The wild-type strain C. globosum NK102, isolated and stocked by our laboratory was used as the host strain for gene deletion experiments in this study. NK102 was grown on potato dextrose agar (PDA) at 28°C. For preparation of protoplasts, approximately (1~4) × 10⁴ ascospores, collected from a single plate were used to inoculate 100 ml of potato dextrose broth (PDB) medium, shaken at 28°C and 200 rpm, for 2 days. Ten grown mycelial balls were transferred into 100 ml fresh PDB medium, continued to grow for 3 days at 28°C with a shaking speed of 160 rpm. For DNA isolation, 5-mm agar plaques containing the fungal hypha were inoculated in 100 ml PDB and incubated for 4 days in a rotary shaker at 28°C and 200 rpm. Under identical culture conditions, mycelium from 4 days cultures were chosen to isolate RNA for RNA-seq profiling, and mycelium from 7 days cultures were chosen for qRT-PCR analysis. For high-performance liquid chromatography (HPLC) analysis, strains were cultured in 100 ml PDB for 7 days with shaking at 28°C and 200 rpm. Escherichia coli, used for constructing plasmids and conjugation, was grown in Luria-Bertani (LB) broth (Difco) or on LB agar plates at 37°C.

Recently, we successfully established a “suicide” CRISPR-Cas9 system to promote gene deletion in C. globosum NK102 (Wang Y. et al., 2016; Xiang et al., 2021). Plasmids for Cgtf1 or Cgtf6 deletion were constructed based on plasmid pCRISPR-Hyg (Supplementary Figure 1). To generate the Cgtf1-targeted gene knockout vector, vector pCRISPR-Hyg was linearized by restriction enzyme Bbs I. The GN₁₉NGG sequence of Cgtf1 was searched for designing the complementary primers N19-F and N19-R. The cohesive terminus of Bbs I was added at the 5′ and 3′ terminal of each primer, respectively. These primers were synthesized, denatured (95°C, 10 min) and anneal naturally (room time) to generate the GN₁₉NGG fragment. The GN₁₉NGG fragment was ligated into the Bbs I site of pCRISPR-Hyg to generate plasmid PCgtf1-N19. The upstream (753 bp) and downstream (736 bp) flanking sequences of the Cgtf1 were amplified from NK102 genomic DNA with primers 01237UP-F/01237UP-R and 01237DOWN-F/01237DOWN-R, respectively. PCgtf1-N19 was cut by Xho I to generate a large fragment containing Cas9 and a small fragment containing a Hyg^R cassette bearing the hygromycin B resistance marker. The upstream of Cgtf1, Hyg^R cassette, downstream of Cgtf1 were fused with the large fragment via In-Fusion HD Cloning kits (Clontech, California, USA). The merged plasmid, designated as pCgtf1, was linearized with Nru I and introduced into the NK102 via PEG mediated protoplast transformation.

To generate the Cgtf6-targeted gene knockout vector, the GN₁₉NGG sequence of Cgtf6 was searched for designing the complementary primers 07161-N19-F and 07161-N19-R. These primers were synthesized, denatured and anneal naturally to generate the GN₁₉NGG fragment. The GN₁₉NGG fragment was ligated into the Bbs I site of pCRISPR-Hyg to generate plasmid PCgtf6-N19. Nucleotide fragments corresponding upstream (631 bp) and downstream (736 bp) sequences based on the Cgtf6 sequence were cloned by PCR using C. globosum NK102 genomic DNA and pair primers 07161-UP-F/07161-UP-R and 07161-DOWN-F/07161-DOWN-R, respectively. Then the upstream of Cgtf6, Hyg^R cassette, downstream of Cgtf6 were fused as mentioned above. The merged plasmid, designated as pCgtf6, was linearized with Nru I and introduced into the NK102 via PEG mediated protoplast transformation. All primers were listed in Supplementary Table 1.

Protoplast preparation and transformation were performed as previously described (Hu et al., 2018). Transformants were selected on PDA plates containing 100 μg/ml hygromycin B (Sigma-Aldrich, St Louis, MO, USA).

