The Histone Demethylase JMJD2‐Dependent Regulation of rDNA Transcription and Integrity of Nucleolus Is Required for Fungal Development and Pathogenicity in the Rice Blast Fungus

ABSTRACT

The nucleolus is a central hub for nuclear functions, including, but not limited to, ribosome biogenesis in mammals and budding yeast. However, how the nucleolus is regulated and contributes to fungal development and pathogenesis is poorly understood. Here we report that a JmjC domain‐containing histone demethylase MoJMJD2 is required for rDNA transcription and nucleolar integrity in the rice blast fungus, Magnaporthe oryzae. MoJMJD2 generates two alternatively spliced isoforms with distinct subcellular localization patterns, suggesting coordinated roles in gene regulation in the nucleolus and nucleoplasm. Deletion of MoJMJD2 rendered the fungus defective in hyphal growth, sporulation and pathogenicity. Notably, loss of MoJMJD2 led to mislocalization of a nucleolar protein to the nucleoplasm and a dramatic reduction of cellular protein synthesis, suggesting the disruption of functional nucleolar structure. Taken together, these results indicate that nucleolar integrity and functions, which are under epigenetic control associated with MoJMJD2, are required for the development and pathogenicity of the rice blast fungus.

Alternative splicing of MoJMJD2 generates nucleolar and nuclear isoforms that coordinate development and pathogenicity in Magnaporthe oryzae. In the nucleolus, MoJMJD2T⁰ promotes rDNA transcription by reducing H3K9me1 at rDNA loci, whereas in the nucleus, MoJMJD2T¹ likely maintains nucleolar integrity and indirectly regulates melanin synthesis, cell wall synthesis and conidiation.

1. Introduction

The nucleolus is a membrane‐less compartment inside the nucleus. A wealth of research in the past few decades has shown that the nucleolus is not just a ribosome factory but a multifunctional domain central to the diverse functions of the nucleus in eukaryotes (Boisvert et al. 2007; Iarovaia et al. 2019). Dynamic changes of the nucleolus in response to the cell cycle, cellular stresses and the metabolic activity of cells reflect the importance of its regulation and the different roles it plays in the cell (Boulon et al. 2010; Hernandez‐Verdun 2011; Leung and Lamond 2003; Shaw and Brown 2012). The target of rapamycin (TOR) protein kinase is a well‐conserved regulator that links the quality and quantity of nutrients to developmental and metabolic processes in eukaryotic cells (Dobrenel et al. 2016; Gonzalez and Rallis 2017). In yeast and animal cells, suppression of TOR by nutrient deprivation and/or rapamycin (TOR inhibitor) treatment leads to a reduction in the size and activity of the nucleolus (Matos‐Perdomo and Machin 2019). Functions of the nucleolus are largely dependent on the repertoire of proteins that enter and/or are released from the nucleolus (Martin et al. 2015). Many, if not all, nucleolar proteins possess nucleolar localization sequences (NoLS) that are composed of positively charged peptides containing a patch of arginine (Scott et al. 2011). Although much has been learned about the nucleolus and transcriptional regulation of rDNA from studies on a budding yeast, Saccharomyces cerevisiae , little is known about the filamentous counterparts to date.

Histone modifications including histone methylation are central to the epigenetic regulation of rDNA transcription in the nucleolus (Iarovaia et al. 2019; Schofer and Weipoltshammer 2018). Interplay between histone methylation and demethylation regulates the number of active rDNA loci and transcription efficiency at a particular locus, indicating the importance of both histone methyltransferases (HMTs) and histone demethylases (HDMs). HDMs are traditionally divided into two types: lysine‐specific demethylase (LSD) and Jumonji C (JmjC) domain‐containing (JMJD) proteins (Dimitrova et al. 2015; Kooistra and Helin 2012). LSD proteins work at mono‐ and dimethylated lysine residues, whereas JMJD proteins target lysine residues with all histone methylation states (mono‐, di‐ and trimethylation) (Shi and Whetstine 2007). In addition, LSD proteins use FAD⁺ as a cofactor, whereas JMJD proteins use Fe²⁺ and α‐ketoglutarate as cofactors. Histone demethylases containing the JmjC domain are classified into seven groups, based on sequence homology and presence of additional domains (Klose et al. 2006). They possess lysine demethylase (KDMs) activity on H3K4, K9, K27 and K36 (Meng et al. 2018). The effects of histone methylation on gene transcription are affected by both the position of the modified residues and the number of methyl groups. In particular, the KDM4 family proteins are known to target H3K9me2/3 and H3K36me2/3 (Labbé et al. 2013).

The importance of histone methylation/demethylation in filamentous fungi has been clearly demonstrated in a wide variety of species including Neurospora crassa, Aspergillus nidulans, Zymoseptoria tritici and Botrytis cinerea (Fraser and Whitehall 2022; Gacek‐Matthews et al. 2015; Hou et al. 2020; Moller et al. 2019; Rountree and Selker 2010). The role of histone methylation in these species was studied primarily in reference to regulation of secondary metabolite gene clusters, genome stability and pathogenicity. For example, in Z. tritici H3K9me3 and H3K27me3 have strong and opposing effects on overall stability of the genome and accessory chromosomes (Fraser and Whitehall 2022; Moller et al. 2019). In A. nidulans , a KDM4 family histone demethylase, KdmA acts as a transcriptional co‐repressor of genes involved in primary metabolism and is required for full expression of secondary metabolism genes via demethylation of H3K36me3 (Gacek‐Matthews et al. 2015). A study in B. cinerea of Jar1‐mediated demethylation of H3K4me3 showed that it is required for regulating fungal development and pathogenesis (Hou et al. 2020).

In this work, we report the involvement of the histone demethylase MoJMJD2, which belongs to the KDM4 family, during fungal development and pathogenesis via regulation of rDNA transcription and nucleolar integrity based on alternative splicing in the rice blast fungus, Magnaporthe oryzae. About 10%–30% of global rice yield is lost annually to rice blast disease (Talbot 2003). Infection of rice plants by M. oryzae starts with the dissemination of asexual spores called conidia. After landing on the leaf surface, the conidium germinates and develops an infection‐specific structure, appressorium, at the tip of the germ tube upon recognition of surface properties such as hydrophobicity (Jeon et al. 2007). The fungus then mechanically ruptures the cuticular layer of the plant by converting turgor pressure generated within the appressorium into mechanical pressure (Foster et al. 2017; Howard and Valent 1996; de Jong et al. 1997). Once inside the plant tissue, M. oryzae develops bulbous invasive hyphae for colonization, during which the fungus has to overcome the plant defence response including reactive oxygen species (ROS), resulting in the development of disease lesions (Kankanala et al. 2007; Wilson and Talbot 2009). In M. oryzae , the involvement of histone methylation in development and pathogenicity has been demonstrated by analyses of a histone methyltransferase, MoSET1 and a histone demethylase, MoJMJ1 (Huh et al. 2017; Pham et al. 2015).

Here, we find that the MoJMJD2 gene generates, via alternative splicing, two protein isoforms, one of which localizes to the nucleolus, binds to the rDNA promoter and favours rDNA transcription in association with reduced H3K9me1 levels. Furthermore, we show that the absence of the MoJMJD2 gene leads to defects in vegetative growth, asexual development and pathogenicity through the dramatic reduction in cellular protein synthesis and disruption of nucleolar integrity. Thus, our results demonstrate MoJMJD2 as a key epigenetic regulator associated with rDNA transcription and nucleolar integrity, and shed light on the importance of controlling nucleolar activity during fungal development and pathogenesis in a filamentous fungus.

2. Results

2.1. MoJMJD2 Produces Two mRNA Species via Alternative Splicing in an Environment‐Dependent Manner

In our previous study, seven genes encoding putative JmjC domain‐containing histone demethylases were identified in M. oryzae (Choi et al. 2015; Huh et al. 2017). Among them, MGG_09186, encoding a putative JMJD2 family histone lysine demethylase (MoJMJD2), was predicted to produce two protein isoforms via alternative splicing. The protein contains the characteristic JmjC, JmjN and PHD‐type zinc finger domains of JMJD2 family proteins (Figure 1A), and key residues required for cofactor binding are highly conserved (Figure S1).

FIGURE 1.

