The experimental studies were carried out on the biocontrol agent T. virens Gv29-8, a GT-producing strain, and on the plant growth promoter T. harzianum T6776 (Fiorini et al., 2016), recently reclassified as T. afroharzianum T6776 (Kubicek et al., 2019). The fungal strains were routinely cultivated on potato dextrose agar (PDA) or potato dextrose broth (PDB) as specified.

The fungal strains, whose genome sequences were downloaded from the National Center for Biotechnology (NCBI) for the in silico study, are presented in Table 1 with their related abbreviations and other information. A. fumigatus Af293 was used as a reference strain and its annotated genome as query. Different basic local alignment search tool (BLAST) analyses (blastn, tblastn, blastp, blastx) were performed using the Af293 gli cluster gene and protein sequences as a query to further assess content and arrangements of genes in TvGv29-8. Successively, the TvGv29-8 gli genes and protein sequences were also used as references to find the putative orthologous gli genes among Trichoderma spp. For the gliH search, the AfgliH described by Schrettl et al. (2010) was used instead of AFUA_5G10320, a gene annotated as gliH in the NCBI database. The highest scoring and covering hit for each gene was verified by the bidirectional best hits (BBH) approach against Af293 and the other Trichoderma genomes and eventually chosen, identified, and localized on the Trichoderma spp. genomes. Each gene orthology was then verified in the OrthoDB catalog of eukaryotic orthologs provided as an option in NCBI. To clearly distinguish between orthologous genes in A. fumigatus, T. virens, T. afroharzianum, T. harzianum, T. reesei, T. gamsii, and T. longibrachiatum, gene designations are preceded, respectively, by the prefix Afgli, Tvgli, Tagli, Thgli, Trgli, Tggli, and Tlgli. The Afgli genes and putative GT-BG of the analyzed Trichoderma strains are listed in Supplementary Table S1.

To verify the correct identification and functionality of the gli clusters in TvGv29-8 and TaT6776, candidate reference genes for gene expression analyses were shortlisted. In detail, gliP, coding the key GT biosynthetic bimodular non-ribosomal peptide synthetase; gliH, unknown function; gliT and gliA, self-protection genes; gliZ and gipA, transcription factors; and gtmA, a gene belonging to a negative regulatory system, were chosen as the main representative genes for GT biosynthesis, self-protection, and regulation, respectively. Calmodulin (TCal) and secretion-associated Ras-related protein (TSar) genes were used as housekeeping genes for gene expression normalization. Real-time RT-PCR protocols were developed for all these genes. Each primer was designed in an intron flanking region. Specific primer pairs were designed using Primer3Plus. The gene names, amplification efficiency, correlation coefficient, amplicon length, and amplicon melting temperature are reported in Supplementary Table S2. The amplification efficiency of quantitative PCR (qPCR) was calculated only for the genes expressed in the absence and the presence of exogenous GT and ranged from 90 to 99.8%, and the R² of the standard curve was 0.997 ± 0.001. Agarose gel electrophoresis and melting curve analyses (data not shown) showed a single temperature for each primer pair that was distinctly different from the non-specific genomic peak, confirming the specificity of the amplification.

TvGv29-8 and TaT6776 conidia were inoculated in PDB at a final concentration of 10⁷ conidia/ml and incubated at 23–25°C under illumination of 16 h light/8 h dark cycles, using daylight tubes 24 W/m², 9,000 lx. After 27 h, fungal RNA was isolated and purified using the PureLink RNA purification kit (Ambion, Thermo Scientific). To avoid genomic DNA (gDNA) contaminations, the RNAs were treated with DNase I (Ambion, Thermo Scientific), and complementary DNA (cDNA) synthesis from messenger RNA (mRNA) (200 ng) was performed using cDNA first-strand synthesis kit (Thermo Fisher) with hexamer primer according to manufacturer’s instruction.

Reactions were performed using the Real-Time PCR PowerUp SYBR Green Master Mix kit (Applied Biosystems) in the ViiA7 Real-Time PCR system (Applied Biosystems). Each RT-PCR reaction was performed in three biological and technical replicates and a no-template control. To determine the specificity of the primer pairs used in the current study, 1.5% agarose gel electrophoresis and melt curve analyses were performed (Supplementary Table S2). To determine the PCR efficiency of each primer pair, a standard curve was generated using linear regression, and the Cq slope was calculated based on the Cq values for all dilutions (5 points, 10-fold dilutions from 50 to 50,000).

