Comparative study of Trichoderma gene expression in interactions with tomato plants using high-density oligonucleotide microarrays

Effect of Trichoderma spp. on the growth and development of tomato plants

To check the effect of different Trichoderma strains on the development of tomato roots, an in vitro assay with 10 different strains was carried out. Trichoderma inoculations affected the root system architecture in tomato seedlings; both positive and negative effects were observed, depending on the strain. The most evident effects were observed for strains T7, T34 and T87, which were selected for further studies (Fig. 1). Strains T7 and T34 produced the most beneficial effects in tomato seedlings in terms of root elongation and lateral root development. In a previous in vitro study (Chacón et al., 2007), strain T34 was also seen to promote the growth of tomato and tobacco plants, increasing leaf area and lateral root formation. In our study, strain T87 had detrimental effects on tomato roots, and the plants also had a smaller leaf area and a lower true leaf number than the controls in three independent experiments. These results are in disagreement with previous studies indicating that strain T87 (Gv29-8) promoted Arabidopsis seedling growth through production of auxin-like compounds (Contreras-Cornejo et al., 2009) and induced higher photosynthetic rates and systemic increases in the uptake of CO₂ in leaves (Vargas et al., 2009). Interestingly, 18-mer peptaibols from this strain elicited systemic defences against a leaf pathogen in cucumber (Viterbo et al., 2007) and the T87-secreted elicitor Sm1 systematically stimulated defence mechanisms in maize (Djonovic et al., 2007). Since the cross-talk between Trichoderma and plants is dynamic, the negative effects that we observed in T87–tomato interactions might be related to the fungal concentration used, the plant material, the developmental stage of the plant, or the timing of the interaction. Moreover, it has been described that T87 is a ‘Q’ strain of T. virens able to produce gliotoxin, a phytotoxic metabolite, under several culture conditions (Howell, 2006); this may have negatively affected the development of tomato plants in our study.

Our in vivo results also showed that strain T87 significantly reduced the stem and root lengths of tomato plants and that tomato plant growth was unaffected by strains T7 and T34 (Table 1). It has recently been reported that the genetic background of the plant affects Trichoderma–tomato interactions, and treatments with biocontrol agents exerting negative effects in some tomato genotypes were also observed (Tucci et al., 2011). Moreover, evidence is being accumulated that the significant advantages of Trichoderma treatments to plants occur when they are under stress (Mastouri et al., 2010).

Overview of the Trichoderma gene expression data from microarray analysis

Plants benefit from the symbiotic interaction with Trichoderma mainly by increases in growth potential, resistance to diseases and tolerance to abiotic stresses (reviewed by Hermosa et al., 2012, this issue) whereas Trichoderma uses the nutrients provided by the plants. It has been demonstrated that plant-derived sucrose is an important nutrient source for Trichoderma and is also involved in the control of root colonization (Vargas et al., 2009). However, little is known about the changes influenced by plants in the transcriptome of Trichoderma. To assess the effects of the presence of tomato plants on the global gene expression of three Trichoderma strains, selected in an in vitro assay as described above, we hybridized Trichoderma HDO microarrays with cRNAs from T. hamatum T7, T. harzianum T34 and T. virens T87 obtained from cultures in the presence and absence of tomato plants. Of the total 34 138 probe sets deposited on the microarray, 1077 (3.15 %) showed at least a 2-fold significant change (FDR 0.25) in expression in the presence of tomato plants. The distribution of the 1077 probe sets across the three sample groups is shown in Fig. 2. Most (1023) of the selected probe sets exhibited expression change (up- or downregulated) in only one culture condition as compared with their control condition, with 593, 336 and 94 selected probe sets displaying changes only in T34–, T87– and T7–tomato interactions, respectively.

One of the most striking findings seen in the Venn diagram (Fig. 2) is that only six differentially expressed probe sets were common to the three Trichoderma–tomato conditions (Table 2). Genes encoding the N-acetylglucosamine-6-phosphate (GlnNAc-6P) deacetylase and glucosamine-6-phosphate (Gln-6P) deaminase proteins, involved in carbohydrate metabolism and also related to cell wall biosynthesis, were upregulated in all three Trichoderma strains. A gene encoding a protein related to lipid metabolism (coenzyme A-transferase) was upregulated in T34 and T87 but was downregulated in T7. Independently of the effects provoked in tomato plants by the different Trichoderma strains, carbohydrate metabolism and transport processes were commonly overrepresented in the three Trichoderma–tomato interactions, which could indicate an active adaptation of the fungus to the rhizosphere niche. A previous study with different endophytic Trichoderma strains also identified an induction of ESTs for enzymes potentially involved in nutritional support of Trichoderma ovalisporum and T. hamatum interacting with cacao seedlings (Bailey et al., 2006).