Domain architecture of MoJMJD2 and alternative splicing of MoJMJD2. (A) Domain architecture of MoJMJD2 and representative orthologues. Mo, Magnaporthe oryzae (XP_003709791; MGG_09180); An, Aspergillus nidulans (XP_050469070); Sc, Saccharomyces cerevisiae (NP_011096); Ce, Caenorhabditis elegans (NP_496969); Dm, Drosophila melanogaster (NP_610331); Hs, Homo sapiens (NP_055478). (B) Schematic of MoJMJD2 splicing isoforms, MoJMJD2T ⁰ and MoJMJD2T ¹ . Predicted nucleolar localization sequences (NoLSs) were identified using the Nucleolar localization sequence Detector (NoD) server. UTR, untranslated region; NLS, nuclear localization sequence. (C) Relative abundances of MoJMJD2T ⁰ and MoJMJD2T ¹ transcripts under different conditions and during germination/appressorium formation, normalized to a single common calibrator across all conditions. CM, complete medium; MM, minimal medium. Because transcript levels varied substantially in magnitude across the 12 groups compared, statistical significance was assessed using Welch's ANOVA followed by the Games–Howell post hoc test, which does not assume homogeneity of variance among groups. Different letters indicate statistically significant differences among means (p < 0.05) based on three independent biological replicates, each analysed with three technical replicates (n = 3).

Among the fungal species examined based on annotation information, alternative splicing was predicted only for MoJMJD2 (Figure S2). The splice variants were confirmed in M. oryzae by cDNA synthesis with specific primer sets followed by sequencing (Figure S3A,B; Table S1). Similar splice variants were also detected in Fusarium graminearum, suggesting that alternative splicing might be conserved among orthologues (Figure S3C–E). Compared with the long form (MoJMJD2T⁰), the short form (MoJMJD2T¹) lacks the predicted C‐terminal NoLS identified by the NoD server (red box in Figure 1B), suggesting distinct subcellular localization of the two proteins (Scott et al. 2011). Analysis of relative transcript abundance using isoform‐specific primers (yellow dumbbell shapes in Figure 1B; Table S2) showed that two splice variants accumulated differentially depending on conditions (Figure 1C). Relative abundance of MoJMJD2T¹ transcript, for example, was much higher in the sporulation‐promoting condition than in the non‐sporulating condition, suggesting a potential role in asexual reproduction (leftmost panel in Figure 1C).

2.2. MoJMJD2T⁰ and MoJMJD2T¹ Proteins Localize to Nucleolus and Nucleus, Respectively

A nuclear localization signal (NLS) at the N‐terminus is found in both MoJMJD2 isoforms, suggesting nuclear localization. Unlike the NLS, the NoLS is missing in MoJMJD2T¹. We therefore hypothesized that MoJMJD2T⁰ localizes to the nucleolus, whereas MoJMJD2T¹ localizes to the nucleus. To test subcellular localization of MoJMJD2T⁰, a construct expressing MoJMJD2T⁰‐GFP under the native MoJMJD2 promoter was introduced into a strain expressing either MoNOP1‐RFP or H1‐RFP. GFP signals co‐localized with the MoNOP1‐RFP in the fungal hyphae (Figure 2A). Because MoNOP1 is a nucleolar marker protein (Franceschetti et al. 2011), this result indicates that MoJMJD2T⁰ is a nucleolar protein.

FIGURE 2.

Localization of MoJMJD2 isoforms. (A) Localization of MoJMJD2T⁰‐GFP in vegetative hyphae expressing MoNOP1‐RFP. Nuclei were stained with Hoechst 33342. (B) Localization of MoJMJD2T¹‐GFP in vegetative hyphae expressing H1‐RFP. (C) Localization of MoJMJD2T⁰‐GFP during spore development. (D) Localization of MoJMJD2T¹‐GFP during spore development. Merged panels were generated by combining all images from fluorescence microscopy. DIC, differential interference contrast. Scale bar = 10 μm.

Unlike MoJMJD2T⁰‐GFP, no signals were detected when MoJMJD2T ¹ ‐GFP was expressed under the native promoter, possibly because MoJMJD2T ¹ transcripts are present at very low abundance except under sporulation‐promoting conditions (Figure 1C). When expressed under a strong constitutive promoter (P _trpC ), green fluorescence from MoJMJD2T¹‐GFP clearly overlapped with the red fluorescence from H1‐RFP in hyphae, indicating that MoJMJD2T¹ is highly likely a nuclear protein in hyphae (Figure 2B).

We also monitored localization of both isoforms during spore germination and apppressorium formation (Figure 2C,D). This revealed patterns distinct from those observed in hyphae. MoJMJD2T⁰ localized to the nucleoli of conidia as expected, although the GFP signals appeared to be very weak and lacked a clear structure (Figure 2C). As the germ tube elongated on a hydrophobic surface, fluorescence inside spores gradually dissipated (2 h), and became almost invisible by the time the appressorium began to develop (4 h). During appressorium formation, the GFP signal initially appeared inside the incipient appressorium in the form of speckles (4 h), and later it was found in the cell periphery (6 h). In contrast to the germinating spores on the hydrophobic surface, GFP signals of MoJMJD2T⁰ in germinating spores on the hydrophilic surface remained associated with the nucleoli, suggesting a role of MoJMJD2T⁰ in the pathogenic development of the fungus (Figure S4).

While MoJMJD2T⁰ was associated with fungal nucleoli, MoJMJD2T¹ did not seem to completely localize to the nucleus. Weak fluorescence was observed mainly in the cytoplasm of spores, germ tubes and appressoria during germination and appressorium development (Figure 2D). The signal was not evenly distributed around the cytoplasm, but rather tended to be observed in and around fungal nuclei. Consistent with these observations, western blot analysis revealed that MoJMJD2T⁰ accumulated as the expected full‐length protein in hyphae, indicating stable production during vegetative growth (Figure S5A). In contrast, MoJMJD2T¹ was also detected as a full‐length protein in hyphae, but only a ~27 kDa band corresponding to free GFP fragments was observed in conidial extracts (Figure S5B), suggesting that MoJMJD2T¹ undergoes stage‐specific instability or cleavage in spores. However, it should be noted that free GFP fragments do not exclude the possibility of GFP‐MoJMJD2T¹ being localized to the nucleus in the asexual spore.

2.3. MoJMJD2 Regulates Steady‐State Level of Protein Synthesis via Epigenetic Control of rDNA Transcription and Is Required for Nucleolar Integrity

To investigate MoJMJD2 function, we generated a deletion mutant (ΔMojmjd2) by replacing the MoJMJD2 gene with a hygromycin resistance gene (HPH) cassette in the wild‐type strain KJ201 (Figure S6A,B; Table S3). Correct replacement was confirmed by inverse PCR (Figure S6C) and Southern blot analysis (Figure S6D,E). The nucleolus is the site of rDNA transcription that is regulated by TOR kinase. Considering that MoJMJD2 is a putative histone demethylase, of which one isoform (MoJMJD2T⁰) localizes to the fungal nucleolus, we first tested whether inhibition of TOR by rapamycin would have any effect on MoJMJD2 transcription by reverse transcription‐quantitative real‐time PCR (RT‐qPCR) analysis using the ΔMojmjd2 as a negative control. This revealed that rapamycin treatment significantly decreased the relative abundance of MoJMJD2 transcript, suggesting that MoJMJD2 may act downstream of TOR kinase signalling (Figure 3A). Next, we compared relative rRNA abundance between the wild type and mutant (Figure 3B), using rapamycin‐treated wild type as a control. As expected, rapamycin treatment dramatically reduced rRNA abundance in the wild type. In the mutant, rRNA levels were even lower than in the rapamycin‐treated wild type, indicating that MoJMJD2 is required to maintain steady‐state rRNA levels.

FIGURE 3.

Regulation of rDNA transcription by MoJMJD2T⁰. (A) Relative MoJMJD2 transcript abundance in wild‐type strain (KJ201) with or without rapamycin and in MoJMJD2 deletion strain (∆Mojmjd2). (B) Relative rRNA abundance in KJ201 and ∆Mojmjd2 with or without rapamycin. NT, no treatment; Rap, rapamycin treatment. In (A) and (B), different letters indicate statistically significant differences (p < 0.05, Tukey's HSD test) based on three independent biological replicates, each analysed with three technical replicates (n = 3). (C) Chromatin immunoprecipitation‐quantitative PCR analysis of MoJMJD2T⁰‐GFP enrichment at rDNA loci. (D) H3K9me1 enrichment at rDNA loci in KJ201 and ∆Mojmjd2. In (C) and (D), asterisks indicate statistically significant differences (*p < 0.05, **p < 0.01, ***p < 0.001; Tukey's HSD test) based on three independent biological replicates, each analysed with three technical replicates (n = 3). (E) SUnSET assay comparing protein synthesis in KJ201 and ∆Mojmjd2 with or without cycloheximide (CHX). The loading control was visualized by Ponceau S staining. (F) Quantification of puromycin incorporation shown in (E). CHX −, without CHX treatment; CHX +, with CHX treatment. Two independent experiments were performed, yielding reproducible results; one representative result is shown here.