For screening of the Δcgtf1 mutants, three PCR reactions per transformant were carried out, primers T01237up-F/THyg-R and THyg-F/T0127down-R were used to detect homologous recombination of the 5′ and 3′ flanks, respectively, and primers L01237Kpn I-F/L01237Hind III-R were used to detect the ORF of gene Cgtf1. For further confirmation, genomic DNA was extracted, digested and subjected to Southern blot analysis as previously described (Hu et al., 2018). A 695 bp 3′-flank of Cgtf1 and a 2,124 bp Hyg^R fragment were labeled as probes.

For screening of the Δcgtf6 mutants, three PCR reactions per transformant were carried out, primers Cgtf6-Up-Hyg-F/Cgtf6-Up-Hyg-R and Cgtf6-Hyg-Down-F/Cgtf6-Hyg-Down-R were used to detect homologous recombination of the 5′ and 3′ flanks, respectively, and primers qCHGG_07161-F/qCHGG_07161-R were used to detect the ORF of gene Cgtf6. For further confirmation, genomic DNA was extracted and subjected to Southern blot analysis as previously described (Hu et al., 2018). Plasmid pCgtf6 was cut by Xho I and generated a 2,131 bp Hyg^R cassette, which was subsequently labeled as the probe. Experiments involving DNA labeling, hybridization and detection were carried out according to the instructions of the DIG High Prime DNA Labeling and Detection Starter Kit II (Roche China, Shanghai, China). All primers were listed in Supplementary Table 1.

Total RNA was extracted from the lyophilized and ground mycelium using an RNAiso Plus kit (TaKaRa, Beijing, China). The first-strand cDNA was generated by reverse transcription in a 20 μl reaction using Transcript First-Strand cDNA Synthesis Super Mix kit (TransGen Biotech, Beijing, China). Quantitative real-time PCR was performed by Light Cycler^® 480 instrument II (Roche China, Shanghai, China). Each reaction of 20 μl PCR was performed with SYBR Green I PCR master mix (Roche China, Shanghai, China). Reactions were set up in three replicates per sample. Controls without addition of the templates were included for each primer set. PCR cycling parameters were: pre-incubation at 94°C for 10 min, followed by 40 cycles of denaturation at 95°C for 10 s, annealing at 55°C for 20 s and extension at 72°C for 20 s. The qRT-PCR data were analyzed using the 2^−ΔΔCt relative quantification method (Livak and Schmittgen, 2001) to calculate relative expression levels of genes. The housekeeping genes encoding actin or glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were served as reference. The amplification efficiencies of the target and reference genes were compared at different template concentrations. The gene-specific pairs of primers used in the amplifications were: q01237-F/q01237-R for Cgtf1, qCHGG_07161-F/qCHGG_07161-R for Cgtf6, qCHGG_01239(s)/qCHGG_01239(as) for CgcheA, qPKS(s)/qPKS(as) for pks-1, qCHGG_01240-F/qCHGG_01240-R for CHGG_01240, qCHGG_01242-F/qCHGG_01242-R for CHGG_01242, qCHGG_01243-F/qCHGG_01243-R for CHGG_01243, qCHGG_01244-F/qCHGG_01244-R for CHGG_01244, qCHGG_02034-F/qCHGG_02034-R for CHGG_02034, GAPDH-up/GAPDH-down for GAPDH, and qActin(s)/qActin(as) for the ACTIN gene (Supplementary Table 1).

Each sample of liquid culture was extracted with an equal volume of ethyl acetate. The organic phase was then concentrated, dissolved and centrifuged as previously reported (Hu et al., 2018). The supernatant was filtered through a 0.45 μm Millipore filter and subjected to HPLC analysis on an Agilent 1,200 HPLC system (Agilent Technologies, CA, USA) with a Kromasil C18 ODS column (4.6 × 250 mm, AKZO Nobel, Gland, Switzerland). The UV detection wavelength was set at 227 nm, and the sample flow rate was set at 1 ml/min. Standard ChA (Sigma, St. Louis, USA) served as control. For quantification of ChA, a standard curve was created with known concentrations of the standard sample.

Ascospores were harvested from 30-day old cultures on PDA at 28°C in triplicate for each strain. Firstly, 2 ml sterile distilled water was added to each plate. Then, ascospores formed on each plate were scrapped using a sterilized spatula, resuspended in 30 ml sterile distilled water and vortex by a Vortex-Genie 2 Mixer (Scientific Industries, New York, USA) for 1 min at the highest speed. After centrifugation at 3,000 × g for 10 min, the supernatant full of hyphae-debris was discarded, and the ascospores pellet was washed twice with sterile distilled water and finally resuspended in 10 ml sterile distilled water. The concentration of ascospores suspension in sterile distilled water was determined by hemocytometry under the Moticam×3 microscope (Motic, Xiamen, China).

