ABSTRACT
Histone modifications are crucial for the regulation of secondary metabolism in various filamentous fungi. Here we studied the involvement of histone deacetylases (HDACs) in secondary metabolism in the phytopathogenic fungus Fusarium fujikuroi, a known producer of several secondary metabolites, including phytohormones, pigments, and mycotoxins. Deletion of three Zn2+-dependent HDAC-encoding genes, ffhda1, ffhda2, and ffhda4, indicated that FfHda1 and FfHda2 regulate secondary metabolism, whereas FfHda4 is involved in developmental processes but is dispensable for secondary-metabolite production in F. fujikuroi. Single deletions of ffhda1 and ffhda2 resulted not only in an increase or decrease but also in derepression of metabolite biosynthesis under normally repressing conditions. Moreover, double deletion of both the ffhda1 and ffhda2 genes showed additive but also distinct phenotypes with regard to secondary-metabolite biosynthesis, and both genes are required for gibberellic acid (GA)-induced bakanae disease on the preferred host plant rice, as Δffhda1 Δffhda2 mutants resemble the uninfected control plant. Microarray analysis with a Δffhda1 mutant that has lost the major HDAC revealed differential expression of secondary-metabolite gene clusters, which was subsequently verified by a combination of chemical and biological approaches. These results indicate that HDACs are involved not only in gene silencing but also in the activation of some genes. Chromatin immunoprecipitation with the Δffhda1 mutant revealed significant alterations in the acetylation state of secondary-metabolite gene clusters compared to the wild type, thereby providing insights into the regulatory mechanism at the chromatin level. Altogether, manipulation of HDAC-encoding genes constitutes a powerful tool to control secondary metabolism in filamentous fungi.
INTRODUCTION
Accessibility of DNA is regulated by posttranslational modifications that define the state of chromatin: either loosely packaged and open for transcription (euchromatin) or more densely packaged, resulting in transcriptional silencing (heterochromatin). Of all known posttranslational modifications to date, histone acetylation seems most abundant (1) and is probably the best-understood one. Acetylation of histones is regulated by the opposing actions of two enzymes: histone acetyltransferases (HATs), which add acetyl groups to defined ε-amino groups of lysine residues, and histone deacetylases (HDACs), which are responsible for their removal. In general, histone acetylation is associated with gene transcription, while histone deacetylation results in gene silencing (2, 3). However, there are increasing numbers of indications that HDACs are also required for gene activation (4–6). To shed more light on the role that histone (de)acetylation plays in biological processes, the enzymes responsible for these reactions have been studied in detail in several organisms. In fungi, the state of histone acetylation has been shown to be crucial for transcriptional regulation of various processes, including secondary metabolism (e.g., see references 7–9). The fact that genes involved in secondary metabolite biosynthesis are often located adjacent to each other in gene clusters and, furthermore, in subtelomeric regions is thought to facilitate their regulation, e.g., by chromatin modifications (10, 11). In general, it is assumed that during active growth, the chromatin landscape of secondary-metabolite gene clusters is enriched with silencing marks to inhibit their biosynthesis under unfavored conditions. Upon reception of inducing external signals, e.g., due to changing environmental conditions or the presence of a respective host, silencing marks are replaced by activating marks, and thus, transcription is initiated (12). A direct link between histone acetylation and secondary-metabolite gene expression in filamentous fungi was first shown for the aflatoxin gene cluster in Aspergillus parasiticus (13). Later on, an enrichment for acetylated histone 3 lysine 9 (H3K9ac), H3K14ac, or H4K12 was also found in the landscape of secondary-metabolite gene clusters under active transcription but was missing in a silencing environment in other fungi, including Fusarium fujikuroi (8, 9, 14–17). These findings suggest an important role of HATs and HDACs in the regulation of genes involved in secondary metabolism.
To date, two groups of HDACs have been identified: (i) the NAD+-dependent silent information regulator 2 (Sir2) family (“sirtuins”), which are homologous to yeast Sir2 (class III), and (ii) the “classical” HDACs, which depend on the presence of Zn2+, which acts as a coactivator for deacetylase activity (18). In fungi, there are two classes of “classical” HDACs: the RPD3-type HDACs (class I) and the HDA1-type HDACs (class II) (3, 19, 20). Both classes can be inhibited by the addition of trichostatin A (TSA) already at low concentrations (21). Thus, treatment with TSA resulted in increased production of various secondary metabolites in different fungi, i.e., Alternaria alternata, Penicillium expansum, Cladosporium cladosporioides, and Diatrype disciformis (22, 23). Besides inhibition of HDACs, replacement of HDAC-encoding genes revealed significant changes of the secondary-metabolite profile in several fungi. Thus, deletion of the class II HDAC hdaA in Aspergillus nidulans and Aspergillus fumigatus resulted in upregulation of penicillin and sterigmatocystin biosynthesis and both the up- and downregulation of several nonribosomal peptide synthetase (NRPS)-encoding genes, respectively (6, 22), while the hdaA homolog HDF2 had only a minor impact on secondary metabolism in Fusarium graminearum. Instead, deletion of the class I HDAC HDF1 resulted in the significant reduction of deoxynivalenol and upregulation of the aurofusarin biosynthetic genes (24). These findings provide strong evidence that HDACs possess both activating and inhibiting roles regarding the regulation of different secondary metabolites in fungi.
F. fujikuroi is well known for the production of the plant hormone gibberellic acid (GA), the causative agent of bakanae disease on rice, for more than 100 years (25). Besides gibberellins, the fungus produces a wide range of other secondary metabolites, such as the red polyketide pigments bikaverin (26, 27) and fusarubins (28, 29) and mycotoxins such as fusarin C (30), fusaric acid (31), or fumonisins (32). Furthermore, recent sequencing of the genome of F. fujikuroi wild-type (WT) strain IMI58289 revealed the presence of about 40 additional putative secondary-metabolite gene clusters, most of which are cryptic and silent under laboratory conditions (17). Possible strategies to awaken these clusters, thereby unraveling the respective products, include various growth conditions, overexpression of key enzymes and/or pathway-specific transcription factors, or modification of histone-modifying enzymes, which in turn leads to alterations of the surrounding chromatin landscape.
This paper describes the impact of three Zn(II)-dependent HDACs, FfHda1 (class II), FfHda2 (class I), and FfHda4 (class II), on growth, differentiation, and secondary metabolism in F. fujikuroi. We applied targeted deletion and constitutive expression of these genes and analyzed the mutants by a combination of chemical and biological approaches. Microarray analysis of the Δffhda1 mutant that has lost the major HDAC in F. fujikuroi indicated its involvement in the regulation of at least four secondary metabolites, bikaverin, fusarubins, fusaric acid, and gibberellins, while deletion of the second class II HDAC-encoding gene, ffhda4, has no impact on the biosynthesis of these secondary metabolites. Comparative chromatin immunoprecipitation followed by high-throughput sequencing (ChIP-seq) provided insights into the regulation of secondary-metabolite genes at the chromatin level. In addition to FfHda1, the class I HDAC FfHda2 is also crucial for secondary-metabolite biosynthesis in F. fujikuroi. Furthermore, rice infection studies revealed that both HDACs FfHda1 and FfHda2 are required for GA-induced bakanae disease of rice.
MATERIALS AND METHODS
Fungal strains and culture conditions.Deletion alleles were constructed in Fusarium fujikuroi wild-type strain IMI58289 (Commonwealth Mycological Institute, Kew, United Kingdom). For cultivation in submersed culture, F. fujikuroi strains were preincubated in 300-ml Erlenmeyer flasks with 100 ml Darken medium (DVK) (33) on a rotary shaker for 72 h at 28°C and 180 rpm. For secondary-metabolite analysis and RNA isolation, a 500-μl aliquot of this culture was used for inoculation of synthetic ICI medium (Imperial Chemical Industries Ltd., United Kingdom) (34) using the desired nitrogen source and cultivated for an additional 2 to 7 days. For transformation of the fungus, protoplasting was carried out with 100 ml ICI medium with 10 g liter−1 fructose instead of glucose and 1 g liter−1 (NH4)2SO4 using 500 μl of preincubated DVK and additional cultivation for up to 16 h on a rotary shaker at 28°C and 180 rpm. For fluorescence microscopy, the fungal strains were grown overnight in liquid ICI medium (6 mM for Hda1::Gfp and 60 mM glutamine for Hda1::Gfp) upon preculturing in DVK. For microarray analysis, the WT and Δffhda1 strains were grown for 3 days in ICI medium with either 6 mM glutamine or 60 mM glutamine upon preculturing. For DNA isolation, the fungus was grown on solid complete medium (CM) with cellophane overlays for 3 days at 28°C. Plate assays were performed on CM as well as on Czapek Dox (CD) medium for up to 7 days (35, 36).
