Transcription Factor RFX1 Is Crucial for Maintenance of Genome Integrity in Fusarium graminearum

Strains and culture conditions.

The F. graminearum strains used in this study are listed in Table 1. Standard laboratory methods and culture media for the Fusarium species were used (27). Cultures were maintained in complete medium (CM) agar plates. Conidia were produced in carboxymethyl cellulose medium (CMC) or on yeast malt agar (YMA) (28, 29). Unless otherwise stated, the growing temperature of the fungal strains was set at 25°C. All strains were stored as conidia and mycelia in 30% glycerol solution at −80°C.

Conidia were harvested from CMC and transferred to CM containing various DNA-damaging agents at the following concentrations: hydroxyurea (10 mM), methyl methanesulfonate (MMS) (0.1%), and bleomycin (20 mU/ml). Conidia were inoculated on CM agar plates and immediately exposed to UV (254 nm, 480 J/m²). The germination rates of the conidia were examined by microscopy 18 h after inoculation, and the percentage of germinated conidia was determined. All statistical analyses were conducted using the R statistical software package (30).

Nucleic manipulation, primers, and PCR conditions.

Genomic DNA was isolated from lyophilized mycelia as previously described (27). Restriction endonuclease digestion, Southern blotting, and hybridization with ³²P-labeled probes were performed in accordance with standard techniques (31). The PCR primers used in this study were synthesized by an oligonucleotide synthesis facility (Bionics, Seoul, South Korea) (see Table S1 in the supplemental material). General PCR procedures were performed in accordance with the manufacturer's instructions (TaKaRa Bio, Inc., Otsu, Japan).

Sexual crosses.

For self-fertilization, cultures were grown on carrot agar plates for 5 days. Sexual reproduction was induced by removing aerial mycelia with sterile 2.5% Tween 60 solution (27). For outcrosses, female strains grown on carrot agar plates were fertilized with conidial suspensions from corresponding male strains 5 days after inoculation. The cultures were incubated under a near-UV lamp (365 nm) at 25°C for an additional 7 days.

Fungal transformation.

For green fluorescent protein (GFP) tagging and complementation, a DNA fragment containing the native promoter and the open-reading frame (ORF) of rfx1 was amplified with RFX1/5F and RFX1_ORF/Rev. The double-joint PCR method (32) was used to fuse the amplicon to the 3′-flanking region of rfx1 and a GFP cassette containing the hygromycin B resistance gene (hyg). Fusion constructs were amplified with nested primers to generate split markers. The resulting constructs were introduced into Δrfx1 protoplasts, using the polyethylene glycol-mediated fungal transformation procedure (33). Transformants were selected under hygromycin B (75 μg/ml).

The ORF of rfx1 was replaced with the GFP cassette to generate strain KH26 (Δrfx1::gfp-hyg). The GFP cassette, 5′-flanking region of rfx1, and 3′-flanking region of rfx1 were fused. The fusion constructs were amplified with nested primers to generate split markers. The resulting constructs were introduced into wild-type protoplasts. rfx1 deletion mutants were selected under hygromycin B (75 μg/ml) and screened for GFP fluorescence in mycelia.

Fluorescence microscopy.

Differential interference contrast and fluorescence images were captured on an Axio Imager A1 microscope (Carl Zeiss, Oberkochen, Germany) with a charge-coupled-device (CCD) camera. Nuclei of conidia were stained with acriflavin and examined with the 470-nm/525-nm (excitation/emission wavelength) filter set (34). The organelle localizations of fusion proteins RFX1-GFP and histone H1 (hH1)-GFP were imaged with the 470-nm/525-nm filter set. Red fluorescence protein (RFP)-tagged histone H1 was examined with the 550-nm/605-nm filter set. The cell wall was stained with 0.2 μl/ml calcofluor white (Sigma-Aldrich, St. Louis, MO) and observed using the 550-nm/605-nm filter set. Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI) (Invitrogen, Carlsbad, CA) and examined with the 365-nm/445-nm filter set (35). Images were analyzed using the AxioVision release 4.7 software package (Carl Zeiss).

Virulence assays.

