Identification of the regulatory protein COH1

In our previous RNA-Seq analysis [9], we found that a gene (MAA_08820, designated as Coh1, as it is involved in colonization of the hemocoel) was highly expressed during hemocoel colonization. In this study, hemolymph collected from last instar Galleria mellonella larvae (used for all assays in this study) was treated with an anticoagulant, and fungal growth in the resulting hemocyte-containing hemolymph (hereafter called hemolymph) was used as an approximation of hemocoel colonization, hereafter called surrogate hemocoel colonization. Penetration of the cuticle was achieved by incubating the fungus on the cuticle of G. mellonella larvae. Unless otherwise indicated, the insect cuticle used in this study was from G. mellonella larvae. Saprophytic growth was achieved by growing the fungus in the liquid medium Sabouraud dextrose broth supplemented with 1% yeast extract (SDY) or on potato dextrose agar (PDA) plates. Root colonization refers to the growth of M. robertsii on the roots of Arabidopsis thaliana. When quantitative reverse transcription PCR (qRT-PCR) analysis was conducted with RNA samples from cuticle penetration, saprophytic growth, and root colonization, the Cq values (the PCR cycle number at which a sample reaction curve intersects the threshold line) were approximately 38 cycles, and no PCR products were detected on the agarose gel (Fig 1A), suggesting that the transcription level of Coh1 was extremely low or was not detectable. However, qRT-PCR analysis and subsequent detection of PCR products on agarose gel confirmed that the Coh1 transcript was expressed during the surrogate hemocoel colonization (Fig 1A). The PCR product of Coh1 was also detected with RNA prepared from live G. mellonella larvae infected by M. robertsii, suggesting that this gene was expressed in the real hemocoel of the insects. When RNA from the insect cadavers that were mummified with the mycelium was used, no PCR product was detected, indicating that Coh1 was not expressed during the necrotrophic stage (Fig 1A).

Download:

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

Fig 1. Expression and regulation of the Coh1 gene.

(A) Agarose gel electrophoresis of quantitative reverse transcription PCR (qRT-PCR) products of Coh1. 1: Saprophytic growth in Sabouraud dextrose broth supplemented with 1% yeast extract (SDY) medium; 2: cuticle penetration; 3: colonization of plant roots; 4: growth in the hemocyte-containing hemolymph; 5: live insects infected with M. robertsii; 6: insect cadavers mummified with M. robertsii mycelium; M: DNA ladder. Images are representative of 3 independent experiments. Upper panel: the gene Coh1; lower panel: the reference gene Gpd encoding glyceraldehyde 3-phosphate dehydrogenase. (B) LT₅₀ (time taken to kill 50% of insects) values when the insects were inoculated by topical application of conidia on the cuticle. ΔCoh1#1, ΔCoh1#2, and ΔCoh1#3 are 3 independent isolates of the deletion mutant ΔCoh1. WT, wild type. The bioassays were repeated 3 times with 40 insects per repeat. Data are expressed as mean ± SE. Values with different letters are significantly different (n = 3, P < 0.05, Tukey’s test in one-way ANOVA). (C) Agarose gel electrophoresis of qRT-PCR products of Coh1 in WT strain and deletion mutants of the histone deacetylase gene Hdac1 (ΔHdac1) and the histone acetyltransferase gene Hat1 (ΔHat1) during saprophytic growth. Note: No qRT-PCR product was seen in the WT strain. (D) qRT-PCR analysis of the expression of Hdac1 and Hat1 in the WT strain during the surrogate hemocoel colonization (Hemolymph) and cuticle penetration (Cuticle) relative to saprophytic growth (SDY). (E) GFP signal in the fungal cells on the cuticle (Cuticle) and in the real hemocoel (Hemocoel) of G. mellonella larvae. The strain with the gfp gene was driven by the promoter of Hat1 (upper panels) or Hdac1 (lower panels). F, fungal cells; H, hemocyte. Scale bar: 10 μm. Images are representative of 3 independent experiments. The mean gray value (MGV) shows the GFP fluorescence intensity in the fungal hyphae. (F) qRT-PCR analysis of Coh1 expression in the WT strain and the mutants ΔHat1 and ΔHdac1 during the surrogate hemocoel colonization. All qRT-PCR experiments in this study were repeated 3 times. For qRT-PCR analyses in this figure, the values represent the fold-change of expression of a gene in treatment compared with expression in its respective control, which is set to 1. The data underlying all the graphs shown in this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3001360.g001

Coh1 is a single-copy gene with a 525-bp open reading frame (ORF) that encodes a protein containing 174 amino acid residues. COH1 contains an Ecl1 domain (PFAM12855) that was also identified in the ECL1 proteins of the saprophytic yeasts Saccharomyces cerevisiae and Schizosaccharomyces pombe; these proteins are considered regulators though their functions have not been characterized [17]. However, BLASTp analysis using the full-length protein sequence of COH1 as a query identified no significant similarity (>e⁻⁵) between M. robertsii COH1 and the yeast ECL1 proteins. No putative signal peptide or nuclear localization signal (NLS) was predicted in COH1. In the chromosome of M. robertsii, Coh1 was distant from its adjacent genes, with its ORF being 111,146 bp away from that of the upstream gene (MAA_10633) and 7,498 bp away from that of the downstream gene (MAA_08821).

COH1 is involved in real hemocoel colonization

To investigate the biological functions of Coh1, we constructed a deletion mutant (ΔCoh1) that had the entire ORF deleted (S1A and S1B Fig). The genes and fungal strains used in this study are listed in Table 1. Southern blot analysis showed that there was only 1 copy of the herbicide-resistance gene Bar (the selection marker gene for fungal transformation) in 3 randomly selected isolates (ΔCoh1#1, ΔCoh1#2, and ΔCoh1#3) (S1C Fig), indicating that the insertion of the foreign DNA only resulted in deletion of Coh1 via homologous recombination, and no ectopic integration occurred. The 3 independent isolates of ΔCoh1 were subjected to the following phenotypic analysis, and no differences were observed between these isolates in all assays (see below). On the PDA plates, ΔCoh1 was not different from the wild-type (WT) strain in colony growth, colony morphology, and conidial yield (S2A–S2C Fig).

Download:

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

Table 1. Plasmids, fusion proteins, and fungal strains used in this study.

https://doi.org/10.1371/journal.pbio.3001360.t001

The pathogenicity of M. robertsii, shown by the time taken for 50% of insects to die (LT₅₀), was assayed in G. mellonella larvae. Inoculations were conducted either by topically applying conidia onto the insect cuticle or by direct injection of conidia into the hemocoel (thus bypassing the cuticle). Via topical application, the LT₅₀ values of the 3 isolates of the mutant ΔCoh1 were approximately 2-fold higher than that of the WT strain (Fig 1B). Via direct injection, the 3 isolates of ΔCoh1 also had higher LT₅₀ values than the WT strain (S2D Fig). At 3 and 4 d after direct injection, the hemocoel of the insects injected with each of the 3 isolates of ΔCoh1 contained significantly fewer (P < 0.05) hyphal bodies than the WT strain (S2E Fig). Quantitative PCR (qPCR) analysis further showed that the fungal burden in the dying insects infected by ΔCoh1 was significantly lower than that of insects infected with the WT strain (P < 0.05) (S2F and S2G Fig). On a normally inductive milieu (i.e., the hydrophobic surfaces of plastic petri dishes) in the presence of low levels of nitrogenous nutrients, no difference in appressorium formation was found between the WT strain and the mutant ΔCoh1 (S2H Fig). The ability of the mutant ΔCoh1 to penetrate the insect cuticle did not differ from the WT strain (S2I Fig).

We then looked at the involvement of COH1 in fungal interaction with host immune defense. Infection with the WT strain decreased phenoloxidase activity in G. mellonella larvae, but no significant difference was found between the mutant ΔCoh1 and the WT strain (S2J Fig). qRT-PCR analysis showed that infection with the WT strain up-regulated the expression of the antimicrobials gallerimycin and defensin in the insects, but again there was no significant difference between the insects infected by the mutant ΔCoh1 and the WT strain (S2K Fig).

