FpKmt6 and H3K27me3 are not essential in F. proliferatum NRRL62905

Loss of Kmt6 appears lethal in Fusarium fujikuroi [23]. To determine whether this trait is unique for F. fujikuroi or is conserved also in other FFSC members, an attempt to remove the KMT6 ortholog was approached in the two closely related fusaria, Fusarium proliferatum NRRL62905 and Fusarium mangiferae MRC7560, from now on referred to as FpWT and FmWT, respectively. The putative KMT6 orthologs were identified by QuartetS [35]. Function- and phylogeny-based metrics based on large-scale comparison predicted FPRN2_01368 and FMAN_01372 as the true orthologs of the H3K27-specific methyltransferase FfKmt6 (FFUJ_00719) from F. fujikuroi [23]. Pairwise sequence alignment using ClustalW showed 99.04% and 98.95% identity of FPRN_01368 and FMAN_01372, respectively, with FfKmt6 on the protein level [36]. The typical domains i.e., EZH2-N, EZH2-MCSS, CXC (pre-SET), and SET domains are present in both species (Fig 1A) [37]. Deletion of FPRN2_01368 and FMAN_01372 was approached by homologous integration of a hygromycin resistance cassette into the respective wild-type strains, FpWT and FmWT. Several transformants were obtained that showed correct in situ integration of the resistance cassette for both transformation approaches (S1 and S2 Figs). While the wild-type gene was present and remained detectable even after several consecutive rounds of single spore isolation in the case of F. mangiferae (S1C Fig), several homokaryotic Δfpkmt6 mutants were obtained that showed correct in situ integration of the resistance cassette and absence of the native FpKMT6 wild-type gene (S2B Fig). Three independent mutant strains, Δfpkmt6_T1, T7, and T9, were selected for further analyses. Southern blot analyses verified the absence of multiple integrations (S2C Fig). All mutants had an identical phenotype. Hence, Δfpkmt6_T1 was arbitrarily chosen for complementation. For this, FpKMT6 driven by its native promoter and followed by the artificial glucanase terminator from Botrytis cinerea, BcTgluc, was targeted to the native locus (in situ, S3A Fig). Obtained mutants, Δfpkmt6/FpKMT6, were analyzed by diagnostic PCR (S3B Fig), and reconstitution of FpKMT6 expression was verified by real-time quantitative polymerase chain reaction (RT-qPCR, Fig 1B). Finally, to verify the involvement of FpKmt6 in H3K27 methylation, western blot analysis was performed using whole protein extracts of FpWT, Δfpkmt6, and Δfpkmt6/FpKMT6 strains [28]. While abundant in FpWT and Δfpkmt6/FpKMT6, H3K27me3 levels were not detectable in Δfpkmt6 (Figs 1C and S4). To analyze if H3K27me3 is replaced by H3K27ac in Δfpkmt6, overall H3K27ac levels were determined by western blot analysis using an H3K27ac-specific antibody. H3K27ac levels were increased about 2.5-fold in Δfpkmt6 as compared to FpWT, and largely restored to wild-type level in Δfpkmt6/FpKMT6 (Figs 1C and S4). Thus, FpKmt6 is the sole H3K27-specific methyltransferase in F. proliferatum NRRL62905, and loss of H3K27me3 is not lethal in this strain.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. FpKmt6 is the sole H3K27-specific methyltransferase in Fusarium proliferatum.

(A) Graphical representation of the domain structure of Neurospora crassa Set7 and orthologs in other fungal species including F. proliferatum NRRL62905. The conserved domains are indicated and InterPro accession numbers are given in the domain description (B) For FpKMT6 expression analysis, FpWT as well as two independent Δfpkmt6 and Δfpkmt6/FpKMT6 strains each were grown on solid CM for 3 days at 30°C. For the determination of transcript levels, RNA was extracted and transcribed into cDNA before RT-qPCR. The FpWT FpKMT6 expression was arbitrarily set as 1. Mean values and standard deviations are shown. n.d., not detected; RE, relative expression. (C) Western blot analysis of FpWT as well as two independent Δfpkmt6 and Δfpkmt6/FpKMT6 strains each using H3K27me3- and H3K27ac-specific antibodies. For referencing an H3pan-specific antibody was used. Roughly, 25 μg of total protein extracts were separated on a SDS gel before western blotting. For quantification, densitometric analysis was performed and the respective wild-type strain was arbitrarily set as 1. n.d., not detected via the Image J analysis tool.

https://doi.org/10.1371/journal.pgen.1011075.g001

FpKmt6 is required for fungal growth and asexual development

The availability of a homokaryotic Δfpkmt6 strain allowed for a detailed functional characterization of FpKmt6 in a member of the FFSC. The impact of FpKmt6 on fungal growth was assessed on complete (vegetable V8-juice, V8, and complete medium, CM), and minimal (synthetic ICI) media supplemented with 6 mM glutamine as the sole nitrogen source. For this, plates were point-inoculated with agar plugs of FpWT, Δfpkmt6, or Δfpkmt6/FpKMT6 and incubated at 30°C in the dark. Fungal growth was determined five days post-inoculation (dpi). Radial hyphal growth was severely impacted in Δfpkmt6 compared to FpWT and restored to wild-type level in the complemented strain Δfpkmt6/FpKMT6 (Fig 2A) on all tested media conditions. To analyze the effect of H3K27me3 depletion on asexual development, conidia formation of FpWT and Δfpkmt6 was assessed on V8 agar plates, and conidia were quantified seven dpi. The overall number of conidia is significantly reduced in Δfpkmt6 as compared to FpWT but when compared to the radial hyphal growth conidia count in FpWT and Δfpkmt6 was similar (S5 Fig). Noteworthy, the growth behavior of Δfpkmt6 differed from FpWT significantly thereby impeding a proper assessment of the asexual development. Therefore, transcription of conidiation-related genes was quantified by RT-qPCR. In A. nidulans, conidiation is regulated via the so-called ‘fluffy’ proteins FlbB-FlbE, which are required for efficient expression of the central regulator of asexual development BrlA. BrlA in turn activates abaA, which further activates wetA. Both abaA and wetA are required for conidia formation and maturation, respectively, in this fungus [38]. While BrlA appears to be absent from the genus Fusarium [25, 39], orthologs of flbB-flbE (FLB2-FLB5), as well as abaA (ABA1) and wetA (WET1) have been identified and functionally characterized in the closely related F. fujikuroi [25]. Loss of FLB3, FLB4, and ABA1 completely abolished conidia formation in this fungus, while the amounts of formed conidia were wild type-like in Δflb2, Δflb5, and Δwet1 [25]. To analyze whether expression of the conidiation-related genes is affected by deletion of FpKMT6 in F. proliferatum, the respective orthologs were determined by QuartetS [35], using the entries for F. fujikuroi [25]. The FpWT, Δfpkmt6, and Δfpkmt6/FpKMT6 were grown for 4 days under conidia-inducing conditions, and expression was determined by RT-qPCR for FpABA1, FpFLB3, FpFLB4, and FpWET1. Expression of FpFLB3 (FPRN2_04427), FpABA1 (FPRN2_21361), and also FpWET1 (FPRN2_07602) was reduced to 28%, 23% and 30%, respectively, and largely restored to wild-type level (1.9, 2.6 and 1.4) in Δfpkmt6/FpKMT6, while FpFLB4 (FPRN2_08233) expression remained unaffected (Fig 2B). Next to this, deletion of FpKMT6 did not affect conidia maturation as all harvested conidia had a wild type-like appearance under the microscope (Fig 2C). The results are in line with the literature. Declined or even aberrant growth as well as impaired sexual and asexual reproduction was observed also for other fungal mutants lacking a functional Kmt6, including Ustilaginoidea virens [40], Podospora anserina [41], and also in other fusaria [20, 23]. Only a slight growth reduction and conflicting results with regard to asexual development were reported for M. oryzae [34, 42]. Interestingly, for E. festucae and Zymoseptoria tritici, only subtle changes in growth and development were observed, implying that H3K27me3 governs different roles in other fungi [16, 43].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Impact of FpKMT6 deletion on radial hyphal growth and asexual development.

