Verification of TRI3 Acetylation of Trichodermol to Trichodermin in the Plant Endophyte Trichoderma taxi

Trichodermin, a trichothecene first isolated in Trichoderma species, is a sesquiterpenoid antibiotic that exhibits significant inhibitory activity to the growth of many pathogenic fungi such as Candida albicans, Rhizoctonia solani, and Botrytis cinerea by inhibiting the peptidyl transferase involved in eukaryotic protein synthesis. Trichodermin has also been shown to selectively induce cell apoptosis in several cancer cell lines and thus can act as a potential lead compound for developing anticancer therapeutics. The biosynthetic pathway of trichodermin in Trichoderma has been identified, and most of the involved genes have been functionally characterized. An exception is TRI3, which encodes a putative acetyltransferase. Here, we report the identification of a gene cluster that contains seven genes expectedly involved in trichodermin biosynthesis (TRI3, TRI4, TRI6, TRI10, TRI11, TRI12, and TRI14) in the trichodermin-producing endophytic fungus Trichoderma taxi. As in Trichoderma brevicompactum, TRI5 is not included in the cluster. Functional analysis provides evidence that TRI3 acetylates trichodermol, the immediate precursor, to trichodermin. Disruption of TRI3 gene eliminated the inhibition to R. solani by T. taxi culture filtrates and significantly reduced the production of trichodermin but not of trichodermol. Both the inhibitory activity and the trichodermin production were restored when native TRI3 gene was reintroduced into the disruption mutant. Furthermore, a His-tag-purified TRI3 protein, expressed in Escherichia coli, was able to convert trichodermol to trichodermin in the presence of acetyl-CoA. The disruption of TRI3 also resulted in lowered expression of both the upstream biosynthesis TRI genes and the regulator genes. Our data demonstrate that T. taxi TRI3 encodes an acetyltransferase that catalyzes the esterification of the C-4 oxygen atom on trichodermol and thus plays an essential role in trichodermin biosynthesis in this fungus.