To test whether MoJMJD2 associates with rDNA, we performed chromatin immunoprecipitation (ChIP) using chromatin from the wild type and a strain expressing MoJMJD2T⁰‐GFP, followed by quantitative PCR (qPCR) analysis of different rDNA repeat regions (Figure 3C). We observed significant enrichment of rDNA promoter, 18S rDNA, and 26S rDNA regions in the ChIP assay using anti‐GFP antibodies, with the strongest enrichment at the promoter region of rDNA loci. Promoters of tested nuclear genes were not enriched with MoJMJD2T⁰‐GFP (Figure S7A). These results are in line with the nucleolar localization of MoJMJD2T⁰ and indicate that MoJMJD2T⁰ preferentially associated with the rDNA promoter in fungal hyphae.

As JMJD2 is a putative histone demethylase, such associations of MoJMJD2T⁰ with rDNA loci prompted us to test whether histone methylation status at rDNA differs between the wild type and deletion mutant. JMJD2 family histone demethylases are known to have H3K9me3 and H3K27me3 as a main substrate, and H3K36me3 as a minor substrate (Labbé et al. 2013). When we immunoprecipitated chromatin using H3K9me2, H3K9me3, H3K27me3 and H3K36me3 antibodies, however, no enrichment of these modifications was detected at rDNA loci of the mutant (Figure S7B–E). Unexpectedly, only H3K9me1 showed significant enrichment at rDNA loci, suggesting that H3K9me1 is regulated in a MoJMJD2‐dependent manner (Figure 3D). Western blotting also showed that global H3K9me1 levels were higher in the deletion mutant than in the wild type and complementation strain, supporting H3K9me1 as a major chromatin mark affected by loss of MoJMJD2 (Figure S7F,G). The enrichment of H3K9me1 was consistent with depletion of H3K9ac at rDNA loci, suggesting that the residual H3K9me1 is associated with suppression of rDNA transcription (Figure S7H).

To see if dramatic reduction in rRNA affected production of cellular proteins, we examined the de novo protein synthesis rate in the fungal hyphae using a surface sensing of translation (SUnSET) assay (Van Hoewyk 2016). This non‐radioactive method for measuring protein synthesis detects puromycylated peptides using western blotting with an anti‐puromycin antibody following puromycin incorporation into nascent polypeptides. In the wild type, addition of the protein synthesis inhibitor cycloheximide (CHX) significantly reduced de novo translation (Figure 3E,F). In contrast, ΔMojmjd2 showed much lower de novo protein synthesis even without CHX treatment. These results suggest that MoJMJD2 is required for a steady‐state rRNA synthesis and subsequently protein synthesis, likely by associating with rDNA loci and regulating H3K9me1 levels.

Because nucleolar formation requires synthesis of rRNA and nucleolar proteins (Falahati et al. 2016; Hernandez‐Verdun 2011; Savino et al. 2001), we examined nucleolar integrity in ΔMojmjd2 by expressing MoJMJD2T⁰‐GFP or MoNOP1‐RFP and monitoring the fluorescence signals. In the mutant, both GFP and RFP signals were not limited to nucleoli and instead were distributed throughout nuclei (Figure S8). This result strongly suggests disruption of nucleolar integrity in ΔMojmjd2.

2.4. MoJMJD2 Is Required for Vegetative Growth, Asexual Development and Pathogenicity

To understand the impact of gene deletion on fungal development and pathogenicity, we compared phenotypes of the mutant with those of the wild type. When grown on oatmeal agar, ΔMojmjd2 showed a 38% reduction in vegetative growth, measured as colony diameter, compared to the wild type (upper panels in Figure 4A,B). The mutant colony was also much less pigmented than the wild type. In addition, the mutant exhibited an approximately 66% decrease in aerial hyphae thickness (middle panels in Figure 4A,C).

FIGURE 4.

MoJMJD2 is required for vegetative growth, conidiation and pathogenicity in Magnaporthe oryzae. (A) Colony morphology, pigmentation and conidiophore development of wild‐type KJ201, ΔMojmjd2 and MoJMJD2c on oatmeal agar at 9 days post‐inoculation (dpi). (B) Colony diameter at 9 dpi. (C) Aerial hyphae thickness at 9 dpi. (D) Conidial production at 9 dpi. (E) Lesion development on rice leaves inoculated with mycelial agar plugs at 7 dpi. (F) Lesion development on wound‐inoculated rice leaves at 7 dpi. In (B)–(D), asterisks indicate statistically significant differences (*p < 0.05; Tukey's HSD test). Data represent three biological replicates from three independent experiments (n = 3). In (E) and (F), three pots were used per strain in each of three independent experiments. Data from the three pots within each experiment were averaged, and each independent experiment was considered one biological replicate (n = 3). Asterisks indicate statistically significant differences (*p < 0.05; Tukey's HSD test). (G) Invasive growth in rice sheath cells at 24 and 48 h post‐inoculation (hpi). AP, appressorium; IH, invasive hypha. Scale bar = 10 μm.

Deletion of MoJMJD2 had the strongest effect on asexual reproduction, causing a dramatic decrease in spore production (Figure 4D). Examination of conidiophore development revealed that ΔMojmjd2 formed very few conidiophores, consistent with its thin aerial hyphal (lowermost panels in Figure 4A). However, conidial morphology of ΔMojmjd2 was indistinguishable from those of the wild type (Figure S9A). Likewise, no obvious difference was observed in germination or appressorium formation between conidia of the wild type and mutant on an inductive surface (plastic coverslip) (Figure S9A,B).

Due to the difficulty in obtaining enough spore suspension, we used agar plugs as inoculum for pathogenicity assays (see Experimental Procedures of pathogenicity assay for details). Disease severity caused by the mutant was assessed by measuring the lesion area relative to the wild type (Figure 4E). The mutant was much less virulent, showing more than a 60% decrease in mean lesion area. Wound inoculation, which allows fungal entry without appressorium formation, showed that the mutant was also less capable of growing inside plant tissue than the wild type (Figure 4F). In addition, lesions caused by the mutant expanded only minimally during prolonged incubation.

To scrutinize the reduced pathogenicity of the mutant, we monitored the appressorium‐mediated penetration and invasive growth of ΔMojmjd2 using a rice sheath assay (Figure 4G). Unlike the wild type, which showed penetration at 24 h post‐inoculation (hpi) and expanded into neighbouring cells at 48 hpi, the mutant was generally slower in penetration and remained largely restricted to the first invaded cell at 48 hpi. The same trend was observed in onion epidermal cells (Figure S9C). These results indicate that reduced pathogenicity of the deletion mutant is attributed, at least in part, to the defects in appressorium‐mediated penetration and impaired growth inside the host. To understand why invasive growth of the mutant is compromised, we tested the ability of the mutant to cope with oxidative stresses or cell‐wall degrading enzymes, which are part of plant defence mechanisms, by adding hydrogen peroxide (Mhamdi and Van Breusegem 2018) and lysing enzymes separately or in combination to the media. The mutant showed hypersensitivity to H₂O₂ and lysing enzymes, suggesting that reduced virulence is likely associated with defects in cell wall integrity and increased sensitivity to ROS (Figure S9D–F).

Introduction of the wild‐type copy of MoJMJD2 into the mutant (MoJMJD2c) complemented all observed phenotypic defects of ΔMojmjd2, indicating that MoJMJD2 is required for normal vegetative growth, asexual development and pathogenesis (Figure 4). We also tested the functions of individual isoforms by expressing MoJMJD2T ⁰ or MoJMJD2T ¹ (under the native promoter of MoJMJD2), separately or en masse in the deletion mutant. Expression of MoJMJD2T ⁰ alone or co‐expression of both isoforms partially restored vegetative growth in the mutant, whereas MoJMJD2T ¹ alone failed to rescue the growth defect (Figure S10A,B). Expressing either isoform alone significantly increased conidial production; furthermore, co‐expression led to further enhancement, although it did not fully restore conidiation to the wild‐type levels (Figure S10C).