(A) Radial hyphal growth of F. proliferatum wild type (FpWT), Δfpkmt6, and Δfpkmt6/FpKMT6 was assessed on complete media (FCM, V8) and minimal media (ICI supplemented with 6 mM Gln). For this, strains were grown for 5 days at 30°C in the dark. Experiments were performed in biological triplicates. (B) The asexual development of the different strains was assessed under dark conditions. For this, FpWT, Δfpkmt6, and Δfpkmt6/FpKMT6 were grown on a solid V8 medium for seven days in the dark. Experiments were performed in triplicates. Conidia production of FpWT was arbitrarily set as 1. (C) Microscopic assessment of asexual conidia in FpWT, Δfpkmt6, and Δfpkmt6/FpKMT6. The bar in the right corner represents 10 μm in size.

https://doi.org/10.1371/journal.pgen.1011075.g002

Re-sequencing of FpWT results in near-complete chromosome map

The previous assembly of FpWT was done by whole-genome shotgun sequencing using 454 pyrosequencing and Illumina HiSeq 2000 technologies and resulted in overall 155 scaffolds (43.2 Mb, 98-fold coverage) with an annotated 15,254 gene loci [8]. To generate more complete chromosome sequences and thus to allow for a correct allocation of histone marks, especially concerning larger genomic regions, the MinION sequencing platform from Oxford Nanopore Technologies was applied [44, 45]. This yielded a near-complete chromosome sequence (43.63 Mb, 110-fold coverage) where the previous 155 scaffolds were assembled into twelve scaffolds, representing the twelve chromosomes present in FpWT as determined by pulsed-field gel electrophoresis previously [8], with an annotated 15,362 gene loci.

To determine the genome-wide distribution of H3K27me3, chromatin immunoprecipitation (ChIP) coupled to NGS was performed. For this, the wild-type strain FpWT was grown for 3 days in a synthetic minimal medium (liquid ICI supplemented with 6 mM glutamine as sole nitrogen source), crosslinked, and subsequently used for ChIP using an anti-H3K27me3 antibody. Reads obtained from NGS were mapped on the new assembly. H3K27me3 is highly abundant in FpWT covering nearly 45% of all annotated genes (6,805 of 15,362). Similar to other Fusarium spp. [20, 23], P. anserina [41], U. virens [46], and M. oryzae [34, 42] but different to Z. tritici [43] a high percentage of BGCs are allocated with H3K27me3 in F. proliferatum (Fig 3). No H3K27me3 was detected in strains deleted for FpKMT6 as determined by ChIP-seq, thereby confirming the specificity of the used antibody and the absence of other H3K27me3-specific methyltransferases in this fungus (Fig 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Genomic localization of H3K27me3 and SM key enzyme-encoding genes in F. proliferatum.

The re-sequenced and newly assembled twelve chromosomes are shown in grey i.e., Chr1 –Chr12, and centromeres are shown in black; SM key enzyme-encoding genes are indicated by bars according to the following color code: polyketide synthase (PKS), blue; non-ribosomal peptide synthetase (NRPS), orange; (sesqui-/di-/sester-/tri-/tetra-) terpene cyclase (STC/DTC/PTTS/TrTC/TeTC), light green; dimethylallyl tryptophan synthase (DMATS), red; putative fungal RIPPs, light brown; cyclodipeptide synthase (CDPS), dark brown. Pseudogenes are highlighted with a ψ. Genome-wide distribution of H3K27me3, present in the wild-type strain (FpWT) and absent from the KMT6-deleted strain (Δfpkmt6), are depicted in red bars in the two lanes beneath each respective chromosome.

https://doi.org/10.1371/journal.pgen.1011075.g003

F. proliferatum is known to produce a wide range of SMs including the well-characterized red pigment bikaverin, or the mycotoxins fusaric acid and fumonisins [8]. Given the new gene models, the presence of known but also additional BGCs was re-evaluated using antiSMASH 6.0 [47]. The new genome assembly of FpWT revealed 59 putative key enzyme-encoding genes possibly involved in SM biosynthesis: 22 NRPSs, 17 PKSs and 4 PKS-NRPSs, 11 terpene synthases (TSs), 3 dimethylallyltryptophan synthases (DMATSs), and 1 type III PKS (Table 1). Overall, 42 of the 59 putative BGCs, 5 of the 6 fungal RiPP-like genes and both putative CDPSs fall within facultative heterochromatin (Fig 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. SM key enzyme-encoding genes in FpWT as determined by antiSMASH 6.0.

Genes upregulated in Δfpkmt6 are highlighted in green, and genes downregulated in red. Genes resulting in non-functional proteins are marked by an Ψ.

https://doi.org/10.1371/journal.pgen.1011075.t001

Thus, 9 additional putative SM key enzyme-encoding genes have been identified. These include one type III PKS-encoding gene, FPRN2_09028, six NRPS-encoding genes, designated NRPS53 through NRPS58 (FPRN2_02333, FPRN2_06362, FPRN2_21757, FPRN2_21914, FPRN2_32910, FPRN2_33084), as well as two putative TS-encoding genes, FPRN2_05485 (PTTS1) and FPRN2_26417 (STC10). FPRN2_05485 harbors a C-terminal prenyltransferase (PT) and an N-terminal terpene synthase (TS) domain, hence designated PTTS1 according to [48]. While the products for most of them are still cryptic, PTTS1 has recently been connected with fusaproliferin biosynthesis in F. proliferatum, designated as FpFUP1 [49]. Next to these classical SM key enzyme-encoding genes, FpWT harbors six fungal RiPP-like genes putatively encoding for ribosomal synthesized and post-translationally modified peptides, and two genes encoding putative cyclodipeptide synthases i.e., FPRN2_32462 and FPRN2_15133, designated CDPS1 and CDPS2, respectively, as predicted by antiSMASH 6.0 [47]. CDPSs are involved in the synthesis of cyclodipeptides, also known as diketopiperazines, using two activated amino acid moieties as aminoacyl-tRNAs as substrates [50, 51]. Most CDPSs are promiscuous enzymes, synthesizing one main cyclodipeptide and one or several minor cyclodipeptides. No CDPS-derived product has been identified in Fusarium yet. The genomic distribution of the putative BGCs along the twelve chromosomes is depicted in Fig 3.

H327me3 substitutes for the loss of H3K9me3 in F. proliferatum

Next, we asked whether there is a connection between facultative heterochromatin (H3K27me3) and constitutive heterochromatin (H3K9me3) since both marks seem to take over distinct functions with regard to gene silencing. This includes a possible re-distribution of H3K9me3 in Δfpkmt6 and/or H3K27me3 in strains deleted for the H3K9me3-specific histone methyltransferase-encoding gene FpKMT1, Δfpkmt1, as shown in other fungi e.g., N. crassa and Z. tritici [43, 52]. To analyze whether this is also true for F. proliferatum, we recorded H3K9me3 and H3K27me3 distributions by ChIP-seq in Δfpkmt6 and Δfpkmt1, respectively, also in this species. The FpKMT1 ortholog was identified by QuartetS [35]. Function- and phylogeny-based metrics based on large-scale comparison predicted FPRN2_27438 as the true ortholog of the H3K9me3 histone methyltransferase FmKmt1 (FMAN_07768) from F. mangiferae [28]. Pairwise sequence alignment using EMBOSSNeedle showed 97.7% identity of FPRN2_27438 (from now on referred to as FpKmt1), with FMAN_07768 on protein level [36]. Deletion of FPRN2_27438 was approached by homologous integration of a hygromycin resistance cassette into FpWT. Several transformants were obtained that showed correct in situ integration of the resistance cassette and the absence of the FpKMT1 wild-type gene (S6A Fig). Western blot analysis showed reduced H3K9me3 levels in three individual-Δfpkmt1 strains to about 30% and 40% of FpWT depending on the antibody used (S6B and S7 Figs). This is similar to the results obtained for F. mangiferae [28] and is likely due to the promiscuity of the used antibodies. All obtained mutants showed an identical phenotype on plates (S6C Fig). Hence, Δfpkmt1_T4 was arbitrarily chosen for complementation. For this, FpKMT1 driven by its native promoter and followed by the BcTgluc terminator sequence was targeted to the native locus (in situ, S8A Fig). Obtained mutants, Δfpkmt1/FpKMT1 were verified by diagnostic PCR (S8B Fig) as well as the absence of growth on hygromycin, and reconstitution of FpKMT1 expression was verified by RT-qPCR (S8C Fig). Thus, similar to F. mangiferae but distinct from F. fujikuroi IMI58289 [23, 28], deletion of KMT1 is not lethal in F. proliferatum. The Δfpkmt1 did not deviate significantly in hyphal growth on solid media or conidiation (S9A and S9B Fig), suggesting that H3K9me3 is largely dispensable for fungal development in this Fusarium species. Similar results were observed for F. mangiferae and F. verticillioides [22, 28]. To determine the genome-wide distribution of H3K9me3, ChIP-seq was performed in FpWT using the previous growth conditions (liquid ICI medium with 6 mM glutamine). In contrast to H3K27me3, only a few H3K9me3-rich islands are present in FpWT. All peaks reside in intergenic regions. This is similar to previous findings for the closely related F. fujikuroi [7]. Both heterochromatic marks i.e., H3K27me3 and H3K9me3, appeared mutually exclusive along the F. proliferatum chromosomes in FpWT (S10 Fig), which is in agreement with N. crassa [53] but stands in contrast to other fungi i.e., P. anserina, Leptosphaeria maculans and Z. tritici, where H3K9me3 and H3K27me3 were largely but not overall devoid from each other [41, 54, 55]. The commercially available H3K9me3-specific antibodies proved promiscuous in western blot analysis (S8B Fig). Thus, to cross-validate the obtained data, ChIP-seq was also performed with a H3K9me3-specific antibody in Δfpkmt1. H3K9me3 levels detected in FpWT were nearly abolished in Δfpkmt1 samples and remaining low signals in the kmt1-deleted strain are likely due to mapping issues as H3K9me3 frequently resides in repeated regions, thereby confirming the specificity of the used H3K9me3 antibody in the ChIP-seq experiments (S10 Fig).

