DISCUSSION
The impressive number of TSs we found in the genomes analysed demonstrates that terpenoid biosynthesis has a great impact in the diversity and complexity of SMs in Trichoderma . Indeed, the TS-gene inventory of these species (15-23 genes per genome) clearly outnumbers those found in other fungi considered as rich producers of SMs, such as Aspergillus spp. (2-10 genes per genome) (de Vries et al., 2017; Kubicek et al., 2019). This likely reflects the importance of TSs and terpenoids in the ecology of these fungi. However, most of the Trichoderma TSs have not yet been characterized (Kubicek et al., 2019), and available information about the diversity of TSs inTrichoderma is scarce. Thus, we focused on characterising the TS-gene arsenal of this genus, providing a complete overview of the nature and diversity of the Trichoderma terpenoid biosynthetic inventory.
Although TS-family size is very homogeneous within Trichoderma , we were able to identify clade-specific TSs, which reflect particular portions of the terpenoid inventory shared by phylogenetically close species. Thus, despite their similar terpenoid biosynthetic potential, the species of Trichoderma have adapted their terpene production according to different environmental demands. Species of Viride clade constitute an example, as they are missing in some groups of TSs that are present in the other clades, but have evolved specific TSs which are absent in the other species.
According to our results, Trichoderma spp. have a huge potential for sesquiterpene biosynthesis. We identified eight groups of sesquiTSs, which constitute almost the third of the total number of TSs found in this work. Species of Viride are particularly rich in sesquiTS belonging to the TRI5-superfamily, and they also contain HAD-like TSs which are absent in species of other clades. These HAD-like proteins harbour a DxDTT motif, which is a variant of the DxDD found in Class II diTS (Nakano et al., 2005, 2009). Shinohara, Takahashi, Osada and Koyama (2016), reported that some sesquiTS can contain HAD-like domains and DxDTT motifs, leading to FPP cyclization through a protonation step, instead of by an ionization step. We hypothesized that HAD-like TSs might be particular bifunctional sesquiTS synthesizing specific metabolites of Viride clade.
Previous studies reported that some Trichoderma spp. have the potential to synthesize longiborneol (Bansal & Mukherjee 2016), an intermediate of the culmorin biosynthetic pathway (Bansal & Mukherjee 2016; McCormick, Alexander, & Harris, 2009). According to our results, longiborneol biosynthesis is very widespread in Trichoderma , but is absent in species of the Viride clade. Since culmorin production has not been reported in Trichoderma , longiborneol could be synthesized as a solely compound or as an intermediate of unknown biosynthetic pathways in these species. Most of these proteins show a conserved D(D/E)HFD motif, which is partially conserved (NDHFD) in proteins of T. arundinaceum and T. brevicompactum . Site-directed mutagenesis and crystallography studies on TSs have revealed that the first aspartate residue (D) of the metal-binding motif interacts directly with Mg2+ (Starks, 1997), and its replacement can lead to anomalous cyclization products or a product mixture (Cane, Xue, & Fitzsimons, 1996). According to this, longiborneol synthases of Brevicompactum species could be actually involved in terpenoid blend formation or in the biosynthesis of newly terpenoids.
We found that most Trichoderma spp. can potentially produce presilphiperfolan-8β-ol. This compound is thought to play a central role in the biosynthesis of a wide range of polycyclic sesquiterpenes in fungi (Pinedo et al., 2008). Thus, these TSs may contribute in generating a variety of structurally complex terpenoids inTrichoderma . Proteins sharing similarity with fungal pentalenene synthases were also identified, which are involved in the biosynthesis of the parent hydrocarbon of the pentalenolactone family of fungal antibiotics (Kim, Cheong, & Yoo, 1998). Nevertheless, the variability found on the structure of the active centre of these proteins suggests that some of them probably synthesize sesquiterpenoids others than pentalenene.