TvGv29-8 and TaT6776 were inoculated in PDB by spore suspension at a final concentration of 10⁷ conidia/ml and incubated at 23–25°C under illumination of 16 h light/8 h dark cycles, using daylight tubes 24 W/m², 9,000 lx. After 24 h, exogenous GT (WVR International) or methanol (MeOH, solvent control) was added to the media at 5 μg/ml (final concentration) and incubated for 3 h under the same conditions. Mycelia were harvested through Miracloth filtration, washed with distilled water, and immediately frozen in liquid nitrogen and processed for gene expression analyses as described above. The culture filtrates were stored at −20°C until GT identification and quantification by HPLC analysis.

Conidia (1 × 10⁶ conidia) of TvGv29-8, TaT6776, and TrRUTC30 (as negative control) were inoculated into 25 ml Weindling modified medium (Wilhite and Straney, 1996) and cultured on a rotary shaker (150 rpm) at 25°C of 16 h light/8 h dark cycles, using daylight tubes 24 W/m², 9,000 lx. To investigate the time course of GT production, aliquots of the culture filtrates were taken at incubation times 24, 48, 72, 96, and 162 h and analyzed by HPLC. Three replicates were performed for each fungus. For BmGT quantification, samples were taken at 27 h incubation time.

Culture filtrate aliquots were diluted with MilliQ water (1:1) and extracted three times with half volume of chloroform while shaking (150 rpm) for 10 min. The concentrated chloroform extracts were dried over sodium sulfate, and the solvent was removed by Rotavapor (37°C, vacuum); the residue was dissolved in 200 μl methanol. Only the samples induced with exogenous GT were furtherly diluted 1:10. All the samples were then subjected to HPLC analysis.

Ten microliters of each biological sample was analyzed in triplicate by reverse-phase HPLC with UV detection (RP-HPLC-UV) using a Dionex^TM UltiMate^TM 3000 (Thermo Fisher Scientific) equipped with an LPG-3400SD quaternary analytical pump, a WPS-3000SL analytical autosampler, a VWD-3100 UV–visible detector, a TCC-3000SD thermostatted column compartment, and an AFC-3000 automatic fraction collector. Chromatographic separation was performed using a Waters XSelect CSH C18 column (150 mm × 2.1 mm ID; particle size, 3.5 μm). The specific liquid chromatographic (LC) parameters were as follows: mobile phase (A) water, TFA (99.95:0.05 v/v); (B) acetonitrile, the mobile phase flow rate was 0.3 ml/min; gradient program, 10% B for 2 min, from 10 to 90% B in 18 min, 90% B for 1 min and then from 90 to 10% B in 1 min and re-equilibration to 10% B for 3 min. All analyses were performed at 30°C. The detection wavelength (λ) was set at 254 nm.

As reference standard GT (VWR International) and BmGT (Sigma-Aldrich Co.) were used. For GT quantitation, a calibration curve was generated using five different dilutions of the GT reference standard (550.00, 275.00, 55.00, 27.50, and 5.50 μg/ml) and similarly for BmGT quantitation, using five different dilutions of the BmGT reference standard (50.00, 25.00, 10.00, 5.00, 1.00 μg/ml). Ten microliters of each standard dilution was analyzed by RP-HPLC-UV as reported above. The calibration curve was plotted using the area under peak versus GT or BmGT standard reference concentrations (μg/ml) (Supplementary Figures S1A,B, respectively). Biological sample concentrations were calculated by determining their area under peak and comparing them to the area of the calibration curve.

TvGv29-8 and TaT6776 conidia at a final concentration of 10⁶ conidia/ml were inoculated in PDB containing different concentrations of GT or MeOH (0, 5, 15, 30, 50 μg/ml) in 96-well plates. The optical density (620 nm) of the liquid cultures was read at least three times a day at regular intervals, using an ELISA plate reader.

Three new putative orthologs – TvgliJ, TvgliA, and TvgliH encoding a dipeptidase, a transporter, and an unknown function protein, respectively – were identified in TvGv29-8, at other loci, outside of the previously described gli cluster on scaffold 47. The relative position and characteristics of the Tvgli genes are shown in Table 2 and Supplementary Table S1. In detail, TvgliJ, located on scaffold 51, encodes a protein with a 52.81% sequence identity to AfGliJ; TvgliA, situated on scaffold 4, codes for TvGliA with 45.13% sequence identity to AfGliA; and TvgliH, on scaffold 10, encodes TvGliH with 54.40% sequence identity to AfGliH. Interestingly, of the TvgliT orthologs previously described, TRIVIDRAFT_138628 was found beside TvgliH, while TRIVIDRAFT_31354 was located on the large scaffold 4 where TvgliA is also present. In addition, two related genes, TvgtmA, the GT thiomethyltransferase, and TvgipA, a C₂H₂ transcription factor, were identified on TRIVI scaffolds 77 and 5, respectively.