To determine whether H3K9me3 is re-positioned upon loss of H3K27me3, ChIP-seq was also performed in the Δfpkmt6 strain using a H3K9me3-specific antibody. The first most noticeable result was that H3K9me3 detection was better (signal-to-noise ratio) in Δfpkmt6 strain, possibly attributable to a higher shearing efficiency of the less dense mycelium. However, this was only detectable for H3K9me3, while other used antibodies had similar efficiencies between FpWT and Δfpkmt6. While H3K9me3 patterns in Δfpkmt6 look wild type-like on most of the chromosomes, there are indeed a few peaks gained in previously H3K27me3-rich regions (Figs 4A, S11 and S12). These peaks appear exclusively in intergenic regions and are specific to previous H3K27me3-rich regions in FpWT. The function of these emerging H3K9me3 peaks remains elusive at this point and needs further attention in the future. This, however, stands in marked contrast to P. anserina and Z. tritici. For P. anserina absence of H3K27me3 led to a drastic reduction of H3K9me3 levels [41, 43, 52].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Crosstalk between H3K27me3 and H3K9me3 in F. proliferatum.

(A) Few additional H3K9me3 peaks are gained in previously H3K27me3 intergenic regions in a strain deleted for FpKMT6 (Δfpkmt6) compared to the wild-type strain (FpWT) as determined by ChIP-sequencing. Here, exemplarily shown for chromosomes 3, 4, 11 and 12. (B) Previously, H3K9me3-marked genomic regions become H3K27me3 upon deletion of FpKMT1 (Δfpkmt1) compared to FpWT. Here, exemplarily shown for chromosomes 5, 6 and 8. Selected chromosomes are shown in grey and centromeres are shown in black; SM key enzyme-encoding genes are indicated by bars according to the following color code: polyketide synthase (PKS), blue; non-ribosomal peptide synthetase (NRPS), orange; (sesqui-/di-/sester-/tri-/tetra-) terpene cyclase (STC/DTC/PTTS/TrTC/TeTC), light green; dimethylallyl tryptophan synthase (DMATS), red; putative fungal RIPPs, light brown; cyclodipeptide synthase (CDPS), dark brown; Pseudogenes are highlighted, Ψ; Genome-wide distribution of H3K9me3 and H3K27me3 are shown beneath each chromosome in brown and red, respectively.

https://doi.org/10.1371/journal.pgen.1011075.g004

Next, we analyzed whether H3K27me3 patterns are affected by deletion of FpKMT1. For this, Δfpkmt1 was also included in the experiments, and ChIP-seq was performed using a H3K27me3-specific antibody. H3K27me3 patterns in Δfpkmt1 largely resemble H3K27me3 of the wild-type strain with one marked exception: previous H3K9me3 peaks become decorated with H3K27me3 in Δfpkmt1 (S13 Fig). This is most prominent at chromosomes 5, 6, and 8, that harbor dominant H3K9me3 peaks in FpWT (Figs 4B and S12). This is also true for the fungal plant-pathogen Z. tritici, P. anserina, and N. crassa, where H3K27me3 relocation in former H3K9me3 marked regions was observed [41, 43, 52].

Thus, for F. proliferatum it appears that H3K27me3 can substitute for the loss of H3K9me3 in Δfpkmt1. It is tempting to speculate that this crosstalk is the reason for the subtle phenotype of Δfpkmt1, a phenotype that stands in marked contrast to other fungi. This idea is supported by the fact that attempts to generate homokaryotic ΔΔfpkmt1/fpkmt6 strains failed in our hands, probably because those strains are not viable.

Loss of H3K27me3 results in elevated gene expression of former silent genes accompanied by an increase in H3K27ac

Loss of H3K27me3 led to a strong induction of prior silenced genes, particularly of genes involved in fungal secondary metabolism in Fusarium [20, 23, 24]. To determine if this is also the case for F. proliferatum, whole-transcriptome analysis was performed using RNA-sequencing (RNA-seq). For this, FpWT and Δfpkmt6 were grown in a synthetic ICI medium with 6 mM glutamine as the sole nitrogen source, identical to the previously performed ChIP experiments. Indeed, lack of FpKmt6 resulted in the upregulation of a large set of genes specifically localized to previous H3K27me3-marked regions (Fig 5). Overall, 4,646 and 3,164 genes are down- and upregulated, respectively, upon deletion FpKMT6 with a confidence interval of log₂ ≥ ±1, p-value < 0.01 (S1 Table). We used Gene Ontology (GO) based gene set enrichment analysis on subsets of genes either by association with H3K27me3 alone or in conjunction with significant transcriptional upregulation in Δfpkmt6 to identify functionalities of genes silenced by H3K27me3. Solely H3K27me3 labeling of genes identified as expected the biological processes (BP): secondary metabolic process (GO:0019748 111 out of 132 genes, p = 2.48E-27) and related ones including antibiotic biosynthetic process (GO:0017000) and mycotoxin metabolic process (GO:0043385); see complete S2 Table.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Loss of FpKmt6 results in the de-repression of previous H3K27me3-associated genes accompanied by H3K27ac.

Selected chromosomes are shown in grey and centromeres are shown in black; SM key enzyme-encoding genes are indicated by bars according to the following color code: polyketide synthase (PKS), blue; non-ribosomal peptide synthetase (NRPS), orange; (sesqui-/di-/sester-/tri-/tetra-) terpene cyclase (STC/DTC/PTTS/TrTC/TeTC), light green; dimethylallyl tryptophan synthase (DMATS), red; putative fungal RIPPs, light brown; cyclodipeptide synthase (CDPS), dark brown; pseudogenes are marked by a Ψ. Beneath each chromosome in descending order: genome-wide distribution of H3K27me3 (red) and H3K27ac (green) in the wild-type strain (FpWT); H3K27ac in Δfpkmt6, appearance of new H3K27ac peaks in Δfpkmt6 as compared to FpWT (ΔH3K27ac); differential expression in Δfpkmt6 compared to FpWT (ΔE); genomic region around PKS17, PKS18 and PKS8 on chromosome 11, not de-repressed in Δfpkmt6, is highlighted by an asterisk.

https://doi.org/10.1371/journal.pgen.1011075.g005

Previously we have shown that H3K27me3 is involved in silencing of genes involved in beauvericin biosynthesis in F. fujikuroi [24]. Knock-down of FfKMT6 resulted in reduced H3K27me3 and de-repression of the respective genes, FfBEA1 –FfBEA4, accompanied by elevated H3K27ac as determined by gene-specific ChIP. To evaluate whether this is true also for F. proliferatum and whether re-decoration of previous H3K27me3 regions with H3K27ac is a general scheme, we performed ChIP-seq using an H3K27ac-specific antibody. Generally, H3K27ac is found in the 5’regions of genes throughout the genome and appears mutually exclusive with H3K27me3 in FpWT. Overall, 42% of the genes are enriched for this histone mark (6,385 of 15,362). Upon removal of H3K27me3, previous H3K27me3-rich regions become replaced by H3K27ac concomitant with a de-repression of the underlying genes (Fig 5). This observation was most notable on chromosomes 10 and 11, which are almost completely covered with H3K27me3 in FpWT but hyperacetylated in Δfpkmt6 accompanied by a de-repression of a large set of genes located on both chromosomes (Fig 5). H3K27ac-rich regions in the FpWT largely stayed acetylated also in Δfpkmt6 as determined by ChIP-seq, going in line with an elevated overall H3K27ac level in Δfpkmt6 compared to FpWT as determined by western blot (Fig 1C). These results align with previous reported results in M. oryzae. Similar to F. proliferatum, deletion of MoKMT6 revealed that a large fraction of genes previously decorated with H3K27me3 were now covered with H3K27ac, which was accompanied by an increased gene expression (fold-change) [34].