Plasmid constructions.Plasmid construction for deletion of all three HDACs was accomplished by yeast recombinational cloning (37). Therefore, the 5′ and 3′ regions of the corresponding genes were amplified with appropriate primer pairs: for the 5′ region, primers 5F and 5R were used, and for the 3′ region, primers 3F and 3R were used, based on the available genomic sequence of F. fujikuroi IMI58289 (17). Hygromycin B was used as a resistance marker. The hygromycin resistance cassette, consisting of the hygromycin B phosphotransferase gene hph (38), driven by the trpC promoter, was amplified by using the primer pair hph-F/hph-R from the template pCSN44 (39). All primers used for PCR were obtained from Eurofins GmbH (Ebersberg, Germany) (see Tables S2 to S5 in the supplemental material). The obtained fragments were cloned into Saccharomyces cerevisiae strain FY834 (40) together with the EcoRI/XhoI-restricted plasmid RS426 (41). For double deletion of ffhda1 and ffhda4 and of ffhda1 and ffhda2, the same targeted deletion strategy was used as that used for the single deletions, but instead of hygromycin B, the Geneticin resistance cassette, amplified from pKS-Gen by using primer pair Geni-gpd-F/Geni-tubT-R, and the nourseothricin resistance cassette, amplified from pZPnat by using primer pair hph-F-trpC-P/hphR-trpC-T2, were used, respectively (42). Subsequently, the ffhda1 deletion mutant was transformed with the knockout fragments for ffhda4 and ffhda2. For overexpression of ffhda1, the gene was amplified by using primer pair OE-hdF2_F/OE-hdF2_R. The obtained fragment was cloned into S. cerevisiae together with the HindIII-restricted modified plasmid RS426 additionally containing the hygromycin resistance cassette driven by the trpC promoter. The gene of interest itself was driven by the gpd promoter amplified from pveAgfp (43) by using primer pair gpd-yeast-for/gpd-yeast-rev. For creation of gfp vectors for gfp fusion, ffhda1 and ffhda2 were amplified by using primer pairs hdF2-gfp_F/hdF2-gfp_R and hdF1-gfp_F/hdF1-gfp_R, respectively, and the obtained fragments were cloned into S. cerevisiae Y834 together with the NcoI-restricted plasmid NDN-OGG harboring a nourseothricin resistance cassette driven by the trpC promoter. The genes itself were driven by the oliC promoter (42). For complementation of the Δffhda1 and Δffhda2 mutants, the respective gene was amplified, including about 2 kb of the native promoter sequence in two fragments, by using primer pairs com-hdF2_F1/com-hdF2_R3 and com-hdF2_F3/com-hdF2_R2 for complementation of the Δffhda1 mutant and primer pairs hdF1-com_F1/hdF1-com_R1 and hdF1-com_F2/hdF1-com_R2 for complementation of the Δffhda2 mutant. Subsequently, the fragments were cloned together with SacII/SpeI-restricted pNDN-OGG, which harbors a nourseothricin resistance cassette (42), into S. cerevisiae Y834. Subsequent sequencing of the plasmids verified the correct assembly, yielding vectors pgpd::ffhda1, pffhda2::gfp, pffhda1::gfp, pffhda2C, and pffhda1C.
Standard molecular methods.DNA was isolated from lyophilized mycelium that was ground to a fine powder in liquid nitrogen and prepared according to methods described previously by Cenis (44) and was used directly for PCR amplification. PCRs were carried out by using 25 ng genomic DNA, 5 pmol of each primer, 200 nM deoxynucleoside triphosphates, and 1 unit BioTherm DNA polymerase (GeneCraft GmbH, Lüdinghausen, Germany) in the case of diagnostic PCR and amplification of upstream and downstream regions for yeast recombinational cloning. PCRs were performed as follows: an initial denaturing step at 94°C for 3 min followed by 35 cycles of 1 min at 94°C, 1 min at 56°C to 60°C, and 1 to 2 min at 70°C and a final elongation step at 70°C for 10 min. Amplification of the yeast-derived knockout fragments was done by using the TaKaRa polymerase kit, as indicated. For complementation strategies and gfp fusion, the respective genes were amplified from genomic DNA of F. fujikuroi IMI28589 (17) by using a proofreading polymerase. Under these conditions, 25 ng genomic DNA, 5 pmol of each primer, and 1 unit of Phusion polymerase (Finnzymes, Thermo Fisher Scientific, Finland) were used. Plasmid DNA from S. cerevisiae was extracted by using a yeast plasmid isolation kit (SpeedPrep; DualsystemsBioTech) and was used directly for PCR. In the cases of overexpression, gfp fusion, and complementation, the obtained vectors were cloned into Escherichia coli, subsequently extracted by using the GeneJETTM plasmid miniprep kit (Fermentas GmbH, St. Leon-Rot, Germany), and sequenced by using the BigDye Terminator v3.1 cycle sequencing kit and the ABI Prism 3730 genetic analyzer (Applied Biosystems, Foster City, CA, USA) according the manufacturer's instructions. DNA and protein sequence alignments were done with DNASTAR (Madison, WI, USA). For Southern blot analysis, genomic DNA was digested with the appropriate enzymes, separated on a 1% (vol/vol) agarose gel, and transferred onto nylon membranes (Nytran SPC; Whatman) by downward blotting (45). Subsequent hybridization was accomplished overnight at 65°C by using 32P-labeled probes generated by using the random oligomer-primer method according to the protocol described previously by Sambrook et al. (46). After hybridization, the membrane was washed with 1× SSPE (0.18 M NaCl, 10 mM NaH2PO4, and 1 mM EDTA [pH 7])–0.1% sodium dodecyl sulfate (SDS) at the same temperature. RNA for Northern blot analysis was isolated from lyophilized mycelium by using the RNAagents total RNA isolation kit (Promega, Mannheim, Germany) according to the manufacturer's instructions. Twenty micrograms of RNA per sample was loaded onto a 1% agarose gel and run under denaturing conditions (1% [vol/vol] formaldehyde) (46). Afterwards, samples were transferred onto Hybond-N+ membranes and hybridized with 32P-labeled probes as described above.
Fungal transformations.Protoplasts were prepared from F. fujikuroi IMI58289. Transformation was carried out as previously described (47). About 107 protoplasts were transformed with ∼10 μg of the amplified replacement cassettes for the knockouts. For overexpression, the wild type was transformed with 10 μg of the gpd::ffhda1 plasmids, and for gfp fusion and complementation, the respective deletion mutants were transformed with 10 μg of the ffhda2::gfp, ffhda2-com, ffhda1::gfp, and ffhda1-com plasmids, respectively. Transformed protoplasts were regenerated as described previously (48). The medium contained the appropriate antibiotic. Single conidial cultures were established from either hygromycin B-, nourseothricin-, or Geneticin-resistant transformants and used for subsequent DNA isolation.
Sequence data and phylogenetic analysis.HDAC sequences were retrieved from the following sources: F. graminearum was obtained from the Broad Institute (http://www.broadinstitute.org/), and characterized HDACs of other fungi were obtained from the NCBI (http://www.ncbi.nlm.nih.gov/). Phylogenetic analysis was performed with protein sequences of the respective HDAC domains by using DNASTAR (Madison, WI, USA).
Chromatin immunoprecipitation and next-generation sequencing.For chromatin immunoprecipitation (ChIP), the mycelium of the wild type was grown in ICI medium upon preculturing in DVK. After 3 days, we cross-linked the mycelium with 1% formaldehyde and incubated it for an additional time period of 10 to 30 min. The cultures were then quenched by the addition of 250 mM glycine for a few minutes, and mycelium was harvested by filtration. About 200 mg of the mycelium was used for subsequent ChIP (49, 50). Libraries were prepared as described previously (51). We used an antibody against H3K9ac (catalog number 39137; Active Motif). Sequencing of the obtained DNA was carried out on an Illumina GAII genome analyzer by single-end 36- to 50-nucleotide (nt) sequencing. Reads were mapped to the F. fujikuroi genome with Burrows-Wheeler Aligner and visualized in a gbrowse2 genome browser (50). Quantification of ChIP-seq data was performed by using EpiChIP (49, 50, 52).
Quantitative real-time PCR.Quantitative real-time PCR (qPCR) was performed by using iTaq Universal SYBR green Supermix (Bio-Rad) in an iQ5 Bio-Rad thermocycler. In all cases, the primer efficiency for qPCR was between 90% and 110%, and the annealing temperature was 60°C. Samples were run in duplicate (technical repeats), resembling a third biological replicate of the ChIP-seq data. The results were calculated according to the ΔΔCT method (53). Two different housekeeping genes were used as reference genes, i.e., the related actin gene rac (primer pair FRACRTPCRFW/FRARTPCRRV) and the GDP-mannose transporter gmt (FFGM_07718-f1/FFGM_07718-r1). The genes cps/ks (FCPSKSRTPCRFW/FCPSKSRTPCRRV) and bik5 (bik5_qPCR_F1/bik5_qPCR_R1), encoding the ent-copalyl diphosphate synthase/kaurene synthase and the pathway-specific transcription factor of the gibberellin and bikaverin gene clusters, respectively, both essential for metabolite biosynthesis, were chosen for amplification. All primers used in this study are listed in Table S4 in the supplemental material.