Overall, TvGv29-8 and Af293 showed a gli-gene-related protein sequence identity between 44 and 68% (Table 2). Thus, in addition to the cluster of eight gli genes, a duplet of two gli genes, two single gli genes, situated in different locations in the genome, and two related genes were identified (Figure 1).

When a similar to GT-BG and related genes search was performed in TvFT-333, using the TvGv29-8 gene sequences, a striking nucleotide identity (100%) (Supplementary Table S3A) was found with the corresponding 13 Tvgli genes (Supplementary Table S4). In addition, TvFT-333 showed an identical GT-BG architecture with a cluster of the same eight gli genes with the same order and orientation (on scaffolds 0430–0433) and two single gli genes, gliA and gliJ, situated respectively, in scaffolds 0040 and 0450. Differently, of the duplet of two gli genes, only TvgliT was present on scaffold 0048, while TvgliH was absent from the TvFT-333 genome. This result was confirmed also by blasting the TvGliH and AfGliH sequences.

To verify the distribution of the GT-BG in different loci, the sequences of the flanking regions of the gli cluster, the duplet and the single genes were compared in the two T. virens strains. Actually, in TvGv29-8, the gli cluster, the gliH-gliT duplet, and one end of gliJ are without flanking sequences (Figure 1). Differently, the sequences flanking gliA and the other end of gliJ confirmed their position outside of the gli cluster.

The third T. virens genome present in NCBI belongs to the “P” strain, IMI 30406, which does not contain any gli genes (Sherkhane et al., 2017); thus, it was not included in this study.

Putative gli clusters were successfully identified in seven strains of four different species: TaT6776, ThB97, ThTR274, Th226.95, TrRUTC30, TrQM6a, and TlSMF2 (Table 2 and Supplementary Tables S1,S4). GT-BG of some representative strains are depicted in Figure 1, where the multiple genes shared between Trichoderma species are shown. Interestingly, the gli cluster was not present in the TgT6085 genome. The organization of the GT-BG in these Trichoderma species differed considerably from T. virens, that is, the GT-BG were present in a single cluster with a very similar architecture: length ranging from 33,140 to 35,284 bp, presence of 11 gli genes, and absence of gliH and gliZ (Supplementary Table S5). Notably, the T. reesei GT-BG were located in a single cluster with 11 out of the 12 genes reported by Patron et al. (2007).

Table 2 and Supplementary Table S4 show the highest identities found in protein sequences using BLASTP search. The predicted products of the gli genes were similar to those found in A. fumigatus showing amino acid identities ranging from 23 to 82%.

Moreover, a very high (99.97%) intraspecies nucleotide identity of the gli clusters was found in T. harzianum and T. reesei, while interspecies identity varied from 70% between Tagli/Trgli and Tagli/Tlgli, to 88% between Tagli/Thgli. In these species, one major facilitator superfamily (MFS) transporter highly similar to AfgliA was located outside of the clusters, while an ATP-binding cassette (ABC) transporter was located in the gli clusters and coding for proteins with a low amino acid sequence identity to AfGliA and TvGliA (Supplementary Table S3B). These ABC transporters (gliA^∗) showed very high interspecies identity (70–88%). The MFS and ABC transporter genes were found to belong to two different orthology groups. Lastly, gliZ relics were found in a genomic position located close to gliG and a second putative gliF, which resulted not to be an AfgliF ortholog. All the GT-BG identified in these species were indicated as orthologs of the corresponding Afgli genes by OrthoDB.

GtmA and gipA were present in all the Trichoderma species tested (Table 2 and Supplementary Table S4). Notably, orthologs of TvgtmA (TGAM01_v208885) and of AfgipA and TvgipA (TGAM01_v202426) were also found in TgT6085, where the GT-BP was absent.

The expression levels of gliP (non-ribosomal peptide synthetase), gliT (putative oxidoreductase), gliA (putative gliotoxin transporter), gliZ (putative Zn₂Cys₆ transcription factor), gliH (hypothetical protein), gipA (putative C₂H₂ transcription factor), and gtmA (putative GT bis-thiomethyltransferase) were analyzed (Figure 2). The results were quite different between the two Trichoderma strains. In the GT-producing TvGv29-8, TvgliP, the key gene in GT synthesis, as well as TvgliA, TvgipA, and TvgtmA were expressed. Among the previously identified gliT orthologs, only gliT (TRIVIDRAFT_138628) was expressed during GT production, establishing it as the putative oxidoreductase TvgliT. Otherwise, TvgliZ is still to be identified since neither the previously indicated ortholog nor the other hits revealed by BLAST analyses (data not shown) were expressed. Interestingly, gliH that was retrieved only in the TvGv29-8 genome was expressed in GT permissive conditions (data not shown). In TaT6776, only gipA was transcribed. The TvgliH primers were also tested in TaT6776, but no amplifications were observed.