To determine whether these partial phenotypic effects extended to rDNA regulation and pathogenicity, we quantified rRNA transcript abundance and evaluated pathogenicity in the same set of isoform‐specific complementation strains, MoJMJD2T ⁰ c, MoJMJD2T ¹ c and MoJMJD2T ⁰ & T ¹ c (Figure S10D–F). rRNA transcript abundance remained dramatically reduced in all three complementation strains relative to the wild type, comparable to that of the ΔMojmjd2 mutant (Figure S10D). Among the complementation strains, MoJMJD2T ⁰ & T ¹ c showed a small but statistically significant increase in rRNA abundance relative to the ΔMojmjd2 mutant, whereas MoJMJD2T ⁰ c and MoJMJD2T ¹ c individually did not differ significantly from the mutant. Notably, rRNA abundance in MoJMJD2T ⁰ c was significantly higher than in MoJMJD2T ¹ c, indicating that MoJMJD2T⁰ contributes more strongly to rDNA regulation than MoJMJD2T¹, even though neither isoform alone was sufficient to restore rRNA levels to wild‐type levels. Pathogenicity assays revealed a similar pattern (Figure S10E,F): KJ201 produced substantially larger lesions than all other strains, whereas the ΔMojmjd2 mutant, MoJMJD2T ⁰ c, MoJMJD2T ¹ c and MoJMJD2T ⁰ & T ¹ c did not differ significantly from one another, although MoJMJD2T ⁰ c showed a numerically higher mean lesion diameter, consistent with a weak residual pathogenicity conferred by MoJMJD2T⁰. Taken together, these results indicate that neither isoform alone nor their combined expression is sufficient to restore rRNA levels or pathogenicity to wild‐type levels, and that MoJMJD2T⁰ makes a modest but consistently greater contribution than MoJMJD2T¹ to both phenotypes.

2.5. MoJMJD2‐Dependent Histone Demethylation and Transcriptional Regulation of Nuclear Genes

Given the defects in colony pigmentation and asexual reproduction of the mutant, we tested if genes involved in melanin production and conidiogenesis are regulated at the transcriptional level in a MoJMJD2‐dependent manner. To this end, transcript abundances of melanin biosynthesis and conidiogenesis‐related genes were first measured and compared using reverse transcription (RT)‐qPCR in the wild‐type, ΔMojmjd2 and MoJMJD2c. In parallel, ChIP‐qPCR was performed to check the histone methylation status in H3K9 and H3K36.

For melanin biosynthesis, eight genes including ones encoding polyketide synthase (MoPKS1) and trihydroxynaphthalene reductase (MoTHR1) were examined. All eight genes were down‐regulated in the mutant compared to the wild type and complementation strain, suggesting that the pigmentation‐less phenotype of the mutant is due at least in part to such suppression of melanin biosynthesis‐related genes in the mutant (Figure 5A). The promoter regions of five genes, including MoPKS1 and MoTHR1, were enriched in H3K9me1 (Figure 5B), while either similar or lower levels of H3K9me2, H3K9me3 and H3K36me3 were detected (Figure S11A–C).

FIGURE 5.

MoJMJD2‐dependent transcriptional and chromatin changes in nuclear genes involved in melanin biosynthesis, conidiogenesis and cell wall synthesis. (A) Relative transcript abundance of melanin synthesis‐related genes in wild‐type strain (KJ201), ∆Mojmjd2 and MoJMJD2c. (B) H3K9me1 enrichment of melanin synthesis‐related genes showing > 2‐fold reduced transcript abundance. (C) Relative transcript abundance of conidiogenesis‐related genes in KJ201, ∆Mojmjd2 and MoJMJD2c. (D) H3K9me1 enrichment of conidiogenesis‐related genes showing > 2‐fold reduced transcript abundance. (E) Relative transcript abundance of cell wall synthesis‐related genes in KJ201, ∆Mojmjd2 and MoJMJD2c. (F) H3K9me1 enrichment of cell wall synthesis‐related genes showing > 2‐fold reduced transcript abundance. In (B), (D), and (F), asterisks indicate statistically significant differences (*p < 0.05, **p < 0.01, ***p < 0.001; Tukey's HSD test) based on three independent biological replicates, each analysed with three technical replicates (n = 3).

For asexual reproduction, 13 genes, including MoSTUA, MoAPS1, MoACR1, MoFLBA, MoFLBC and MoFLBD (Kim and Lee 2012; Kim et al. 2009), were examined. Most of these genes were down‐regulated more than 2‐fold in ΔMojmjd2, compared to the wild‐type strain (Figure 5C). At their promoter regions, H3K9me2 and H3K9me3 levels did not differ significantly between ΔMojmjd2 and KJ201 (Figure S11D,E). In contrast, H3K9me1 and H3K36me3 appeared to be enriched in the promoter of some genes in the mutant, compared to the wild type (Figure 5D and Figure S11F). Our data suggest that MoJMJD2 is implicated in regulation of asexual reproduction in association with changes in H3K9me1 and H3K36me3.

Due to the hypersensitivity to hydrogen peroxide treatment and cell wall integrity in the mutant, we checked the transcription of genes involved in ROS and cell wall synthesis‐related genes. Comparison of ROS‐related gene expression among strains showed that none of them except for Catalase‐3 and SOD4 seem to be differentially expressed in the mutant (> 2‐fold criterion) (Figure S12). In contrast, many cell wall synthesis‐related genes were down‐regulated in the mutant, suggesting that disruption in cell wall integrity might be responsible for sensitivity to H₂O₂ and lysing enzyme (Figure 5E). Because expression of these genes depended on MoJMJD2, we examined H3K9 and H3K36 methylation at their loci. ChIP‐qPCR analysis of nine cell wall synthesis‐related genes again showed significant differences between ΔMojmjd2 and KJ201 for H3K9me1, but not H3K9me2, H3K9me3 or H3K36me3 (Figure 5F and Figure S11G–I). Because MoJMJD2T¹‐GFP localizes to the nucleus in fungal hyphae, we reasoned that nuclear genes showing increased H3K9me1 in the mutant might be potential targets of MoJMJD2‐dependent regulation. Contrary to this expectation, however, the immunoprecipitation experiment showed that the promoters of nuclear genes examined were not enriched with MoJMJD2T¹, suggesting that MoJMJD2 affects their transcription indirectly (Figure S13).

3. Discussion

Here, we provide evidence that MoJMJD2, a gene encoding histone demethylase, is important for rDNA transcription and nucleolar integrity in the rice blast fungus. Histone methylation/demethylation is a key epigenetic mechanism for regulation of development and differentiation of cells (Berry and Janknecht 2013; D'Oto et al. 2016). In M. oryzae , methylation and demethylation at H3K4, catalysed by MoSET1 and MoJMJ1, respectively, are required for infection‐specific development (Huh et al. 2017; Pham et al. 2015). Regulation of H3K27me3 is also important for in planta expressed genes and fungal pathogenicity (Wu et al. 2022; Zhang et al. 2021). In this study, we found that loss of MoJMJD2 altered the methylation status of H3K9me1 and H3K36me3, probably through distinct mechanisms. Sequence homology and domain analysis revealed that MoJMJD2 is an evolutionarily conserved histone demethylase (HDM) and is closest to KdmA of A. nidulans in protein length and domain architecture (Gacek‐Matthews et al. 2015; Klose et al. 2006). KdmA demethylases H3K9me and H3K36me. An orthologue of MoJMJD2 in S. cerevisiae also targets H3K9me3 and H3K36me3 for demethylation (Klose et al. 2007; Tu et al. 2007). Despite this evolutionary conservation, our data suggest that MoJMJD2‐dependent control of H3K9me1 may occur directly or indirectly, while its effect on H3K36me3 appears to occur indirectly. Although canonical KDM4/JMJD2 family members are best characterized as H3K9me3/H3K36me3 demethylases, ChIP‐qPCR analysis showed that MoJMJD2T⁰‐GFP is enriched at the rRNA promoter and the 18S and 26S coding regions, but not at the short intervening ITS region (ITS1–5.8S–ITS2, ~450–510 bp), indicating that MoJMJD2 directly occupies specific subregions of the rDNA locus. Consistent with this, H3K9me1 accumulation in the mutant was observed across all examined regions, including the ITS region. Because the ITS region is physically flanked by the 18S and 26S regions where MoJMJD2T⁰ occupancy was detected, and because MoJMJD2 possesses the catalytic residues required for Fe(II)/α‐ketoglutarate‐dependent demethylase activity, we propose that MoJMJD2 is more likely to directly restrain H3K9me1 at the regions it occupies, and that the increase in H3K9me1 within the short ITS region probably reflects spreading of this mark from the adjacent, directly regulated 18S and 26S regions, rather than a direct effect of MoJMJD2 at this site. Further studies, including generation of a catalytically dead MoJMJD2 mutant and in vitro demethylation assays using recombinant protein, will be needed to formally establish the direct enzymatic activity of MoJMJD2 toward H3K9me1. These findings suggest that MoJMJD2 may play mechanistically distinct roles in association with different histone methylation marks. Further studies will be required to elucidate how MoJMJD2 mediates this differential regulation and why its substrate preference appears to differ from that of related fungal demethylases.