We also constructed the complemented strain C-ΔCoh1 of the deletion mutant of Coh1 by ectopically integrating its genomic clone into the genome of ΔCoh1 (S1D Fig). Five transformants were randomly selected for the following analysis. In all analyses, no differences were found between the 5 transformants, and data about only 1 transformant is shown. The complemented strain C-ΔCoh1 was not different from the WT strain in cuticle penetration and LT₅₀ values via inoculation by topical application and injection (S3D–S3F Fig). Although the colony growth of the strain C-ΔCoh1 on PDA plates was not different from the WT strain (S3A Fig), its conidial yield was significantly reduced (S3B Fig). The colony of the strain C-ΔCoh1 started to autolyze 14 d after inoculation of a conidial suspension on the center of a PDA plate (S3C Fig). In contrast to the WT strain, Coh1 was expressed in the complemented strain C-ΔCoh1 during saprophytic growth on the PDA plates (S1E Fig). Therefore, the deletion mutant ΔCoh1 was not successfully complemented by its genomic clone in the strain C-ΔCoh1, suggesting that the native chromosomal position of Coh1 is essential for its transcriptional regulation. To investigate whether the difference in saprophytic growth on PDA plates between the WT strain and C-ΔCoh1 resulted from the expression of Coh1 in C-ΔCoh1, we constructed the Coh1-overexpressing strain Coh1^OE by transforming the coding sequence of COH1, driven by the constitutive promoter Ptef from Aureobasidium pullulans [18], into the WT strain (S1F Fig). As with C-ΔCoh1, Coh1^OE showed reduced conidial yield and its colony autolyzed after 14 d of culture on PDA plates (S3B and S3C Fig). The strain Coh1^OE was not significantly different from the WT strain and C-ΔCoh1 in cuticle penetration and LT₅₀ values with inoculation via topical application or injection (S3D–S3F Fig).

Regulation of Coh1 by a histone deacetylase and a histone acetyltransferase

To identify the mechanisms that regulate Coh1 expression, we compared the expression level of Coh1 in the WT strain with that in mutants with regulator genes deleted [2,9,19]. Since it was impossible to collect sufficient hemolymph from the G. mellonella larvae for screening a large number of mutants, we searched for the regulators that suppress the expression of Coh1 during saprophytic growth in the medium SDY, where Coh1 was not expressed (Fig 1A). We found that Coh1 was expressed in 11 epigenetic mutants with deletion of histone acetyltransferases, deacetylases, and methyltransferases, indicating that these epigenetic regulators negatively controlled Coh1 expression during saprophytic growth (Figs 1C and S4A). Compared with during saprophytic growth, only a histone H3 deacetylase gene (MAA_02098, designated as Hdac1) and a histone H3 acetyltransferase gene (MAA_02282, designated as Hat1) were significantly down-regulated during surrogate hemocoel colonization (Figs 1D and S4B). No significant differences in the expression levels of Hat1 and Hdac1 were found between cuticle penetration and saprophytic growth (Fig 1D). We further compared the expression levels of Hdac1 and Hat1 during cuticle penetration with that during colonization of the real hemocoel (i.e., the hemocoel of the insects infected by M. robertsii). As we could not obtain enough fungal biomass from the real hemocoel to prepare sufficient RNA for qRT-PCR analysis, we analyzed the expression of Hdac1 and Hat1 by tracing the GFP (green fluorescent protein) signal in 2 strains (WT-PHdac1-GFP and WT-PHat1-GFP), in which the gfp gene was driven by Hdac1 or Hat1 promoter in the WT strain (S1J Fig). Consistent with the results obtained from the qRT-PCR analysis with the surrogate hemocoel colonization, in both strains the GFP fluorescent intensity in the fungal cells on the cuticle was stronger than in the real hemocoel (Fig 1E).

In a previous study, we constructed a deletion mutant of Hat1, ΔHat1, and its complemented strain C-ΔHat1 [19]. In this study, we generated an Hdac1 deletion mutant, ΔHdac1, and its complemented strain C-ΔHdac1 (S1G and S1H Fig). During the surrogate hemocoel colonization, the expression level of Coh1 was increased 27.6-fold in the mutant ΔHat1 and 15.2-fold in the mutant ΔHdac1 (Fig 1F). We thus postulated that the reduction in the expression of Hat1 and Hdac1 during hemocoel colonization derepressed their negative regulation of Coh1, resulting in the expression of Coh1. To test this postulation, we first tried to identify the target sites of HAT1 and HDAC1 by immunoblot analysis of the acetylation levels of 7 lysine residues in histone H3. Compared with the WT strain, the acetylation level of histone H3 on lysine 4 (H3K4) was significantly reduced in the mutant ΔHat1, whereas the acetylation level of histone H3 on lysine 56 (H3K56) was increased in the deletion mutants ΔHdac1 and ΔHat1 (Fig 2A).

Download:

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

Fig 2. HDAC1 and HAT1 control Coh1 expression by regulating histone H3 acetylation in its promoter region.

(A) Immunoblot analysis of acetylation levels of 7 lysine residues on histone H3 protein in the wild-type (WT) strain and the mutants ΔHdac1 and ΔHat1. All Western blot images shown in this study are representatives of at least 3 independent experiments. (B) The acetylation level of H3K56 in the Coh1 promoter in the mutant ΔHdac1, its complemented strain C-ΔHdac1, the Hdac1-overexpressing strain Hdac1^OE, and the WT strain during saprophytic growth in Sabouraud dextrose broth supplemented with 1% yeast extract (SDY) and surrogate hemocoel colonization (Hemolymph). (C) The acetylation levels of histone H3, H3K56, and H3K4 in the Coh1 promoter in the mutant ΔHat1 and its complemented strain C-ΔHat1 relative to the WT strain during saprophytic growth. (D) qRT-PCR analysis of Hdac1 expression in the WT strain, the mutant ΔHat1, and its complemented strain C-ΔHat1 during saprophytic growth and surrogate hemocoel colonization. (E) The acetylation levels of histone H3 and H3K4 in the Hdac1 promoter in the mutant ΔHat1 and its complemented strain C-ΔHat1 relative to the WT strain during saprophytic growth. (F) The acetylation levels of histone H3 and H3K4 in the Hdac1 promoter in the WT strain during surrogate hemocoel colonization relative to saprophytic growth. For chromatin immunoprecipitation quantitative PCR (ChIP-qPCR) analyses in this figure, the values represent the fold-change of the acetylation level of histone H3, H3K4, or H3K56 compared with the level in its respective control, which is set to 1. All ChIP-qPCR experiments were repeated at least 3 times. The data underlying all the graphs shown in this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3001360.g002

We then investigated whether Hat1 and Hdac1 regulated the acetylation of histone H3 in the Coh1 promoter. Using chromatin immunoprecipitation quantitative PCR (ChIP-qPCR) analysis, we found that the acetylation level of histone H3 in the Coh1 promoter in the WT strain during surrogate hemocoel colonization was 13.8-fold higher than during saprophytic growth. More specifically, the acetylation level of H3K56 in the Coh1 promoter during surrogate hemocoel colonization was 3.5-fold higher than during saprophytic growth, but no significant difference in the acetylation level of H3K4 was observed between saprophytic growth and surrogate hemocoel colonization. During saprophytic growth, the acetylation level of H3K56 in the Coh1 promoter in the deletion mutant ΔHdac1 was 3.3-fold higher than in the WT strain, but no significant difference was found between the WT strain, the complemented strain C-ΔHdac1, and the Hdac1-overexpressing strain Hdac1^OE (Fig 2B). Overexpression of Hdac1, driven by the constitutive promoter Ptef from A. pullulans [18], was confirmed by qRT-PCR (S1I Fig). During surrogate hemocoel colonization, H3K56 in the promoter of Coh1 in the WT strain was 5-fold more acetylated than in the strain Hdac1^OE, but no difference was found among the WT strain, the deletion mutant ΔHdac1, and the strain C-ΔHdac1 (Fig 2B).

As described above, HAT1 is a histone acetyltransferase. Unexpectedly, the acetylation levels of histone H3, H3K4, and H3K56 in the Coh1 promoter in the WT strain were all lower than in the mutant ΔHat1; however, no significant differences were observed between the WT strain and the strain C-ΔHat1 (Fig 2C). It is likely that HAT1 does not directly acetylate histone H3 in the promoter of Coh1 but instead could regulate other components, which in turn alter the histone H3 acetylation level in the Coh1 promoter. Since the target of HDAC1 is H3K56, and HAT1 negatively regulated the global acetylation level of H3K56 (Fig 2A), we investigated whether HAT1 regulated Hdac1 expression by controlling the acetylation of histone H3 in its promoter. qRT-PCR analysis showed that the expression level of Hdac1 in the WT strain was significantly higher than in the mutant ΔHat1 during saprophytic growth and surrogate hemocoel colonization (Fig 2D), suggesting that HAT1 positively regulated Hdac1 transcription. ChIP-qPCR analysis further showed that the acetylation levels of histone H3 and H3K4 in the Hdac1 promoter in the mutant ΔHat1 were both lower than in the WT strain (Fig 2E). In the WT strain, the acetylation levels of histone H3 and H3K4 in the promoter of Hdac1 during saprophytic growth were significantly higher than during surrogate hemocoel colonization (Fig 2F). To further confirm that HAT1 regulated the expression of Hdac1, which in turn controlled Coh1 expression, we constructed the strain ΔHat1-Hdac1^OE with Hdac1 overexpressed in the mutant ΔHat1 (S1I Fig). qRT-PCR analysis showed that the expression level of Coh1 in ΔHat1 during surrogate hemocoel colonization was 3.3-fold higher than that of ΔHat1-Hdac1^OE (S4C Fig).