Despite the similar beneficial effects on tomato seedlings observed for strains T7 and T34, only seven probe sets were commonly altered in this strain pair (Fig. 2, Table 2), the highest number of them corresponding to genes encoding hypothetical proteins and genes that had no matches in databases. The other probe sets corresponded to genes encoding a 40S ribosomal protein, related to translation, and a protein homologous to the translationally controlled tumour protein (TCTP), also called fortilin or P23, a protein with chaperoning activity related to mitotic proliferation during the early stages of cell differentiation (Felts & Toft, 2003). TCTP/P23 overexpression in human cells leads to cell elongation resembling hyphal growth in fungi. In a previous study, the induction of the gene encoding the protein homologous to TCTP in T. harzianum T34 in contact with tomato roots was also observed (Samolski et al., 2009). These results suggest that P23 could be related to hyphal growth and the plant growth benefit derived from T7 and T34 root colonization. We also observed that a hexokinase activity, which phosphorylates hexoses and is mainly involved in the glycolytic pathway, leading to energy and metabolic building block availability, was commonly overrepresented in the T7 and T34 strain pair. However, the differences in plant development observed in biological assays between this strain pair and T87 may not be attributable only to such activity.

The largest number of common probe sets between strain pairs was observed between T34 and T87, transport and redox metabolism being the best-represented processes. This could be due to the closer phylogenetic distance between these two strains, although very different effects were observed in tomato seedlings when these strains were tested in in vitro and in vivo assays.

To check the reliability of the microarray-based data, we used quantitative real-time PCR to analyse the expression of selected genes: one upregulated gene and one downregulated gene in the T7–tomato interaction, two upregulated genes in the T34–tomato interaction, and three upregulated genes and one downregulated gene in the T87–tomato interaction (Fig. 3). The Gln-6P deaminase gene was upregulated in all three Trichoderma–tomato interactions. We also observed an upregulation of the 4-aminobutyrate transaminase gene in the T34–tomato interaction, and of cutinase and cellobiohydrolase I in the T87–tomato interaction, but a downregulation of the VIR1 chitinase gene in the T7–tomato and of the hydrophobin II gene in the T87–tomato interaction in comparison to their controls.

Our study may have certain limitations and it should be noted that several probe sets could correspond to the same gene; that GO categories were only assigned to some 45–52 % of the probe sets with significant expression change; that the expression changes were analysed under hydroponic culture conditions; and that with the present approach genes that rapidly and transiently change during Trichoderma–tomato interaction may have been missed. Nevertheless, several strengths of our study could be mentioned. Firstly, our chip contained 374 824 probes representative of a total of 34 138 gene transcripts from 14 Trichoderma strains. Although no T. hamatum genes were included in the microarray, a substantial representation of the Trichoderma transcriptome can be expected, considering that already sequenced Trichoderma genomes have been estimated to contain 9143–12 518 predicted genes (Martinez et al., 2008; Kubicek et al., 2011). Secondly, the robustness of our experimental approach was tested by using three biological replicates for each of the six conditions considered, and a hierarchical clustering of selected probe sets (Supplementary Fig. S1), based on the different expression patterns, grouped the three biological replicates of each experimental condition together, indicating a high quality of samples and overall comparability. Thirdly, eight quantitative real-time PCR results confirmed the microarray data (Fig. 3). Furthermore, we used a corrected P-value of <0.05 to identify the significantly overrepresented biological processes within the differentially expressed (FDR 0.25) probe sets in each sample.

