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.

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 Pfam³ (Finn et al., 2014).

The gene deletion vector was constructed based on the double-joint 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.

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 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^–ΔΔCt method (Livak and Schmittgen, 2001). Each reaction was repeated three times independently.

The wild-type ZJUF0986 and the mutant ΔTtTRI3 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).

The fermentation broth of T. taxi strains was harvested from 7-day-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.

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 OD₆₀₀ 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 3-s 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 0.1 M of potassium phosphate buffer (pH 8.0) by centrifugation using Amicon^® Ultra centrifugal filters (Millipore, Billerica, MA, United States). Protein concentration was determined by the NanoDrop^TM 2000 UV-Vis spectrophotometer (Thermo Fisher Scientific, Waltham, MA, United States).

The activity of the acetyltransferase assay was performed according to Garvey et al. (2009) with slight modifications. Briefly, reaction mixtures were prepared at room temperature by combining 150 μl of 1.5 mM acetyl-CoA (Sigma-Aldrich Corp., St. Louis, MO, United States) with 50 μl of 4 mM trichodermol and 2 μl of 5 mg/ml bovine serum albumin (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.

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).

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 (Cardoza et al., 2011), 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 C₂H₂-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.

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 (Cardoza et al., 2011), 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 ΔTtTRI3 (data not shown). However, in contrast to the wild type, ΔTtTRI3 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 ΔTtTRI3 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 ΔTtTRI3 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 ΔTtTRI3 strain and remained at the same level as in the wild-type 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 ΔTtTRI3 mutant, resulting in the complementation strain ΔTtTRI3–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 of R. solani and 20% resulted in complete inhibition (Figure 4). In addition, GC-MS analysis of the fermentation broth from ΔTtTRI3–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.

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.

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 ΔTtTRI3. 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.