COH1 physically interacts with the transcription factor COH2

As a putative regulatory protein, COH1 could interact with other proteins to control hemocoel colonization. Using a pull-down assay with the strain WT-COH1-HA expressing the fusion protein COH1::HA (a protein with COH1 tagged with HA [hemagglutinin]) in the WT strain (S5B Fig), we failed to identify the proteins that interact with COH1. Unless otherwise indicated, all genes encoding fusion proteins were driven by the constitutive promoter Ptef from A. pullulans in this study. Previous studies showed that alteration of expression of a gene can change the expression pattern of its interacting components [20–23], so we used RNA-Seq to profile the differentially expressed genes (DEGs) between the Coh1-overexpressing strain Coh1^OE and the WT strain to identify candidates interacting with COH1. There were 1,893 DEGs, with 1,227 genes up-regulated and 666 genes down-regulated in the Coh1^OE strain. As a regulatory protein, it is more likely that COH1 interacts with other regulators to control gene expression. Compared with the WT strain, 5 transcription factors were up-regulated in the strain Coh1^OE. Yeast 2-hybrid assays showed that COH1 interacted with 1 transcription factor (MAA_07838) (Fig 3A), which is designated as COH2, as it also regulates colonization of the hemocoel (see below). Coh2 has an 849-bp ORF that encodes a protein containing 282 amino acid residues with a deduced molecular weight of 30.9 kDa. COH2 is a bZIP transcription factor with a DNA binding domain (COH2-DBD; Ser-30 to Thr-50), NLS (Ser-16 to Asp-39), and dimer interface domain (DID) (His-53 to Leu-88).

Download:

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

Fig 3. COH1 physically interacts with the transcription factor COH2 to reduce COH2 stability.

(A) Yeast 2-hybrid analysis confirms the physical interaction of COH1 with COH2. Left panel: colonies were grown in SD/−Ade/−His/−Leu/−Trp + X-α-gal + AbA. Right panel: COH1 lacks autoactivation activity. The Y2HGold cells with pGBKT7-COH1 cannot grow in SD/−Ade/−His/−Trp + X-α-gal. BD, binding domain; NC, negative control (yeast cells containing the plasmid pGADT7-T and pGBKT7-Lam); PC, positive control (yeast cells containing the plasmid pGADT7-T and pGBKT7-53). (B) Coimmunoprecipitation confirmation of the physical interaction of COH1 with COH2. The fusion proteins COH1::HA and COH2::Myc (molecular weight = 44.6 kDa) were simultaneously expressed in the strain COH1-HA/COH2-Myc. The control was the strain WT-COH2-Myc expressing the protein COH2::Myc. Immunoprecipitation was conducted with anti-HA antibody. Proteins were detected by immunoblot (IB) analysis with anti-HA or anti-Myc antibodies. The dimer COH2::Myc is indicated by red arrows. M, protein ladder. (C) Differential accumulation of the COH2::FLAG protein in 2 isolates of the strain WT-COH2-FLAG and 5 isolates of the strain Coh1^OE-COH2-FLAG. Equal loading of proteins was confirmed by the β-tubulin protein that was detected by the anti-β-tubulin antibody. Numbers indicate band intensity for COH2::FLAG relative to β-tubulin. The values of the #1 isolate of the strain WT-COH2-FLAG were set to 1. (D) Histoimmunochemical staining of the COH2::FLAG protein in fungal cells in the real hemocoel of G. mellonella larvae. Top panel: the strain WT-COH2-FLAG; middle panel: ΔCoh1-COH2-FLAG; bottom panel: Coh1^OE-COH2-FLAG. Scale bar represents 10 μm. Images are representative of 3 independent experiments. F, fungal cells; FITC, fluorescein isothiocyanate; H, hemocyte; MGV, mean gray value. (E) Confirmation of degradation of the COH2::FLAG protein by the proteasome pathway. The mycelium grown in Sabouraud dextrose broth supplemented with 1% yeast extract (SDY) medium was treated with the 26S proteasome inhibitor MG132. (F) The ubiquitination level of the COH2::FLAG protein increased due to its interaction with COH1. The COH2::FLAG protein was pulled down from the MG132-treated mycelium with an anti-FLAG antibody and immunoblotted with an anti-ubiquitin (Ubi) antibody (left) and the anti-FLAG antibody (right). IP, immunoprecipitation.

https://doi.org/10.1371/journal.pbio.3001360.g003

We further conducted a coimmunoprecipitation (Co-IP) assay using a strain (COH1-HA/COH2-Myc) constitutively expressing COH1::HA and COH2::Myc (a protein with COH2 tagged with Myc) and confirmed that COH1 physically interacted with COH2 in vivo (Fig 3B). In the Co-IP assay, the Western blot analysis with an anti-Myc antibody always showed that the mass of the detected protein was approximately 2-fold greater than the predicted molecular weight (44.6 kDa) of the protein COH2::Myc (Fig 3B), suggesting that COH2 formed a homodimer. To confirm this, a further Co-IP assay using a strain (COH2-FLAG/COH2-Myc) constitutively expressing COH2::FLAG (a protein with FLAG fused to COH2) and COH2::Myc showed that these 2 fusion proteins formed a dimer in vivo (S6A Fig). The DID in COH2 contains 7 leucine residues that could be responsible for the dimerization of this transcription factor. To test this, we constructed a strain (COH2^ΔDID-Myc) expressing the protein COH2^ΔDID::Myc with the leucine residues in the DID substituted to alanine in the fusion protein COH2::Myc. Western blot analysis showed that the fusion protein COH2^ΔDID::Myc did not form dimers (S6B Fig).

We then assayed the region of COH2 that physically contacted COH1 using Co-IP assays. COH2 was divided into the N-terminus section (Met-1 to Thr-100) and the C-terminus section (Ser-101 to Arg-282). The N-terminus section contained the predicted DNA binding domain COH2-DBD, the DID, and the NLS. For Co-IP assays, we constructed 2 strains: COH1-HA/COH2-N-Myc, expressing HA-tagged COH1 and the fusion protein with the N-terminus of COH2 (COH2-N) fused with Myc, and COH1-HA/COH2-C-Myc, expressing the HA-tagged COH1 and the COH2 C-terminus (COH2-C) fused with Myc. Co-IP assays showed that COH1 can physically interact with the N-terminus of COH2, but did not with the C-terminus (S6C and S6D Fig). As with the protein COH2::Myc, COH2-N::Myc also formed dimers (S6C Fig).

The protein COH2::FLAG was used as a representative of the COH2 fusion proteins used in this study (see below) to assay whether the COH2 protein fused with a tag retained its WT activity. To this end, we constructed a COH2::FLAG expression cassette with its coding sequence driven by the Coh2 promoter, which was then transformed into the mutant ΔCoh2 (see below) to produce the strain ΔCoh2-PCoh2-COH2-FLAG. No differences in saprophytic growth and pathogenicity were found between the WT strain, ΔCoh2-PCoh2-COH2-FLAG, and the complemented strain C-ΔCoh2, indicating that the COH2::FLAG protein functioned as COH2 (S7A and S7B Fig). However, this method could not be used to investigate whether the COH1 fusion proteins used in this study retained their WT activity because the native chromosomal position of Coh1 is essential for its promoter activity (S1D and S1E Fig). However, the Coh1-overexpressing strain Coh1^OE was not different in colony growth and pathogenicity from the strain constitutively expressing the fusion protein COH1׃׃Myc (S3C Fig), indirectly showing that COH1 and COH1׃׃Myc had the same functions.