Histone isolation and immunoblotting.For histone isolation, the fungal strains were grown for 3 days in ICI medium with 6 mM and 60 mM glutamine upon preculturing in DVK. The mycelium was then harvested, lyophilized, and subsequently used for histone isolation according to methods described previously by Honda and Selker (54). Therefore, the mycelium was ground to a fine powder and suspended in 20 ml lysis buffer (0.3 M sucrose, 0.04 M NaHSO3, 25 mM Tris-HCl [pH 7.4], 0.01 M MgSO4, 0.5 mM EDTA, 0.5% NP-40, 1 mM phenylmethylsulfonyl fluoride [PMSF], 1 μg/ml leupeptin, 1 μg/ml pepstatin), pelleted, and washed in the same buffer. The nuclear fraction was then resuspended in 2 ml ice-cold CW buffer (0.15 M NaCl, 10 mM Tris-HCl [pH 8.0], 1 mM β-mercaptoethanol, 1 mM PMSF, 1 μg/ml leupeptin, 1 μg/ml pepstatin) and mixed with 2 ml 0.4 M H2SO4. After incubation overnight on a rotary shaker, debris was pelleted and removed. The supernatant that contained the histone proteins was mixed with a one-quarter volume of trichloroacetic acid and incubated on ice for 60 min. Histone proteins were recovered by pelleting, washed in ice-cold acetone, and dried in a vacuum. Purified histones were resuspended in Tris (pH 8), the protein concentration was determined, and 10 μg protein of each sample was subjected to SDS-polyacrylamide gel electrophoresis and immunoblotting using antibodies against histone 3 lysine 9 (H3K9ac) (catalog number 39137; Active Motif) and the overall acetylation of histone 3 (H3ac) (catalog number 06-599; Millipore).
Microscopy.For fluorescence microscopy, a Zeiss AxioImager M1 instrument was used. Differential interference contrast (DIC) microscopy was performed by using bright-field images when indicated. Green fluorescent protein (GFP) fluorescence was examined by using filter set 38 (excitation bandpass [BP] 470/40, beam splitter Farbteiler 495, and emission BP 525/50). Images were taken with a Zeiss AxioCam MRm camera and analyzed by using the Axiovision Rel 4.8 software package.
HDAC activity assay.The method for isolation of the histone-enriched proteins was modified from the method described previously by Ding et al. (55). Protoplasts of the WT and the single HDAC deletion mutants were resuspended to about 5 × 109 protoplasts per ml in lysis buffer (10 mM Tris-HCl [pH 7.5], 10 mM NaCl, 15 mM MgCl2, 250 mM sucrose, 0.5% NP-40, 0.1 mM EGTA, 200 μM PMSF), mixed thoroughly, and kept on ice for 15 min. Subsequently, 1 ml of the lysate was layered over 4 ml suspension buffer (10 mM Tris-HCl [pH 7.5], 10 mM NaCl, 3 mM MgCl2) with 30% sucrose and centrifuged for 10 min at 4°C at 1,300 × g. Nuclear pellets were washed in suspension buffer without sucrose, resuspended in 100 μl extraction buffer (50 mM HEPES [pH 7.5], 420 mM NaCl, 0.5 mM EDTA, 0.1 mM EGTA, 10% glycerol, 200 μM PMSF), and sonicated for 30 s by using a Bioruptor Plus instrument (Diagenode, Liège, Belgium). After 30 min on ice, the nuclear suspension was centrifuged at 10,000 × g for 10 min at 4°C, and the supernatant was used for HDAC activity assays with the colorimetric HDAC activity assay kit (Active Motif, Carlsbad, CA, USA), according to the manufacturer's instructions. The deacetylation reaction was carried out at 37°C for 120 min. Fluorescent signals were detected with a plate reader at 405 nm (catalog number 550; Bio-Rad, Munich, Germany). For positive and negative controls, 25 μg of an untreated HeLa cell nuclear extract and a HeLa cell nuclear extract treated with 1 μM TSA was added, respectively. The concentration of the deacetylated compound was calculated by using the standard curve provided with the kit. HDAC activity is given in pmol min−1 mg−1. HDAC activity was related to the overall protein concentration of the samples, determined by a Bradford protein assay.
Microarray analysis.Expression data analyses were carried out with the NimbleGen F. fujikuroi custom microarray, which was designed based on the genome annotation of F. fujikuroi wild-type strain IMI58289 (17). Hybridization of microarrays was done at Arrows Biomedical (Münster, Germany) and Roche NimbleGen (Iceland). For each experimental condition, two biological replicates were created. We processed the retrieved tif images to numeric raw data by using NimbleScan v2.6 software (Roche NimbleGen, Inc.). After that, we utilized the oligo R package (56) and its implementation of the RMA algorithm (57) for background correction, normalization, and summarization of probe intensities. Differentially expressed genes were determined with the help of the limma R package (58) by fitting linear models for each gene and computing moderated t statistics by using the empirical Bayes method. The resulting P values were corrected for multiple testing by applying the Benjamini-Hochberg procedure (59). In comparing two experimental conditions, we regard genes with a log2 fold change above 1 and a false discovery rate (FDR) below 0.05 as being differentially expressed.
Pathogenicity assay.Seeds from Oryza sativa were prepared and surface sterilized as described previously (43). A 5-mm-diameter mycelial plug of the indicated strains was transferred into a 3- by 20-cm test tube filled with 25% vermiculite (Deutsche Vermiculite Dämmstoff GmbH, Sprockhövel, Germany). Subsequently, germinated rice seedlings were transferred onto another layer of vermiculite and watered with 10 ml Gamborg B5 solution (3.16 g liter−1) (Duchefa Biochemie, Haarlem, the Netherlands). H2O was used as a negative control, and 100 ppm GA3 was used as the positive control. Rice and fungus were then incubated for 7 days at 28°C with a 12-h-light–12-h-dark cycle.
Analysis of secondary metabolites.For gibberellin analysis, 20 ml of culture fluid of 7-day-old cultures was extracted over a C18 cartridge and measured by using high-performance liquid chromatography coupled to UV detection (HPLC-UV) as described previously (60). For analysis of bikaverin, fusarubins, fusarins, and fusaric acid, culture fluid of 7-day-old cultures was filtered over a 0.2-μm membrane filter (Millex; Millipore) and used directly for analysis without further preparation, as described previously (29). Bikaverin formation was detected at 512 nm, fusarubin formation was detected at 450 nm, fusarin formation was detected at 363 nm, and fusaric acid formation was detected at 268 nm. For metabolite quantification, the concentration of the desired secondary metabolites was related to the biomass accumulation to exclude falsification of the obtained results due to differences in growth behavior. For direct comparison of the mutant to the wild type, the production level of the wild type was set to 100%.
Statistical analysis.Statistical analyses of the mutant strains compared to the F. fujikuroi wild-type strain were accomplished by two-sample t tests in Excel (Microsoft).
Accession numbers.The microarray data are available from the NCBI Gene Expression Omnibus (GEO) under series accession number GSE43768. The ChIP-seq data are available from the NCBI Sequence Read Archive under accession numbers SRR826542, SRR1011532, SRR1011533, and SRR1011534 (Bioproject PRJNA223306).
RESULTS
Inhibition of Zn(II)-dependent HDACs affects gibberellin biosynthesis in Fusarium fujikuroi.To gain insights into the role that the Zn(II)-dependent HDACs play in regulating secondary metabolism of F. fujikuroi, we studied the influence of the HDAC inhibitor TSA on the biosynthesis of gibberellins, the most prominent secondary metabolites produced by this fungus. The wild type was grown with and without TSA (1 μM) under gibberellin biosynthesis-inducing (low-nitrogen) conditions (61). After 7 days, accumulation of gibberellic acid (GA3) and its bioactive precursors, GA4 and GA7, was quantified by high-performance liquid chromatography coupled to UV detection (HPLC-UV). Gibberellin biosynthesis was decreased in TSA-treated samples by about 75% compared to the untreated control (Fig. 1). To exclude the possibility that the TSA solvent (dimethyl sulfoxide [DMSO]) was responsible for the significant reduction, the wild type was also treated with DMSO in equal amounts. However, no difference in gibberellin yield was observed compared to that of the untreated wild-type samples, suggesting that HDAC inhibition is indeed responsible for the reduced gibberellin biosynthesis (data not shown). These results suggest that the level of histone (de)acetylation is an important parameter for the regulation of gibberellins and probably other secondary metabolites in F. fujikuroi.
Inhibition of Zn(II)-dependent histone deacetylases (HDACs) results in decreased gibberellin (GA) biosynthesis in F. fujikuroi. (A) The wild type was grown in liquid synthetic medium under GA-inducing conditions (ICI medium with 6 mM glutamine) with 1 μM (red line) and without (black line) trichostatin A (TSA). After 7 days, the culture filtrate was extracted and analyzed as described in Materials and Methods. The quantified bioactive GA3 and its precursors, GA4 and GA7, are labeled in the chromatogram. mAU, milli-absorbance units. (B) Quantification of gibberellin accumulation in the WT without TSA and in TSA-treated WT samples. The amount of gibberellins was related to the biomass in order to exclude falsification in the quantification of gibberellin biosynthesis due to altered growth behavior. The experiment was performed in triplicate; mean values and standard deviations (referring to overall gibberellin accumulation) are given. Asterisks above the bars denote significant differences in the measurement of the TSA-treated samples compared to the untreated WT. ∗∗, P < 0.01.