INTRODUCTION
Trichothecenes are a group of naturally occurring sesquiterpenoids capable of inhibiting the function of eukaryotic ribosomes, thereby causing hemorrhagic lesions, dermatitis, and immunological problems to people and animals (Pace et al., 1998;McLaughlin et al., 2009;McCormick et al., 2011). All trichothecenes contain a core tricyclic 12,13-epoxytrichothec-9-ene (EPT) structure with an epoxide substituent and could be divided into four different types according to the pattern of esterification and oxygenation (Ueno, 1984;Ueno and Hsieh, 1985;Cole et al., 2003). Trichodermin belongs to the type that lacks the oxygen substitution at C-8 while containing a hydroxyl residue at C-4 and is renowned as a very potent inhibitor of protein synthesis in eukaryotic cells (Carrasco et al., 1973;Cundliffe and Davies, 1977;Kubicek and Druzhinina, 2015). It primarily interferes in the chain-termination process on the ribosome by blocking the interaction of the peptidyl transferase with the peptide release factor (Wei et al., 1974). Trichodermin was first isolated from a Trichoderma species as an antifungal antibiotic that is active against a variety of pathogenic fungi including Candida albicans (Godtfredsen and Vangedal, 1965). This sesquiterpenoid metabolite is known to be produced by only two Trichoderma species (Trichoderma brevicompactum and Trichoderma arundinaceum; Kubicek and Druzhinina, 2015;Nielsen et al., 2005) and exhibits significant inhibition to several human pathogenic Candida spp. and a nosocomial strain of the filamentous fungus Aspergillus fumigatus (Tijerino et al., 2011). The ability of these Trichoderma species to produce trichodermin, to secrete extracellular hydrolytic enzymes, and to directly interact with plants has made them effective biocontrol agents against plant pathogens like Rhizoctonia solani, Botrytis cinerea, and some soil pathogens (e.g., Fusarium spp.; Howell, 2003;Kulikov et al., 2006;Malmierca et al., 2012). A silkworm infection assay demonstrated that trichodermin yielded therapeutic effects on silkworms infected by C. albicans (Uchida et al., 2016). Recently, trichodermin has drawn attentions from cancer biologists because it can induce cell apoptosis through mitochondrial dysfunction and endoplasmic reticulum stress in human chondrosarcoma cells and by mitotic arrest and DNA damage in human p53-mutated pancreatic cancer cells (Su et al., 2013;Chien et al., 2016). These new findings suggest that trichodermin is a potential therapeutic agent worthy of further development to a clinical trial candidate for treating cancer. To this end, a thorough understanding of the biosynthetic pathway leading to trichodermin is essential.
Trichothecene biosynthesis in Trichoderma shares several features with that in Fusarium yet exhibits some distinct differences in gene clustering, gene arrangement, and gene functions (Malmierca et al., 2013). For example, orientation of part of TRI genes is opposite to that in Fusarium , and TRI5 gene-which encodes a trichodiene synthase that in T. arundinaceum coverts farnesyl pyrophosphate (FPP) into trichodiene, consistent with the function of FsTRI5 in trichothecenes biosynthesis (Hohn and Middlesworth, 1986)-is not present in the TRI cluster . In addition, some enzymes encoded by the TRI cluster show several catalytic differences: the cytochrome P450 monooxygenase TaTRI4 exhibits a different pattern of oxidation: in T. arundinaceum, TaTRI4 introduces oxygen into trichodiene at positions C-2, C-11, and C-13 , whereas the Fusarium orthologue TRI4 further targets position C-3 (McCormick et al., 2006;Tokai et al., 2007); TaTRI11 of T. arundinaceum functions as a hydroxylase, which catalyzes the hydroxylation of EPT at the C-4 position in contrast to that of the C-15 position in Fusarium sporotrichioides (Alexander et al., 1998). TRI3 of Fusarium is responsible for acetylation of hydroxyl group at C-15 in T-2 toxin biosynthesis (McCormick et al., 1996;Cardoza et al., 2011), while the role of TRI3 in Trichoderma has not been characterized until recently. TRI3 had been proposed to take part in the acetylation of C-4 hydroxyl on trichodermol in Trichoderma  and was evidenced to be responsible for trichodermin biosynthesis in an endophytic T. brevicompactum by mutational analysis (Shentu et al., 2018). Yet evidence showing direct participation of TRI3 in trichodermol biotransformation is still lacking, and study of trichodermin biosynthetic pathway in other Trichoderma species rather than T. arundinaceum and T. brevicompactum has not been reported.
We have previously identified a novel trichoderminproducing endophytic fungus from Chinese yew Taxus mairei, Trichoderma taxi ZJUF0986 (Zhang et al., 2007). It was then shown to inhibit the growth of several fungal plant pathogens, including (but not limited to) B. cinerea and R. solani (Zhang et al., 2007;Chen et al., 2008;Shentu et al., 2013). The objective of this study was thus to identify the core genes responsible for the synthesis of trichodermin in the species and-more importantly-to investigate the role of TRI3 in producing trichodermin in Trichoderma. The results reveal that TRI clusters in T. taxi are significantly similar to those in T. arundinaceum and T. brevicompactum and that TtTRI3 encodes an acetyltransferase that performs the step by acetylating trichodermol at C-4 to trichodermin.

Microbial Strains Used
Endophytic T. taxi ZJUF0986 was isolated from T. mairei (Zhang et al., 2007) and is available in the China Center for Type Culture Collection under CGMCC No. 1672. R. solani (ACCC 3614) was used as a test pathogen as described previously (Wang et al., 2012). T. taxi and its derivatives and R. solani were maintained on potato dextrose agar (PDA) plates at 25 • C in the dark. Escherichia coli DH5α was used for regular cloning and plasmid maintenance. E. coli M15 (Qiagen, Hilden, Germany) was adopted as protein expression host.