The COH1 and COH2 interaction reduces COH2 stability

The impacts of the physical interaction between COH1 and COH2 on the activity of the transcription factor COH2 were then investigated. We first assayed the impact of the interaction on COH2 stability. To this end, the protein level of the fusion protein COH2::FLAG in the strain WT-COH2-FLAG (the protein COH2::FLAG expressed in the WT strain) was compared with the strain Coh1^OE-COH2-FLAG (COH2::FLAG expressed in the Coh1-overexpressing strain Coh1^OE). Five isolates of the strain Coh1^OE-COH2-FLAG and 2 isolates of the strain WT-COH2-FLAG were selected, as they had the same transcription level of the fusion gene encoding the protein COH2::FLAG (S8A Fig). In the SDY medium, where COH1 was expressed in the strain Coh1^OE-COH2-FLAG but not in WT-COH2-FLAG, 2 isolates of the strain WT-COH2-FLAG both accumulated much higher levels of the protein COH2::FLAG than all 5 isolates of the strain Coh1^OE-COH2-FLAG (Fig 3C). We further assayed whether COH1 reduced the amount of the COH2::FLAG protein in the real hemocoel of insects infected by M. robertsii. With histoimmunochemical staining of the fungal cells collected from the real hemocoel with anti-FLAG antibody, the COH2::FLAG protein was found to be more abundant in the strain ΔCoh1-COH2-FLAG than the strain WT-COH2-FLAG (Coh1 expressed in the real hemocoel), which in turn had more COH2::FLAG protein than the Coh1-overexpressing strain Coh1^OE-COH2-FLAG (Fig 3D). To investigate whether reduction of the COH2::FLAG protein caused by the expression of COH1 was due to degradation by the proteasome, the strains WT-COH2-FLAG and Coh1^OE-COH2-FLAG were grown in the SDY medium supplemented with MG132, a specific inhibitor of the 26S proteasome. As shown in Fig 3E, MG132 treatment increased the level of the protein COH2::FLAG in the strain Coh1^OE-COH2-FLAG, but not in the strain WT-COH2-FLAG, indicating that the interaction between COH1 and COH2 induces the degradation of the COH2::FLAG protein by the proteasome. We further found that the ubiquitination level of the COH2::FLAG protein in the strain Coh1^OE-COH2-FLAG was higher than in WT-COH2-FLAG, indicating that the ubiquitin–proteasome pathway was responsible for the COH2 degradation (Fig 3F).

We then tested the impact of COH1 on COH2 dimerization. To do this, we constructed a strain (Coh1^OE-COH2-FLAG/COH2-Myc) with the fusion proteins COH2::FLAG and COH2::Myc expressed in the strain Coh1^OE (S5 Fig). As in the strain COH2-FLAG/COH2-Myc, with the fusion proteins COH2::FLAG and COH2::Myc expressed in the WT strain (S5 Fig), COH2 dimerized in the strain Coh1^OE-COH2-FLAG/COH2-Myc (S8B Fig), showing that COH1 did not impact COH2 dimerization.

We also assayed the impact of COH1 on the entry of COH2 into the nucleus. To this end, we tried to construct a strain expressing the fusion protein COH2::GFP in the strain Coh1^OE, but this attempt was not successful because the protein COH2::GFP was always separated into the 2 proteins: COH2 and GFP. However, we successfully expressed a fusion protein COH2-N::GFP (GFP fused to the COH2 N-terminus containing the NLS, COH2-DBD, and DID) in the WT strain and the strain Coh1^OE to produce WT-COH2-N-GFP and Coh1^OE-COH2-N-GFP (S5C Fig), respectively. In these 2 strains during saprophytic growth in SDY medium and during real hemocoel colonization, the GFP fluorescence intensity was strongest in the nucleus (S8C Fig). However, using the strain COH1-HA/COH2-N-GFP that simultaneously expressed COH1::HA and COH2-N::GFP, a Co-IP analysis showed that these 2 fusion proteins could not physically contact each other (S8D Fig). In addition, the protein COH2-N::GFP did not undergo dimerization (S5C Fig). Therefore, COH2-N::GFP is not suitable for assaying the impact of COH1 on the entry of COH2 into the nucleus.

COH2 is involved in pathogenicity

To investigate the biological functions of Coh2, we first assayed its expression pattern. qRT-PCR analysis showed that Coh2 was constitutively expressed during saprophytic growth, cuticle penetration, and surrogate hemocoel colonization (S9A Fig). For the strain WT-PCoh2-GFP, with gfp driven by Coh2 promoter in the WT strain, the GFP fluorescent signal in the fungal cells on the insect cuticle was as strong as that in the real hemocoel of infected insects (S9B Fig), showing that Coh2 was constitutively expressed during cuticle penetration and real hemocoel colonization.

We then constructed a Coh2 deletion mutant (ΔCoh2) and its complemented strain C-ΔCoh2 (S1G and S1H Fig). On PDA plates, no significant differences in colony growth, colony phenotype, or conidial yield were seen between ΔCoh2, C-ΔCoh2, and the WT strain (S9C and S9D Fig). With topical application of conidia on the insect cuticle, the LT₅₀ value of ΔCoh2 (17.4 ± 0.59 d) was 2-fold higher than that of the WT strain (8.7 ± 0.05 d) (P < 0.05), but no significant difference (P > 0.05) was found between the WT strain and the complemented strain C-ΔCoh2 (8.6 ± 0.07 d). With direct injection, the LT₅₀ value of ΔCoh2 (4.5 ± 0.24 d) was slightly higher than that of the WT strain (3.6 ± 0.09 d) (P < 0.05), and again the strain C-ΔCoh2 (3.7 ± 0.13 d) was not significantly different (P > 0.05) from the WT strain. On the hydrophobic surfaces of plastic petri dishes, appressorial formation was delayed in the mutant ΔCoh2, whereas no significant difference was observed between the WT strain and C-ΔCoh2 (S9E Fig). Compared with the WT strain, the ability of the mutant ΔCoh2 to penetrate the cuticle was decreased, and again no difference was found between the WT strain and C-ΔCoh2 (S9F Fig).

To investigate how Coh1 interacted with Coh2 to regulate pathogenicity, we constructed the double gene deletion mutant ΔCoh1::ΔCoh2 (S1G Fig). WT-COH2-FLAG and ΔCoh1-COH2-FLAG were strains that overexpressed Coh2 in the WT strain and the mutant ΔCoh1, respectively. With topical application, the LT₅₀ value of the mutant ΔCoh1::ΔCoh2 was not significantly different from those of ΔCoh1 and ΔCoh2, but was significantly higher than those of the strains WT-COH2-FLAG and ΔCoh1-COH2-FLAG. Compared to the WT strain, the virulence of the strains WT-COH2-FLAG and ΔCoh1-COH2-FLAG was also significantly reduced (S9G Fig). With direct injection, the LT₅₀ value of the strain ΔCoh1::ΔCoh2 was not significantly different from those of ΔCoh1, ΔCoh2, WT-COH2-FLAG, and ΔCoh1-COH2-FLAG (S9H Fig). With both inoculation methods, no significant difference in fungal burden was found between the insects infected by ΔCoh1, ΔCoh2, ΔCoh1::ΔCoh2, WT-COH2-FLAG, and ΔCoh1-COH2-FLAG, which had significantly lower fungal burden than insects infected with the WT strain (S2F and S2G Fig). Compared to insects infected by the WT strain, the phenoloxidase activity in insects infected by the mutant ΔCoh2 was increased 2.3-fold (P < 0.05). No significant difference in phenoloxidase activity was found between the insects infected by the WT strain, ΔCoh1, C-ΔCoh2, WT-COH2-FLAG, and ΔCoh1-COH2-FLAG (S2J Fig). No significant differences in the expression levels of the antimicrobials gallerimycin and defensin were found between the insects infected by the WT strain and the mutants ΔCoh1, ΔCoh2, WT-COH2-FLAG, and ΔCoh1-COH2-FLAG (S2K Fig).

Determination of the DNA motif bound by COH2

To investigate how the transcription factor COH2 regulates pathogenicity genes with the regulatory protein COH1, we first used ChIP-Seq analysis with the strain WT-COH2-FLAG to identify the genes directly regulated by COH2. In the ChIP-Seq analysis, the peak caller MACS (Model-based Analysis of ChIP-Seq) identified a total of 562 peaks that were bound by COH2 (S10A Fig), 66.4% of which contained a consensus 7-nucleotide motif (TGA[C/G]T[C/A][G/A]) with multiple possible nucleotides at position 4, 6, and 7 (Figs 4A and S10B). This motif is designated as COH2-BM (COH2 binding motif). Consistent with COH2 forming a dimer, the motif COH2-BM contains a palindrome sequence: TGA and TCA spaced by 1 nucleotide at the fourth position. Electrophoretic mobility shift assay (EMSA) was then used to confirm the binding of COH2 to the motif COH2-BM (TGAGTCT) in the promoter of the gene MAA_04430, a representative of genes with the motif COH2-BM in their promoters determined by the ChIP-Seq analysis. We failed to express the whole protein of COH2 in Escherichia coli, but the portion (Met-1 to Thr-100) containing the predicted DNA binding domain, designated as COH2-DBD, was successfully expressed and the recombinant protein was then purified for EMSA (S10C and S10D Fig). The gel shift assay showed the DNA probe containing the motif COH2-BM (biotin-labeled) bound to the recombinant COH2-DBD protein (Fig 4B). The specific competitor (unlabeled DNA probe) almost completely abolished the DNA band shift (Fig 4B). To assay the importance of each position in the motif COH2-BM for binding to COH2-DBD, unlabeled COH2-BM mutants were constructed by replacing the consensus base with each of the other 3, and these mutants were subsequently used as competitors for the binding of the biotin-labeled DNA probe to the recombinant protein COH2-DBD. Consistent with the data obtained from the ChIP-Seq analysis, only the probe with G to A mutation at position 7 almost completely abolished the band shift of the labeled DNA probe, and the probe with C to A mutation at position 6 and all 3 mutated probes at the space nucleotide (at position 4) also significantly altered the band shift (S10E and S10F Fig). Mutations at other positions also impacted the band shift, but to a lesser extent than the mutated probes described above (S10F Fig).