The virulence of fungal strains on wheat heads was assessed in the wheat cultivar Eunpamil as previously described (36, 37). Conidia were harvested from CMC cultures and resuspended to 10⁶ spores/ml in 0.01% Tween 20 solution. The center spikelet was drop inoculated midanthesis with 10 μl of conidial suspension. Inoculated plants were incubated in a humidity chamber at 25°C for 3 days and grown in a greenhouse for an additional 18 days. Infected wheat heads were imaged, and the spikelets with head blight symptoms were counted.

Toxin analysis.

The conidial suspension was inoculated in defined medium containing 5 mM agmatine (MMA) and incubated for 7 days as previously described (38). Culture filtrates were extracted with an ethyl acetate-methanol mixture (4:1, vol/vol). The extracts were dried (39), and the residues were derivatized with trimethylsilylating reagent [N,O-bis(trimethylsilyl)acetamide (BSA), trimethylchlorosilane (TMCS), and 1-(trimethylsilyl)imidazole (TMSI), 3:2:3; Supelco, Bellefonte, PA]. The resulting samples were analyzed with a Shimadzu QP-5000 gas chromatograph-mass spectrometer (GC-MS; Shimadzu, Kyoto, Japan) as previously described (40).

Neutral comet assay.

Fungal conidia produced in CMC culture were inoculated into 50 ml of YPG liquid medium (3 g yeast extract, 10 g peptone, and 20 g glucose per liter) at 10⁶ spores/ml and grown for 12 h with shaking (33). Mycelia were harvested by filtration and incubated in 35 ml of 1 M NH₄Cl containing Driselase (10 mg/ml) (Sigma-Aldrich) at 30°C to generate protoplasts. For bleomycin treatment of wild-type cells, protoplasts were generated in Driselase solution containing 20 mU/ml bleomycin. Protoplasts were collected by centrifugation 4 h after incubation and resuspended in 1 M NH₄Cl.

Low-gelling-temperature agarose (1%; Sigma-Aldrich) was molten in 1 M NH₄Cl and kept at 40°C. The protoplast suspension was mixed with molten agarose at 6 × 10³ cells/ml. The agarose was poured onto frosted slides and allowed to gel. Slides were submerged in neutral lysis solution for double-strand break (DSB) detection, in accordance with a standard protocol (41). After overnight lysis, the slides were washed three times in Tris-acetate-EDTA buffer for 30 min each. Electrophoresis was conducted in Tris-acetate-EDTA buffer for 25 min at 0.6 V/cm. The slides were rinsed with deionized water, stained in ethidium bromide solution, rinsed again with deionized water, and dried at room temperature. Dried agarose gels were rehydrated before fluorescence microscopy.

Comet images were captured on an Axio Imager A1 microscope with a CCD camera and the 550-nm/605-nm filter set. Comet images were composed of the comet head and comet tail. The percentage of DNA in the tail was analyzed for individual comet images by using CometScore image analysis software (TriTek Corp., Sumerduck, VA).

Quantitative real-time (qRT)-PCR.

Wild-type and Δrfx1 conidia produced in CMC culture were inoculated into 50 ml of CM. The wild-type culture was incubated for 24 h and the Δrfx1 culture for 32 h with shaking. Mycelia were harvested and subcultured in fresh CM or CM containing 20 mU/ml bleomycin. The cultures were incubated for an additional 2 h. Total RNA was isolated from mycelia that were ground in liquid nitrogen using an Easy-Spin total RNA extraction kit (Intron Biotech, Seongnam, South Korea). First-strand cDNA was synthesized using SuperScript III reverse transcriptase (Invitrogen).

qRT-PCR was performed with SYBR green supermix (Bio-Rad, Hercules, CA) and a 7500 real-time PCR system (Applied Biosystems, Foster City, CA). An endogenous housekeeping gene, ubiquitin C-terminal hydrolase (ubh; FGSG_01231.3), was used as a reference gene (42). The transcript levels of target genes under different conditions were compared by the cycle threshold (2^−ΔΔCT) method (43). The qRT-PCR analysis was repeated three times with two biological replications.

RNA-sequencing analysis.