T. hamatum T7 genes differentially expressed in response to tomato plants

A total of 113 probe sets differed significantly (FDR 0.25) in expression by at least 2-fold in 20 h T. hamatum T7–tomato plant interactions: 90 were upregulated whereas 23 were downregulated. This was the lowest number of differentially expressed probe sets among the three Trichoderma–tomato interactions analysed. Since the T. hamatum genome was not present on the HDO microarray, it could be expected that only a partial representation of the changes occurring in its transcriptome was being detected. However, we found that most of the differentially expressed probe sets (Table 3), 68 % from upregulated and 74 % from downregulated probe sets, were due to hybridization with EST-based unique sequences derived from 12 Trichoderma strains. Since these percentages were higher than those found for T. harzianum T34 (65 % from upregulated and 50 % from downregulated probe sets), represented by 3826 EST-based transcripts on this chip, we consider that this HDO microarray is an adequate tool for obtaining information about gene expression in T. hamatum T7.

GO categories were assigned to 59 of the 113 probe sets. Next we examined whether these differentially expressed probe sets were associated with similar GO categories. This analysis revealed that six categories were significantly overrepresented in T. hamatum T7 in the presence of tomato plants. Because several GO terms overlapped, based on similar groups of the probe sets we identified 14 different overrepresented probe sets (Supplementary Table S2) that were grouped into two physiological processes: carbohydrate metabolism and transport. As indicated above, both processes were overrepresented in the three Trichoderma strains in interaction with tomato plants. Strain T7 showed the highest number of differentially expressed probe sets related to chitinases. We identified an upregulation of genes with homology to glycoside hydrolase family 18 class III chitinases G1, 18-13 and 18-17 (phylogenetic group B), whereas a downregulation of a gene encoding the glycoside hydrolase family 18 class V chitinase VIR1 (phylogenetic group A) (Seidl et al., 2005) was only observed in this strain. Considering that group A chitinases have been associated with mycelial autolysis (Brunner et al., 2003), downregulation of proteins of this kind in the early Trichoderma–plant interaction could be expected. Similarly, group B chitinases with a typical cellulose-binding domain, such as 18-13 and 18-17, have been related to Trichoderma–plant recognition and attachment (Seidl et al., 2005), and their upregulation could be related to the beneficial effect of strain T7 on plants. A peptide with a cellulose-binding domain has been identified in T. asperelloides as a microbe-associated molecular pattern able to induce defence responses in cucumber (Brotman et al., 2008).

Chitinases are also involved in the metabolism of chitin, a polysaccharide not present in plant roots. However, only a few chitinase genes are induced by chitin in Trichoderma, suggesting that non-chitinous cell wall components can also act as inducers (Gruber et al., 2011). This is in agreement with the upregulation of genes encoding GlnNAc-6P deacetylase, Gln-6P deaminase and chitinase activities in hydroponic Trichoderma–plant interaction observed in the present study and may indicate that chitooligomers, able to trigger plant defence responses (Woo et al., 2006), could be generated by the chitin-hydrolytic system of Trichoderma without the need for chitin induction.

Regarding transport processes, it could be thought that an upregulation would be needed for nutrient exchange between the fungus and the plant. However, we observed a downregulation of major facilitator superfamily (MFS) transporters, which could indicate that after hyphal root attachment and colonization, Trichoderma nutrient uptake, at least at an early stage, is limited.

T. harzianum T34 genes differentially expressed in response to tomato plants

We identified 641 probe sets that differed significantly (FDR 0.25) in expression by at least 2-fold in 20 h T. harzianum T34–tomato plant interaction (Fig. 2); 65 % of the 519 upregulated and 50 % of the 122 downregulated probe sets were from EST-based unique sequences (Table 3). GO categories were assigned to 325 of the 641 probe sets. Subsequent analysis revealed that 235 GO categories were significantly overrepresented in T34 in the presence of tomato plants vs T34 alone. We identified 186 different overrepresented probe sets (Supplementary Table S3) that were grouped into several physiological processes, the transport and carbohydrate metabolism processes being those most evidently affected.