Download:

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

Fig 4. Identification of hemocoel-colonizing genes regulated by COH1 and COH2.

(A) Chromatin immunoprecipitation sequencing (ChIP-Seq) analysis identified the DNA motif COH2-BM that was bound by the transcription factor COH2. (B) Electrophoretic mobility shift assay (EMSA) confirms the in vitro binding of the biotin-labeled motif COH2-BM (Bio-probe) to the recombinant protein COH2-DBD (COH2 DNA binding domain). The binding activity was demonstrated by the shift of the labeled DNA band prior to the addition of the specific competitor (Cold-probe: the unlabeled motif COH2-BM) in 50-, 100-, 150-, 200-, or 300-fold excess. The tested DNA motif COH2-BM is from the promoter of the gene MAA_04430, which was shown to have the motif COH2-BM by the ChIP-Seq analysis. (C) Chromatin immunoprecipitation quantitative PCR (ChIP-qPCR) analysis confirms that COH2 in vivo binds to the motif COH2-BM in the MAA_04430 promoter in the strain WT-COH2-FLAG, expressing the fusion protein COH2::FLAG. WT-FLAG: a strain expressing the tag FLAG only. (D) ChIP-qPCR confirmation of the in vivo binding of COH2 to the promoters of the 4 destruxin biosynthesis genes (DtxS1, DtxS2, DtxS3, and DtxS4). (E) ChIP-qPCR analysis shows that during surrogate hemocoel colonization, deleting the Coh1 gene increased the binding of COH2 to the promoters of the 4 destruxin biosynthesis genes. ΔCoh1-COH2-FLAG: a strain expressing the protein COH2::FLAG in the mutant ΔCoh1. (F) ChIP-qPCR analysis confirms that during saprophytic growth, COH1 reduced the in vivo binding of COH2 to the promoters of the 4 destruxin biosynthesis genes. Coh1^OE-COH2-FLAG: a strain expressing the protein COH2::FLAG and overexpressing the Coh1 gene. The data underlying all the graphs shown in this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3001360.g004

To validate the ChIP-Seq and EMSA results, we performed ChIP-qPCR with primers designed to cover the motif COH2-BM in the promoter of the gene MAA_04430. After immunoprecipitation, qPCR analysis showed that the copy number of the DNA fragment containing the motif COH2-BM from the strain WT-COH2-FLAG was 29.4-fold higher than that from the strain WT-FLAG, which expressed the FLAG tag only (Fig 4C).

The COH1 and COH2 interaction derepresses COH2-mediated repression of hemocoel colonization genes

As COH1 was only expressed during hemocoel colonization, we thus investigated how COH1 and COH2 interacted to regulate this infection stage. To this end, RNA-Seq analysis was first used to compare the transcriptomes of the WT strain and the mutants ΔCoh1 and ΔCoh2 during surrogate hemocoel colonization. Compared with the WT strain, 68 genes were up-regulated and 207 genes down-regulated in the mutant ΔCoh2 (S11A Fig). Compared with the WT strain, 52 genes were down-regulated and 97 genes up-regulated in the mutant ΔCoh1 (S11A Fig). RNA-Seq analysis showed that neither Coh1 nor Coh2 regulated Hat1expression, and subsequent qRT-PCR analysis further confirmed that there was no significant difference in the expression level of Hat1 between the WT strain and the mutants ΔCoh1 and ΔCoh2, and between the Coh1-overexpressing strain Coh1^OE and the Coh2-overexpressing strain WT-COH2-FLAG (S11C Fig). Remarkably, among the 15 genes that were down-regulated in ΔCoh1 but up-regulated in ΔCoh2 (S11B Fig), 10 were involved in hemocoel colonization, including 4 genes in the destruxin biosynthesis pathway, a siderophore iron transporter (MAA_10457), and a laccase (MAA_00990). To investigate whether these 15 genes are directly regulated by COH2, we searched for the motif COH2-BM in their promoters, which were determined as the approximately 2-kb DNA fragments upstream of the ORF start sites or as the regions, if shorter than 2 kb, between the ORFs of the 15 genes and the ORFs of their respective adjacent genes. The motif COH2-BM was found in the promoters of all the 15 genes.

Using the 4 destruxin biosynthesis genes (DtxS1, DtxS2, DtxS3, and DtxS4) as representatives, we investigated the mechanisms for COH1 and COH2 to regulate the genes involved in hemocoel colonization. qRT-PCR further confirmed that during surrogate hemocoel colonization, compared with the WT strain, these 4 destruxin biosynthesis genes were down-regulated in the mutant ΔCoh1 and up-regulated in the mutant ΔCoh2 (S11D Fig). Using the gene DtxS3 as a representative, we assayed the impact of Coh1 and Coh2 on the expression of the destruxin biosynthesis genes in the real hemocoel of insects infected by M. robertsii. To this end, an expression cassette with gfp driven by the DtxS3 promoter PDtxS3 was transformed into the WT strain and the mutants ΔCoh1 and ΔCoh2 to produce the strains WT-PDtxS3-GFP, ΔCoh1-PDtxS3-GFP, and ΔCoh2-PDtxS3-GFP (S1J Fig). GFP fluorescent intensity in ΔCoh2-PDtxS3-GFP cells collected from the real hemocoel appeared to be stronger than in WT-PDtxS3-GFP, which was in turn stronger than in ΔCoh1-PDtxS3-GFP (S11E Fig).

ChIP-qPCR analysis showed that COH2 occupancy at DtxS1, DtxS2, DtxS3, and DtxS4 promoters in the strain WT-COH2-FLAG was 102.8-, 29.3-, 62.0-, and 13.9-fold higher, respectively, than in the strain WT-FLAG (Fig 4D). Based on the qRT-PCR and ChIP-qPCR data described above, we postulated that COH2 is a repressor of the destruxin biosynthesis genes and that this repression is mitigated by COH1 when the fungus enters the insect hemocoel. To test this, we assayed the impact of the presence of COH1 on the binding of COH2 to the promoters of the 4 genes. To this end, we constructed a strain (ΔCoh1-COH2-FLAG) with the fusion protein COH2::FLAG expressed in the mutant ΔCoh1 (S5A Fig). During surrogate hemocoel colonization, the enrichments of the protein COH2::FLAG in the promoters of Dtxs1, Dtxs2, Dtxs3, and Dtxs4 in the strain ΔCoh1-COH2-FLAG were all higher than in WT-COH2-FLAG (Fig 4E). Conversely, compared with the strain WT-COH2-FLAG, the enrichment of COH2::FLAG at the promoter of Dtxs1, Dtxs2, Dtxs3, and Dtxs4 was reduced 5.3-, 3.4-, 5.3-, and 5.3-fold, respectively, in the strain Coh1^OE-COH2-FLAG (Fig 4F).

COH2 induces the expression of cuticle-degrading genes

As described above, in addition to regulating hemocoel colonization along with the regulator COH1, COH2 also controlled cuticle penetration. To investigate how COH2 regulates cuticle penetration, we used RNA-Seq to compare the transcriptomes of the mutant ΔCoh2 and the WT strain when they were grown on locust hindwings. Compared with the WT strain, 549 genes were down-regulated and 236 genes up-regulated in ΔCoh2 (S11A Fig). The COH2 binding motif COH2-BM was found in 368 genes out of the 549 down-regulated genes, and 171 out of the 236 up-regulated genes. In this study, the G. mellonella larva is used as the host for pathogenicity analysis. Therefore, by qRT-PCR analysis of 5 representative genes, we validated the results obtained with locust hindwings on the cuticle of G. mellonella larva, and no difference in their expression was found between the locust hindwings and the G. mellonella cuticle (S11G Fig).