Phylogenetic analysis reveals four classical HDACs in the genome of Fusarium fujikuroi.In contrast to the genome of Saccharomyces cerevisiae, which contains five “classical” HDAC-encoding genes, Hda1, Hos1, Hos2, Rpd3, and Hos3 (62, 63), filamentous fungi in general harbor four HDAC-encoding genes that are highly conserved throughout the different species (64). HDAC proteins in F. fujikuroi were identified by BLASTp analyses (65) in F. fujikuroi wild-type strain IMI58289 by using the protein sequences for HDA-1, HDA-2, HDA3, and HDA-4 from Neurospora crassa (GenBank accession numbers XP_956974, XP_964451, XP_964367, and XP_961839) (49). Accordingly, the proteins in F. fujikuroi were designated FfHda1 (FFUJ_09787), FfHda2 (FFUJ_01551), FfHda3 (FFUJ_00456), and FfHda4 (FFUJ_08772). Phylogenetic analysis using protein sequences of studied HDACs from other fungi grouped the F. fujikuroi HDACs into the two known classes: FfHda2 and FfHda3 belong to class I HDACs, and FfHda1 and FfHdF4 belong to class II HDACs (see Fig. S1 in the supplemental material). As in other filamentous fungi, Hos1 from S. cerevisiae has no homolog in F. fujikuroi.
Histone deacetylase activity of single HDACs in Fusarium fujikuroi.In order to determine the contribution of each HDAC to the overall HDAC activity of the cell and to study their role in growth and secondary metabolism in F. fujikuroi, we generated single ffhda1, ffhda2, and ffhda4 deletion mutants. Transformation of the respective knockout fragments with the F. fujikuroi wild-type strain gained at least three independent mutants for each gene. Correct integration of the resistance cassette was subsequently verified by diagnostic PCR (data not shown) and Southern blot analyses (see Fig. S2 in the supplemental material). Several attempts to generate homokaryotic ffhda3 deletion mutants failed, suggesting an essential role of FfHda3 in growth and development of F. fujikuroi. To study the contribution of each HDAC to the total HDAC activity of the cell, an HDAC activity assay was performed. The assay detects deacetylation capacity by utilizing a short acetylated peptide that is deacetylated by most HDAC enzymes. The wild type and the Δffhda1, Δffhda2, and Δffhda4 single deletion mutants were grown in liquid synthetic medium with 6 mM glutamine, which was also used for TSA treatment (Fig. 1). While deletion of ffhda4 showed only minor differences in total HDAC activity, deletion of ffhda1 and ffhda2 resulted in 64% and 25% reductions of total HDAC activity, respectively, indicating that FfHda1 is the major HDAC in F. fujikuroi (Fig. 2). Therefore, the biological role of FfHda1 was analyzed in more detail.
Contribution of single Zn(II)-dependent histone deacetylases (HDACs) in Fusarium fujikuroi to the overall HDAC activity. The wild type (WT) and the three single HDAC deletion mutants, Δffhda1, Δffhda2, and Δffhda4, were grown for 24 h in synthetic ICI medium under gibberellin-inducing conditions (6 mM glutamine). A crude nuclear extract was isolated as described in Materials and Methods and directly used for quantification of HDAC activity. A HeLa cell nuclear extract served as a positive control (+), and for the negative control (−), the HeLa cell nuclear extract was directly incubated with the HDACi trichostatin A (TSA) according to the manufacturer's instructions. HDAC activity is given in pmol min−1 mg−1. Experiments were done in quadruplicate. Mean values and standard deviations are given. Asterisks above the bars denote significant differences in the measurements of the indicated strains compared to the WT. ∗∗, P < 0.01; ∗∗∗, P < 0.001.
Microarray analysis with the Δffhda1 mutant reveals differential expression of several secondary-metabolite gene clusters in Fusarium fujikuroi.To determine the overall impact of FfHda1 on expression of secondary-metabolite biosynthetic genes in F. fujikuroi, microarray analyses were performed with the wild type and the Δffhda1 mutant. Both strains were grown for 72 h in liquid synthetic medium under nitrogen-limiting (6 mM glutamine) and nitrogen-sufficient (60 mM glutamine) conditions, as all so-far-studied secondary-metabolite gene clusters are regulated by nitrogen availability in F. fujikuroi (17). Overall, 22 and 242 genes were significantly (2-fold change in expression; 95% confidence interval) up- and downregulated under low-nitrogen conditions, respectively, and 392 and 493 genes were up- and downregulated under high nitrogen conditions, respectively, in the ffhda1 deletion mutant (data not shown). Surprisingly, only some of the 45 putative secondary-metabolite gene clusters present in the genome of F. fujikuroi (17) showed differential expression in the Δffhda1 mutant compared to the wild type (see Table S1 in the supplemental material). Noteworthy, in some cases, all cluster genes showed an altered expression pattern, while in others, only a few genes or even only one single gene of a putative cluster was affected by deletion of ffhda1. Altogether, four and six gene clusters showed differential expression under low- and high-nitrogen conditions, respectively (see Table S1 in the supplemental material) (85, 86). Four genes of the gibberellin gene cluster (FFUJ_14331, FFUJ_14335, FFUJ_14336, and FFUJ_14337) and four genes of the fumonisin gene cluster (FFUJ_09243, FFUJ_09246, FFUJ_09252, and FFUJ_09254) were significantly downregulated, while only one gene of the cryptic PKS13 mini-gene cluster (FFUJ_12020), consisting most likely only of one to two genes (19), was upregulated under these low-nitrogen conditions (see Table S1 in the supplemental material). Under nitrogen-sufficient conditions, three genes out of the fusaric acid gene cluster (FFUJ_02105, FFUJ_02107, and FFUJ_02109) and the key enzyme-encoding gene of the cryptic NRPS2 (FFUJ_04614) gene cluster with a so-far-unknown function (17) were downregulated (see Table S1 in the supplemental material). A significant upregulation under high-nitrogen conditions was found only for fus5 (FFUJ_10054), one of the fusarin C biosynthetic genes (see Table S1 in the supplemental material). An interesting phenomenon was observed for the bikaverin gene cluster: while the bikaverin genes (FFUJ_06742 to FFUJ_06747) were expressed only under low-nitrogen conditions in the wild type, they were slightly downregulated under these normally inducing conditions (66, 67) but significantly upregulated under normally repressing (high-nitrogen) conditions in the Δffhda1 mutant (see Table S1 in the supplemental material).
To verify the altered expression data at the production level, accumulation of the respective secondary metabolites was quantified by HPLC coupled to UV and diode array detection (HPLC-UV/DAD). In accordance with the expression data, levels of production of gibberellins and bikaverin were reduced to about 10% and 25% of the wild-type level in the ffhda1 deletion mutant, respectively, under nitrogen-limiting conditions (Fig. 3A). Complementation of the Δffhda1 mutant with the ffhda1 wild-type gene, driven by its native promoter, fully restored wild-type production levels (see Fig. S3 in the supplemental material). No significant differences were observed regarding fusaric acid and fusarin production under high-nitrogen (60 mM glutamine) conditions (data not shown), although some of the biosynthetic genes were affected in the Δffhda1 mutant (see Table S1 in the supplemental material). However, when grown with adequate amounts (120 mM) of sodium nitrate instead of glutamine, generating optimal alkaline culture conditions for fusaric acid biosynthesis (E.-M. Niehaus and B. Tudzynski, unpublished data), its production was significantly reduced in the mutant (Fig. 3A).
Deletion and constitutive expression of ffhda1 have diverse effects on secondary-metabolite biosynthesis in Fusarium fujikuroi. (A) The wild type (WT) and the Δffhda1 mutant were grown for 7 days under biosynthesis-inducing conditions for the following five secondary metabolites: bikaverin (BIK) (6 mM glutamine), fusarubins (FSR) (6 mM sodium nitrate), gibberellic acids (GA) (6 mM glutamine), fusaric acid (FU) (120 mM sodium nitrate), and fusarins (FUS) (60 mM glutamine). After 7 days, a culture filtrate was taken, and samples for HPLC analyses were prepared as described in Materials and Methods. Experiments were performed in triplicate for each metabolite. Accumulation of metabolites was related to the biomass to exclude falsifications in the quantification due to an altered growth behavior of the mutant compared to the WT. Production of the WT was set to 100%. Mean values and standard deviations are given. Asterisks above the bars denote significant differences in the measurements of the indicated strains compared to the WT. ∗∗, P < 0.01; ∗∗∗, P < 0.001. (B) Detection of bikaverin under normally repressing conditions (60 mM glutamine) in the Δffhda1 mutant. (C) Northern blot analysis of the WT and the Δffhda1 mutant after 3 days of growth in ICI medium under conditions of nitrogen starvation (6 mM glutamine) and nitrogen sufficiency (60 mM glutamine). The biosynthetic gene bik2 (O-methyltransferase) was used for probing. (D) The WT and an overexpression mutant (gpd::ffhda1) were grown under biosynthesis-inducing conditions as described above for panel A. Experiments were performed in triplicate for each metabolite. Mean values and standard deviations are given. Production of the WT was set to 100%. Asterisks above the bars denote significant differences in the measurements of the indicated strains compared to the WT. ∗, P < 0.05; ∗∗, P < 0.01; ∗∗∗, P < 0.001.