The relevance of transport processes in the T34–tomato interaction was highlighted by the upregulation of 67 probe sets related to proteins involved in such processes, e.g. MFS domain proteins (24), amino acid transporters (18), mitochondrial carrier proteins (7) and the maltose ABC transporter (1). An upregulation of the genes involved in sugar and amino acid transport, using macroarray analysis (Chacón et al., 2007), and of the genes related to mitochondrial carrier proteins, using an HDO microarray analysis (Samolski et al., 2009), was also observed in the T34–tomato interaction. Moreover, this strain showed upregulation of the highest number of genes encoding enzymes related to carbohydrate metabolism and involved in nutritional support of the fungus (β-galactosidase, cellobiohydrolase, endo-1,4-β-glucanase, xylanase, etc.) and in basic cell functions (glycerol-3P-dehydrogenase, alcohol dehydrogenase, phosphoenolpyruvate synthetase, etc.). The data indicate activation of glycolysis/gluconeogenesis, the tricarboxylic acid cycle, the pentose phosphate pathway, and other processes related to the generation of building-block compounds and energy and also involved in cell wall synthesis; these responses support the notion that nutrient exchange between the fungus and the plant is increased during the T34–tomato interaction. Since cuticle is not present in roots and since cutinase probe sets were upregulated in T34, a global regulation of the Trichoderma hydrolytic system could occur in order to improve the acquisition of carbon for saprophytic growth. Moreover, it has also been reported that the ethylene-inducing xylanase EIX1 elicits hypersensitive responses and other plant defence responses independently of its xylanolytic activity (Rotblat et al., 2002).

An upregulation of genes related to amino acid, nucleotide, energy and vitamin metabolism, detoxification and regulation, as well as of several genes involved in signalling, transcription, translation and post-translational processes was observed in T34–tomato microarrays, as shown in previous studies addressing this interaction (Chacón et al., 2007; Samolski et al., 2009). Although the upregulated genes were different from those observed in previous studies, our data confirm that modifications in the above physiological processes occur during the early stages of interaction between T34 and tomato in order to support the fungus’s higher demand for metabolic energy.

Although at least one gene encoding a β-ketoacyl synthetase, an enzyme that catalyses the condensation of malonyl-acyl carrier protein with the growing fatty acid chain, was downregulated, other genes involved in the biosynthesis of lipids and fatty acids (acetyl-CoA acetyltransferase, lipoic acid synthetase and 3-ketoacyl-acyl carrier protein reductase) were upregulated. These data are consistent with the upregulation of genes involved in lipid biosynthesis and catabolism observed in the T34–tomato interaction (Chacón et al., 2007). However, genes encoding laccases, which are enzymes with polyphenol oxidase activity related to melanin-like pigment synthesis in conidiospores, the induction of fruiting bodies and pathogenic interactions with plants (Hölker et al., 2002; Kiiskinen et al., 2004), were downregulated in T34.

An additional feature of the T34–tomato interaction observed in this work was the upregulation of genes encoding two enzymes involved in γ-aminobutyric acid (GABA) metabolism (succinate-semialdehyde dehydrogenase and 4-aminobutyrate transaminase). The GABA carbon is converted into succinic semialdehyde and then to succinate in reactions catalysed by these mitochondrial enzymes. An induction of both enzymes has been observed previously in Cladosporium fulvum–tomato and Pseudomonas–Arabidopsis interactions (Solomon & Oliver, 2002; Park et al., 2010), and in the symbiosis of Rhizobium leguminosarum with the pea host plant (Prell et al., 2002). The authors suggested that during the interactions the micro-organisms could alter the physiology of the plant, resulting in an enhanced production of GABA, which in turn plays a signalling role in the induction of the fungal enzymes responsible for its degradation. Thus, T. harzianum T34 could use GABA of plant origin as a source of nutrients.

Although one gene related to hydrophobin proteins was upregulated, another three hydrophobin genes were downregulated, as discussed below for the T87–tomato interaction.

T. virens T87 genes differentially expressed in response to tomato plants

We identified a total of 383 probe sets that showed at least 2-fold significant (FDR 0.25) changes in expression in the 20 h T. virens T87–tomato plant interaction. Whereas 65 % of the 205 upregulated probe sets were from the T. virens genome, the majority (85 %) of the 178 downregulated probe sets were from EST-based unique sequences (Table 3). It could be expected that the majority of the downregulated probe sets were from the T. virens T87 genome but a high level of cross-hybridization with heterologous probe sets was detected. We know that 991 ESTs from T. virens T59 were also used in the microarray design and this could explain some of the high percentage of downregulated probe sets coming from the 12 662 Trichoderma unique sequences in strain T87. GO categories were assigned to 173 of these 383 differentially expressed probe sets. Further analysis revealed that 58 GO categories were significantly overrepresented in T. virens T87 in the presence of tomato plants vs T. virens T87 alone. We identified 67 different probe sets (Supplementary Table S4) and then, based on similar groups of probe sets, we reduced the 58 overrepresented GO categories to six physiological processes. The highest proportions of the overrepresented probe sets corresponded to genes encoding proteins involved in carbohydrate metabolism and cell communication processes.