Our analyses revealed that single-copy SQS and OSC enzymes provide the linear and cyclic precursors required for triterpene biosynthesis in most Trichoderma species. However, the presence of an additional SQS in the genome of T. pleuroti indicates that pathway-specific SQSs could be present as well. In the same way, additional GGPP and FPP synthases may act as donors of terpenoid precursors in specific biosynthetic pathways in members of the clades Harzianum and Brevicompactum. This most likely reflects specific portions of the terpenoid inventory of these species, and guarantees an efficient distribution of terpenoid precursors between primary and secondary metabolism.
Harziandione was the first diterpene isolated from Trichodermaspp. (Ghisalberti et al., 1993), and a number of these compounds have been recently reported in these species (Adelin et al., 2014; Chen et al., 2019; Miao et al., 2012; Song et al., 2018; Zhao et al., 2019). Nevertheless, the low number of diTS found here indicates the ability ofTrichoderma spp. for diterpene biosynthesis is scarce and not very widespread within the genus.
Interestingly, our analyses revealed that Trichoderma spp. have the potential to produce sesterterpenes and indole diterpenes. Sesterterpenes are rare among terpenoids, and their antimicrobial and nematocidal properties (Tian, Deng, & Hong, 2017) could confer competitive advantages to some Trichoderma spp. In the other hand, production of indole diterpenes have been reported in some Sordariomycetes being involved in protecting their reproductive structures from fungivores (Saika et al., 2008). Many of the indole diterpene-producer fungi stablish symbiotic relations with plants, thus, biosynthesis of these compounds may confer ecological advantages onTrichoderma -host associations as well (Parker & Scott 2004).
No monoTS were found among the genomes analysed in this work, although production of monoterpenes has been reported in T. virens(Crutcher et al., 2013; Inayati, Sulistyowati, Aini, & Yusnawan, 2019). No bona fide monoTS have been identified in fungi (Schmidt-Dannert, 2014), and the scarce availability of sequences of these enzymes probably leaded to miss-predict them. Biochemical studies have shown that fungal sesquiTS are able to cyclize GPP (Lopez-Gallego et al., 2010); thus, we hypothesize that enzymes involved in monoterpene biosynthesis in Trichoderma could be actually included within “uncharacterized group 4”, since they are phylogenetically close to sesquiTS, and other uncharacterized proteins were found restricted to some clades not including T. virens or fell into the TRI5-superfamily.
We found highly conserved tri5 orthologs in non-trichothecene-producer Trichoderma species missing on the entire TRI cluster, some of which have been previously reported (Gallo, Mulè, Favilla, & Altomare, 2004). This leaves open some questions about the role of tri5 in beneficious Trichoderma spp. non producer of trichothecenes. Unlike other tri5 -containing species,tri5 is embedded in a BGC in T. gamsii , enclosing tailoring enzymes, a TF and a transporter. The presence of a TF within the cluster suggests a pathway-specific regulation, while the presence of a transporter suggests the production of a sesquiterpenoid with extra-cellular functions. Some of these genes were likely transferred toTrichoderma spp. by HGT from Eurotiomycetes, but the entire cluster is only present in T. gamsii . suggesting the origin of a novel tri5 -associated BGC in this species. This cluster could lead to an uncharted trichodiene-derived sesquiterpenic biosynthetic pathway producing novel metabolites with potential agronomic interest. Since tri5 seems to be functionally associated to two different BGCs (TRI and that found in T. gamsii ), we hypothesize this sesquiTS is involved in different metabolic pathways inTrichoderma .
The striking genomic potential for terpenoid production ofTrichoderma spp. found in this work suggests that functional differentiation of gene family members is the driver for the high TS gene numbers of these species. Assessing changes in the relative expression of TS when the fungus grows under different environmental conditions, or when interacts with other organisms, enables individuating genes that could play a role in these frameworks and hypothesizing about processes they could be involved in. Here, we provide a picture showing that different TS genes are differentially regulated, a strong indication of different biological functions.