Identification and Sequence Analysis of TRI Gene Cluster in Trichoderma taxi
The genome of ZJUF0986 was obtained by sequencing (manuscript in preparation). Nucleotide and amino acid sequences of TRI genes from T. arundinaceum and T. brevicompactum were retrieved from National Center for Biotechnology Information (NCBI). Initially, TaTRI3 gene was used for BLASTn search against ZJUF0986 genome to identify the contig containing the TRI3 orthologue in ZJUF0986. The contig was then used as a query to search for other TRI genes by BLASTx. 1 The organization of the encoded open reading frames (ORFs) was illustrated by pDRAW32. 2 Alignment of the amino sequences of all the putative T. taxi TRI proteins with those from T. arundinaceum and T. brevicompactum was performed by MEGA5.0 (Tamura et al., 2011) and then reviewed by Genedoc (Nicholas et al., 1997). All TRI amino sequences were subjected to protein domain analysis by Pfam 3 (Finn et al., 2014).

Generation of TtTRI3 Gene Knockout Mutants and Complementation
The gene deletion vector was constructed based on the doublejoint PCR strategy (Yu et al., 2004). Briefly, approximate 1 kb upstream and 1 kb downstream sequences flanking TtTRI3 locus were amplified with the primers TRI3-upF/R and TRI3-downF/R, respectively. A 1.4-kb hygromycin B resistance (HPH) cassette was cloned from pCB1003 (Carrol et al., 1994) with primers HPH-F/R. The three fragments were mixed together in the second-round PCR, the product of which acted as template for the final amplification with nested primers TRI3-NestF/R. The double-jointed PCR product was purified and subcloned into the commercial TA-cloning vector pMD18-T (TaKaRa, Dalian, China). The resulting plasmid and the Agrobacterium binary vector for plant transformation, pCAMBIA1300 (GenBank: AF234296; Hajdukiewicz et al., 1994), were both cut with HindIII/XbaI and then joined together by T4 DNA ligase (TaKaRa), leading to the final TtTRI3 gene knockout vector, pCAMBIA-TRI3. To obtain the TtTRI3 null mutants, pCAMBIA-TRI3 was transformed into the wild strain using Agrobacterium tumefaciens-mediated transformation (ATMT) (Rho et al., 2001). The transformants with resistance to hygromycin B (40 µg/ml) were selected and screened by PCR with the primers ATMT-F/R and ATMT2-F/R to verify deletion by replacement. To complement the deletion strain, a fragment containing TtTRI3 gene locus with its native promoter and terminator region was amplified by primers TRI3comp-F/R and ligated into a modified pCAMBIA1300 vector in which the original hygromycin B resistance gene was replaced with a G418-resistant gene. The final complementation vector was introduced in the mutant by ATMT and randomly inserted into the genome.

Nucleic Acid Extraction and qRT-PCR
Vegetative mycelia were harvested from 5-day-old culture in a 250-ml flask containing 100 ml of liquid potato dextrose broth (PDB) incubated at 25 • C with shaking (200 rpm). Genomic DNA and total RNA were extracted by the cetyl trimethylammonium bromide (CTAB) method (Del Sal et al., 1989) and the TRIzol reagent (TaKaRa), respectively. qRT-PCR was performed using SYBR Premix ExTaq (TaKaRa) on a Mastercycler Ep realplex TRI12RT-R GTGGATCAATTTTCACCGCTG thermo cycler (Eppendorf, Hamburg, Germany). Actin gene was chosen as the normalizing gene, and the expression levels of TRI genes were calculated by the 2 − Ct method (Livak and Schmittgen, 2001). Each reaction was repeated three times independently.

Extraction and Analysis of Trichodermin
The wild-type ZJUF0986 and the mutant TtTRI3 were cultured in 150 ml of PDB medium in 500-ml Erlenmeyer flasks at 25 • C in the dark. After 7 days' incubation with shaking (200 rpm), the fermentation broth was filtered through three layers of lens paper. The filtered medium was extracted three times with equal volumes of ethyl acetate. The extract was dried by evaporation and dissolved in methanol and then analyzed by gas chromatography (GC)-MS as described by Cardoza et al. (2011).