The most striking result of the microarray analysis was the unexpected deregulation of the bikaverin genes: high expression levels under repressing conditions and low expression levels under normally inducing conditions. This contrasting expression pattern was confirmed by Northern blot and HPLC-DAD analyses (Fig. 3B). Recently, we have shown that the biosynthesis of the second group of red pigments, the perithecial pigments fusarubins, is regulated opposingly to bikaverin regarding the optimal pH or the impact of the Gα subunit FfG1 and the adenylyl cyclase FfAc (29, 68). To show whether fusarubins are produced instead of bikaverin in the mutant, we analyzed the accumulation of this pigment under both bikaverin-inducing (6 mM glutamine at acidic pH) and fusarubin-inducing (6 mM sodium nitrate at alkaline pH) conditions (29). However, fusarubins were not detectable under repressing conditions (data not shown) and almost not detectable under inducing conditions (Fig. 3A), indicating that the presence of FfHda1 is essential for the formation of the perithecial pigments. Since deletion of ffhda1 resulted in downregulation of gibberellins, bikaverin, fusarubins, and fusaric acid under normally inducing conditions, we investigated whether the constitutive activation of FfHda1 might result in an activation of these secondary metabolites. Therefore, we generated an ffhda1 overexpression mutant by exchanging the native ffhda1 promoter for the strong constitutively active gpd promoter from A. nidulans in the wild-type background. Northern blot analysis verified significant overexpression of ffhda1 compared to the wild type (data not shown). Surprisingly, expression of gibberellin, bikaverin, fusarubin, and fusaric acid biosynthetic genes and accumulation of the respective secondary metabolites did not reveal an upregulation of any pathway but instead revealed the same tendency as in the ffhda1 deletion mutant: gibberellin, bikaverin, fusarubin, and fusaric acid biosynthesis levels were reduced to about 20%, 20%, 60%, and 45% compared to the wild-type levels, respectively (Fig. 3D). The only difference from the deletion mutant was that wild-type-like regulation of bikaverin gene expression and biosynthesis were restored, as bikaverin was not observed anymore in nitrogen sufficiency (data not shown).
To determine the contribution of the second class II HDAC, FfHda4, we analyzed secondary-metabolite biosynthesis in the Δffhda4 mutant. The wild type and the Δffhda4 mutant were grown under biosynthesis-favoring conditions for the five well-studied secondary metabolites in F. fujikuroi (gibberellins, bikaverin, fusarubins, fusaric acid, and fusarins) (29, 61, 66, 67, 69, 70; Niehaus and Tudzynski, unpublished). However, no significant differences were observed for the Δffhda4 mutant compared to the wild type, suggesting that FfHda4 is not involved in the regulation of these secondary metabolites (see Fig. S4A in the supplemental material). To further investigate if FfHda1 and FfHda4 have some overlapping functions in F. fujikuroi, as previously shown for A. nidulans (22), a Δffhda1 Δffhda4 double mutant was generated. However, the secondary-metabolite profile of the Δffhda1 Δffhda4 double mutant resembled that of the Δffhda1 single mutant (see Fig. S4B and S4C in the supplemental material), underlining that FfHda4 is dispensable for secondary metabolism, in contrast to its homolog in A. nidulans.
In summary, both ffhda1 deletion and overexpression resulted in significantly reduced amounts of gibberellins, bikaverin, and fusarubins and slightly reduced amounts of fusaric acid under optimal production conditions, while the second class II HDAC, FfHda4, does not play an important role in biosynthesis of the secondary metabolites studied here. Furthermore, deletion of ffhda1 resulted not only in downregulation of bikaverin gene expression under inducing conditions but also in derepression under normally repressing, nitrogen-sufficient conditions.
ChIP-seq analysis reveals distinct alterations of histone acetylation in the Δffhda1 mutant.Recently, we have shown that highly expressed secondary-metabolite gene clusters (e.g., gibberellin and bikaverin gene clusters) are enriched for H3K9 acetylation marks in F. fujikuroi (17). To study whether the reduction of secondary-metabolite production in the Δffhda1 mutant is due to alterations of the H3K9 acetylation state, we performed chromatin immunoprecipitation assays followed by high-throughput sequencing (ChIP-seq) using H3K9ac-specific antibodies. To do so, the wild type and the Δffhda1 mutant were grown for 3 days under nitrogen-limiting (6 mM glutamine) and nitrogen-sufficient (60 mM glutamine) conditions, thus resembling the conditions of the microarray approach. Interestingly, H3K9 acetylation at the gibberellin gene cluster was reduced upon deletion of ffhda1, while the same region was highly acetylated in the wild type under gibberellin-inducing conditions (6 mM glutamine). Acetylation was absent under conditions of nitrogen sufficiency (60 mM glutamine) in both strains (Fig. 4B). A different scenario was observed for the bikaverin gene cluster, where H3K9 acetylation was even increased under biosynthesis-favoring conditions (6 mM glutamine) compared to the wild type, despite the reduction of bikaverin production due to the deletion of ffhda1 (Fig. 4C). Moreover, the bikaverin gene cluster was also enriched for H3K9 acetylation under repressing conditions (60 mM glutamine) in the Δffhda1 mutant, thus explaining the unexpected upregulation of bikaverin genes and concomitant bikaverin formation under these conditions (Fig. 4C). Subsequent ChIP-qPCR experiments for cps/ks and bik5, encoding the ent-copalyl diphosphate synthase/kaurene synthase and the bikaverin-specific transcription factor, respectively, both essential for metabolite biosynthesis, confirmed these results (Fig. 4D and E).
Chromatin immunoprecipitation reveals distinct alterations in histone acetylation at the gibberellin (GA) and bikaverin (BIK) gene clusters in Fusarium fujikuroi. The wild type (WT) and the Δffhda1 mutant were grown in synthetic ICI medium with either 6 mM (nitrogen starvation) or 60 mM (nitrogen sufficiency) glutamine. After 3 days, the mycelium was used for ChIP as described in Materials and Methods. ChIP assays were performed by using an H3K9 acetylation-specific antibody. Subsequent next-generation sequencing was carried out on an Illumina GAII genome analyzer by single-end 36- to 50-nt sequencing, and quantitative real-time PCR was carried out as described in Materials and Methods. (A) Chromosomal location of the gibberellin and the bikaverin gene clusters. (B and C) ChIP-seq analysis of the GA gene cluster (B) and the bikaverin gene cluster (C) in the ffhda1 deletion mutant compared to the WT under both nitrogen conditions (6 mM glutamine and 60 mM glutamine). Experiments were done twice, each in duplicate. Red lines indicate significantly enriched reads. (D and E) Quantification of precipitated DNA at the gibberellin (D) and the bikaverin (E) gene clusters. The gibberellin biosynthetic gene cps/ks and the pathway-specific transcription factor bik5 were chosen for verification of the ChIP-seq data. In each case, the amount of precipitated DNA in the WT under low-nitrogen conditions (6 mM glutamine) was arbitrarily set to 1. Mean values and standard deviations are given.
In summary, deletion of ffhda1 resulted in reduced H3K9 acetylation across the gibberellin gene cluster under inducing conditions, explaining the significantly decreased gibberellin production levels. In contrast, H3K9 acetylation marks are enriched at the bikaverin gene cluster under both repressing and inducing conditions, possibly leading to bikaverin gene expression under repressing conditions in the Δffhda1 mutant.
The class I HDAC-encoding gene ffhda2 functions as an activator for secondary metabolism in Fusarium fujikuroi.Besides FfHda1, the class I HDAC FfHda2 contributes to overall HDAC activity: deletion of ffhda2 resulted in a 25% reduction of total HDAC activity compared to that of the wild type (Fig. 2). To determine if FfHda2 is also involved in regulation of secondary metabolite biosynthesis in F. fujikuroi, formation of gibberellins, bikaverin, fusarubins, fusaric acid, and fusarins was analyzed by HPLC-UV/DAD. Four out of the five analyzed secondary metabolites were significantly affected by the deletion of ffhda2: while the amounts of gibberellins, bikaverin, fusaric acid, and fusarins were decreased to about 26%, 35%, 0.1%, and 20% of the amounts in the wild type, respectively, fusarubin biosynthesis was wild-type-like (Fig. 5A). Northern blot analyses of the wild type and the Δffhda2 mutant confirmed the obtained results (data not shown). Similarly to the upregulation of bikaverin genes under repressing, high-nitrogen conditions in the Δffhda1 mutant, fusarubin gene expression and production were observed under repressing, acidic pH conditions (6 mM glutamine) in the Δffhda2 mutant, conditions that repress fusarubin biosynthesis in the wild type (29) (Fig. 5B). Complementation of the Δffhda2 mutant with the wild-type gene copy driven by its native promoter restored the wild-type phenotype with regard to secondary metabolism (see Fig. S5 in the supplemental material).