Remarkably, 44 cuticle-degrading genes were down-regulated in the deletion mutant ΔCoh2, including 30 proteases, 3 chitinases, 2 lipases, and 9 cytochrome P450 enzymes (S11F Fig). Among the 30 protease genes, 27 had the motif COH2-BM in their promoters. The COH2-BM motif was also found in the promoters of the 3 chitinase genes, 1 lipase gene, and 7 cytochrome P450 genes. Two protease genes (MAA_10199 and MAA_10350) and 1 chitinase gene (MAA_10456) were used as representatives to confirm that COH2 directly regulated the cuticle-degrading genes described above. During penetration of the G. mellonella cuticle, qRT-PCR confirmed that these 3 genes were down-regulated in the mutant ΔCoh2 compared with the WT strain (Fig 5A). Since it was not possible to obtain enough biomass to get sufficient DNA for ChIP-qPCR analysis during cuticle penetration, i.e., growing M. robertsii on the insect cuticle, an approximation of cuticle penetration was prepared by growing the fungus in a cuticle medium with the cuticle of G. mellonella larvae as the sole carbon and nitrogen source (called cuticle medium hereafter). When fungus was grown in the cuticle medium, ChIP-qPCR analysis showed that in the strain WT-COH2-FLAG, the COH2::FLAG protein bound to the promoters of these 3 genes (Fig 5B).

Download:

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

Fig 5. COH1 inactivates the COH2-mediated induction of cuticle-degrading genes during hemocoel colonization.

(A) Quantitative reverse transcription PCR (qRT-qPCR) analysis of the expression of cuticle-degrading genes during cuticle penetration in the mutant ΔCoh2 relative to the wild-type (WT) strain. The chitinase MAA_10456 and the proteases MAA_10199 and MAA_10350 are used as representatives. (B) Chromatin immunoprecipitation quantitative PCR (ChIP-qPCR) analysis shows the occupancy of the COH2::FLAG protein in the promoters of the chitinase and protease genes in the strain WT-COH2-FLAG relative to occupancy in the control strain WT-FLAG, which is set to 1. The fungal strains were grown in the cuticle medium. (C) qRT-PCR analysis of the expression levels of the 3 genes during surrogate hemocoel colonization relative to the expression level during cuticle penetration, which is set to 1. (D) ChIP-qPCR analysis of the occupancy of COH2 in the promoters of the 3 genes during surrogate hemocoel colonization (Hemolymph) relative to the growth in the cuticle medium (Cuticle medium), which is set to 1. The strain WT-COH2-FLAG was used. (E) qRT-PCR analysis of the expression levels of the 3 genes during surrogate hemocoel colonization in the mutant ΔCoh1 relative to the expression level in the WT strain, which is set to 1. (F) ChIP-qPCR analysis of the occupancy of the COH2::FLAG protein in the promoters of the 3 genes during surrogate hemocoel colonization in the strain ΔCoh1-COH2-FLAG (the mutant ΔCoh1 with COH2::FLAG expressed) relative to occupancy in the strain WT-COH2-FLAG, which is set to 1. (G) GFP signal in the fungal cells of 2 strains (WT-PMAA_10199-GFP and ΔCoh1-PMAA_10199-GFP) with the gfp gene driven by the promoter of the protease gene MAA_10199. Top panel: WT-PMAA_10199-GFP on the cuticle (Cuticle); middle panel: WT-PMAA_10199-GFP in the real hemocoel of G. mellonella larva infected by a fungal strain (Hemocoel); bottom panel: ΔCoh1-PMAA_10199-GFP in the real hemocoel. F, fungal cells; H, hemocyte; MGV, mean gray value. Scare bar, 10 μm. The data underlying all the graphs shown in the figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3001360.g005

We further investigated the impact of the cuticle-degrading genes regulated by COH2 on the ability of the fungus to degrade the insect cuticle. To this end, we quantified cuticle degradation products following 12 h of fungal growth in the cuticle medium. Although the growth rates of the WT strain and the mutant ΔCoh2 in the cuticle medium were similar, the mutant secreted significantly less total extracellular protease than the WT strain (P < 0.05) (S11H Fig), and it thus released significantly fewer amino acids (S11I Fig) and peptides (S11J Fig) from the cuticle (P < 0.05). Likewise, the WT strain produced significantly (P < 0.05) more chitinase than ΔCoh2 (S11K Fig).

COH1 represses expression of cuticle-degrading genes during hemocoel colonization

The expression of cuticle-degrading genes is inhibited once the fungus enters the insect hemocoel, otherwise they cause hyperactive immunity of the insects, which reduces fungal reproducibility in the insect cadavers [9,24]. We thus investigated whether COH2-mediated induction of the cuticle-degrading genes was inactivated by COH1 when the fungus entered the hemocoel. The 2 proteases MAA_10199 and MAA_10350 and the chitinase MAA_10456 were again used as representatives for cuticle-degrading genes regulated by COH2 during cuticle penetration. Compared with cuticle penetration, the expression levels of the 3 cuticle-degrading genes were all significantly reduced during surrogate hemocoel colonization (Fig 5C). Consistent with the expression patterns of these the 3 genes, ChIP-qPCR analysis showed the that in vivo binding of COH2 to their promoter regions during growth in the cuticle medium was significantly greater than during surrogate hemocoel colonization (Fig 5D). During surrogate hemocoel colonization, the expression levels of these 3 cuticle-degrading genes in the mutant ΔCoh1 was higher than in the WT strain (Fig 5E), and in vivo binding of COH2 to their promoters in ΔCoh1 was also greater than in the WT strain (Fig 5F).

Using the protease gene MAA_10199 as a representative of the 3 cuticle-degrading genes described above, the results about their expression obtained with surrogate hemocoel colonization were validated in the real hemocoel of infected insects by tracing GFP signal in the strains WT-PMAA_10199-GFP and ΔCoh1-PMAA_10199-GFP, which were constructed by transforming the gfp gene driven by the MAA_10199 promoter into the WT strain and the mutant ΔCoh1, respectively (S1J Fig). The GFP fluorescent intensity in WT-PMAA_10199-GFP cells on the cuticle was stronger than in the real hemocoel (Fig 5G). In the real hemocoel, the GFP fluorescent intensity in the strain ΔCoh1-PMAA_10199-GFP was stronger than in WT-PMAA_10199-GFP (Fig 5G).

Gene deletion

M. robertsii ARSEF2575 was obtained from the Agricultural Research Service Collection of Entomopathogenic Fungal Cultures (US Department of Agriculture). E. coli DH5α was used for constructing plasmids. Transformations of M. robertsii were mediated by Agrobacterium tumefaciens AGL1 [30].

Gene deletion and complementation of gene deletion mutants were conducted as previously described [30]. The plasmids for gene deletion were constructed using pPK2-Bar-GFP or pPK2-Sur-GFP as master plasmids [30]. To construct a plasmid for complementing a gene deletion mutant, the genomic clone of the gene—containing the promoter region, ORF, and termination region—was amplified by PCR and inserted into the plasmid pPK2-Sur-GFP [30]. High-fidelity Taq DNA polymerase (KOD Plus Neo; Toyobo, Osaka, Japan) was used in PCR reactions, and all PCR products were confirmed by DNA sequencing. The primers for the construction of gene deletion plasmids and other primers used in this study are presented in S1 Table.

Construction of strains overexpressing Coh1 or Hdac1

The coding sequences of Coh1 and Hdac1 were cloned by PCR and inserted downstream of the constitutive promoter Ptef in the plasmid pPK2-Bar-Ptef [2] to produce plasmid pPK2-Bar-Ptef-Coh1 and pPK2-Bar-Ptef-Hdac1, respectively. The resulting plasmids were then transformed into the WT strain to produce the strains Coh1^OE and Hdac1^OE. The overexpression of the 2 genes was confirmed by qRT-PCR.

Assays of phenoloxidase activity and antimicrobial expression

To assay the impact of fungal infection on phenoloxidase activity and expression of antimicrobial-encoding genes in the G. mellonella larvae, 5 μL of a conidial suspension (1 × 10⁷ conidia/mL) was injected into each larva. After 12 h, the hemolymph of 3 larvae was collected into PBS (phosphate buffered saline) buffer at 4°C, and the protein concentration was determined using the BCA Protein Assay Kit (Meilune, Dalian, China). The hemolymph (5 μg of total protein) was suspended in a CaCl₂ solution (5 mM) to achieve a final volume of 20 μl, which was then mixed with 80 μL of L-DOPA (L-3,4-dihydroxyphenylalanine) solution (20 mM, pH 6.6). After 30 min of incubation at 26°C in the dark, the absorbance at 492 nm was measured. Phenoloxidase activity in the hemolymph was expressed as the absorbance per microgram of protein. L-DOPA solution was used as the blank [31,32]. The control was the hemolymph collected from the uninoculated insects. This experiment was repeated 3 times.

qRT-PCR analysis of the expression levels of antimicrobial-encoding genes was conducted with RNA prepared from the fat bodies of the infected G. mellonella larvae. The G. mellonella 18S rRNA gene (GenBank accession number: AF286298) was used as an internal standard.