Antifungal Activity Test
The fermentation broth of T. taxi strains was harvested from 7day-old cultures grown at conditions described above. R. solani was incubated on PDA containing 10 or 20% (v/v) harvested fermentation broth, using PDA containing 10 or 20% of non-inoculated fermentation broth as a reference set. Growth of R. solani on plates was monitored and recorded after 3 days' incubation.

Bacterial Cloning and Expression of TtTRI3 Gene and Enrichment of the Protein Product
Reverse transcription and subsequent PCR was carried out with the PrimeScript TM RT reagent Kit with gDNA Eraser (TaKaRa) and the primers TRI3cDNA-F/R. Purified PCR product was double-digested by BamHI and HindIII and ligated into pQE30 (Qiagen) (Figure 6A). The resulted plasmid pQE30-TRI3 was sequencing verified and introduced into E. coli M15 (Qiagen), leading to strain M15-TRI3. Culture of M15-TRI3 was initiated and allowed to grow to OD 600 at 0.6 with aeration at 37 • C; 1 mM of IPTG was then added, and the culture was incubated overnight at 25 • C with shaking. Cells were harvested by centrifugation at 10,000 g for 5 min at 4 • C, re-suspended in buffer (pH 7.5) containing 50 mM of Tris, 200 mM of NaCl, 1 mM of β-mercaptoethanol, and 5 mM of imidazole. The cell suspension was subjected to sonication for a total of 15 min on ice, with 40% duty cycle (2-s working and 3s pause). The cell lysate was clarified at 20,000 g, and cell debris was discarded. The His-tagged TtTRI3 protein in the lysate was enriched using Ni-NTA beads (Invitrogen, Carlsbad, CA, United States) as recommended in the suppliers' manual and concentrated and buffer-changed to 0. Enzymatic Activity Assay of Acetyltransferase Activity of TtTRI3 The activity of the acetyltransferase assay was performed according to Garvey et al. (2009)  (BSA). All components were dissolved in 0.1 M of potassium phosphate buffer (pH 8.0). The reaction was initiated by the addition of 10 µl of 290 µg/ml enriched enzyme described above. After incubation at 25 • C for 5 h, the reaction mixture was extracted with an equal volume of ethyl acetate, and the extract was subjected to GC-MS analysis to verify the formation of trichodermin.

Primer Sequences and Deposition of Nucleotide Sequences
All sequences of the primers used in the study are listed in Table 1. The nucleotide sequence of the contig containing TRI genes was directly submitted to GenBank under the accession number KY670722. The coding sequences of the individual TRI genes were also deposited with the following assigned accession numbers: KY860616 (TtTRI3), KY860617 (TtTRI4), KY860618 (TtTRI6), KY860619 (TtTRI10), KY860620 (TtTRI11), KY860621 (TtTRI12), and KY860622 (TtTRI14).