While alternative splicing is prevalent in animals, it is considered much rarer in fungi (Fang et al. 2020). Nevertheless, an increasing number of studies have revealed that alternative splicing might be more widespread than previously thought. In our work, we showed that the MoJMJD2 gene produces two splicing variants (MoJMJD2T⁰ and MoJMJD2T¹), the ratio of which appears to be regulated in response to environmental conditions. MoJMJD2T⁰ contains a nucleolar localization sequence (NoLS) that is absent in MoJMJD2T¹. Consistent with this, MoJMJD2T⁰ localized to the nucleolus, while MoJMJD2T¹ localized to the nucleus in fungal hyphae, indicating distinct roles for the two isoforms. A gene deletion approach combined with RT‐qPCR analysis and ChIP analysis revealed that maintaining a steady‐state level of cellular rRNA is dependent on MoJMJD2T⁰ and is associated with H3K9me1 regulation at the rDNA loci. Such low level of rRNA compromised de novo cellular protein synthesis. To further dissect the individual contribution of each isoform, we generated isoform‐specific complementation strains expressing MoJMJD2T ⁰ alone, MoJMJD2T ¹ alone, or both isoforms together in the ΔMojmjd2 background. Neither isoform alone nor their combined expression fully restored rRNA abundance or pathogenicity to wild‐type levels, in clear contrast to the full rescue achieved by the native gDNA construct, although the strain expressing MoJMJD2T ⁰ alone conferred a modest but consistent partial rescue of vegetative growth, conidiation and pathogenicity. These results indicate that neither isoform alone is sufficient to substitute for the other, and that coordinated expression of both MoJMJD2T ⁰ and MoJMJD2T ¹ under their native regulatory context is required for full MoJMJD2 function. Notably, rRNA abundance was significantly higher in the MoJMJD2T ⁰ c strain than in the MoJMJD2T¹ c strain, further supporting a greater contribution of MoJMJD2T⁰ to rDNA regulation, consistent with its nucleolar localization. However, although MoJMJD2T¹ is localized to the nucleus and may influence nuclear gene expression, ChIP‐qPCR analysis did not detect appreciable enrichment of MoJMJD2T¹‐GFP at the promoters of nuclear genes that showed differences in both transcription and histone methylation levels (H3K9me1 and H3K36me3) between the wild type and mutant, suggesting that MoJMJD2T¹ may not directly regulate transcription of these genes through promoter occupancy. Nevertheless, because the MoJMJD2T¹‐GFP complementation strain conferred only partial phenotypic rescue, this interpretation should be considered in light of the possibility that differences in the expression or regulation of the fusion protein may have limited our ability to detect its association with chromatin. Therefore, we cannot fully exclude a direct regulatory role for MoJMJD2T¹ at these loci. Although MoJMJD2T¹ transcript abundance increased under sporulation‐promoting conditions, its subcellular localization under this condition could not be determined; therefore, its isoform‐specific contribution to asexual development remains to be clarified. In some cases, it has been shown that two splicing isoforms play different roles in the cell. For example, in Lachancea kluyver, a single SKI7/HBS1 gene carries out two different functions via alternative splicing, generating functional diversity in fungal proteome (Marshall et al. 2013). In Nile tilapia, temperature regulates the ratio of alternative splicing variants of histone demethylase Kdm6bb that mediates sex determination (Yao et al. 2023). In the rice blast fungus, recently it was shown that MoYPD1, which is important for osmoregulation, produces isoforms in a signal‐specific manner, and that full virulence requires presence of all the isoforms (Buhring et al. 2024). It should be noted that MoJMJD2 produces two isoforms that are different not only in function but also in localization: one in the nucleolus and the other in the nucleus. It is therefore tempting to conjecture that regulation of the isoform ratio in combination with their distinct localization pattern enables the fungus to coordinate nucleolar activity including rRNA production with transcription of nuclear genes in response to environmental conditions.

In mammalian cells, HDMs such as KDM2A, KDM4A and JHDM1B were shown to localize in the nucleoli and regulate rDNA transcription in response to nutrients, stresses and growth factors (Frescas et al. 2007; Salifou et al. 2016; Tanaka et al. 2010). To our best knowledge, epigenetic regulation of the nucleolus has not been reported to date in filamentous fungi. In our previous study, we described the dynamics of the nucleolus under different nutrient conditions and during pathogenic development (Cho et al. 2023). This work showed that nucleolar size and morphology of the rice blast fungus are distinct from those of animals and S. cerevisiae . In our current study, we revealed that one pivotal mechanism associated with regulation of nucleolar function and integrity in the fungus involves MoJMJD2‐dependent chromatin regulation at rDNA loci. Because H3K9me1 plays a role as a repressive mark for gene transcription, deletion of MoJMJD2 appears to increase H3K9me1 retention at rDNA loci and subsequently suppress rRNA production. This was in concordance with depletion of H3K9ac in rDNA loci of the mutant. TOR kinase is likely to act upstream of MoJMJD2, as rapamycin treatment reduced the transcription of the MoJMJD2 gene. This is also biologically plausible, because rDNA transcription and subsequent ribosome production should be adjusted to the nutrient status of the cell, for which the TOR signalling pathway is responsible. Notably, a list of genes undergoing alternative splicing of pre‐mRNA upon knockout of a gene encoding arginine methyltransferase, MoHMT1, includes MoJMJD2, suggesting the possibility that MoHMT1 is one of the components regulating transcription and alternative splicing of MoJMJD2 (Fang et al. 2020).

The deletion mutant of MoJMJD2 was defective in vegetative growth, pigmentation, asexual reproduction and pathogenicity. We found that in the ΔMojmjd2, not only rDNA loci but also the genes involved in melanin biosynthesis, asexual reproduction, and cell wall synthesis were down‐regulated, which may at least partly explain why the mutant displays the observed phenotypes. Notably, while the MoJMJD2c complementation strain fully restored the major phenotypes associated with MoJMJD2 function, including vegetative growth, pigmentation, asexual reproduction and pathogenicity, the transcript levels of some of these downstream genes in MoJMJD2c did not always fully match those of the wild‐type strain KJ201. We consider this residual quantitative discrepancy likely to reflect strain‐to‐strain variability commonly associated with fungal transformation and complementation, such as differences in the genomic context of the reintroduced construct or minor variation in gene copy number rather than incomplete functional complementation of MoJMJD2 itself, given that all major phenotypes were fully rescued. Such incomplete quantitative restoration of individual downstream gene expression despite full phenotypic rescue is not uncommon in complementation studies using reintroduced genomic constructs. Because deletion of MoJMJD2 results in reduced global translation, it is conceivable that the fungus cannot efficiently sustain growth‐related processes due to the depletion of cellular proteins. Such perturbation in cellular proteostasis might impact cell cycle and autophagy, which are required for hyphal growth and fungal development including asexual development. Deletion of the MoJMJD2 gene resulted in a significant decrease in pathogenicity as well, compared to the wild‐type strain, indicating that MoJMJD2 is required for full virulence of the rice blast fungus. Our analysis of pathogenicity in the ΔMojmjd2 revealed that such decrease in virulence is associated with defects in appressorium‐mediated host penetration and invasive growth. The dynamic redistribution of MoJMJD2T⁰ observed during appressorium formation—from the nucleolus to punctate structures and subsequently to the cell periphery—raises the question of its underlying mechanism and functional significance, which remain to be elucidated. In mammalian cells, nucleolar proteins are known to relocalize away from the nucleolus in response to nutrient limitation and other cellular stresses, a phenomenon termed the nucleolar stress response, in which such proteins frequently acquire extraribosomal functions outside the nucleolus (Boulon et al. 2010). Given that appressorium formation involves substantial nutrient depletion and cellular reprogramming, it is possible that MoJMJD2T⁰ undergoes a similar stress‐induced relocalization. Because appressorium morphogenesis in M. oryzae is critically dependent on conidial autophagy (Veneault‐Fourrey et al. 2006), it is also conceivable that the punctate structures observed during the intermediate stage of relocalization reflect autophagy‐related compartments. Furthermore, the eventual accumulation of MoJMJD2T⁰ at the cell periphery coincides with the site where the septin‐scaffolded F‐actin network assembles to organize the penetration peg (Dagdas et al. 2012), raising the possibility that MoJMJD2T⁰ may perform a chromatin‐independent, moonlighting role at this location. These possibilities are speculative at present, and future studies—such as co‐localization analysis with autophagy or septin markers—will be needed to clarify the mechanism and biological relevance of this dynamic relocalization. Detailed analysis pointed out that increased sensitivity to ROS and cell wall stress appeared to be responsible for the reduction in invasive growth of the mutant. This was further supported by RT‐qPCR analysis of genes involved in ROS detoxification and cell wall synthesis. Among seven genes associated with the detoxification of ROS, the transcript levels of only two genes were down‐regulated in the mutant, whereas the transcript levels of seven genes among nine genes involved in cell wall synthesis were down‐regulated in the mutant. Although many genes showed enrichment in H3K9me1, MoJMJD2T¹ did not appear to bind the promoter regions of those genes. These observations suggest that MoJMJD2T¹ regulates these genes indirectly, rather than through direct promoter occupancy, possibly via effects on nucleolar function, translational capacity, or broader chromatin organization. It is noteworthy that many phenotypes of ΔMojmjd2 are reminiscent of those observed in the deletion mutant of MoMCK1 gene, which is a MAP kinase kinase kinase that is implicated in the cell wall integrity signalling pathway (Jeon et al. 2008). It would be interesting to see whether the regulation of cell wall synthesis genes by MoJMJD2 is mediated by the cell wall integrity signalling pathway. Clearly, further studies are required to determine how exactly relative transcript abundances of nuclear genes change in ΔMojmjd2. Although analysis of an additional independent deletion mutant would further strengthen the genetic evidence, the ΔMojmjd2 mutant was generated by targeted gene replacement, and single‐copy, locus‐specific integration was confirmed by inverse PCR and Southern blot analysis (Figure S6). These analyses support the conclusion that the observed phenotypes are associated with the targeted disruption of the MoJMJD2 locus, although they cannot by themselves exclude the possibility of additional background or second‐site mutations elsewhere in the genome. Importantly, complementation with the native MoJMJD2 gDNA construct fully restored the major phenotypes of the ΔMojmjd2 mutant, including vegetative growth, pigmentation, asexual reproduction and pathogenicity. Together, these genetic data provide strong evidence that the major phenotypes observed in ΔMojmjd2 are attributable to loss of MoJMJD2, while the absence of an additional independent deletion mutant remains a limitation that should be addressed in future studies.