RNA-seq profiling was carried out by a commercial provider to monitor the consequences of the Cgtf1 knock-out. Illumina HiSeq™ sequencing of total mRNA from the wild-type or deficient mutants Δcgtf1-10 and Δcgtf1-14 was conducted by BGI (Shenzhen, China; http://en.genomics.cn/navigation/index.action). Raw reads were counted using SOAPnuke version 1.4.0 (BGI, Shenzhen, China) and then were trimmed using trimmomatic (version 0.36). The resulted clean reads were mapped to the C. globosum reference genome (CBS 148.51) using HISAT(Hierarchical Indexing for Spliced Alignment of Transcripts), (version 2.1.0) and were aligned using Bowtie 2 (version 2.2.5). Expression values were calculated as fragments per kilo base of transcript per million mapped reads (FPKM) using RSEM (version 1.2.8), (Li and Dewey, 2011). P-values were used to evaluate expression differences at a statistically significant level (Mortazavi et al., 2008). A false discovery rate (FDR)-corrected p-value ≤ 0.001 and |Log2 (Ratio value)| ≥1 were used to identify the differentially expressed genes (DEGs) and differentially expression tags (DETs).

The 1611-bp Cgtf1 gene was cloned, subjected to sequencing and submitted to the GenBank database under accession number MT050051. The predicted protein CgTF1 encoded by Cgtf1 is a 536 amino acids sequence comprises two overlapping domains, a GAL4-like Zn(II)₂/Cys₆ binuclear cluster DNA-binding domain (GAL4, cd00067) and a fungal Zn(II)₂Cys₆ binuclear cluster domain (Zn_clus, pfam00172), (Figure 1). The amino acid sequence of CgTF1 is identical to CgCheR. Blastp analysis revealed that CgTF1 had 77.19, 38.10, 30.56, 26.10 and 24.64% identity to a GAL4-like Zn₂Cys₆ binuclear cluster DNA-binding domain protein (C. globosum E-C-2, QHD43140.1), CheC (P. expansum, CAO91859.1), CheF (P. expansum, CAO91864.1), GAL4 (Aspergillus sclerotialis, RJE27110.1) and Gal4p (S. cerevisiae, NP_015076.1), respectively. To characterize the functional roles of CgTF1 on ChA biosynthesis, CgTF1 was deleted using an established “suicide” CRISPR-Cas9 system (Xiang et al., 2021). All gene-targeted plasmids were constructed based on plasmid pCRISPR-Hyg (Supplementary Figure 1), which contain the Cas9 and gDNA expression cassettes and a hygromycin B resistance marker. To target Cgtf1, the GN₁₉NGG fragment was ligated into pCRISPR-Hyg to generate plasmid PCgtf1-N19, two flanking sequences of the Cgtf1 were then fused with the hygromycin-resistant gene, respectively. The resulted construct, designated pCgtf1 (Supplementary Figure 2A), was linearized and introduced into the C. globosum NK102 via PEG mediated protoplast transformation. PCR amplification and DNA sequencing revealed that the correct gene deletion occurred at the target locus in transformants Δcgtf1-10 and Δcgtf1-14 (Supplementary Figures 2A,B). Southern blotting and RT-PCR were employed for further verification (Supplementary Figures 2C–F).

CgcheR mutation resulted in the almost complete loss of spores as well as decreased radial growth in C. globosum CBS148.51 (Cheng et al., 2021). Consistent with CgCheR, CgTF1 is essential for sporulation. No significant difference was observed in hypha growth rate between CgTF1 null mutants and the wild-type (Supplementary Figure 3). However, CgTF1 deletion resulted in delayed generation of perithecium and inhibition of ascospore development (Figure 2A). No ascospore was observed under the microscope even for a 30 days culture, indicating that the CgTF1 null mutant has lost the ability to produce ascospores (Figure 2B). As pks-1 was previously proven to be essential for sporulation and pigmentation in C. globosum NK102 (Hu et al., 2012), we determined the mRNA levels of pks-1 in Cgtf1 deficient mutants. As anticipated, Cgtf1 deletion resulted in down regulation of pks-1 transcription, e.g., the mRNA level was dramatically reduced to 44% in Δcgtf1-10 (Figure 2C). These results suggested that CgTF1 regulates transcription of pks-1.