Availability of C source had contrasting effects on the expression of TS genes, as observed in SMs genes from other fungi (Calvo, Wilson, Bok, & Keller, 2002; Jiao, Kawakami, & Nakajima, 2008). Similarly, TS genes were regulated in opposite ways in response to oxidative stress. Association of oxidative stress with SMs biosynthesis in fungi has been extensively demonstrated, and it has been suggested that it is induced to prevent fungi from ROS damage (Hong, Roze, & Linz, 2013). In particular, up-regulation of ts9 suggests that biosynthesis of indole diterpenes might occur in T. gamsii T6085 under oxidative stress conditions, as it has been observed in Aspergillus spp. (Fountain et al., 2016). Interestingly, addition of 0.9% sucrose or 0.5 mM H2O2 to the medium did not inducetri5 expression in T. gamsii T6085, differently from what observed in T. brevicompactum when grown in presence of 1% or 2% sucrose or in presence of 0.5 mM H2O2 (Tijerino et al., 2011a). This suggests different types of regulation of tri5 in T. gamsii T6085 and T. brevicompactum .
Nitrogen availability has a considerable impact on secondary metabolism in fungi (Hautbergue et al., 2018); in Fusarium fujikuroi , it affected the expression of 3 out of 10 TS genes (Wiemann et al., 2013). Similarly, saline stress modulates changes in SMs production, as it has been shown in T. harzianum (Bualem, Mohamed, & Moulay, 2015). N starvation and saline stresses negatively regulated TS genes inT. gamsii T6085, suggesting terpenoid biosynthesis does not confer particular advantages to the fungus to overcome these stresses.
In T. arundinaceum , tri gene expression is affected when grown in dual cultures with B. cinerea , while polyketides and harzianum A (HA) produced by the first induce changes in some B. cinerea genes linked to its virulence (Malmierca et al., 2015). SinceT. gamsii T6085 is able to suppress F. graminearum on wheat spikes and to reduce DON production by the pathogen (Matarese et al., 2012; Sarrocco et al., 2019, 2013), we investigated whether a differential expression of TS genes of T. gamsii T6085 occurs when the fungus interacts with F. graminearum on wheat spikes. However, our results show that the presence of the pathogen did not induce prominent changes in TS expression in T. gamsii T6085 when both fungi were on wheat spikes, although ts11 was found slightly up-regulated. Gene expression patterns are highly dynamic, and more extensive time-course experiments are needed to provide more information about the role of TSs in this scenario.
Root colonization by Trichoderma is an intimate relationship involving a tightly regulated exchange of molecular signals including SMs (Hidangmayum & Dwivedi, 2018). When Trichoderma colonizes the roots, it releases a variety of SMs that promote substantial changes in plant biochemistry, which in turn, cause changes in the fungal physiology (Contreras-Cornejo et al., 2018). Our results indicate that a significative reprogramming of terpene biosynthesis occurs in T. gamsii T6085 when colonizes wheat roots. Down-regulation of ts9 , suggests that root colonization induces a repression in indole diterpene biosynthesis in the fungus. In addition, the contrasting effects observed in the expression of sesquiTS and the up-regulation of the SQS, suggest that root colonization induces a modulation on sesquiterpene and triterpene biosynthesis in T. gamsii T6085 through FPP as central node. The absence of tri5 expression in response to wheat spikes and the strong up-regulation observed during root colonization was remarkable, as it suggests that signals from the roots induce the expression of this gene in T. gamsii T6085. Although activation of tri5 usually leads to the production of phytotoxic compounds, such as trichodermin in T. brevicompactum (Tijerino et al., 2011a), HA produced by T. arundinaceum lacks on phytotoxic activity and has a crucial role in plant protection against B. cinerea (Malmierca et al., 2013). This example illustrates how different ecological demands led to an adjustment in metabolic pathways governed by a single signature gene in species within the same fungal genus (Mukherjee et al., 2013). In this context, we can imagine the involvement of tri5 in the biosynthesis of a sesquiterpenoid involved in promoting a beneficial relationship between T. gamsiiT6085 and wheat plants, a question that must be further addressed by using tri5 -deletion mutants of the fungus. Results strongly indicate the involvement of TS genes in the interaction of T. gamsii T6085 with plant roots, and further studies determining their impact on the plant physiology will provide more information about the roles of these genes in the Trichoderma -plant beneficial interaction.