The relevance of carbohydrate metabolism was reflected by the upregulation of genes corresponding to enzymes such as GlnNAc-6P deacetylase, Gln-6P deaminase and chitinase, also observed in strains T7 and T34 during the colonization of tomato roots as discussed above, as well as β-glucosidase, xylanase, glycerol-3P-dehydrogenase and cutinase enzymes, also observed in T34 during the colonization of tomato roots. A major characteristic of the T87–tomato interaction was the large number of downregulated hydrophobin probe sets (24). Hydrophobins are small proteins that are found on the outer surfaces of both hyphal and conidial cell walls, where they mediate interactions between the fungus and the environment (Seidl-Seiboth et al., 2011), and some Trichoderma hydrophobin genes have been implicated in hyphal development and conidiation (Askolin et al., 2005; Mendoza-Mendoza et al., 2007; Muñoz et al., 1997) and root colonization (Viterbo & Chet, 2006). In previous studies the upregulation of the Hfb2 gene of T. reesei in media containing complex plant polysaccharides (Nakari-Setälä et al., 1997), the activation of a hydrophobin of T. atroviride in interactions with bean roots (Marra et al., 2006) and the upregulation of a hydrophobin homologous gene in T. harzianum T34 interacting with tomato roots (Samolski et al., 2009) have been reported. We might thus have expected to see an enhanced expression of one or more fungal hydrophobin genes during the Trichoderma–tomato interaction in our experiments. However, the hydrophobin genes appeared to be downregulated due to the presence of the plants. A differential regulation of 10 different class II hydrophobin genes in T. atroviride has been also documented (Mikus et al., 2009). Whether the downregulation of hydrophobin genes observed by us might be due to a direct influence of the plant or beneficial for the root colonization by the fungus remains to be further investigated. The redundancy of hydrophobin genes raises the question of different physiological functions. In this context, the downregulation of hydrophobin genes observed in T87 in this study could be involved in the negative effects caused by this strain in tomato seedlings in terms of root elongation and lateral root development.

Genes involved in lipid metabolism and in transport of phosphate, amino acids and general substrates were also upregulated. Our data are in agreement with the facilitation of nutrient exchange between the fungus and the plant and the reorganization of the membranes and cell wall of the fungus (Chacón et al., 2007).

However, we also observed downregulation of a gene encoding glutamine synthetase, an enzyme of importance in ammonia assimilation and the regulation of nitrogen metabolism, and one related to redox processes (d-lactate dehydrogenase, which modulates pyruvate flow when glucose is required) – both involved in fungal growth – and genes with homology to myosin I-binding proteins with important roles in the movements of organelles (Lee et al., 2000). The three probe sets assigned to the myosin I-binding GO category showed homology with genes encoding an l-amino acid oxidase, an oxidoreductase that catalyses the oxidative deamination of l-amino acids. Taking the T87 data together, it is difficult to understand the role played by the downregulation of genes involved in amino acid metabolism, although the turn-off of glutamine synthetase and l-amino acid oxidase could indicate a sufficient availability of nitrogen. In bacteria, lactate dehydrogenases can be regulated by the fructose 1,6-bisphosphate/phosphate ratio (van Niel et al., 2004). The downregulation of lactate dehydrogenase in T87 could be due to the accumulation of fructose 6-phosphate obtained via the activation of Gln-6P deaminase observed in the T87–tomato interaction.

Conclusions

During early Trichoderma–tomato interactions, effects on tomato seedling development varied depending on the Trichoderma strain. The transcriptomic responses of Trichoderma to the presence of tomato plants in the medium were also strain-dependent. Although few differentially regulated genes were common to T. hamatum T7, T. harzianum T34 and T. virens T87 in the presence of tomato plants, the overall transcriptomic changes detected in the three strains pointed to a significant activation of fungal carbohydrate metabolism and transport processes in the mutualistic Trichoderma–tomato interaction. The gene encoding TCTP/P23 protein was upregulated in the plant-growth-promoting T7 and T34 strains, relative to T87; this could be related to active hyphal growth in the contact with tomato roots. By contrast, the high number of downregulated hydrophobin probe sets observed in T87 could indicate that, under the experimental conditions of our study, this strain had a limited root attachment, affecting its capacity to promote plant growth.