H3K36me3 patterns at deregulated BGCs appear distinct between FpWT and Δfpkmt6

In N. crassa it was previously shown that proper H3K27me2/3 states depend on a catalytically active version of the lysine methyltransferase NcAsh1. NcAsh1 facilitates H3K36me3 at otherwise silent or poorly transcribed genomic regions significantly co-localizing with H3K27me2/3 [33]. To determine whether H3K36me3 depends on H3K27me3 also in F. proliferatum, we recorded the genome-wide association of H3K36me3 and H3K27me3 in FpWT and Δfpkmt6 by ChIP-seq using the same conditions as previously described. While all genes were associated with H3K36me3 in FpWT, some genes appear to have lost this association in the Δfpkmt6 strain (S14 Fig), the same is true for M. oryzae [34]. This observation prompted us to have a closer look at the H3K36me3 patterns in F. proliferatum. In N. crassa, Ash1- and Set2-catalyzed genes exhibit distinct methylation patterns: while Ash1-catalyzed H3K36me2 was prominent across the promoter and body of the genes, Set2-catalyzed H3K36me2 was depleted from transcriptional start sites (TSS) but enriched over gene bodies [33], which is also true for F. fujikuroi [26]. Indeed, H3K36me3 patterns at deregulated BGCs in Δfpkmt6 deviate from FpWT: BGCs that are silent in FpWT are enriched with H3K36me3 over the TSS and gene bodies, thereby resembling the Ash1-associated methylation pattern, while the TSS of BGCs that are de-repressed in Δfpkmt6 resemble more the Set2-associated methylation pattern (Fig 6A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. H3K36me3 pattern at de-regulated genes appear distinct between FpWT and Δfpkmt6.

(A) Chromatin marks for selected genomic regions of the F. proliferatum wild-type strain NRRL 62905 (FpWT) and a strain deleted for FpKMT6 (Δfpkmt6). In descending order are H3K27me3 (red), H3K36me3 (orange), input (grey), and transcription as determined by RNA-seq (black) for FpWT and the same for Δfpkmt6. Below this, the selected BGC is indicated as follows: the PKS- and NRPS-encoding genes are highlighted in blue and yellow, respectively. Other cluster genes are shown in black, and genes not belonging to the FUB cluster are in grey. (B) Grey bar plots represent the differential expression of genes with at least an average transcript level of 0 RPKM. Positive bars represent genes upregulated in Δfpkmt6 negative bars represent downregulated genes in Δfpkmt6, indicated by arrow up and down, respectively. Orange bars in the upper lane show genes with H3K36me3-enriched promoters in the wild-type strain (FpWT). Orange bars in the lower lane depict H3K36me3-enriched promoters in Δfpkmt6. (C) H3K36me3 occupancy with respect to the ATG and STOP codons of transcribed genes (mean transcription level > 0 RPKM) and separated into H3K27me3 (facultative heterochromatin, 4420 genes) and euchromatin (not marked with H3K27me3, 7438 genes). The top row is the mean occupancy of all transcribed genes, the middle row shows genes that are significantly upregulated in Δfpkmt6 (3016 facultative heterochromatic genes, 445 euchromatic genes), and the third row depicts genes that are transcriptionally silenced (2381 facultative heterochromatic genes, 1114 euchromatic genes) and show high H3K36me3 levels in their promoter regions. Green and red lines depict patterns of FpWT and Δfpkmt6 (in replicates), respectively.

https://doi.org/10.1371/journal.pgen.1011075.g006

Next, when comparing the differential transcription between FpWT and Δfpkmt6 with the differential H3K36me3 occupancy in the promoter regions of genes (minimal [CPM] coverage in 1 kb upstream of the ATG), it became apparent that the majority of genes showing high H3K36me3 in their promoter regions were associated with facultative heterochromatin (S15A Fig). The difference in H3K36me3 between Δfpkmt6 and FpWT is negatively correlated with the differential transcription between these two strains i.e., high transcription is correlated with reduced H3K36me3 promoter occupancy in Δfpkmt6 as compared to FpWT (S15B Fig). Euchromatic genes (Fig 6B, on the right) also show a different transcriptional response in respect to Δfpkmt6 with a majority of genes being downregulated with respect to FpWT. This presumably secondary effect correlates with increased H3K36me3 promoter occupancy of euchromatic genes in Δfpkmt6. However, most of the changes in H3K36me3 occurs in previous H3K27me3-rich regions in Δfpkmt6 (Fig 6B). Overall, H3K36me3 patterns in euchromatic genes appear similar in both, FpWT and Δfpkmt6, with increased occupancy towards the terminator region to prevent non-sense transcription (Fig 6C, top right), and this pattern is mimicked in genes assigned to H3K27me3-rich regions in Δfpkmt6 whereas H3K36me3 promoter occupancy is elevated in FpWT. (Fig 6C, top left). This is even more pronounced in differentially expressed genes (Fig 6C, middle row right). In genes showing low or no transcription (Fig 6C, bottom row) H3K36me3 is high in promoter regions which is true for both, facultative heterochromatin and euchromatin. Thus, high H3K36me3 promoter occupancy levels co-occur in facultative heterochromatic regions suggesting that this histone mark is involved in preventing transcription of these genes. The roles of the two H3K36-specific histone methyltransferases, FpSet2 and FpAsh1, are yet unclear, and the general mechanism of targeted promoter H3K36me3 requires further investigation. However, it is intriguing to speculate that, similar to N. crassa [33], Ash1-mediated H3K36me3 marks TSS and, thus, contributes to SM-gene silencing in FpWT, while FpSet2-catalyzed H3K36me3, generally enriched over the gene bodies but absent from TSS in N. crassa [33], marks actively transcribed genes at previous H3K27me3 sites in Δfpkmt6.

Loss of H3K27me3 results in increased BGC expression in F. proliferatum

Of the 59 predicted key enzyme-encoding genes 16 and 29 were down- and upregulated, respectively, in Δfpkmt6. Next to this, 5 of the 6 fungal RiPP-like genes as well as CDPS1 and CDPS2 were also significantly upregulated in Δfpkmt6 within the confidence interval of log₂ ≥ ±1, p-value < 0.01 (Table 1).

All SM key enzyme-encoding genes identified as significantly upregulated in Δfpkmt6 are enriched with H3K27me3 in FpWT, while this is the case for only 4 of the 16 downregulated SM key enzyme-encoding genes (Table 1). Expression of 3 of them, i.e., DMATS1 (FPRN2_13086) involved in r-N-DMAT biosynthesis [56] and NRPS36 (FPRN2_15183) and NRPS54 (FPRN2_06362) was generally very low in both FpWT and Δfpkmt6. It is noteworthy, that the removal of H3K27me3 and/or its replacement with H3K27ac in Δfpkmt6 is not sufficient to induce transcription in all gene sets, and that not all previously H3K27-methylated genes become acetylated, which is in accordance with results shown in M. oryzae [34]. For 19 key enzyme-encoding genes expression was not significantly changed upon deletion of FpKMT6 although all of them are allocated with H3K27me3 in FpWT. For most of these, expression was generally very low suggesting that additional factors are required to induce their transcription. This is most prominent in the case of the two PKS-encoding genes, PKS17 (FPRN2_14992) and PKS18 (FPRN2_15000), and the remnant of PKS8 (FPRN_15017) [32], all located in close proximity to one another on chromosome 11. None of those genes is de-repressed upon relief of H3K27me3 in Δfpkmt6 (Fig 5, highlighted with an asterisk). Thus, the removal of H3K27me3 allowed for the expression of 29 SM key enzyme-encoding genes (49%), five out of six fungal RiPP-like genes, and the two putative CDPSs, named CPDS1 and CPDS2. All of them are localized in H3K27me3-rich regions in FpWT (Fig 3 and Table 1).