FfHda1 and FfHda2 have common and distinct effects on secondary-metabolite biosynthesis in Fusarium fujikuroi. (A) The wild type (WT) and the Δffhda2 mutant were grown for 7 days under biosynthesis-inducing conditions for the following five secondary metabolites: bikaverin (BIK) (6 mM glutamine), fusarubins (FSR) (6 mM sodium nitrate), gibberellic acids (GA) (6 mM glutamine), fusaric acid (FU) (120 mM sodium nitrate), and fusarins (FUS) (60 mM glutamine). Experiments were performed in triplicate for each secondary metabolite. Accumulation of the respective metabolites was related to the biomass to exclude falsifications in the quantification due to an altered growth behavior of the mutant compared to the WT. The WT value was set to 100%. Mean values and standard deviations are given. Asterisks above the bars denote significant differences in the measurements of the indicated strains compared to the WT. ∗∗, P < 0.01; ∗∗∗, P < 0.001. (B and C) The WT and the Δffhda2 mutant were grown for 4 days (in the case of Northern blot analysis) and 7 days (for HPLC analysis) under fusarubin biosynthesis-repressing conditions (acidic pH) (ICI medium with 6 mM glutamine). Accumulation of fusarubins and bikaverin in the liquid culture is highlighted by red boxes. The biosynthetic genes fsr1 (PKS) and fsr2 (O-methyltransferase) were used for probing in the Northern blot. (D) The WT and the Δffhda1 Δffhda2 mutant were grown for 7 days under biosynthesis-inducing conditions and used for HPLC analysis as described above for panel A. Experiments were done in triplicate for each secondary metabolite. Standard deviations are given. Production of the WT was set to 100% for each secondary metabolite investigated. Asterisks above the bars denote significant differences in the measurements of the indicated strains compared to the WT. ∗∗, P < 0.01; ∗∗∗, P < 0.001.
To determine if both FfHda1 and FfHda2 affect secondary metabolism independently of each other, Δffhda1 Δffhda2 double mutants were generated and analyzed for their potential to produce the five secondary metabolites in comparison to the wild type and the single deletion mutants. All strains were grown under optimal conditions for production of gibberellin, bikaverin, fusarubin, fusaric acid, and fusarins and analyzed by HPLC-UV/DAD. Deletion of both ffhda1 and ffhda2 showed additive and distinct phenotypes compared to the single deletion mutants: levels of accumulation of gibberellins and bikaverin were decreased to about 2% and 31%, respectively (Fig. 5D), and bikaverin formation was again observed under repressing (nitrogen-sufficient) conditions, similarly to the Δffhda1 single mutant and the Δffhda1 Δffhda4 double mutants (data not shown). On the one hand, fusarubin formation was dramatically reduced in the Δffhda1 Δffhda2 double mutant under inducing conditions (Fig. 5D), resembling the Δffhda1 mutant. On the other hand, fusarubins accumulated under repressing conditions (6 mM glutamine), thereby resembling the phenotype of the Δffhda2 mutant (Fig. 5B and C). Furthermore, the nitrogen-induced secondary metabolites fusaric acid and fusarins were significantly reduced to about 0.1% and 8%, respectively, as in the Δffhda2 mutant (Fig. 5D).
Influence of HDACs on vegetative growth and asexual development in Fusarium fujikuroi.To study the involvement of the three HDACs in hyphal growth, we performed plate assays on complete medium (CM) and minimal medium (CD). While deletion of ffhda1 and ffhda2 showed wild-type-like hyphal growth, deletion of ffhda4 resulted in growth reduced to about 88% and 70% on CM and CD medium, respectively, compared to the growth of the wild type (see Fig. S6 in the supplemental material). In addition, we analyzed a possible involvement of the three Zn(II)-dependent HDACs in asexual spore formation. While the Δffhda1 mutant produced wild-type-like levels of microconidia, their formation was significantly reduced to about 56% compared to the wild type in the ffhda4 deletion mutant (see Fig. S6 in the supplemental material). Contrary to this, microconidium formation of the Δffhda2 mutant was increased about 3.6 times compared to that of the wild type (see Fig. S6 in the supplemental material). These results indicate that FfHda1 and FfHda2 are dispensable for hyphal growth but that FfHda2 is a repressor of microconidium formation, while FfHda4 functions as an activator for both hyphal growth and asexual development.
Localization of FfHda1 and FfHda2 within the cell.In order to deacetylate histones, HDACs must be localized to the nucleus. However, an ever-increasing number of nonhistone proteins (e.g., transcription factors) have been shown to be subject to acetylation and deacetylation by HATs and HDACs (71), thus implying that HDACs can also be localized in the cytoplasm at some point. In mammals, class I HDACs are found mostly in the nucleus, while class II HDACs are able to shuttle between the nucleus and the cytoplasm upon specific as-yet-unknown signals (19). To determine the subcellular localization of FfHda1 and FfHda2 in F. fujikuroi, both genes were fused to gfp and driven by either the native or the strong A. nidulans oliC promoter. Subsequent transformation of the resulting ffhda1::gfp or ffhda2::gfp plasmid into the Δffhda1 or Δffhda2 background, respectively, resulted in several transformants, which were analyzed by epifluorescence microscopy. Constructs were functional and showed secondary-metabolite patterns comparable to those of either the wild type or the respective overexpression mutants (data not shown).
No fluorescence was obtained with fusion constructs driven by the native ffhda1 and ffhda2 promoters, probably due to low expression levels of the respective genes (data not shown). However, strong GFP signals were observed for the constructs with the oliC promoter. Both FfHda1::GFP and FfHda2::GFP predominantly localized to the nucleus, as also previously shown for HDF1 in F. graminearum and its homolog Hos2 in Magnaporthe oryzae (24, 72). In addition, a faint GFP signal was also visible in the cytoplasm of vegetative hyphae in both mutants (see Fig. S7 in the supplemental material). Whether the GFP signal in the cytoplasm is indeed due to the ability of both HDACs to shuttle between the nucleus and the cytoplasm, thereby being able to deacetylate both histones and nonhistone proteins, or if the presence of the proteins in the cytoplasm is due to the exchange of the native promoter for the constitutive oliC promoter awaits proof.
Importance of FfHda1 and FfHda2 for bakanae disease on rice.To determine a possible impact of FfHda1, FfHda2, and FfHda4 on virulence, rice seedlings were infected with the wild type and the generated HDAC mutants. The wild type and the GA3 standard were used as positive controls, and H2O was used as a negative control. After 7 days postinfection, the rice plants were screened for typical infection symptoms, such as gibberellin-induced elongation of internodes. While elongation of the internodes was reduced in both the Δffhda1 and Δffhda2 mutants compared to the wild type, elongation of rice plants was even more reduced and looked similar to the H2O control when infected with the Δffhda1 Δffhda2 mutant. No differences were observed between the Δffhda4 mutant and the wild type (Fig. 6). These results indicate the importance of both HDACs FfHda1 and FfHda2 for gibberellin-induced bakanae disease.
FfHda1 and FfHda2 are both required for gibberellin-induced bakanae disease on rice. Rice seedlings and the indicated fungal strains, the wild type (WT) and the Δffhda1, Δffhda2, Δffhda4, and Δffhda1 Δffhda2 mutants, were cocultivated for 7 days at 28°C with 80% humidity, as described in Materials and Methods. Noninfected rice seedlings and rice seedlings treated with 100 ppm of the bioactive gibberellic acid GA3 served as negative and positive controls, respectively. For determination of bakanae symptoms on grown rice plants, the length between internodes was measured (highlighted by black arrows). (A) Pictures of rice plants after 7 days of cocultivation with the indicated fungal strains. (B) Measurement of internode elongation. The experiment was performed in triplicate. Mean values and standard deviations are given. Asterisks above the bars denote significant differences in the measurements of the indicated strains compared to the WT. ∗∗∗, P < 0.001.
DISCUSSION
A useful approach to determine the general impact of histone acetylation in an organism is the inhibition of HDACs by specific inhibitors (HDACi) such as TSA. Studies of A. alternata and P. expansum revealed an increased production of numerous yet unidentified secondary metabolites upon TSA treatment (22). Similarly, inhibition with other HDACi, including suberohydroxamic acid, valproic acid, and suberoylanilide hydroxamic acid (SAHA), resulted in an increased production of various secondary metabolites in several fungi (23). Treatment with SAHA revealed the production of the novel secondary metabolite nygerone A in A. niger (73) and expression of the orsellinic acid gene cluster in A. nidulans, which is not transcribed under standard laboratory conditions (8). Conversely, addition of the HAT inhibitor anacardic acid led to orsellinic acid gene silencing (8). These results underline the well-accepted model that chemical or genetic inactivation of HDACs is associated with gene transcription due to hyperacetylation of the respective DNA sequence. However, this is not always the case. Thus, in yeast, TSA treatment resulted in both activation and inactivation of gene expression of only a few out of 340 genes analyzed (74). In F. fujikuroi, treatment with TSA resulted in decreased gibberellin biosynthesis (this study), suggesting that regulation of secondary metabolism by histone (de)acetylation is more complex. Genome-wide microarray analyses with a deletion mutant of the major HDAC in F. fujikuroi, Δffhda1, revealed altered expression patterns for some, but not all, gene clusters involved in secondary metabolite biosynthesis. Subsequent HPLC analyses of the respective products confirmed these findings: production of gibberellin, bikaverin, fusarubin, and fusaric acid is reduced under optimal production conditions in the Δffhda1 mutant, while bikaverin formation is induced under normally repressing conditions. These findings suggest that contrary to the general assumption, FfHda1 is needed for sufficient transcription of the respective genes. A similar finding was previously reported for the FfHda1 homolog HdaA in A. fumigatus: four NRPS-encoding genes, including gliP, involved in the formation of gliotoxin, were downregulated upon deletion of hdaA (6). Furthermore, microarray and genetic studies with HDAC-encoding genes (HosB and Rpd3) demonstrated that HDACs are required for both transcriptional activation and silencing in yeast, and deletion of the class I HDAC-encoding gene HDF1 (the homolog of ffhda2) resulted in reduced accumulation of deoxynivalenol in F. graminearum (4, 5, 24). However, while the mechanistic background leading to silencing of secondary metabolite gene clusters by HDAC activity is quite well understood, not much is known about the mechanism leading to HDAC-mediated gene activation.