In A. fumigatus, AfgliP, AfgliT, AfgliA, and AfgipA are involved both in GT production and self-protection against GT toxicity (Schrettl et al., 2010; Dolan et al., 2015), while AfgtmA indirectly and negatively regulates the production of GT (Dolan et al., 2017), which is known to be involved in a positive regulation of its own synthesis (Cramer et al., 2006). To evaluate the role of exogenous GT on gliP, gliT, gliA, gipA, and gtmA expression, RT-qPCR results were compared between TvGv29-8 and TaT6776 grown in different conditions (PDB, 5 μg/ml GT, and methanol), and the significance was evaluated by ANOVA (Figure 2).

In detail, in TvGv29-8, TvgliP and TvgtmA expression levels were not significantly different in all the tested conditions. In contrast, TvgliT and TvgliA genes were upregulated by 1.6- and 6-fold, respectively, in the presence of exogenous GT. The expression levels of the transcription factor gipA were increased in both TvGv29-8 and TaT6776 in the presence of exogenous GT, but this effect was statistically significant only in TaT6776 (p ≤ 0.001). Strikingly, when the fungus TaT6776 was exposed to exogenous GT, the expression of the TagliA^∗ and TagtmA genes was detectable and measured by 200- and 16,000-fold, respectively, compared to samples not exposed to GT (p ≤ 0.0001). The expression of the tested genes in the solvent control was comparable to that observed without exogenous GT.

TvGv29-8 is a well-known GT-producing strain, which was confirmed in this study by HPLC analyses. The GT and BmGT concentrations referred to the level present in the original culture media. GT production began at 24 h after inoculation in Weindling medium, and it rapidly increased to a concentration of 74.9 ± 32.3 μg/ml at 48 h (Figure 3). Subsequently, the GT secreted was slowly increased to 93.1 ± 11.3 μg/ml at 162 h.

No GT- and BmGT-related peaks were present for TaT6776 (Figure 4A), demonstrating that it does not produce GT, despite the presence of the gli cluster in its genome. No information was available on GT production by TaT6776 before this study. As expected, comparable HPLC results were found for TrRUTC30, confirming that it does not produce GT (Figure 4A).

The production of GT and BmGT in TvGv29-8 and in TaT6776 was analyzed further after adding exogenous GT to their culture media. In this case, both GT and BmGT were present also in the culture filtrates of TaT6776 (Figure 4B). The amount of recovered GT (4.98 ± 0.9 μg/ml) corresponded to that added to the medium, while 0.19 ± 0.07 μg/ml of BmGT was produced by TaT6776.

In TvGv29-8, the increased amount of synthesized GT, from 67.09 ± 28.6 μg/ml without the addition of exogenous GT to 93.4 ± 37 μg/ml with the addition of exogenous GT, was not statistically significant (Figure 4C). Similarly, in TvGv29-8, BmGT production was observed at 27 h after inoculation at a concentration of 0.98 ± 0.4 μg/ml and 1.57 ± 0.8 μg/ml (Figure 4A), respectively, without and with the addition of exogenous GT.

The ability of TvGv29-8 and TaT6776 to grow at different GT concentrations in the culture medium was verified in a multiwell assay (Figures 5A,C). The fungi differed in their sensitivity to exogenous GT. TvGv29-8 growth was only slightly affected by any GT concentration up to 50 μg/ml. In contrast, the growth of TaT6776 was completely suppressed at 50 μg/ml GT and significantly inhibited (p = 0.005) at 30 μg/ml GT. Nonetheless, TaT6776 did not show a dramatic sensitivity to exogenous GT, since the inhibition of its growth at 15 μg/ml GT was small (19%), although statistically significant (p = 0.003), compared to its growth without GT. Similarly, its growth was inhibited by 25% at 30 μg/ml GT after 96 h of incubation (data not shown). Interestingly, the presence of methanol in the media did not appear to cause a relative reduction in the growth of both TvGv29-8 and TaT6776 even at the volumes used at the highest tested GT concentrations, 30 and 50 μg/ml (Figures 5B,D).