Wild-type and Δrfx1 conidia were harvested from CMC culture and inoculated into CM. The wild-type culture was incubated for 24 h, and the Δrfx1 culture was incubated for 32 h with shaking. Total RNA was extracted as stated above from mycelia that were ground in liquid nitrogen. RNA-sequencing libraries were constructed using the Illumina TruSeq RNA sample preparation kit in accordance with the standard low-throughput protocol. Samples were run on an Illumina HiSeq2000 instrument using the reagents provided in the Illumina TruSeq paired-end (PE) cluster kit V3-cBot-HS and the TruSeq SBS kit v3-HS (200 cycles).

Genome-wide transcript levels of genes were quantified in reads per kilobase of exon per million mapped sequence reads (RPKM) (44). When the RPKM value was 0, it was changed to 1 to calculate the fold change of the transcript level. Genes for which differential transcript levels were detected were functionally characterized using the Munich Information Centre for Protein Sequences FunCat functional classification and annotation system (45). Over- and underrepresentation analysis of FunCat categories was performed by using Pearson's χ² test.

Microarray data accession number.

RNA sequencing data were deposited in NCBI's Gene Expression Omnibus and are accessible through GEO Series accession number GSE52198.

Molecular characterization of the rfx1 gene.

The Fusarium Comparative Database (Fusarium Comparative Sequencing Project, Broad Institute of Harvard and MIT, http://www.broadinstitute.org) annotated the rfx1 gene (locus FGSG_07420) based on the presence of an RFX DNA-binding domain. We aligned protein sequences from RFX1 homologs using ClustalW in the MEGA 5.2 program and BoxShade. The alignment showed that the RFX DNA-binding domain of FgRFX1 is highly conserved among all characterized RFX proteins (Fig. 1A). Phylogenetic analyses of RFX homologues were performed using a neighbor-joining algorithm, with bootstrap values calculated from 100 iterations. The phylogenetic tree revealed that RFX homologs of filamentous fungi clustered into a separate group relative to those of yeasts and animals (Fig. 1B).

Effects of rfx1 deletion on nucleus and septum formation.

The rfx1 deletion strain was provided by a mutant library of TFs in F. graminearum (37). For genetic complementation and GFP tagging of rfx1, a construct containing the rfx1 ORF fused with the GFP cassette was introduced into the rfx1 deletion mutant. The Southern hybridization results showed that the construct replaced gen in the genome of the complementation strain, resulting in strain KH23 (see Fig. S1 in the supplemental material). The Δrfx1 strain showed severely reduced radial growth on CM agar plates and did not produce aerial mycelia (Fig. 2 and Table 2).

Strain KH24 (Δrfx1::gen hH1-gfp-hyg) was generated by crossing strain mat1g (Δmat1-1::gen hH1-gfp-hyg) and the Δrfx1 strain (Table 1). KH24 showed the same phenotypes as the parental Δrfx1 strain and exhibited green fluorescence in nuclei. Strain KH25 (Δrfx1::rfx1-gfp-hyg hH1-rfp-hyg) was generated by crossing strain mat1r (Δmat1-1::gen hH1-rfp-hyg) and the complementation strain (Δrfx1::rfx1-gfp-hyg). KH25 had both green and red fluorescence in its nuclei. Conidia of the histone H1 (hH1)-GFP strain, KH24, and KH25 were produced in CMC and inoculated into CM to examine the nucleus and septum formation in hyphae. Mycelia were harvested 12 h after inoculation (except for mycelia of KH24, which were harvested 24 h after inoculation). Septa and nuclei of hyphae were observed by calcofluor white staining and fluorescent histone H1 proteins, respectively. The formation of septa in Δrfx1 hyphae was severely inhibited compared to their formation in the wild type (Fig. 3). The Δrfx1 hyphae were largely aseptate, whereas wild-type hyphae possessed numerous septa. The Δrfx1 nuclei were irregular in size, and the Δrfx1 strain produced small fragmented nuclei, called micronuclei.