CgCheR has been reported to be involved in ChA biosynthesis by activating the transcription of chaetoglobosin biosynthetic genes in a pathway-specific manner (Cheng et al., 2021). To determine the function of CgTF1 in ChA biosynthesis, HPLC analysis was performed to quantify the concentration of ChA in Cgtf1 deletants. In 7-day-old fermentation broths, no ChA was detected in CgTF1 null mutants (Figure 3A). Subsequently, qRT-PCR was performed to detect the mRNA level of genes within the ChA biosynthetic gene cluster in Cgtf1 deletants. As shown in Figure 3B, the core gene, CgcheA (CHGG_01239) was significantly down regulated in Cgtf1 deletants. The mRNA level of CgcheA in Δcgtf1-10 and Δcgtf1-14 was dramatically reduced to 10.8 and 2.3%, respectively, of that in the wild-type strain. Similarly, CgTF1 deletion also led to a general decrease of transcription of the other genes within the cluster. Most of all, CgcheB (CHGG_01240) and CgcheG (CHGG_01243) were barely expressed in CgTF1 null mutants (Figure 3C). These results confirmed that CgTF1 regulates transcription of CgcheA gene cluster to control ChA biosynthesis as well as CgCheR.

To assess a global profile of genes regulated by CgTF1, an RNA-Seq profiling analysis was performed to identify the differentially expressed genes associated with the biosynthesis of ChA in the Cgtf1 deletants Δcgtf1-10 and Δcgtf1-14. Total RNA was extracted from mycelium grown in PDB medium for 4 days for the Illumina HiSeq™ sequencing as described in experimental procedures. The resulting sequences were aligned to the reference genome of C. globosum CBS 148.51 (assembly ASM14336v1) and the information was used to analyze the DEGs between the wild-type strain and CgTF1 null mutants. DEGs were selected based on the FDR-corrected p-value ≤ 0.001 and |Log2 (Ratio value)| ≥ 1. In a total of 1,388 differentially expressed genes, 548 genes were up-regulated, while 840 genes were down-regulated in Cgtf1 deletants (data in Supplementary Table 2). Consistent with the qRT-PCR results mentioned above, genes within the ChA biosynthetic gene cluster were down-regulated simultaneously in Cgtf1 deletants (Table 1). However, the gene with the highest gene expression variation (log2 ratio = −4.76) was CHGG_01244, not genes encoding enzymes for synthesis of the skeleton structure of ChA, such as CgcheA or pks-1. Transcription of genes previously reported to be involved in regulation of ChA biosynthesis was also affected by CgTF1. For example, the expression of CglaeA was increased with a gene expression variation of 0.89 in Δcgtf1 compared to the wild-type, while the expression of gna-1 was decreased with a gene expression variation of −0.41 in Δcgtf1 compared to the wild-type.

Gene ontology (GO) analysis for the differentially expressed genes was performed to identify their function. Based on the Blast2GO analysis of sequence homology, 989 annotated sequences that had received Blast hits from the non-redundant NCBI protein database were classified into 28 functional groups under the main categories of the GO classification (Figure 4A and Supplementary Table 2). The top four enriched GO terms in biological process were cellular process (220 genes), biological regulation (72 genes), localization (99 genes), and metabolic process (243 genes). The top four enriched GO terms in cellular component were membrane part (359 genes), cell (165 genes), organelle (118 genes) and membrane (365 genes). Moreover, catalytic activity (508 genes), binding (443 genes), transporter activity (88 genes) and transcription regulatory activity (58 genes) were significantly enriched in the molecular function subgroup. It should be noted that most of the genes in GO term of transcription regulatory activity were putative zinc-binding protein-encoding genes (Figure 4B). It is suggested that the CgTF1 may act as a regional regulator and involve in other secondary metabolism pathways except for ChA biosynthesis. Thus, we screened for genes with high expression variation (|log2 ratio| ≥ 2.00). A total of 335 genes, of which were 7 putative zinc finger transcription factor genes were picked out (Supplementary Table 2 and Table 2). In order to dig the downstream effectors of CgTF1 and reveal the crosstalk between CgTF1 and these zinc finger transcription factors, we selected two genes CHGG_ 02034 (log2 ratio = −3.17) and CHGG_07161 (log2 ratio = −2.68) that with the highest gene expression variation to delete (Table 2).