To cross-verify the results obtained by RNA-sequencing, FpWT, Δfpkmt6, and the complemented strain (Δfpkmt6/FpKMT6) were grown under identical conditions for subsequent SM analysis. After seven days of growth at 30°C and 180 rpm in the dark, aliquots were taken and analyzed for the most prominent fusarial SMs by targeted metabolomics using LC-MS/MS. To account for the aberrant growth of Δfpkmt6, single cultures were inoculated with twice the volume compared to FpWT and (Δfpkmt6/FpKMT6), resulting in a similar final biomass of all strains (S16 Fig). Going in line with the expression data (S1 Table), biosynthesis of the red pigment bikaverin was almost abolished in Δfpkmt6 and restored nearly to wild-type level in the complemented strain Δfpkmt6/FpKMT6, while fusaric acid was increased about 140-fold as compared to FpWT and Δfpkmt6/FpKMT6 (S17 Fig). Fusaric acid is normally induced by high nitrogen [57, 58]. Noteworthy, fusaric acid levels (most upregulated genes in the transcriptome) reached amounts in 6 mM glutamine (normally repressing) in Δfpkmt6 that were comparable with the amounts produced in FpWT under normally inducing high nitrogen conditions (60 mM glutamine), suggesting that H3K27me3 is responsible for fusaric acid gene-silencing under non-inducing conditions. The distribution of H3K27me3 under different culture conditions was already studied in the plant pathogen Verticillium dahliae. Here, ChIP-seq analyses under three different growth conditions revealed overall globally stable H3K27me3 patterns. Expression analysis under these different conditions were correlated with the respective H3K27me3 patterns and results indicated that genes, which are stably covered can be differentially expressed regardless of the chromatin status [59]. If fungi can remove, re-locate H3K27me3 or unwind the tightly packed chromatin structure and, if so, which mechanisms are involved in this phenomenon is currently unknown. Further work is required to prove this assumption unequivocally. Next, two SMs not detectable in FpWT and Δfpkmt6/FpKMT6 cultures were found to be produced in Δfpkmt6 i.e., siccanol (the deacetylated form of fusaproliferin) and beauvericin (S17 Fig). The latter was also increased in the RNA-seq experiments though not significantly and hence is not listed in Table 1. This discrepancy is likely attributable to the different time points of measurement and the crippled growth of Δfpkmt6. The BGC involved in fusaproliferin biosynthesis has recently been elucidated [49]. Overall, five genes are involved in fusaproliferin biosynthesis encoding two cytochrome P450 monooxygenases (FPRN2_05484, FpFUP2 and FPRN2_05488, FpFUP3), the sesterterpenoid synthase composed of a C-terminal prenyltransferase and N-terminal terpene synthase (PTTS) (FPRN2_05485, FpFUP1), a FAD-linked oxidase (FPRN2_05486, FpFUP4), and an acyltransferase (FPRN_05487, FpFUP5). Two putative transcription factor-encoding genes are located directly downstream of the BGC, but their involvement in fusaproliferin biosynthesis is still unknown. All five FUP genes are silenced by H3K27me3 in FpWT and significantly de-repressed in Δfpkmt6 accompanied by H3K27ac under the chosen conditions (S1 Table and S18 Fig). It remains unknown, why only siccanol but not fusaproliferin was detected in the Δfpkmt6 samples. One possible explanation is that the relatively low levels that might potentially be produced are below the limit of detection (LOD) for fusaproliferin, which exhibits a comparatively poor sensitivity in ESI-MS/MS. Next to this, four genes are involved in beauvericin biosynthesis in F. fujikuroi, encoding a NRPS (FPRN2_12971, FpBEA1), the ketoisovalerate reductase KivR (FPRN2_12972, FpBEA2), an ABC transporter (FPRN2_32324, FpBEA3) and a Zn2Cys6 transcription factor (FPRN2_32320, FpBEA4) that was shown to negatively impact beauvericin biosynthesis [24]. In line with this, only FpBEA1 –FpBEA3 are upregulated in F. proliferatum, while expression of FpBEA4 is reduced in Δfpkmt6 compared to FpWT (S1 Table). It is possible that up- and downregulation of FfBEA1-FfBEA3 in Δffbea4 and OE::FfBEA4, respectively, in F. fujikuroi [24], is due to a yet unknown secondary effect. Nonetheless, expression patterns of the FpBEA1-FpBEA4 were in agreement with the production levels (S17 Fig).

F. proliferatum is well-known to produce fumonisins [60]. Yet, fumonisins were not detected in FpWT samples although grown under low nitrogen conditions that were shown to induce fumonisin biosynthesis in the closely related F. fujikuroi [61]. In line with these results, the fumonisin BGC is not expressed in FpWT, and low expression levels are likely due to H3K27me3-mediated gene silencing as FUM genes are located in a H3K27me3-rich region (Fig 3). As a consequence, removal of FpKMT6 results in an upregulation of FpFUM1 (FPRN2_42368) the key enzyme-encoding gene involved in fumonisin biosynthesis under these conditions in Δfpkmt6. Yet, next to FpFUM1 only 5 of the 15 additional biosynthetic genes, i.e., FpFUM21 (FPRN2_32368), FpFUM6 (FPRN2_32367), FpFUM7 (FPRN2_32366), FpFUM8 (FPRN2_32365) and FpFUM15 (FPRN2_13016) are significantly upregulated in Δfpkmt6 (log₂ ≥ ±1, p-value < 0.01) thereby possibly explaining the lack of the final product also in Δfpkmt6 samples (Tables 1 and S1).

H3K27me3 is decisive for gibberellin biosynthesis in F. proliferatum NRRL62905

Members of the FFSC have the genetic potential to produce three classes of phytohormones, i.e., auxin, cytokinins, and gibberellins. Both auxin and cytokinins are produced in liquid cultures in FpWT, while gibberellins are not detectable [8]. Production of gibberellins (GA) is a hallmark of the rice pathogen F. fujikuroi [7]. Though the seven biosynthetic genes involved in the production of these phytohormones appear conserved in the FFSC (including F. proliferatum NRRL62905), the production of GAs is mainly restricted to F. fujikuroi with few exceptions. F. proliferatum ET1 isolated as an endophyte from orchids produces ent-kaurene and GAs in axenic culture though only in small amounts [8, 62]. The reasons for that are not well understood. Pseudogenization is likely responsible for the failure to detect GAs in the F. proliferatum strains D02945 and D00502 despite of expression of the respective BGC [63], while the genes are silent in other FFSC members in axenic cultures including the here used F. proliferatum wild-type strain, FpWT ([8]; this study). Intriguingly, the GA biosynthetic genes were among the most highly upregulated SM genes in Δfpkmt6 (Tables 1 and S1). In line with the transcript data, GA biosynthetic genes are enriched with H3K27me3 in FpWT, and H3K27 becomes acetylated upon deletion of FpKMT6 (Fig 7A). To evaluate transcript data and verify whether deletion of FpKMT6 indeed results in detectable GA amounts, FpWT, Δfpkmt6 and Δfpkmt6/FpKMT6 were grown under identical, F. fujikuroi GA-inducing, conditions as for ChIP and RNA-seq for subsequent chemical analysis. In agreement with previous data [8] and transcript data, no GAs were detectable in FpWT and Δfpkmt6/FpKMT6 samples, while GAs were very well detected in the Δfpkmt6 strain (Fig 7B). The most prominent bioactive GAs are GA1, GA3, GA4 and GA7, with GA3 being the main product in F. fujikuroi GA biosynthesis [64]. However, GA1 but not GA3 was the main product in Δfpkmt6 (Fig 7B). GA biosynthesis is mediated by a seven-gene cluster, encoding a pathway-specific geranylgeranyl diphosphate synthase (FpGgs2, FPRN2_08493), the bifunctional ent-copalyl diphosphate synthase/ent-kaurene diphosphate synthase (FpCps/Ks, FPRN2_08492), four cytochrome P450 monooxygenases (FpP450-1 to FpP450-4, FPRN2_31371, FPRN2_08494, FPRN2_31367 and FPRN2_31372), and a desaturase (FpDes, FPRN2_08497). Starting from acetyl-CoA, GAs are synthesized via the mevalonate pathway resulting after several biosynthetic steps in the first biologically active C19 GA, GA4. Desaturation of GA4 at C1,2 by the GA4 1,2 desaturase (Des) yields GA7. In the final step, both GA4 and GA7 are converted by the C13 oxidase P450-3 to GA1 and GA3, respectively, whereby GA3 is the main product and only minor amounts of GA1 are formed in F. fujikuroi (Fig 7C; Studt and Tudzynski, 2014). Thus, Des activity is decisive for the proportion of GA3 to GA1 in the fungus. Indeed, FpDES is the only GA biosynthetic gene that is not affected by deletion of FpKMT6 and remains silent, while the remaining six are significantly upregulated in Δfpkmt6 as compared to FpWT (Fig 7A and S1 Table), thereby explaining the absence of GA7 and GA3 in the Δfpkmt6 samples. Thus, FpKmt6-mediated H3K27me3 is responsible for GA-gene silencing in F. proliferatum NRRL62905.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Gibberellin biosynthesis is silenced by H3K27me3 in F. proliferatum.