The level of conidium production of the Δrfx1 strain was 9 times lower than that of the wild type (Table 2). In addition, the Δrfx1 conidia were abnormally shaped (Fig. 4). The average conidium length in the Δrfx1 strain (71 μm) was much longer than that in the wild type (45 μm) (P < 0.01). Deletion of rfx1 resulted in more severe defects in the nucleus and septum formation of conidia than of hyphae. Most wild-type conidia (∼70%) had four or five cells with one nucleus per cell; in contrast, most Δrfx1 conidia (∼70%) had one or two cells with multiple nuclei. In the Δrfx1 conidia, the nuclei were irregularly shaped and many were micronuclei. The hyphal growth and conidial morphology of KH23 were restored to the wild-type phenotypes.

Virulence and toxin production.

The virulence of the fungal strains was examined by point inoculation of wheat spikelets. The Δrfx1 strain was unable to infect the inoculated spikelets, whereas the wild-type and complemented strains readily colonized the inoculated spikelets and spread to the neighboring spikelets. The disease index (number of diseased spikelets per wheat head) of the Δrfx1 strain was zero (Table 2). We also examined the production of trichothecene, which is an important virulence factor. Trichothecene biosynthesis was induced in MMA, which contains agmatine as a nitrogen source. GC-MS detected two trichothecenes, deoxynivalenol and 15-acetyldeoxynivalenol. The production of total trichothecenes was not markedly different (P > 0.05) among the strains on the basis of the biomass (g) of each sample.

Sexual development.

The Δrfx1 strain produced few immature perithecia, whereas the wild-type and the complementation (Δrfx1::rfx1-gfp) strains produced abundant mature perithecia 7 days after sexual induction (Fig. 5). The perithecia did not contain any ascus structures. The heterothallic F. graminearum strain carrying the MAT1-1 deletion with histone H1-GFP (mat1g) was crossed with the Δrfx1 mutants to investigate the male fertility of the Δrfx1 strain. We randomly isolated 144 ascospores, but 57 ascospores did not germinate. The genotypes of the germinated 87 ascospores were analyzed. The segregation ratio of the genotypes was 39:27:8:13 (rfx1 hH1-gfp/rfx1 hH1/Δrfx1 hH1-gfp/Δrfx1 hH1), indicating that most ascospores carrying the rfx1 deletion were not viable. Progenies carrying the rfx1 deletion were many fewer than those carrying wild-type rfx1 (21:66). However, the ratio of hH1-GFP to non-GFP segregation was 1:1, indicating that the outcross underwent normal sexual recombination.

We further investigated role of rfx1 in sexual development, replacing the rfx1 gene with the cytoplasmic GFP gene and thus generating a Δrfx1::gfp-hyg strain (KH26). KH27 was generated by crossing mat1r and KH26 (Table 1). A heterothallic F. graminearum strain carrying the MAT1-1 deletion with histone H1-RFP (mat1r) was crossed with KH27 to examine the fertility of the Δrfx1 strain as a male (Fig. 6). When mature asci were dissected, four out of eight ascospores expressed GFP, indicating that they were rfx1 deletion mutants. In F. graminearum, ascospores underwent two mitoses, producing four cells with one nucleus after spore delimitation. Ascospores with the rfx1 deletion were composed of one cell with one or two nuclei, whereas the non-GFP ascospores were composed of four cells with one nucleus. Thus, deletion of rfx1 blocked mitosis in ascospores after spore delimitation. Defects of spore maturation may cause inviability of ascospores carrying the deletion of rfx1.

Localization of GFP-tagged RFX1 proteins.

To determine the subcellular distribution of RFX1, we complemented the Δrfx1 strain by introducing the RFX1-GFP-hyg construct, which was GFP tagged at the C terminus of RFX1. The resulting Δrfx1::rfx1-gfp strain (KH23) exhibited wild-type phenotypes under all conditions, as previously described. Conidia of KH23 were inoculated into CM and examined after 12 h of incubation. Conidiation of KH23 was induced in CMC and examined 3 days after induction. Although KH23 does not contain any predictable nuclear localization signals, RFX1-GFP localized to the nuclei. Intense fluorescence was found in nuclei of growing hyphae and developing conidiophores (Fig. 7). However, the fluorescence was too weak to detect in mature hyphae and conidia.

Sensitivity of rfx1 deletion strain to DNA damage.