In summary, here we propose a novel role of MoJMJD2 in the spatial regulation of epigenetic changes inside the fungal nucleolus and of nuclear genes (Figure 6). MoJMJD2 achieves this feat through alternative splicing, which allows functional diversity and spatial coordination in an environment‐dependent manner. MoJMJD2T⁰ moves into the nucleolus and is associated with reduced H3K9me1 levels and activation of rDNA transcription, while MoJMJD2T¹ stays in the nucleus (hyphae) or nucleus‐associated structure (conidia). Lack of MoJMJD2 disrupted nucleolar integrity, leading to the mislocalization of nucleolar proteins such as MoNOP1. We demonstrated via genetic analysis that MoJMJD2 is required for vegetative growth, asexual reproduction, appressorium‐mediated penetration and invasive growth. Our results thus support a strong link between nucleolar function/integrity and fungal development/pathogenesis and imply that components involved in nucleolar functions and regulation may represent promising targets for devising crop disease control strategies.

FIGURE 6.

Schematic representation of the role of MoJMJD2 in controlling development and pathogenicity in the rice blast fungus. MoJMJD2 produces two protein isoforms that are localized to the nucleolus or nucleus. One isoform regulates the transcriptional activity of rDNA loci via demethylation of H3K9me1 in the nucleolus, and the other form appears to regulate nuclear genes via demethylation of H3K9me1 and H3K36me3. Solid arrows represent the regulation/link that is supported by experimental evidence, while the dotted arrow represents the one that is hypothesized by circumstantial evidence.

4. Experimental Procedures

4.1. Fungal Isolates and Culture Conditions

The wild‐type strain KJ201 and RFP‐fused histone H1 (H1‐RFP) were obtained from the Center for Fungal Genetic Resources (CFGR), South Korea. All strains, including mutants, were grown on oatmeal agar (OMA; 5% oatmeal and 2.2% agar, w/v) at 25°C under constant white light with aeration to promote conidiation. Mycelia for DNA, RNA and protoplast extraction were cultured in complete medium (CM; 0.6% yeast extract, 0.6% casamino acid and 1% sucrose, w/v) at 25°C with shaking at 120 rpm. Transformants resistant to hygromycin, geneticin or bialaphos were selected on TB3 agar (0.3% yeast extract, 0.3% casamino acid, 1% glucose, 20% sucrose and 0.8% agar, w/v) supplemented with hygromycin B (200 ppm), geneticin (400 ppm) or bialaphos (1250 ppm), respectively. ROS and cell wall stress assays were performed on CM agar supplemented with H₂O₂ (1, 3, or 5 mM) or lysing enzyme from Trichoderma harzianum (10 mg/mL; Sigma‐Aldrich).

4.2. Sequence and Domain Architecture Analysis

Nucleotide and amino acid sequences were retrieved from National Center for Biotechnology Information (NCBI). Multiple sequence alignment was performed using Clustal Omega (Sievers and Higgins 2018), domain architecture was analysed using InterProScan (Finn et al. 2017), and alternatively spliced isoforms were predicted using the Comparative Fungal Genomics Platform (CFGP). Nucleolar localization sequences were predicted using the NoD server (Scott et al. 2011).

4.3. Targeted Gene Deletion and Complementation

Deletion constructs were generated by double‐joint PCR (Yu et al. 2004) using approximately 1 kb of the 5′ and 3′ flanking regions of MGG_09186 fused to either side of the hygromycin resistance gene (HPH) cassette. Constructs were transformed into the wild‐type protoplasts, and candidate transformants were screened on hygromycin‐containing TB3 agar and verified by PCR using gene‐ and marker‐specific primers (Figure S5; Table S3). Correct replacement and absence of ectopic integration were confirmed by inverse PCR and Southern blotting. For complementation, genomic MGG_09186 including its upstream region was co‐transformed with a geneticin resistance marker into the deletion mutant, and complemented strains were selected on geneticin‐containing TB3 agar and verified by PCR using a primer pair for the MoJMJD2 gDNA.

4.4. Construction of Fluorescence‐Fused Proteins

To analyse subcellular localization, MoJMJD2T⁰ and MoJMJD2T¹ were fused to GFP under their predicted native promoter or the constitutive TrpC promoter, respectively, using the EZ‐Fusion HT Cloning Kit (Enzynomics). MoNOP1 (MGG_09195) was fused to RFP under its predicted native promoter as a nucleolar mark. In all fusion constructs, including those used for complementation, a flexible linker consisting of six glycine residues (6 × Gly) was inserted between the coding sequence of each protein (MoJMJD2T⁰, MoJMJD2T¹, or MoNOP1) and the corresponding fluorescent tag. Constructs were cloned into the T&A cloning vector (RBC). Primers used to amplify ORFs and predicted promoter regions are listed in Table S4.

4.5. Fluorescent Microscopy

Localization of MoJMJD2 isoforms was examined using strains expressing MoJMJD2T⁰‐GFP/MoNOP1‐RFP, MoJMJD2T⁰‐GFP/H1‐RFP or MoJMJD2T¹‐GFP/H1‐RFP. Conidia from 9‐day‐old OMA cultures were suspended in water, adjusted to 5 × 10⁴ conidia/mL, and 40 μL was placed on hydrophobic or hydrophilic surfaces for incubation at 25°C in a humid chamber for 0, 0.5, 2, 4 or 6 h. For hyphal observations, strains were grown on CM‐coated slide glasses for 5 days at 25°C; nuclei were stained with Hoechst 33342 (Invitrogen) where indicated. Differential interference contrast and fluorescence images were obtained using a DM2500 microscope (Leica) equipped with a DFC7000 T camera (Leica) and processed with LAS X software (Leica).

4.6. Developmental Phenotype Analysis

Strains were cultured on OMA at 25°C under continuous light for 9 days. Colony growth, aerial hyphal formation, conidiation, conidial germination and appressorium formation were examined in three biological replicates from three independent experiments. Aerial hyphae were observed microscopically after scraping mycelia from agar blocks. Conidia were counted with a haemocytometer. Germination and appressorium formation were assessed after 8 h incubation of 40 μL conidial suspension (5 × 10⁴ conidia/mL) on hydrophobic coverslips at 25°C under high humidity. Images were captured with a DM2500 microscope (Leica) and DFC7000 T camera (Leica).