Identification and Sequence Analysis of TRI Genes in Trichoderma taxi
In order to understand the underlying biosynthetic pathway of trichodermin in T. taxi, we made use of the recently sequenced genome of T. taxi (manuscript in preparation). With the use of T. arundinaceum TRI3 gene as a query to search against the genome, BLASTn identified the TtTRI3-containing contig as a 42.5-kb-long assembly (Figure 1). Further BLASTp search found that the contig encodes several other TRI orthologues, namely, TtTRI4, TtTRI6, TtTRI10, TtTRI11, TtTRI12, and TtTRI14, and that they were located in immediate vicinity on the contig. Along with TtTRI3, they appeared in the same order and orientation as those in T. arundinaceum and T. brevicompactum (Figure 1). Notably, the size of the genomic region between TRI11 and TRI12 in T. taxi is more like in T. arundinaceum than in T. brevicompactum, where the size is much larger (Figure 1). Consistent with previous observation in Trichoderma , TRI5 is absent from the core TRI cluster in T. taxi. It is noteworthy that we are unable to recover the complete TtTRI5 gene sequence from the contigs since querying TaTRI5 against the assembled genome resulted in three separated short contigs (data not shown). As expectable from the close phylogenetic relationship between T. taxi and T. brevicompactum/T. arundinaceum (Jaklitsch and Voglmayr, 2015), the proteins encoded by the TRI cluster display a high degree of identity, suggesting that they are structurally and functionally conserved ( Table 2). Results from protein domain search identified TtTRI3 as a TRI3 family 15-O-acetyltransferase, TtTRI4 and TtTRI11 as cytochrome P450 domain-containing oxygenases, TtTRI12 as a fungal trichothecene efflux pump, TtTRI6 as a C 2 H 2 -type zinc finger domain-containing transcriptional regulator, and TtTRI10 likely to be a transcription factor. No match in protein domain database was found for TtTRI14. Though orthologues of TRI14 have been found in other Hypocreales, such as Fusarium spp. (Peplow et al., 2003), Stachybotrys spp., and Cordyceps spp., the function of the protein is currently unknown. Disruption of TtTRI3 Impaired Production of Trichodermin and Antagonism Against Rhizoctonia solani TRI3 had not been characterized in Trichoderma spp. when we initiated this study. It is highly conserved in the genus Trichoderma and other Sordariomycetes, but the distance (i.e., branch length) between Trichoderma TRI3 and other fungi is high, suggesting potential differences in function (Figure 2). In Fusarium, TRI3 catalyzes the acetylation of the hydroxy group at C-15 of the trichothecene skeleton (McCormick et al., 1996). In T. brevicompactum, TRI3 was proposed to be responsible for the acetylation of the hydroxy group at C-4 of the trichothecene skeleton , but this hypothesis has not been experimentally tested yet. We therefore generated a TtTRI3 null mutant by replacing the TRI3 ORF by a 1.4-kb HPH cassette. Successful gene replacement was confirmed by PCR screening (Figure 3).
There were no significant phenotypic changes observed in colonial morphology and growth rate between wild type and TtTRI3 (data not shown). However, in contrast to the wild type, TtTRI3 lost the ability to secrete molecules that inhibited growth of R. solani (Figure 4): when 10% of the filtered fermentation broth from the parent strain was added to a PDA plate, growth of R. solani on the plate was strongly inhibited, and complete inhibition was observed by addition of 20%. In contrast, even adding 20% of the fermentation broth of the knockout strain TtTRI3 to the plate did not impair growth of R. solani. This suggests that TRI3 catalyzes an essential step in the generation of the compound that inhibits R. solani growth.
To test whether deletion of TtTRI3 indeed affects the production of trichodermin, the fermentation broth from both the wild-type culture and the TtTRI3 culture was subject to GC-MS analysis. This showed that the production of trichodermin was significantly decreased in the TtTRI3 strains ( Figure 5A). Interestingly, the concentration of trichodermol, the immediate trichodermin precursor, did not accumulate to higher levels in the TtTRI3 strain and remained at the same level as in the wildtype strain (Figure 5B), suggesting the operation of an effective feedback control in its biosynthesis. These data demonstrate that TRI3 plays an essential role in the conversion of trichodermol to trichodermin and that trichodermin confers T. taxi most (if not all) of its ability to inhibit growth of R. solani.
To further confirm that TRI3 is responsible for converting trichodermol to trichodermin in T. taxi and that trichodermin is the main component conferring inhibitory effect to R. solani, wild-type TtTRI3 gene with its native promoter and terminator was introduced into the TtTRI3 mutant, resulting in the complementation strain TtTRI3-TRI3.
The TRI3 complementation restored the knockout strain's ability to inhibit growth of R. solani to that of the wild-type level; i.e., 10% of added fermentation broth largely inhibited growth FIGURE 5 | Trichodermin and trichodermol production in strain ZJUF0986, TtTRI3, and TtTRI3-TRI3. Amount of trichodermin (A) and trichodermol (B) in fermentation broth was detected by gas chromatography (GC)-MS. Peaks for trichodermin and trichodermol were determined by the corresponding reference chemicals. The concentration of reference chemicals in the prepared samples was 188 ppm (part per million) for trichodermin and 42 ppm for trichodermol. of R. solani and 20% resulted in complete inhibition (Figure 4).