To investigate the function of rfx1 in the DNA damage response, we examined the germination rate of conidia under conditions of DNA damage (Fig. 8). Conidia of the fungal strains were exposed to UV and several DNA-damaging agents, including hydroxyurea, MMS, and bleomycin. The conditions induced low-level DNA damage, such that the wild-type and Δrfx1::rfx1-gfp conidia germinated well. However, the germination rate of Δrfx1 conidia was markedly reduced under all DNA damage conditions, indicating that deletion of rfx1 resulted in sensitivity to DNA damage, regardless of the type of damage.

Interestingly, we found that bleomycin (10 mU/ml), which induced low levels of DNA damage, inhibited septum formation in the wild-type hyphae. Septum formation was severely delayed and micronuclei were found when a higher concentration (20 mU/ml) of bleomycin was used (Fig. 3). Bleomycin is known to induce DNA DSBs (46, 47). Both bleomycin and deletion of rfx1 triggered aseptation and micronuclei in hyphae, implying that deletion of rfx1 might induce spontaneous DNA strand breaks.

Accumulation of spontaneous DNA DSBs.

A neutral comet assay was performed to measure the amount of DNA DSBs in fungal cells. The comet assay combines DNA gel electrophoresis with fluorescence microscopy to visualize the migration of DNA from individual agarose-embedded cells (41). In the neutral comet assay, the comet head contains undamaged DNA, and the comet tail contains DSB-generated fragments. Therefore, cells with more DSBs show a higher percentage of DNA in the tail. Because bleomycin produces DNA DSBs (46, 47), we treated wild-type cells with 20 mU/ml bleomycin as a positive control.

The bleomycin-treated cells contained a higher percentage of DNA in the tail than the untreated cells, indicating that the neutral comet assay detected DNA DSBs (Fig. 9). The percentage of DNA in the tail was increased in Δrfx1 cells compared to the result for wild-type cells (P < 0.01). The occurrence of DSBs in Δrfx1 cells was similar to that in bleomycin-treated cells. The results suggested that deletion of rfx1 triggered DNA DSBs in the absence of a DNA-damaging agent.

Effects of rfx1 deletion on genome-wide transcription profiles.

We obtained and analyzed genome-wide transcription profiles generated from the RNA sequencing data of the wild-type and Δrfx1 strains. The RPKM values of 13,820 recently reannotated genes in F. graminearum (48) were obtained and compared (see Dataset S1 in the supplemental material). Genes with transcript levels showing a 3-fold or greater difference between the Δrfx1 and wild-type strains were functionally characterized (45), as shown in Table 3. The results revealed that 38% of all genes were upregulated in the Δrfx1 strain, whereas only 1% were downregulated. These findings implied a role for RFX1 as a transcriptional repressor. There was a significant enrichment of genes in the “Unclassified protein” category, which showed a 3-fold increase in transcript abundance compared to the transcription of the genome as a whole (P < 0.001). However, most of the FunCat categories showed significant underrepresentation of 3-fold-upregulated genes (P < 0.001), except for FunCat categories 32 and 41. The transcript levels of 35 genes involved in DNA repair (FunCat 10.01.05.01) were increased in the deletion mutants. This implied that deletion of rfx1 triggered derepression of many genes whose functions have not been annotated.

There are two complementary mechanisms for DNA DSB repair: homologous recombination (HR) and nonhomologous end joining (NHEJ). We selected several proteins required for HR and NHEJ (49), including RAD51, RAD52, and RAD54, which are required for HR, KU70, KU80, and DNA ligase IV, which are required for NHEJ, and RAD50, which is required for both HR and NHEJ. Because Δrfx1 cells had spontaneous DNA DSBs, the expression of DSB repair genes was examined in detail (Table 4). The Δrfx1 cells showed greater expression of DSB repair genes than the wild-type cells in RNA-sequencing analysis.

The transcriptome analyses were validated for the selected genes using qRT-PCR (Table 4). Consistent with the RNA sequencing data, the DSB repair genes were upregulated in the Δrfx1 cells in the qRT-PCR analysis. In response to bleomycin, the expression of DSB repair genes was increased in the wild-type and Δrfx1 cells. The Δrfx1 cells with bleomycin exhibited the highest expression of the genes, indicating an additive effect. These findings indicate that both bleomycin and deletion of rfx1 produced DSBs in DNA and induced the expression of DSB repair genes.