(A) Chromosome 8 is shown in grey and centromere in black; SM key enzyme-encoding genes are indicated by bars according to the following color code: polyketide synthase (PKS), blue; non-ribosomal peptide synthetase (NRPS), orange; (sesqui-/di-/sester-/tri-/tetra-) terpene cyclase (STC/DTC/PTTS/TrTC/TeTC), light green. Below the chromosome, the H3K27me3 coverage in the F. proliferatum wild-type strain NRRL62905 (FpWT) is shown. Chromatin marks at the BGC involved in GA biosynthesis in the F. proliferatum wild-type strain NRRL62905 (FpWT) and the strain deleted for FpKMT6 (Δfpkmt6). In descending order are H3K27me3 (red), H3K27ac (green), and transcription as determined by RNA-seq (black) for FpWT and the same for Δfpkmt6. Below this, the selected BGC is indicated as follows: the key enzyme-encoding gene is depicted in light green; cluster genes are shown in black, and bordering genes in grey. (B) Chemical analysis of GA biosynthesis. GA1_, but not GA3, was detected by targeted MS-based analysis in Δfpkmt6 but absent from FpWT and the complemented strain Δfpkmt6/FpKMT6. (C) Simplified scheme of the GA biosynthetic pathway (modified from Studt and Tudzynski, 2014).

https://doi.org/10.1371/journal.pgen.1011075.g007

Intrigued by these findings, we wanted to test whether H3K27me3 is also involved in GA gene silencing in another FFSC member. For this, we included F. mangiferae MRC7560 (FmWT) in our analyses. First, the genome-wide distribution of H3K27me3 was recorded via ChIP-seq using an H3K27me3-specific antibody and identical culture conditions as for FpWT. In line with low expression levels, GA biosynthetic genes are enriched for H3K27me3 also in FmWT (Fig 8A). To evaluate the relevance of H3K27me3 for GA gene regulation, downregulation of FmKMT6 was approached using the TetOff system [65], as deletion of FmKMT6 did not result in homokaryotic Δfmkmt6 strains (S1 Fig). At least three independent F. mangiferae mutant strains, i.e., TetOff::FmKMT6_T7, T8 and T10, were gained that showed correct integration of the TetOff construct as verified by diagnostic PCR (S19A and S19B Fig), and reduced H3K27me3 levels in western blot upon supplementation with 25 μg/mL doxycycline-hyclate (Sigma-Aldrich, DOX) (S19C and S20 Figs). All three TetOff::FmKMT6 mutants showed severe growth defects when supplemented with 10 to 50 μg/mL DOX (S19D Fig). To test whether GA genes are affected by the downregulation of FmKMT6 in F. mangiferae, relevant strains were grown in GA-inducing conditions (synthetic ICI with 6 mM glutamine) and GA transcript levels were analyzed and compared to FmWT. While only low and even lower GA expression was detected in FmWT and TetOff::FmKMT6 without the addition of DOX, GA transcript levels were significantly increased upon the addition of DOX in TetOff::FmKMT6 but remained the same in FmWT (Fig 8B). Chemical analysis via LC-MS/MS under the same culture conditions confirmed these results. Indeed, GA3 was detected in cultures inoculated with TetOff::FmKMT6 upon supplementation with DOX, while GAs biosynthesis is still absent from FmWT (Fig 8C). In contrast to F. proliferatum, all seven GA biosynthetic genes are increased upon knock down of FmKMT6 about 10- to 50-fold in F. mangiferae. Thus, it is intriguing to speculate that H3K27me3 functions as an adaptive layer for GA gene expression in F. proliferatum and F. mangiferae and possibly also other FFSC members but not F. fujikuroi. Here, the complete GA cluster is devoid of H3K27me3, which explains the fact, that F. fujikuroi can utilize these genes more frequently than other species [23]. These results again demonstrate how the chromatin landscape can shape fungal lifestyles.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Gibberellic acid (GA) biosynthesis is also silenced by H3K27me3 in Fusarium mangiferae.

(A) Detailed depiction of H3K27me3 coverage at the GA gene cluster (highlighted in light green) located on Scaffold12 in FmWT. Below the chromosome, the H3K27me3 coverage in the F. mangiferae wild-type strain MRC7560 (FmWT) is shown. Below this, the genes involved in GA biosynthesis are indicated as follows: the key enzyme-encoding gene is depicted in light green; cluster genes are shown in black, and bordering genes in grey. (B) Gene expression of GA biosynthetic genes in FmWT and two TetOff::FmKMT6 mutants under non-inducing (0 DOX) and inducing (25 DOX) conditions. Fungi were cultivated under low nitrogen conditions (F. fujikuroi GA-inducing conditions). Transcript levels were determined using RT-qPCR from prior transcribed cDNA. While GA biosynthesis is silent under both conditions for FmWT, the partial reduction of H3K27me3 led to the expression of all GA biosynthetic genes. (C) GA biosynthesis in FmWT and the TetOff::FmKMT6 knock-down strains. Fungi were cultivated under low nitrogen conditions (liquid ICI and 6 mM glutamine) for 7 days post inoculation at 30°C. Experiments were performed in triplicates. HPLC-HRMS measurement revealed that GAs are absent from cultures containing FmWT, while the TetOff::FmKMT6 strains were very well able to produce GAs in liquid cultures.

https://doi.org/10.1371/journal.pgen.1011075.g008

Strains, media, and culture conditions

The F. proliferatum strain NRRL62905, FpWT, was used as parental strain for the generation of FpKMT6 and FpKMT1 deletion (Δfpkmt6 and Δfpkmtl), and in situ complementation (Δfpkmt6/FpKMT6 and Δfpkmt1/FpKMT1). The F. mangiferae MRC7560 wild-type strain, FmWT, originated in Israel and deposited in the culture collection of the Medical Research Council (MRC) (Tygerberg, South Africa) was used as parental strain for the deletion attempt (Δfmkmt6), and the inducible knock-down using the tetracycline-sensitive TetOff system (TetOff::FmKMT6) [65]. The uracil-auxotrophic Saccharomyces cerevisiae FGSC 9721 (FY 834) [66] was provided by the Fungal Genetics Stock Center (Kansas State University) and used for yeast recombination cloning [67]. Escherichia coli strain DH5α (Invitrogen™) was used for plasmid propagation.