4.7. Pathogenicity Assay

For pathogenicity assay, 4‐mm OMA plugs from 9‐day‐old cultures were placed on leaves of 3‐week‐old rice plants or injected into wound sites. Inoculated plants were incubated at 25°C in a humid chamber in the dark for 24 h, then under 80% humidity with a 16 h light period. Lesions were assessed at 7 days post‐inoculation (dpi). Three pots were used per strain in each of three independent experiments. Data from the three pots within each experiment were averaged to obtain one biological replicate, resulting in three biological replicates per strain. For rice sheath assays, 4‐week‐old sheaths were inoculated with conidial suspensions (5 × 10⁴ conidia/mL) and incubated at 25°C under a 16 h light period. Sheaths were examined microscopically at 24 and 48 hpi.

4.8. Nucleic Acid Manipulation and Gene Expression Analysis

Genomic DNA for transformant screening was extracted as described previously (Chi et al. 2009). Unless otherwise indicated, strains used for gene expression analysis were cultured in CM for 3 days at 25°C with shaking prior to RNA extraction. For gene expression analysis of conidiation‐related genes (Figure 5C), strains were instead cultured on OMA at 25°C under continuous light for 9 days prior to RNA extraction, consistent with the conidiation‐inducing conditions described in Section 4.6. Total RNA was isolated using the Easy‐BLUE Total RNA Extraction Kit (Intron Biotechnology), and cDNA was synthesized using the ImProm‐II Reverse Transcription System (Promega). Quantitative real‐time PCR was performed in 20 μL reactions containing 2 μL cDNA (25 ng input RNA), 1 μM primers and 10 μL Power SYBR Green PCR Master Mix (Applied Biosystems) using a StepOnePlus Real‐Time PCR system (Applied Biosystems) with 40 cycles of 95°C for 15 s, 60°C for 30 s and 72°C for 30 s. Relative transcript abundance was normalized to β‐tubulin and calculated using the 2^−ΔΔCt method (Livak and Schmittgen 2001). Primer sequences are listed in Table S2. Three technical replicates were performed for each of three independent biological replicates.

4.9. ChIP‐qPCR

Mycelia grown in CM for 3 days were cross‐linked in 100 mL buffer A (1 mL 1 M Tris‐Cl; pH 8.0, 20 mL 2 M sucrose, 1 mL 1 M sodium butyrate, 100 μL 0.1 M phenylmethylsulfonyl fluoride (PMSF), 2.7 mL 1% formaldehyde, 36 μL 14.1 M β‐mercaptoethanol and 75 mL distilled water) for 20 min at 20°C with shaking (120 rpm), and the reaction was quenched with 10 mL of 1 M glycine. For ChIP‐qPCR analysis of the conidiation‐related gene promoter examined in Figure 5D, mycelia were instead obtained from strains grown on OMA at 25°C under continuous light for 9 days, consistent with the conidiation‐inducing conditions described in Section 4.6. Samples were filtered through Miracloth, frozen in liquid nitrogen and ground to a powder. Nuclei were isolated using the CelLytic PN Isolation/Extraction Kit for plant (Sigma‐Aldrich). Chromatin was sheared by sonication (cycle = 1, amplitude = 20) and ChIP was performed using the Pierce Agarose ChIP Kit (Thermo Fisher Scientific). Antibodies used were anti‐histone H3K9me1 (ab9045), H3K9me2 (ab1220), H3K9me3 (ab8898), H3K9ac (ab4441), H3K27me3 (ab6002), H3K36me3 (ab9050) and anti‐GFP (ab290) (all from Abcam). ChIP‐enriched DNA and input DNA were subjected to qPCR using the same cycling conditions as described above for qPCR, with gene‐specific primer pairs listed in Table S5. The threshold cycle values obtained from 5% input samples were adjusted to represent 100% input by subtracting log₂(20)≈4.32. Enrichment of specific genomic regions was calculated using the percent input method (Lacazette 2017), as follows: % Input = 100 × 2^{(Adjusted Ct input—Ct IP)}. ChIP‐qPCR was performed using three independent biological replicates, each analysed with three technical replicates.

4.10. SUnSET Assay

To detect newly synthesized proteins, 3‐day‐old mycelia grown in CM were treated with 1 mM puromycin for 20 min and analysed using the surface sensing of translation (SUnSET) (Cary et al. 2014; Lee et al. 2023). Total protein was extracted with Mg/NP‐40 buffer (Kim et al. 2001), separated by 12% SDS‐PAGE, transferred to a PVDF membrane, and stained with Ponceau S to confirm equal loading. Membranes were blocked in 5% skim milk in Tris‐buffered saline with 0.1% Tween 20 (v/v) (TBST) for 1 h, followed by incubation with anti‐puromycin antibody (1:10,000; MABE343, Sigma‐Aldrich) for 1 h. After washing three times with TBST for 10 min each, the membrane was incubated with anti‐mouse IgG‐horseradish peroxidase antibody (1:10,000; ab6789, Abcam) for 1 h. The membrane was washed three times with TBST for 10 min each, and the immunoblot signals were detected using ECL. As a control, mycelia were treated with 20 μM cycloheximide for 2 h before protein extraction.

Author Contributions

Song Hee Lee: conceptualization, investigation, formal analysis, visualization, validation, project administration, writing – original draft. Taehyeon Kim: conceptualization, formal analysis, visualization, validation, investigation. Joo Hyun Lee: investigation. Jiyeun Park: investigation. Ki‐Tae Kim: data curation, investigation. Hokyoung Son: investigation. Junhyun Jeon: conceptualization, funding acquisition, project administration, supervision, writing – review and editing.

Funding

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (RS‐2024‐00407104 and RS‐2025‐16071113).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Alignment of sequence motifs in JmjC domain of MoJMJD2 and the JMJD2 group of histone demethylases. Asterisks indicate Fe²⁺ binding sites. Pound signs indicate α‐ketoglutarate binding sites. Black shadows indicate the conserved amino acid sequence in histone demethylase.

Figure S2: Phylogenomic tree of the selected species and the orthologues of MoJMJD2. The species tree only shows topology (bootstrap value over 50%). The Pfam domains of the homologues are colour‐coded. The alternative splicing form of JMJD2 was only predicted for Magnaporthe oryzae, Arabidopsis thaliana , Homo sapiens and Sarcophilus harrisii .

Figure S3: Alternative splicing of JMJD2 in Magnaporthe oryzae and Fusarium graminearum. (A) Gel electrophoresis image showing the amplification of partial cDNA for alternative spliced isoforms of MoJMJD2T ⁰ and MoJMJD2T ¹ using isoform‐specific primer sets. (B) Partial sequence of MoJMJD2T ⁰ and MoJMJD2T ¹ cDNA. (C) Schematic representation of JMJD2 alternative splicing isoforms in F. graminearum. Two types of alternative splicing isoforms, JMJD2 isoform 1 and isoform 2, were predicted. (D) Electrophoresis image showing the amplification of partial gDNA and cDNA for the alternative spliced isoforms of JMJD2 in F. graminearum. (E) Partial sequence of JMJD2 gDNA and cDNA of isoform 2 in F. graminearum.

Figure S4: Localization of MoJMJD2T⁰‐GFP during spore development on a hydrophilic surface. Nuclei are marked with H1‐RFP. Merged panels were generated by combining all images from fluorescence microscopy. DIC, differential interference contrast. Scale bar = 10 μm.

Figure S5: Verification of MoJMJD2 protein expression in Magnaporthe oryzae. (A) Western blot analysis of GFP‐tagged MoJMJD2T⁰ (expressed under the native promoter in the KJ201 strain) and MoJMJD2T¹ (expressed under the TrpC promoter in the KJ201 strain). Proteins were extracted from hyphae and detected using an anti‐GFP antibody. (B) Western blot analysis of GFP‐tagged MoJMJD2T¹ (expressed under the TrpC promoter in the KJ201 strain) using protein extracts from hyphae and spores. Detection was preformed with an anti‐GFP antibody. Ponceau S staining was used as a loading control.