We first examined expression levels of CHGG_02034 and CHGG_07161 in the wild type of C. globosum NK102. qRT-PCR analysis indicated that CHGG_07161 indeed expressed at a normal level and down regulated in mutant Δcgtf1 (Figure 4C), while the expression of CHGG_02034 was repressed in the wild-type (data not shown). Thus, CHGG_07161 was selected for further investigation. CHGG_07161 was located at contig_NT165980.1 of scaffold_5, which was different from CgcheR (located at contig_NT165976.1 of Scaffold_1). A single-copy homolog of CHGG_07161 was found in the genome of C. globosum NK102, named Cgtf6. The 1772-bp gene Cgtf6 was cloned, subjected to sequencing and submitted to the GenBank database under accession number MT050052. We conducted a Blastp analysis against EggNOG database (http://eggnogdb.embl.de/) for more information about Cgtf6. It was indicated that the predicted protein CgTF6 encoded by Cgtf6 is a 400 amino acids sequence comprises a ZnF_C₂H₂ domain (SM00355), which shares a high sequence identity (52.64%) with the gene XP_003662155.1 of Myceliophthora thermophila ATCC 42464 (Supplementary Figure 4A).

In order to investigate the function of CgTF6 and figure out the correlation between CgTF1 and CgTF6, we tried to knock out the gene Cgtf6 by CRISPR-Cas9 system as mentioned above. In order to improve the efficiency of homologous recombination, the distance between two flanking fragments was shortened to 69 bp (Supplementary Figure 4B). The gene deletion construct pCgtf6 was linearized and introduced into the C. globosum NK102 via PEG mediated protoplast transformation. Six transformants were identified by diagnostic PCR (Supplementary Figure 4C). Two of them were subsequently subjected to Southern blot analysis. Only Δcgtf6-6 was confirmed the desired disruption of Cgtf6 by the marker Hyg^R via double crossover at the target site, and with a single-copy of integration (Supplementary Figures 4D,E).

An 8-day comparison on colony diameter as strains growing indicated that CgTF6 deletion has no significant effect on hypha growth rate of C. globosum (Supplementary Figure 5). However, CgTF6 deletion significantly inhibited sporulation of C. globosum. The number of ascospores produced by the wild type was (8.67 ± 0.16) × 10⁷ per plate, while dramatically reduced to (1.19 ± 0.03) × 10⁷ per plate as CgTF6 deleted (Figures 5A,B). Consistent with this phenotype, the mRNA level of pks-1 in Δcgtf6 was significantly decreased to 19.9% of that in the wild-type strain (Figure 5C).

We examined the effect of CgTF6 deletion on ChA biosynthesis. HPLC analysis demonstrated that the yield of ChA in the mutant Δcgtf6 was significantly increased to 1.69-fold of that in the wild-type (Figures 6A,B). Considering down regulation of CgTF6 in Δcgtf1 mutants (Figure 4C), we speculated CgTF6 acted as an end product feedback effector and interacted with CgTF1 to maintain a tolerable concentration of ChA for cell survival. Thus, deletion of Cgtf6 probably removed this inhibition effect and promoted ChA biosynthesis. Subsequently qRT-PCR analysis confirmed this speculation. As shown in Figure 6C, transcript level of Cgtf1 in the Δcgtf6 mutant was significantly increased to 2-fold of that in the wild-type. In order to reveal the regulation mechanism of CgTF6 on ChA biosynthesis, we then examine the mRNA levels of other genes within ChA biosynthesis gene cluster in mutant Δcgtf6 (Figure 6D). Transcript levels of core genes CgcheA and CgcheB were down regulated, while two oxidative modification enzymes encoding genes CHGG_01242 and CHGG_01243 demonstrated markedly increase in different degrees. In addition, the mRNA level of unknown gene CHGG_01244 was also increased sharply in Δcgtf6. To sum up, CgTF6 probably inhibited transcription of CgTF1, and then inhibited transcription of modification enzyme encoding genes to suppress ChA biosynthesis.