Strains were maintained on solid complete medium (CM, [68]). Fungal growth assays were conducted on solid CM, vegetable V8 juice (30 mM CaCO₃, 20%, v/v, vegetable juice; Campbell Food, Puurs, Belgium), and synthetic ICI [69] medium supplemented with 6 mM glutamine (Carl Roth). Plates were inoculated either with 10 μL of a 10⁵ conidia/mL solution or agar plugs and incubated at 30°C in the dark for up to seven days. In the case of TetOff::FmKMT6 mutant strains, solid CM was supplemented with either 0–50 μg/mL doxycycline (DOX) to induce silencing of FmKMT6. For genomic DNA isolation, relevant strains were grown for three days on CM covered with cellophane sheets (Folia Bringmann) at 30°C in the dark. Conidia production was assessed on solid V8. Medium was inoculated with a 5 mm² agar plug each and incubated at 20°C for 7 days under the presence of 18 h light and 6 h dark (L/D) or dark conditions (D) and 70% humidity. Conidia were counted using a Neubauer improved hemocytometer under a light microscope (Carl Zeiss). Microscopy of the fungal conidia was performed using a confocal laser scanning microscope (CLSM, Leica STELLARIS 5). For fungal liquid cultivations, mycelia were pre-cultured in 100 mL Darken medium [70] in a 300 mL Erlenmeyer flask for 3 days in the dark (30°C, 180 rpm). Then, 0.5 mL of the pre-culture was transferred to 100 mL ICI media supplemented with either 6 mM or 60 mM glutamine (Carl Roth) as sole nitrogen source. In case of the TetOff strains, cultivation was performed similar as described earlier [71]. Briefly, 150 μL of Darken medium of the respective wild-type strain and mutants were transferred to 30 mL synthetic liquid ICI supplemented with 6 mM glutamine as sole nitrogen source and 25 μg/mL DOX to induce continuous silencing of fmKMT6. All fungal cultures were incubated on an orbital shaker at 180 rpm, 30°C for 3 days for RNA- or ChIP-sequencing as well as protein extraction in case of the TetOff mutant strains and 7 dpi for SM analysis. Protein extraction of all other mutant strains were performed with mycelia grown on solid CM at 30°C in the dark and harvested 4 dpi. For Nanopore sequencing, FpWT was grown for 3 days in liquid YPG [72].

Plasmid construction and strain generation

Plasmids for deletion (pΔfpkmt6, pΔfmKMT6, and pΔfpkmt1), complementation (pΔfpkmt6/FpKMT6 and pΔfpkmt1/FpKMT1^s), and inducible downregulation of FmKMT6 (TetOff::FmKMT6) were generated using yeast-recombinational cloning as described previously [67]. Primers were obtained from Merck (Austria) and are listed in S3 Table. Fragments for cloning and subsequent transformation were amplified using a high-fidelity polymerase (Q5-polymerase, New England Biolabs or Phusion High-Fidelity DNA polymerase, Thermo Scientific). In the case of targeted gene deletion, ~1 kb upstream (5′) and downstream (3′) region of the gene of interest were amplified from FpWT gDNA using the primer pairs 5F/5R and 3F/3R, respectively, and the hygR resistance cassette amplified from the template vector pCSN44 [73] with the primer pair hphF/hphR, was used as selection marker. Complementation plasmids were generated as follows: FpKMT6 was amplified together with ~1kb upstream (5’) region in two fragments using the primer pairs KMT6_5F/Dia_KMT6_WT_R and KMT6_F1/FpKMT6_Cil-Tgluc_R2, followed by the artificial terminator sequence, Tgluc, from Botrytis cinerea B05.10 (BcTgluc) and the nourseothricin resistance cassette (natR) using the primers Tgluc_F2//hphF amplified from pffccl1_IL_C [74]. In the case of KMT1, FpKMT1 was amplified together with ~ 1kb upstream (5’) region using the primer pair KMT1_5F/ FpKMT1_Cil-Tgluc_R, followed by BcTgluc and the geneticin resistance cassette (genR) with the primer pair TglucF2//hphF from pΔfmkmt1/FmKMT1^Ces [28]. Amplification of the downstream (3′) region for both was performed with the respective primer pairs 3F/3R. In the case of the in situ inducible knock-down construct (TetOff::FmKMT6), the TetOff elements (Pgpda, tTA-Advanced, AFcgrA and the TetO7 sequence) coupled to the hygromycin resistance cassette was amplified from pTetOff::FfH2A.Z, using the primer pair TtrpC_R/ Tet-off_R. Next, ~1 kb of the 5’ and 3’ regions were amplified from FmWT gDNA using the primer pair KMT6_F/TetOff_KMT6_5R and TetOff_KMT6_3F/KMT6_3R, respectively. The Saccharomyces cerevisiae FY834 strain was transformed with the respective PCR fragments, the resistance cassettes and the EcoRI/XhoI-digested shuttle vector pRS426, yielding pΔfpkmt6/FpKMT6, Δfpkmt1/FpKMT1 and TetOff::FmKMT6 (S3A, S8A and S19A Figs). Correctness of the complementation and the silencing plasmids was verified by sequencing (LGC Genomics, Germany).

Fungal transformations were performed as described previously [32]. For gene deletion, linear fragments were amplified from pΔfpkmt6, pΔfmkmt6 and pΔfpkmt1 with the respective primer pairs 5F//3R (S1 Table), using a proof-reading polymerase (Q5-polymerase, New England Biolabs or Phusion Flash High-Fidelity PCR Master Mix, Thermo Fisher Scientific). In the case of complementation, 10 μg each of pΔfpkmt6/FpKMT6 and pΔfpkmt1/FpKMT1 were digested with XbaI/XhoI prior to transformation (S3A and S8A Figs). Transformation of the TetOff::FmKMT6 construct was approached using 10 μg AanI/Eco105I digested plasmid DNA (pDNA, S19A Fig). Fungal transformants were selected on regeneration medium supplemented with the appropriate antibiotic i.e., hygromycin B (Merck Millipore), geneticin (Fermtech Garching) or nourseothricin (Jena Bioscience). Homologous integration events of Δfpkmt6 and Δfmkmt6 was verified by diagnostic PCR using GoTaq G2 DNA Polymerase (Promega) and the primer pairs dia_kmt6_5’/trpC-T and dia_kmt6_3’/trpC-P, and absence of the wild-type gene was analyzed using the primer pairs dia_KMT6_WT_F/R (S1B, S1C and S2B Figs and S3 Table). In the case of Δfpkmt1, dia_kmt1_5’/trpC-T and dia_kmt1_3’/trpC-P proved homologous integration, and absence of the wild-type gene was verified using the primer pair dia_KMT1_WT_F/R (S6 Fig and S3 Table). Successful complementation of Δfpkmt6 (S3 Fig) with the respective wild-type gene, FpKMT6, was verified with the primer pairs dia_KMT6_5’/trpC-T (absence of hphR), dia_KMT6_3’/trpC-P2 and dia_KMT6_WT_F/dia_KMT6_WT_R (presence of natR and FpKMT6, respectively. In the case of Δfpkmt1, attempts to complement the deletion strain Δfpkmt1 by in situ integration of FpKMT1 failed in our hands, but several transformants that showed ectopic integration of the complementation fragment were gained as verified with the primer pairs KMT1_dia_WT_F/R (presence of FpKMT1), dia_kmt1_5’/KMT1_dia_WT_R and dia_kmt1_3’/genR_R (would show in situ integration, S8 Fig). Complemented strains were additionally verified by reverse transcriptase-quantitative PCR (RT-qPCR, S3C and S8C Figs).

Standard molecular techniques

Yeast pDNA was extracted with the GeneJET Plasmid Miniprep Kit (Thermo Fisher Scientific) according to the manufacturer’s protocol and directly used as templates for PCR in the case of the targeted gene deletion. In the case of complementation, yeast pDNA was re-transformed in E. coli DH5α (Invitrogen) according to the manufacturer. Subsequent extraction of pDNA was again performed with the GeneJET Plasmid Miniprep Kit (Thermo Fisher Scientific. Genomic DNA (gDNA) for subsequent PCRs was extracted from lyophilized, ground material as described elsewhere [75]. For gene expression analysis, RNA was isolated from lyophilized and ground mycelium with the RNA reagent TRIzol (Thermo Fisher Scientific) according to the manufacturer’s instructions. For cDNA synthesis, 1 μg RNA was treated with DNaseI (Thermo Fisher Scientific) and transcribed into cDNA using LunaScript RT SuperMix Kit (NEB). For subsequent RT-qPCR analysis the iTaq Universal SYBR Green Supermix (Bio-Rad) was used and cDNA was quantified on an iCycler iQ Real-Time PCR System (Bio-Rad). In all cases the primer efficiency was kept between 90 and 110%, Ct values greater than 32 were taken as not expressed. Results were calculated according to the ΔΔCt [76]. Expression of all tested genes were normalized to the expression of actin (FPRN2_04233), Glyceraldehyde-3-phosphate dehydrogenase (GPD, FPRN2_25921), and β-tubulin (FPRN2_07708). All primers used are listed in S3 Table. Western blot analyses was performed as described previously [28]. For western blot analysis roughly 25–50 μg of total protein extract was separated on a SDS gel and the following primary antibodies were used for subsequent probing: anti-H3 C-Term (Active motif, 91299), anti-H3K9me3 (Active motif, 39161 / Abcam, 8988), anti-H3K27me3 (Active motif, 39155), and anti-H3K27ac (Cell Signaling, D5E4 XP), and anti-rabbit HRP-conjugated secondary antibody (Jackson ImmunoResearch) was used for subsequent detection of chemiluminescence signals. Membranes were developed with Clarity ECL Western Substrate (Bio-Rad) and visualized with a ChemiDoc XRS (Bio-Rad) system. Subsequent densitometric analyses of western blot signals were performed using the ImageJ software. Signals were normalized to the histone H3 C-term signal and the wild-type signal ratio was set to 100% for referencing.