Figure S6: Deletion of targeted gene, MoJMJD2 via gene replacement. (A) Schematic diagram of the targeted gene replacement strategy (top figure). The knockout construct was obtained by double‐joint PCR. The MoJMJD2 gene was replaced with a hygromycin resistance gene (HPH) cassette by homologous recombination. The resulting mutants from protoplast transformation were screened by PCR using primer sets specific to genomic DNA and the HPH cassette (figure in the middle, relevant to (B)). In addition, inverse PCR was performed to verify correct gene replacement and ensure the absence of ectopic integration (bottom figure). (B) Electrophoresis image showing the amplification of the fusion construct at each flanking site of MoJMJD2 and the HPH cassette. The band in the first lane represents the amplicon from the 5′ flanking site of MoJMJD2 and the 5′ site of the HPH cassette. The band in the second lane represents the amplicon from the 3′ site of the HPH cassette and the 3′ flanking site of MoJMJD2. (C) Electrophoresis image of inverse PCR performed using the restriction enzyme, XbaI. The result showed a single amplicon of the predicted size, indicating the correct gene replacement event. (D) Schematic representation of the BamHI restriction sites in the MoJMJD2 locus (upper panel) and the HPH‐replaced locus (lower panel). The positions of the probes used for Southern blot hybridization are indicated by grey bars. White boxes represent flanking sequences, grey arrows indicate gene regions and the direction of the arrows represents gene orientation. (E) Southern blot analysis of the MoJMJD2 gene replacement. Genomic DNA was completely digested with BamHI (left panel) and hybridized with the respective probe shown in (D) (right panel). The altered band size in the mutant compared to the wild type indicates successful gene replacement.

Figure S7: Enrichment of JMJD2T⁰ at nuclear genes and histone modifications in rDNA loci. (A) Enrichment of JMJD2T⁰‐GFP at the promoter regions of melanin synthesis‐, conidiogenesis‐ and cell wall synthesis‐related genes. (B) Enrichment of H3K9me2 at rDNA loci in KJ201 and ∆Mojmjd2. (C) Enrichment of H3K9me3 at rDNA loci in KJ201 and ∆Mojmjd2. (D) Enrichment of H3K27me3 at rDNA loci in KJ201 and ∆Mojmjd2. (E) Enrichment of H3K36me3 at rDNA loci in KJ201 and ∆Mojmjd2. (F) Western blot analysis of global H3 methylation states in KJ201, ∆Mojmjd2 and MoJMJD2c strains. Total histone proteins were probed with antibodies specific to H3K9me1, H3K9me2, H3K9me3 and H3. Total histone H3 was used as a loading control. (G) Quantification of histone methylation levels shown in (F). Band intensities were measured and normalized to total histone H3. Two independent western blot experiments were performed, yielding reproducible results; one representative result is shown here. (H) Enrichment of H3K9ac at rDNA loci in KJ201 and ∆Mojmjd2. In (A)–(E) and (H), asterisks indicate statistically significant differences (*p < 0.05, **p < 0.01, ***p < 0.001; Tukey's HSD test) based on three independent biological replicates, each analyzed with three technical replicates (n = 3).

Figure S8: Localization of MoJMJD2T⁰ and MoNOP1 in ΔMojmjd2. (A) Localization of MoJMJD2T⁰‐GFP in vegetative hyphae of ΔMojmjd2. (B) Localization of MoNOP1‐RFP in vegetative hyphae of ΔMojmjd2. Nuclei are stained with Hoechst 33342. Merged panels were generated by combining all images from fluorescence microscopy. DIC, differential interference contrast. Scale bar = 10 μm.

Figure S9: Germination, appressorium formation, pathogenicity and hyphal growth on media supplemented with hydrogen peroxide or lysing enzyme. (A) Conidial germination and appressorium formation in KJ201, ΔMojmjd2 and MoJMJD2c, observed under a bright‐field microscope. Scale bar = 50 μm. (B) Quantification of conidia germination and appressorium formation rate at 8 h post‐incubation (hpi). No statistically significant differences among strains were detected (one‐way ANOVA followed by Tukey's HSD test) based on three independent biological replicates (n = 3). (C) Inoculation of onion epidermal cells with KJ201, ∆Mojmjd2 and MoJMJD2c. Invasive growth in onion epidermal cells was monitored under the microscope at 24 and 48 hpi. AP, appressorium; IH, invasive hyphae. Scale bar = 50 μm. (D) Sensitivity of KJ201 and ∆Mojmjd2 to cell wall lysing enzyme. Left panels show protoplasts produced from KJ201 and ΔMojmjd2 using a lysing enzyme. Right panel shows the number of protoplasts in KJ201 and ΔMojmjd2 measured 2 h after lysing enzyme treatment. Asterisks indicate statistically significant differences (p < 0.05, Tukey HSD test) based on three independent biological replicates (n = 3). (E) Growth of KJ201, ∆Mojmjd2, and MoJMJD2c on media supplemented with hydrogen peroxide and cell wall lysing enzyme, separately or in combination. Growth was assessed on complete medium (CM) agar at 9 days post‐inoculation (dpi). (F) Bar graph showing growth reduction in KJ201, ∆Mojmjd2 and MoJMJD2c relative to the control. NT, no treatment; LE, lysing enzyme. Asterisks indicate statistically significant differences (p < 0.05, Tukey HSD test) based on three independent biological replicates (n = 3).

Figure S10: Vegetative growth, asexual reproduction, rRNA transcript abundance, and pathogenicity of Magnaporthe oryzae wild‐type strain (KJ201), MoJMJD2 deletion mutant (∆Mojmjd2), MoJMJD2T ⁰ complementation strain (MoJMJD2T ⁰ c), MoJMJD2T ¹ complementation strain (MoJMJD2T ¹ c) and dual isoform complementation strain (MoJMJD2T ⁰ and T ¹ c). (A) Colony morphology of the indicated strains at 9 days post‐inoculation (dpi). (B) Colony diameter measurements of the strains shown in (A) at 9 dpi. (C) Quantification of asexual reproduction by counting conidia produced by the strains in (A) at 9 dpi. In (B) and (C), because the magnitude and variance of the wild‐type strain differed substantially from those of the other strains, statistical significance was assessed using Welch's ANOVA followed by the Games–Howell post hoc test, which does not assume homogeneity of variance among groups. (D) Relative transcript abundance of rRNA in the strains shown in (A), measured by reverse transcription‐quantitative PCR. Different letters indicate statistically significant differences in mean values (p < 0.001, Tukey's HSD test) based on three independent biological replicates, each analyzed with three technical replicates (n = 3). (E) Rice leaves inoculated with the indicated strains, photographed at 7 dpi. (F) Quantification of lesion diameter from the leaves shown in (E). Because lesion diameter values were comparable in variance across strains, statistical significance in (F) was assessed using one‐way ANOVA followed by Tukey's HSD test. In (B), (C), and (F), different letters above the bars indicate statistically significant differences among means (p < 0.05) based on three independent biological replicates (n = 3). All strains were grown on oatmeal agar (OMA) medium at 25°C.

Figure S11: Enrichment of histone methylation at the promoter regions of nuclear genes. (A) Enrichment of H3K9me2 at melanin synthesis‐related genes. (B) Enrichment of H3K9me3 at melanin synthesis‐related genes. (C) Enrichment of H3K36me3 at melanin synthesis‐related genes. (D) Enrichment of H3K9me2 at conidiogenesis‐related genes. (E) Enrichment of H3K9me3 at conidiogenesis‐related genes. (F) Enrichment of H3K36me3 at conidiogenesis‐related genes. (G) Enrichment of H3K9me2 at cell wall synthesis‐related genes. (H) Enrichment of H3K9me3 at cell wall synthesis‐related genes. (I) Enrichment of H3K36me3 at cell wall synthesis‐related genes. In (A)–(I), asterisks indicate statistically significant differences in mean values (*p < 0.05, **p < 0.01; Tukey's HSD test) based on three independent biological replicates, each analyzed with three technical replicates (n = 3).

Figure S12: Relative transcript abundance of genes involved in response to reactive oxygen species (ROS) in wild type strain (KJ201), MoJMJD2 deletion strain (∆Mojmjd2) and complemented strain (MoJMJD2c).

Figure S13: Enrichment of MoJMJD2T¹‐GFP in nuclear genes and rDNA loci. The nuclear genes include genes involved in conidiogenesis, melanin synthesis and cell wall synthesis. Asterisks indicate statistically significant differences in mean values (**p < 0.01; Tukey HSD test) based on three independent biological replicates, each analyzed with three technical replicates (n = 3).

Table S1: List of oligonucleotide sequences for detecting and sequencing MoJMJD2 alternative splicing isoforms.

Table S2: List of oligonucleotide sequences for real‐time PCR.

Table S3: List of oligonucleotide sequences for MoJMJD2 deletion strain.

Table S4: List of oligonucleotide sequences for construction of fluorescence‐fused genes.

Table S5: List of oligonucleotide sequences for chromatin immunoprecipitation quantitative PCR (ChIP‐qPCR).

Data Availability Statement

The data that support the findings of this study are included in this article and its Figures [Link], [Link] and Tables [Link], [Link].