Two different mechanisms are responsible for downregulation of gibberellin and bikaverin genes in the Δffhda1 mutant.Histone acetylation was found to be enriched in the chromatin landscape of various secondary-metabolite gene clusters under production conditions. Thus, the aflatoxin gene cluster in A. parasiticus was found to be enriched for H3K14 acetylation (13), activation of the sterigmatocystin and orsellinic acid gene clusters depends on SAGA/Ada-mediated acetylation of both H3K9 and H3K14 in A. nidulans (8, 15), and in F. fujikuroi, some secondary metabolite gene clusters, e.g., those for gibberellin and bikaverin biosynthesis, were also found to be enriched for H3K9 acetylation under optimal production conditions (17). Remarkably, comparative ChIP analysis of the wild type and the Δffhda1 mutant using the H3K9 acetylation-specific antibody did not reveal the expected increase but revealed a decrease in H3K9 acetylation across the gibberellin gene cluster in the Δffhda1 mutant, thereby providing an explanation for the unexpected downregulation of gibberellin gene expression and biosynthesis. To rule out the possibility that the loss of one HDAC results in overexpression of another one, e.g., ffhda2, leading to hyperdeacetylation and concomitant downregulation of the gibberellin biosynthetic genes, expression of ffhda2 was analyzed in Δffhda1 mutants. No differences were observed, disproving overcompensation of the loss of ffhda1 by FfHda2 (data not shown). Noteworthy, to this point, we cannot rule out the possibility that the second class I HDAC, FfHda3, or class III sirtuin proteins cause the hypoacetylation at the gibberellin gene cluster.
Whether the reduced production of secondary metabolites in the hdaA and HDF1 deletion mutants in A. fumigatus and F. graminearum, respectively, is also due to hypoacetylation in the chromatin landscape of the downregulated gene clusters remains to be elucidated. In A. fumigatus, overexpression of hdaA resulted in increased production of gliotoxin, suggesting that HdaA is directly required for GLI gene expression (6). Contrary to this, constitutive activation of FfHda1 in F. fujikuroi did not result in upregulation but resulted in downregulation of secondary-metabolite production, including gibberellin biosynthesis. A similar effect was observed for cphos2 in Claviceps purpurea: both deletion as well as overexpression of cphos2 (homolog of ffhda2) resulted in a decreased accumulation of ergot alkaloids (L. Neubauer and P. Tudzynski, personal communication). However, nothing is known regarding the underlying mechanism. The lack of H3K9 hyperacetylation in the chromatin landscape of the gibberellin gene cluster in the Δffhda1 mutant suggests that in this case, not histones but rather nonhistone proteins are the main target of FfHda1, e.g., transcription factors or coactivators. One possible explanation is that FfHda1 activity is required for HAT-mediated histone acetylation and, thus, initiation of transcription. HATs are recruited to their reaction sites via global transcription factors that are localized to these genomic regions upon environmental stimuli. Sufficient histone acetylation by recruited HATs then enables binding of chromatin-remodeling complexes, which further open the chromatin landscape, thus exposing the DNA sequence of interest. As a consequence, a pathway-specific transcription factor can bind to the DNA sequence, which in turn recruits the entire transcriptional machinery essential for subsequent gene transcription. The histone acetylation state is balanced due to the opposing action of HATs and HDACs (Fig. 7A). Hyperacetylation of such a nonhistone factor due to deletion of ffhda1 might result in steric hindrance or a decrease of the DNA binding affinity, thereby leading to improper binding and subsequently to failure of HAT recruitment and thus hypoacetylation (75) (Fig. 7A). Lack of histone acetylation at the gibberellin gene cluster in turn might hinder binding of chromatin-remodeling complexes and a pathway-specific transcription factor(s) and, thus, recruitment of the whole transcriptional machinery needed for sufficient gene activation. A similar hypothesis was proposed previously for the unexpected downregulation of COX-2 expression in human tumor cells. ChIP analysis indicated that expression of the transcription factor-encoding gene c-jun, which is required for COX-2 transcription, is suppressed by TSA treatment due to an inhibition of recruitment of the preinitiation complex to the c-jun promoter (75, 76). However, nothing is known about the mechanism of HDAC-mediated gene activation in fungi so far. Consequently, overexpression of FfHda1 would then result in effective HAT recruitment and subsequent acetylation of the gibberellin gene cluster. However, the constitutively active FfHda1 constantly removes added acetyl groups in the vicinity of the gibberellin gene cluster, thereby lowering gibberellin gene expression levels and subsequent biosynthesis (Fig. 7A). This hypothesis would in part explain why gibberellin accumulation in the ffhda1 deletion mutant (∼10% of the wild type) is decreased compared to that in the overexpression mutant (∼50%), since gene transcription is probably decreased due to less acetylation. To our knowledge, this is the first report demonstrating that deletion of an HDAC-encoding gene results in hypoacetylation of a secondary-metabolite gene cluster and, furthermore, that both deletion and overexpression may result in gene silencing.
Hypothetical model for decreased gibberellin (A) and bikaverin (B) biosynthesis in the Δffhda1 and gpd::ffhda1 mutants of Fusarium fujikuroi. (A) For sufficient gene activation in the wild type (WT), an as-yet-unknown transcription factor (TF) is deacetylated upon reception of an environmental stimulus by FfHda1 (1). Only the deacetylated transcription factor, e.g., the GATA transcription factor AreA, which is required for sufficient gibberellin gene expression, can then bind to the gibberellin (GA) gene cluster, resulting in the recruitment of the histone acetyltransferase (HAT) complex (2). The HAT is now able to acetylate histone proteins in the vicinity of the gibberellin gene cluster (3), resulting in accessibility of the genomic region for chromatin-remodeling complexes that further open the chromatin landscape, enabling the binding of further pathway-specific transcription factors and the transcription machinery, thus leading to gene activation. The concomitant presence of FfHda1 at the gibberellin gene cluster keeps the balance between histone acetylation and deacetylation (4). Upon deletion of ffhda1, the global transcription factor cannot be deacetylated anymore, resulting in hypoacetylation and the inability to bind to the cluster due to steric hindrance, thus resulting in gene silencing (1). When ffhda1 is overexpressed, the gibberellin-specific transcription factor is deacetylated (1), localizes to the gibberellin gene cluster, and recruits the HAT (2). While the HAT acetylates the histone in the vicinity of the gibberellin gene cluster, leading finally to gene activation (3), overrepresentation of FfHda1 deacetylates the histones (4), resulting putatively in hypoacetylation and thereby in downregulation of gene expression. (B) A similar mechanism of gene transcription happens at the bikaverin gene cluster. A bikaverin-specific transcription factor binds to the bikaverin gene cluster (1) and recruits a HAT (2), which then acetylates histones (3) in the vicinity of the bikaverin gene cluster, thereby allowing accessibility of the genomic DNA for chromatin-remodeling complexes that further open the genomic region. This enables binding of further transcription factors, e.g., the pathway-specific regulator BIK5, and subsequently the transcriptional machinery, finally resulting in gene activation. Hyperacetylation of the histones due to HAT activity is counteracted by HDAC activity (4). The global transcription factor is not deacetylated by FfHda1 prior to binding; however, whether either another HDAC is involved in this process or the transcription factor can bind in the acetylated state or is not subject to acetylation in the first place is not known (indicated by “?”). Therefore, acetylation of histone in the vicinity of bikaverin genes still takes place upon deletion of ffhda1 (1 to 3); however, histones are not deacetylated anymore to create a balanced state between HAT and HDAC activities, thereby leading to hyperacetylation (4). The lack of ffhda1 possibly also results in hyperacetylation of histones in the vicinity of the hypothetical bikaverin-specific repressor, resulting in its activation (5). The repressor then binds to the bikaverin gene cluster, thereby leading to gene silencing (6). Similar to the gibberellin gene cluster, overexpression of ffhda1 possibly results in hypoacetylation and thus downregulation of bikaverin gene expression due to overrepresentation of FfHda1.