Chemical analyses

Supernatants for chemical analyses were retrieved from fungi-inoculated 7-day-old liquid ICI cultures supplemented either with 6 mM or 60 mM glutamine. Detection of siccanol in FpWT, Δfpkmt6 and Δfpkmt6/FpKMT6^Cis as well as GA₃ in samples from F. mangiferae were analyzed as previously described [77]. For all other samples, supernatants were filtered through the 0.2 μm syringe filters (RC membrane, 15 mm, non-sterile, PP housing, luer/slip, Phenomenex, Aschaffenburg, Germany). An aliquot of 100 μL filtrate was diluted 1:1 with acetonitrile (MeCN), the samples were stored at 4°C overnight, filtered a second time, and directly used for screening analysis. Bikaverin, fusaric acid, and beauvericin were obtained from Merck (Darmstadt, Germany), Gibberellins (GAs) GA1, GA3, GA4 and GA7 were obtained from Serva (Heidelberg, Germany), fumonisins B1, B2, and fusarin C were obtained as described in previous work [78] all in analytical grade.

For the targeted screening an LC-MS/MS method was developed. The chromatographic separation and subsequent MS/MS analysis was achieved with a 1260 LC system (Agilent, Waldbronn, Germany), coupled to a QTRAP 5500 mass spectrometer equipped with a Turbo V Ion Source (SCIEX, Darmstadt, Germany). Data acquisition and subsequent data analysis was done with the Analyst 1.6.2. software (SCIEX, Darmstadt, Germany). Chromatographic separation was performed on a Nucleodur C18 Gravity SB column (100 mm * 2 mm i.d., particle size 3 μm) equipped with a 4 * 2 mm i.d. guard column of the same material (Macherey-Nagel, Düren, Germany). The gradient started with 10% MeCN + 0.1% FA (solvent A) and 95% H₂O + 0.1% FA (solvent B) for 1.5 min and a flow rate of 0.50 mL/min at 40°C column oven temperature. Within 15.5 min, A was increased linearly to 100%. For the next 1.5 min, A was held at 100%. Equilibration was performed for 3 min. Electrospray ionization was performed in both positive mode at a source voltage of 4.5 kV, and in negative mode at -4.5 kV. Source temperature was set to 450°C; curtain gas flow to 35 psi, nebulizer gas to 35 psi, heater gas to 45 psi. Both declustering potential (DP) and collision energy voltage (CE) were optimized along with the analyte specific transitions and measurement modes and are given in S4 Table. For qualitative detection of the formation of the target compounds, precursor ion scans were implemented to the LC-MS/MS method based on the fragmentation pattern of the respective analyte. Additionally, a standard solution was prepared containing all analytes at a concentration of 50 μg/mL, which was measured before sample measurement was commenced and after completion of the sample sequence. In all cases, obtained results were normalized to the biomass of the respective strains to account for growth differences of the different strains.

Nanopore sequencing and new assembly

High molecular weight DNA was extracted from ground lyophilized mycelium using Genomics Buffer Set (Qiagen) and purified using QIAGEN Genomic-Tips 20/G as described previously [45]. Small DNA fragments were removed from the sample with Circulomics Short Read Eliminator XS in accordance with manufacturers recommendations prior to sequencing on a MinION R9.4.1 flow cell using the ligation sequencing protocol (SQK-LSK109) from Oxford Nanopore Technologies. Basecalling, demultiplexing, and adaptor removal were performed with Guppy v5.0.12 [79] in GPU mode using the r941_min_sup_g507 model. The reads were filtered to a minimum length of 20 kb and a minimum quality of 80 (Q7) using Filtlong v0.2.0 [80] followed by assembling of the reads as described previously (Petersen et al., 2022) with the software versions: Minimap2 v2.17 [81], Miniasm v0.3 [82]. Polishing was performed by Racon v1.3.3 [83] and Medaka v1.4.3 (Oxford Nanopore Technologies, 2018). An identity between utg000012l and utg000004l was observed and BLAT v37x1 [84] was used to calculate the overlap; thus, the first 878,742 pb from utg000012l were then added to utg000004l on the 5’ end creating the new utg000004l. The fungal telomeric repeat sequence of TTAGGG/CCCTAA was used to identify telomeric region in the assembly of FpWT as described previously [45]. The genome sequence data is available on NCBI Genbank CP128300-CP128312.

RNA sequencing, mapping and quantification

Total RNA for subsequent sequencing was extracted using TRIzol Reagent as described above, and sent to the Vienna BioCenter (Vienna, Austria) for quality control, library preparation and sequencing. Experiments were performed in duplicates. Library prep and sequencing was performed using poly-A enrichment kit (NEB) and Nextera Library prep kit. 50 bp single end sequencing was performed using a HiSeq v4 Illumina sequencer. Obtained sequences were de-multiplexed, quality controlled, filtered using trimmomatic 0.36 [85] and mapped on the new genome assembly. Mapping was performed using BWA [86] and reverse transcripts were counted using python script HTSeq [87]. Normalization and statistics were done using R/Bioconductor and the limma and edgeR packages, using mean-variance weighting (voom) and TMM normalisation [88]. A significance cut-off of p < 0.01 and differential expression of +/-1 (2-fold) was applied for analysis. Transcription levels are log2 read counts per kilobase of exon per million library reads (RPKM). The RNA-seq data has been submitted to NCBI’s Gene Expression Omnibus [89] and are accessible through GEO Series accession number GSE235901. GO annotation done using PANNZER2 [90] and processed using R library “GOstats”, “GSEABase”; GO term reduction was done using library “rrvgo”. Only GO-terms enriched in the selected gene sets with p < 0.05 were consider for further investigation.

Chromatin immunoprecipitation and ChIP-sequencing

Chromatin immunoprecipitation (ChIP) was performed essentially as described [23, 74]. The following antibodies were used for immunoprecipitation: anti-H3K9me3 (Active motif, 39161), anti-H3K27me3 (Active motif, 39155), anti-H3K27ac (Cell Signaling, D5E4-XP(R)), H3K36me3 (Abcam, 9050). Experiments were performed at least in biological duplicates, and input controls were added for each strain. Obtained DNA was sent to the Vienna BioCenter (Vienna, Austria) for library preparation and sequencing. Paired-end sequencing was performed on HiSeq v4 Illumina sequencer and quality filtering, trimming and mapping was performed as described for RNA-seq experiments. Quantification of mapping in specific regions around genes were performed in R using the GenomicRanges Biostrings [91, 92]. Regions of interest and ChIP coverage were determined as follows: For H3K27me3 the maximum coverage in the region 100 bp upstream and 1000 bp downstream of ATG of each gene, coverages above 4 were considered as positives; for H3K27ac the max coverage in the region 2000 bp upstream and 100 bp downstream with a threshold of 6; for H3K36me3 the region 1000bp upstream and 0 bp downstream and minimum coverage of above 2 were considered positive; and for H3K9me3 peaks above a height of 2 were considered positives, peaks present in both (Δfpkmt6 and FpWT) were excluded only peaks emerging in kmt6 were considered and genes within regions +/- 10kb and +/- 50 kb around them were indicated. Peak-calling was done in R functions used are available via github (https://github.com/symbiocyte/MNase). The ChIP-seq data is available under BioProject PRJNA986687.