In addition, GC-MS analysis of the fermentation broth from
TtTRI3-TRI3 showed that trichodermin biosynthesis was recovered and that its abundance in the complementation strain was even higher than that in the wild-type strain ( Figure 5A). These data confirmed that the results obtained with the TtTRI3 knockout strain are valid, and TRI3 plays an essential role in converting trichodermol into trichodermin in T. taxi.

TtTRI3 Converts Trichodermol to Trichodermin
To examine the potential conversion of trichodermol to trichodermin by TtTRI3 also in vitro, TtTRI3 was recombinantly produced with a His-tag in E. coli and purified to >70% homogeneity (estimated by ImageJ: https://cnij.imjoy.io/) by Ni-NTA resin batch treatment (Figure 6B). A biochemical assay combining trichodermol, TtTRI3, and acetyl-CoA was established to test if this enriched TtTRI3 was able to catalyze acetylation of trichodermol in vitro. GC-MS analysis of the ethyl acetate extract of the reaction mixture showed that a new peak appeared at the position of trichodermin, while the peak was missing when the cell lysate was prepared from E. coli cells where the empty vector was introduced instead (Figure 6C). Exclusion of acetyl-CoA or the recombinantly produced TtTRI3 protein from the mixture also prevented the formation of trichodermin (data not shown), thus proving that TRI3 and acetyl-CoA are required for the reaction to happen.