Analysis of fungal burden in insects

Absolute qPCR analysis was used to assay the fungal burden in the insects infected by M. robertsii by quantifying the fungal genomic DNA. The 18s rDNA internal transcribed spacer (ITS) sequence was used for qPCR analysis to estimate fungal burden. To construct a standard curve for the absolute qPCR analysis, the genomic DNA of M. robertsii was extracted from mycelium grown in SDY medium, and a serial dilution was made as templates for qPCR. To extract DNA from the infected insects, 5 larvae were surface sterilized with sodium hypochlorite solution (1%) and pooled for DNA extraction. The fungal burden was expressed as the amount of fungal genomic DNA per 1 gram of infected larvae.

Co-IP analysis

In addition to the master plasmids pPK2-Sur-Ptef-FLAG, pPK2-Sur-Ptef-HA, pPK2-Bar-Ptef-FLAG, and pPK2-Bar-Ptef-HA that were used for producing a protein fused with FLAG or HA [33], in this study we also constructed the master plasmid pPK2-Bar-Ptef-Myc by inserting the coding sequence corresponding to 13 repeats of the Myc peptide in the plasmid pPK2-Bar-Ptef [2]. All fusion genes were driven by the constitutive promoter Ptef from A. pullulans [18].

To assay the in vivo physical interaction of COH1 with COH2, the strain COH1-HA/COH2-Myc simultaneously expressing the fusion proteins COH1::HA and COH2::Myc was constructed. The coding sequence of COH1 (stop codon excluded) was cloned by PCR and inserted upstream of the coding sequence of the HA tag in the plasmid pPK2-Sur-Ptef-HA to produce the plasmid pPK2-Sur-Ptef-COH1-HA, which was then transformed into the WT strain to produce a strain called COH1-HA. The coding sequence of COH2 (stop codon excluded) was also amplified by PCR and inserted upstream of the coding sequence of the Myc tag in the plasmid pPK2-Bar-Ptef-Myc, producing the plasmid pPK2-Bar-Ptef-COH2-Myc. This plasmid was then transformed into the strain COH1-HA to produce the strain COH1-HA/COH2-Myc for Co-IP analysis. Similarly, to identify the section of COH2 that physically contacted COH1, the protein was divided into 2 portions: the N-terminus (Met-1 to Thr-100) containing the DNA binding domain COH2-DBD and NLS, and the C-terminus (Ser-101 to Arg-282). The coding sequences of the N-terminus and C-terminus were cloned by PCR and inserted upstream of the coding sequence of the Myc tag in the plasmid pPK2-Bar-Ptef-Myc. The resulting plasmids were then transformed into the strain Coh1-HA to produce the strains COH1-HA/COH2-N-Myc and COH1-HA/COH2-C-Myc for Co-IP analysis.

To assay whether COH2 formed a homodimer, the strain COH2-FLAG/COH2-Myc was constructed. To do this, the plasmid pPK2-Bar-Ptef-COH2-Myc was transformed into the WT strain to produce a strain called COH2-Myc. The coding sequence of COH2 (stop codon excluded) was amplified by PCR and inserted upstream of the coding sequence of the FLAG tag in pPK2-Sur-Ptef-FLAG, producing the plasmid pPK2-Sur-Ptef-COH2-FLAG. This plasmid was then transformed into the strain COH2-Myc to produce the strain COH2-FLAG/COH2-Myc for Co-IP analysis. To assay the impact of COH1 on COH2 dimerization, the coding sequence of COH1 was cloned and inserted downstream of the promoter Ptef in the plasmid pPK2-NTC-GFP-Ptef [26] to produce the plasmid pPK2-NTC-GFP-Ptef-COH1, which was then transformed into the strain COH2-FLAG/COH2-Myc to produce the strain Coh1^OE-COH2-FLAG/COH2-Myc.

Co-IP analysis was performed as previously described [33]. Protein concentration was determined using the BCA Protein Assay Kit (Meilune, Dalian, China). The mouse anti-Myc and anti-HA antibodies were purchased from ABclonal (Wuhan, China), and the rabbit anti-FLAG antibody was purchased from Sigma-Aldrich (St. Louis, MO, US). The Dynabeads protein G beads were purchased from Invitrogen (Carlsbad, CA, US).

Confirming the expression of the fusion proteins in the fungal strains and assaying the presence of target proteins after Co-IP were conducted using standard Western blot analysis. Total fungal proteins were extracted as previously described [33]. All Co-IP assays were repeated at least 3 times.

ChIP-Seq and ChIP-qPCR analysis

The strain WT-COH2-FLAG expressing the fusion protein COH2::FLAG was used for ChIP-Seq to identify the DNA motif bound by the transcription factor COH2. Conidia (10⁸) were inoculated in 100 mL of SDY medium and incubated at 26°C for 36 h with 220 rpm shaking. The mycelium was then collected and subjected to ChIP-Seq analysis that was conducted by the company Wuhan Igenebook Biotechnology (Wuhan, China). The antibody used for immunoprecipitation was anti-FLAG (Huabio, Hangzhou, China), and ChIP-enriched DNA was sequenced using the Illumina HiSeq 2000 sequencing system. After initial quality control to remove sequencing adaptors and low-quality bases, the clean reads were mapped to the genome of M. robertsii using the software BWA (version 0.7.15-r1140). The peak caller MACS (version 2.1.1.20160309; q value < 0.05) was used to localize the potential binding sites of COH2 [34].

ChIP-qPCR was performed as previously described [33], with some modifications. The cross-linked mycelium was ground into fine powder in liquid nitrogen and then resuspended in a lysis buffer. After centrifugation, the chromatin DNA was sheared using M220 (Covaris, Woburn, MA, US) with PIP 75 W, duty factor 10, cycles/burst 200 count, and time 1,500 s. Immunoprecipitation was performed using antibodies (anti-FLAG [Sigma-Aldrich, St. Louis, MO, US], anti-H3Ac, anti-H3K4Ac, or anti-H3K56Ac [Active Motif, Carlsbad, CA, US]) together with Dynabeads protein G beads (Invitrogen, Carlsbad, CA, US). ChIP-enriched DNA was used as a template for qPCR analysis using Thunderbird SYBR qPCR Mix without ROX (Toyobo, Osaka, Japan). Relative enrichment values were calculated by dividing the immunoprecipitated DNA by the input DNA. All ChIP-qPCR analyses were repeated 3 times.

To prepare the mycelium during the surrogate hemocoel colonization, the hemolymph (250 μL) was collected from the last instar G. mellonella larvae and mixed with an equal volume of anticoagulant solution (60 mM trisodium citrate, 52 mM citric acid, 120 mM NaCl, 200 mM glucose, 20 mM EDTA), into which 10 μL of a conidial suspension (1 × 10⁷ conidia/mL) was inoculated. After incubation at 26°C for 36 h with gentle shaking, the culture was treated with a cycle of freezing (−80°C) and thawing (26°C), and then subjected to centrifugation. The pellet was resuspended in sterile water and collected again with centrifugation. This washing with water was repeated 7 times to remove insect hemolymph, and the mycelium was then collected for ChIP-qPCR analysis. To obtain enough biomass to get sufficient DNA for ChIP-qPCR analysis during cuticle penetration, fungal mycelium was grown in the cuticle medium with the cuticle of G. mellonella larvae (1%, V/W) as the sole carbon and nitrogen source. Preparation of the G. mellonella cuticle was conducted as previously described [26]. To this end, the mycelium grown in the SDY medium described above was collected by filtration and then washed with sterile water 3 times. The mycelium was then grown at 26°C for 1 h in the cuticle medium, then collected for ChIP-qPCR analysis.

EMSAs

EMSAs were conducted as previously described [33] to analyze the binding of the COH2 DNA binding domain (COH2-DBD) to a DNA probe containing COH2-BM. The DNA probe was a portion of the promoter of the gene MAA_04430, which was shown to be bound by COH2 by the ChIP-Seq analysis. To express COH2-DBD (Ser-30 to Thr-100) in E. coli, its coding sequence was cloned by PCR and inserted into the EcoRI/HindIII sites of the plasmid pET28a (Invitrogen). The resulting plasmid was transferred into the E. coli BL-21 strain, and the expression of the recombinant protein was induced by isopropyl β-D-1-thiogalactopyranoside at 18°C for 16 h as described in the manufacturer’s instructions (Novagen, Madison, WI, US). The recombinant COH2-DBD protein was then purified to homogeneity with HisPur Ni-NTA resin (Thermo Fisher Scientific, Waltham, MA, US) and gel filtration chromatography with HiLoad 16/600 and 26/600 Superdex 75 prep grade (GE Healthcare Bio-Sciences, Uppsala, Sweden).