While deletion of ffhda1 resulted in hypoacetylation at the gibberellin gene cluster, the bikaverin gene cluster was found to be hyperacetylated under both biosynthesis-inducing as well as -repressing conditions, suggesting that in this case, histone proteins are the main targets of FfHda1. Surprisingly, bikaverin gene expression and product accumulation were significantly reduced under optimal production conditions, despite H3K9 hyperacetylation. Moreover, ChIP-seq analysis revealed H3K9 hyperacetylation across the bikaverin gene cluster under normally repressing conditions in the Δffhda1 mutant, explaining the unexpected bikaverin biosynthesis under these high-nitrogen conditions. Thus, H3K9 hyperacetylation of the chromatin landscape seems to be essential but not sufficient for expression of bikaverin genes. We propose that in addition to hyperacetylation at the bikaverin gene cluster, histones in the vicinity of a gene that encodes a putative bikaverin-specific transcriptional repressor is hyperacetylated upon deletion of ffhda1 and thereby activated under production conditions but not under normally repressing conditions. Repressor binding to the bikaverin gene cluster in turn hinders binding of activating transcription factors, e.g., BIK5, and the transcriptional machinery, resulting in decreased bikaverin biosynthesis. The global transcription factor is not deacetylated by FfHda1 prior to binding; however, whether either another HDAC is involved in this process or the transcription factor can bind in the acetylated state or is not subject to acetylation in the first place remains to be elucidated. A similar hypothesis was proposed for the unexpected downregulation of alkaloid biosynthesis in C. purpurea or deoxynivalenol production in A. nidulans (77, 78).
Acetylation of histones and nonhistone proteins depends on available acetyl coenzyme A (acetyl-CoA) pools generated by acetyl-CoA synthetase activity during carbon metabolism (79). However, acetyl-CoA is also required for de novo synthesis of fatty acids during primary metabolism (80). Galdieri and Vancura (81) showed that decreased expression of the acetyl-CoA carboxylase gene ACC1, responsible for the conversion of acetyl-CoA into malonyl-CoA, results in hyperacetylation of histones and nonhistone proteins and subsequently in altered transcription. Thus, a limited acetyl-CoA pool might also explain the altered acetylation pattern: some gene clusters are hypoacetylated (gibberellin gene cluster), while others are hyperacetylated (bikaverin gene cluster). This hypothesis is supported by our finding that overall H3K9 acetylation and H3 acetylation were not significantly altered in the F. fujikuroi Δffhda1 mutant compared to the wild type (see Fig. S8 in the supplemental material).
Noteworthy, not all regulatory mechanisms are overruled in the Δffhda1 mutant: while nitrogen repression of bikaverin genes by a putative, as-yet-unknown repressor was circumvented by the Δffhda1 mutant, PacC-mediated repression of bikaverin genes (67) was maintained under alkaline pH conditions (6 and 120 mM sodium nitrate). In the case of gibberellin genes, AreA-mediated expression under low-nitrogen conditions was abolished under inducing low-nitrogen conditions in the Δffhda1 mutant, but repression of gibberellin genes under high-nitrogen conditions was maintained. Fungal GATA factors are known for the recruitment of HAT complexes (82). Berger and colleagues (83) demonstrated that H3K9K14 acetylation parallels AreA occupancy at promoters during transcription and, furthermore, that AreA is required for sufficient niaD and niiA gene activation by the specific transcription factor NirA in A. nidulans. To determine whether AreA is also involved in the recruitment of HAT complexes in F. fujikuroi (e.g., at the AreA-dependent gibberellin gene cluster), comparative ChIP experiments using an areA-deficient mutant are currently in progress.
In addition to gibberellins and bikaverin, formation of fusarubin was also drastically reduced in the Δffhda1 mutant under inducing low-nitrogen, alkaline conditions. Further investigations using different chromatin marks and various growth conditions will shed light on chromatin alterations in the Δffhda1 mutant and other mutants regarding other secondary-metabolite gene clusters, as shown here for the gibberellin and bikaverin gene clusters.
FfHda2 is also important for the regulation of secondary metabolism in Fusarium fujikuroi.In addition to FfHda1, deletion of the class I HDAC-encoding gene ffhda2 also resulted in decreased biosynthesis of gibberellins, bikaverin, fusaric acid, and fusarins in Δffhda2 mutants, suggesting that FfHda2 is essential, either directly or indirectly via deacetylation of nonhistone factors, for sufficient transcription of the respective biosynthetic genes. On the contrary, production of fusarubins was wild-type-like under optimal conditions but was increased in Δffhda2 mutants under normally repressing production conditions, suggesting that in this case, hyperacetylation is indeed concomitant with higher gene transcription levels. Similarly to the deregulation of bikaverin gene expression in the Δffhda1 mutant, only one repressive effect was overruled: while repression by an acidic pH could be circumvented by the Δffhda2 mutant, nitrogen repression (29) was maintained, suggesting additional repressors under either condition. Future analyses similar to those described here for the Δffhda1 mutant will give more insights into how deletion of ffhda2 affects the chromatin landscape of these gene clusters.
FfHda1 and FfHda2 show distinct effects on pigment biosynthesis in Fusarium fujikuroi.While the Δffhda1 Δffhda2 double mutant showed the same deregulated phenotype as the single ffhda2 deletion mutant with regard to fusarubin biosynthesis under repressing acidic conditions, fusarubin biosynthesis was dramatically reduced under inducing alkaline conditions, thus resembling the phenotype of the ffhda1 deletion mutant. These results suggest that FfHda1 is indeed directly required for the activation of fusarubin biosynthesis under inducing conditions, possibly via deacetylation of nonhistone factors, similar to gibberellin gene expression. However, this is not the case under repressing acidic pH conditions. In addition, the Δffhda1 Δffhda2 mutant resulted in bikaverin formation under nitrogen-sufficient conditions, as was observed for the Δffhda1 single and Δffhda1 Δffhda4 double mutants, suggesting that only FfHda1, but not FfHda2, is involved in the deacetylation of a postulated bikaverin-specific repressor.
FfHda1 and FfHda2 are both required for GA-induced bakanae disease on rice.Deletion of HDC1 and hos2 (ffhda2 homolog) in Cochliobolus carbonum and Magnaporthe oryzae, respectively, resulted in mutants that were greatly impaired in pathogenicity, while deletion of the ffhda1 and ffhda4 homologs hda1 and hos3 had no or only weak effects on M. oryzae (72, 84). Similarly, deletion of HDF1 (ffhda2 homolog) had the greatest impact on virulence in F. graminearum. Also, double deletion of both hdF1 and hdF2 as well as hdF1 and hdF3 resulted in even less-pathogenic strains than single deletions, indicating that the HDACs have overlapping functions regarding pathogenicity in this fungus (24). However, nothing is known regarding the underlying mechanism of pathogenicity so far. In F. fujikuroi, deletion of ffhda1 and ffhda2 had comparable effects regarding virulence on rice, as infected rice plants showed reductions in hyperelongation of about 25% and 22%, respectively. This reduction is most likely due to the decreased formation of gibberellins. However, similar to F. graminearum, the double deletion of ffhda1 and ffhda2 resulted in even more reduced virulence: the internodes were less elongated, probably due to even less production of gibberellins than the single deletion mutants. Recently, we showed that gibberellins are not necessary for initial penetration of the rice roots but rather are involved in invasive growth. Microscopic analyses of a gibberellin-deficient strain showed significantly fewer penetration events of the root cells within the rice plants than the wild type (17). Similarly, deletion of ppt1 in F. fujikuroi, which resulted in the complete loss of polyketide synthase (PKS)- and NRPS-derived products, had no impact on rice root penetration (60). Once penetration occurs, minor amounts of gibberellins are likely enough to induce bakanae symptoms. However, the impairment of other as-yet-unknown virulence factors in the double deletion mutant compared to the Δffhda1 and Δffhda2 single deletion mutants cannot be excluded. In conclusion, we provide strong evidence that both FfHda1 and FfHda2 are crucial for regulation of some, but not all, secondary metabolites, whereas FfHda4 is rather involved in fungal development (conidiation) and growth but is dispensable for secondary metabolism in F. fujikuroi. Pathogenicity assays revealed that both FfHda1 and FfHda2 are required for gibberellin-induced bakanae disease of rice. The obtained results demonstrate that deletion of HDAC-encoding genes can result in both gene activation and repression and, furthermore, in bypassing normally repressing regulatory mechanisms, as in the case of bikaverin and fusarubin biosynthesis. We propose that HDAC activity by FfHda1 is essential for sufficient activation of gibberellin biosynthesis, while downregulation of bikaverin genes is likely due to the activation of an as-yet-unknown transcriptional repressor. ChIP and ChIP-seq analyses using the H3K9 acetylation-specific antibody revealed insights into alterations of the acetylation pattern upon deletion of ffhda1 and moreover indicated that the scenario for each secondary metabolite might be different despite similar outcomes. Noteworthy, we are just beginning to unravel the complexity of posttranslational modifications and their involvement in secondary metabolism given the number of possible histone modifications and the increasing number of modified nonhistone factors. Thus, the obtained results substantially contribute to the understanding of how secondary metabolites are regulated by histone-modifying proteins, thereby adding another level of complexity to the simplified model of histone (de)acetylation.
ACKNOWLEDGMENTS
This work and the research fellowship of L.S. were supported by funds of the Deutsche Forschungsgesellschaft (DFG), Graduiertenkolleg 1409 (GRK1409, Germany). Work in Oregon was supported by funds from the American Cancer Society (RSG-08-030-01-CCG) and the NIH (GM097637) to M.F.
We thank Kristina Smith for bioinformatic analyses, development of the F. fujikuroi ChIP gbrowse site, and submission of data to NCBI collections.
FOOTNOTES
- Received 13 May 2013.
- Accepted 30 September 2013.
- Accepted manuscript posted online 4 October 2013.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/AEM.01557-13.
- Copyright © 2013, American Society for Microbiology. All Rights Reserved.
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