Disruption of TtTRI3 Gene Downregulated the Expression of Other TRI Genes
We have noted above that the deletion of TtTRI3 did not result in significant over-accumulation of trichodermol, suggesting a tight feedback control. To test whether this would involve regulation of gene expression, we therefore quantified the expression level of TtTRI4, TtTRI6, TtTRI10, TtTRI11, and TtTRI12 in the parent strain ZJU0986 and its knockout mutant TtTRI3. As illustrated in Figure 7, we indeed found that the expression of all structural genes was decreased in the mutant by more than 50%, whereas the expression of the regulator genes TtTRI10 was unaffected and that of TtTRI6 reduced by only 30%. These data suggest the operation of a feedback loop in the trichodermin biosynthesis pathway in T. taxi. FIGURE 7 | Comparison of TRI genes expression level between wild-type strain and TtTRI3. Expression level for TRI genes was determined by qRT-PCR. Expression level of genes in ZJUF0986 was normalized as 1. Calculation was based on data obtained from three independent experiments.
Although chemical synthetic routes for trichodermin had been developed not long after its discovery (Colvin et al., 1973), it was not until recently that the biosynthetic pathway was elucidated . Like other trichothecenes, biosynthesis of trichodermin starts from the cyclization of FPP to form trichodiene by the TRI5 trichodiene synthase. TRI4 then oxidizes trichodiene to isotrichodiol, which subsequently forms EPT by non-enzymatic dehydration. TRI11 then hydroxylates EPT to trichodermol (Figure 8), and TRI3 was proposed to catalyze the acetylation of trichodermol to trichodermin . Recently, TRI3 has been shown to be involved in trichodermin synthesis in an endophytic T. brevicompactum by mutational analysis (Shentu et al., 2018). In this study, for the first time, we present direct evidence that TtTRI3 is capable of performing the acetylation of trichodermol to trichodermin, using acetyl-CoA as acetyl donor. Our results also imply that biosynthetic pathway of trichodermin is conserved in Trichoderma species capable of producing it.
Some Trichoderma species such as T. brevicompactum were reported to produce both trichodermin and harzianum A (HA), depending on the strains and the cultivation conditions (Nielsen et al., 2005). HA differs from trichodermin in that it is esterified at C-4 with octa-2Z,4E,6E-trienedioic acid while trichodermin is esterified with acetic acid. In HA-producing strains, octa-2Z,4E,6E-trienedioic acid was also detected, suggesting that HA is synthesized in one single step from trichodermol and an octa-2Z,4E,6E-trienedioic group donor (Nielsen et al., 2005), which has been proved in T. arundinaceum by gene disruption (Proctor et al., 2018). Here, we showed that acetylation of trichodermol is also catalyzed by TRI3.
In Fusarium, TRI3 functions to acetylate the hydroxyl group at C-15 of the trichothecene skeleton (McCormick et al., 1996). Gene TRI3 in F. sporotrichioides (FsTRI3) shares 48% identity in translated amino acid sequence with TRI3 from Trichoderma ( Table 2; Cardoza et al., 2011), which is rather low but consistent with a functional difference (shift of acetylation site from C-15 in Fusarium to C-4 in Trichoderma) between the two TRI3 enzymes. Intriguingly, a similar situation has been observed in the case of TRI11/TRI11, where amino acid sequence identity between the two genera is 39%, much lower than that of other TRI proteins, such as TRI4 (73%), TRI5 (57%), and TRI12 (55%) ( Table 2; Cardoza et al., 2011), consistent with the fact that Fusarium TRI11 oxygenates trichothecene skeleton at C-15, while Trichoderma TRI11 does so at C-4 position and that other TRI proteins are functionally more conserved (Alexander et al., 1998;Cardoza et al., 2011). It is reasonable that the two genes are co-evolved since the substrate upon which TRI3 acts depends on TRI11.
In this study, the disruption of TtTRI3 gene severely reduced the expression of other trichodermin biosynthesis genes, while the regulatory gene TtTRI6 was less and TtTRI10 not at all affected. The results were in accordance with the observation that deletion or silencing of TaTRI4 in T. arundinaceum reduced expression level of TaTRI5 (Malmierca et al., 2012). Given the sampling time points in our study, our results are also in good compliance with those obtained by Shentu et al. (2018) where TRI genes were downregulated at midlate growth phase. Surprisingly, the disruption of TaTRI5 in the HA biosynthetic pathway upregulated all other TaTRI genes except TaTRI4 (Malmierca et al., 2013). Our data suggest that the expression of TRI genes is subjected to metabolite feedback repression, a common phenomenon in biosynthetic pathways (Vasanthakumar et al., 2013;Pratelli and Pilot, 2014;Ness, 2015), likely by TRI10. In T. taxi, a decrease in the concentration of trichodermin may elicit this regulation, because the concentration of trichodermol was not significantly altered in the knockout mutant. However, other metabolites of the biosynthetic pathway, which have not been measured, may also be involved. In Fusarium, TRI6 and TRI10 function as regulatory genes to control expression of other genes in trichothecene biosynthesis (Proctor et al., 1995;Tag et al., 2001;Nasmith et al., 2011). FsTRI6 binds to the consensus sequence TNAGGCCT that is found in the promoter of most TRI genes . The disruption of Fusarium TRI6 greatly reduced the expression of TRI5 and TRI4, and the mutant failed to accumulate T-2 toxin (Proctor et al., 1995). FsTri10 was also found to regulate all other TRI genes and several putative isoprenoid biosynthetic genes (Peplow et al., 2003). How TRI6 and TRI10 cooperate to regulate the expression of other TRI genes in Trichoderma, possibly in coordination with metabolite feedback, would be an interesting topic for the future.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ supplementary material.

AUTHOR CONTRIBUTIONS
HC performed TRI3 expression, purification, and the in vitro reaction and prepared the original manuscript. LM carried out all the GC-MS experiments and analyzed the data. NZ sequenced the genome and constructed the mutant strains. CX performed the antifungal activity test. JL and CZ designed and supervised the research and revised the manuscript. WW assisted with chemical structures and manuscript revision. SX and CK perfected the language and revised the manuscript. All authors discussed, edited, and approved the final version.

ACKNOWLEDGMENTS
We are grateful to Li Boyan from Guizhou Medical University and Zhang Xiaotao from China Tobacco Guizhou Industrial Co., Ltd. for carefully revising the original manuscript.