed in Supplemental Table S8. Relative enrichment was calculated as fold enrichment (ChIP signal divided by mock signals). All experiments were conducted with three independent biological replicates.

Identification of cryptic transcription events

To identify genes potentially affected by cryptic transcription initiation in Δset2, a region-based intragenic coverage analysis was performed as described previously (McCauley et al. 2021). Gene bodies were divided into three segments: 5′ (upstream third), middle (central third), and 3′ (downstream third). Read counts for each segment were obtained using featureCounts v2.0.1 (Liao et al. 2014) with parameters -t exon -g transcript_id -f -O -T 8 -p -C, and normalized to TPM. For each gene, a middle-to-edge fold-change ratio was calculated as the read density in the middle segment divided by the average of the 5′ and 3′ segments, reflecting internal transcript accumulation indicative of cryptic initiation. Fold-change values were calculated separately for PH-1 and Δset2. Genes with a middle-to-edge ratio >1 in Δset2 and <1 in PH-1, and a log₂ Δset2/PH-1 >1 were considered candidates exhibiting cryptic transcription initiation in Δset2.

Aurofusarin treatment assays in F. graminearum

To evaluate the biological effect of aurofusarin against F. graminearum, PH-1 conidia were germinated in 2% sucrose at 25°C in the dark for 5 h, after which aurofusarin was added at a final concentrations of 10 or 100 ppm. After 48 h, mycelia were harvested, dried, and weighed. For conidiation assays, PH-1 was cultured in CMC liquid medium supplemented with aurofusarin (10 or 100 ppm) at 25°C and 180 rpm in the dark for 5 days. Conidial yield and septation were quantified microscopically. To assess germination, conidia were incubated in 2% sucrose containing 10 or 100 ppm aurofusarin for 5 h under the same conditions. Samples were stained with calcofluor white (CFW) and examined under a fluorescence microscope.

Statistical analysis

All statistical analyses were conducted using IBM SPSS Statistics v27. Data are presented as mean ± standard deviation (s.d.). Differences between two groups were evaluated by a two-tailed Student's t-test. For multiple comparisons, one-way ANOVA followed by Tukey's multiple-range test was applied. P-values and representation factor (RF) for gene-set overlaps were calculated using Fisher's exact test in R (R Core Team 2025), assuming a total F. graminearum genome size of 14,145 genes. Correlation analysis between the total protein number and SET domain proteins across fungal species was performed using Pearson's correlation (linear regression) via the Bioinformatics online platform (https://www.bioinformatics.com.cn; last accessed 10 Dec 2024).