To prepare the DNA probe labeled with biotin, one of its strands was biotin-labeled (Tsingke Biological Technology, Hangzhou, China), which was annealed to its complementary strand to form double-strand DNA according to the manufacturer’s instructions. Two unlabeled strands of the DNA probe were commercially synthesized and annealed to form the unlabeled DNA fragment. A mutated DNA probe was prepared exactly as the WT DNA probe with 1 nucleotide in the motif COH2-BM substituted to each of the other 3 ones. The details of the mutated DNA probes are shown in S10E Fig.

EMSAs were conducted using the LightShift Chemiluminescent EMSA Kit (Thermo Fisher Scientific). In the competition assays, an unlabeled DNA probe was added in 300-fold excess.

RNA-Seq and qRT-PCR analysis

Total RNA was extracted with the TRIzol reagent (Thermo Fisher Scientific). The mycelium during saprophytic growth and surrogate hemocoel colonization was prepared as described above for ChIP-qPCR analysis, and was then subjected to RNA extraction. To prepare the mycelium during cuticle penetration, a conidial suspension was applied to locust hindwings placed on a water agar plate. After 30 h of incubation at 26°C, the hindwings and fungal biomass were subjected to RNA extraction.

RNA-Seq analysis was performed by Personal Gene Technology (Nanjing, China). Paired-end sequencing was conducted on the Illumina HiSeq 2000 sequencing platform. Clean reads were mapped to the M. robertsii genome using software HISAT2 (https://daehwankimlab.github.io/hisat2/). Reads that aligned uniquely to the reference sequence were used for gene expression quantification using the FPKM (fragments per kilobase per million fragments) method. Differential expression analysis was performed with DESeq software with an adjusted P value 0.05 (Benjamini–Hochberg method). For the RNA collected from the mycelium grown in the hemolymph, the reads that originated from the insect RNA were filtered, and the remaining reads were mapped to the M. robertsii genome for subsequent analyses.

For qRT-PCR, complementary DNAs (cDNAs) were synthesized with total RNAs with ReverTra Ace qPCR RT Master Mix (Toyobo, Osaka, Japan). qRT-PCR analysis was conducted using Thunderbird SYBR qPCR Mix without ROX (Toyobo, Osaka, Japan). The genes Gpd and tef were used as internal standards as previously described [35]. The relative expression level of a gene was determined using the 2^−ΔΔCt method [36]. All qRT-PCR experiments were repeated 3 times.

Assay of the acetylation level in histone H3

To analyze the acetylation level of histone H3, 100 μg of total protein was electrophoresed on a 12% SDS-PAGE gel and then transferred to a PVDF membrane (Bio-Rad, Hercules, CA, US). The global acetylation level of histone H3 was detected by anti-acetyl histone H3 (Sigma-Aldrich). Antibodies against the acetylation levels of different lysine sites in histone H3 (H3K4, H3K9, H3K14, H3K18, H3K23, H3K27, and H3K56) were all purchased from Active Motif (Carlsbad, CA, US) and diluted 1:2,000 for Western blot analysis. Histone H3 protein was assayed with anti-H3 (Merck Millipore; 1:1,000 dilution). All blots were imaged by the chemiluminescence detection system (Clarity Western ECL, Bio-Rad), and the signal intensities of all bands were quantified using Image J software (https://imagej.nih.gov/ij). The experiments were repeated 3 times.

Yeast 2-hybrid and autoactivation assays

Yeast 2-hybrid assays were performed according to the manufacturer’s instructions (Clontech, Japan). The interactions between COH1 and 5 transcription factors were assayed. The coding sequence of each of the 5 transcription factors (MAA_08013, MAA_06777, MAA_07566, MAA_06937, and MAA_07838) was cloned by PCR and inserted into the plasmid pGADT7. The coding sequence of COH1 was also cloned by PCR and inserted into the plasmid pGBKT7 to produce the plasmid pGBKT7-COH1. All pGADT7-based plasmids were transformed into Y187 cells, whereas pGBKT7-COH1 was transformed into Y2HGold cells. The Yeastmaker Yeast Transformation System 2 (Takara Bio, Tokyo, Japan) was used to prepare yeast-competent cells. After mating, the resulting strains were grown on the medium SD/−Ade/−His/−Leu/−Trp with X-α-gal and AbA (Aureobasidin A; Takara Bio). The autoactivation of COH1 was tested by inoculating the strain containing the plasmid pGBKT7-COH1 on the medium SD/−Ade/−His/−Trp with X-α-gal. Yeast 2-hybrid and autoactivation assays were repeated 3 times.

Inhibition of the activity of the ubiquitin–proteasome pathway

Conidia (10⁸ in total) were inoculated into 100 mL of SDY medium and grown at 26°C for 30 h, and the 26S proteasome inhibitor MG132 (200 μM) was then added. After 4 h of incubation, the mycelium was collected by filtration for protein preparation. To assay the ubiquitination level of the protein COH2::FLAG, immunoprecipitation with the mouse anti-FLAG (HUABIO, China) was performed as previously described [30]. The rabbit anti-ubiquitin antibodies were purchased from ABclonal (Wuhan, China).

Histoimmnunochemical staining

Histoimmnunochemical staining analysis of the fusion protein COH2׃׃FLAG was conducted as previously described [2]. The primary antibody was anti-FLAG (HUABIO, China), and fluorescein isothiocyanate (FITC)–conjugated goat anti-mouse IgG (HUABIO, China; dilution 1:200) was used for secondary labeling.

Analysis of gene expression by tracing GFP signal

To investigate expression of a gene in the real hemocoel of the infected insects by tracing GFP signal, we constructed a strain with GFP coding sequence driven by the promoter of this gene. The conidia (10⁵ conidia in 10 μL of 0.01% Triton X-100 solution) of this fungal strain were injected into a G. mellonella larva. After 36 h, the fungal cells in the hemocoel were collected for GFP signal observation and quantification. To analyze the expression of the gene during cuticle penetration, 200 μL of the conidial suspension was inoculated on the G. mellonella cuticle, and GFP signal was observed and quantified after 36 h of incubation.

The fluorescence intensity across a hypha in the images described above was quantified using ImageJ software (https://imagej.nih.gov/ij). To this end, an image was saved in a Tiff format, and its bit depth was then changed to 8 bits, followed by subtracting the general background using a 50-pixel “rolling ball diameter” [37]. The hyphae in the image were then selected, and the area, the mean gray value (MGV), and the MGV’s standard deviation for the selected regions were then calculated. The fluorescence intensity of the selected hyphae is shown as the MGV.

Pathogenicity assays

The G. mellonella larvae (RuiQing Bait Company, Shanghai, China) were used to assay the pathogenicity of M. robertsii. Inoculations were conducted via topical application of conidia on the insect cuticle or by injection of conidia into the insect hemocoel [10]. All bioassays were repeated 3 times with 40 insects per repeat. Assays of appressorial formation on the hydrophobic surface of a petri dish (Corning, Corning, NY, US) were conducted as previously described [2].

The ability to penetrate the insect cuticle was assayed as previously described [33]. Briefly, 2 μL of a conidial suspension (1 × 10⁷ conidia/mL) was applied on the center of an intact cuticle that was placed on a PDA plate. After incubation at 26°C for 48 h, the cuticle was removed, and the PDA plate was further incubated to allow the growth of the fungi, if any, that had breached the cuticle and reached the PDA medium. The stronger the ability of the fungi to penetrate the insect cuticle, the more fungi reached the PDA medium, i.e., the larger the colony that appeared on the PDA plate following removal of the cuticle and incubation.

To assay the ability of the fungus to produce blastospores in the insect hemocoel, 5 μL of a conidial suspension (1 × 10⁵ conidia/mL) was injected into the hemocoel of G. mellonella larvae. Then, 3 and 4 d after injection, 10 larvae per treatment were surface sterilized with 1% bleach and individually bled as previously described [10]. Aliquots of hemolymph (50 μL) from each larva were mixed with 50 μL of the anticoagulant solution and spread onto a selective medium [10]. After incubation at 26°C for 4 d, the colony forming units (CFU) were determined. All assays were repeated at least 3 times.