A new species of Trichoderma hypoxylon harbours abundant secondary metabolites

Some species of Trichoderma are fungicolous on fungi and have been extensively studied and commercialized as biocontrol agents. Multigene analyses coupled with morphology, resulted in the discovery of T. hypoxylon sp. nov., which was isolated from surface of the stroma of Hypoxylon anthochroum. The new taxon produces Trichoderma- to Verticillium-like conidiophores and hyaline conidia. Phylogenetic analyses based on combined ITS, TEF1-α and RPB2 sequence data indicated that T. hypoxylon is a well-distinguished species with strong bootstrap support in the polysporum group. Chemical assessment of this species reveals a richness of secondary metabolites with trichothecenes and epipolythiodiketopiperazines as the major compounds. The fungicolous life style of T. hypoxylon and the production of abundant metabolites are indicative of the important ecological roles of this species in nature.

were also explored, which inferred from sequence analyses of the combined internal transcribed spacer (ITS), partial RNA polymerase II subunit (RPB2) and translation elongation factor 1 alpha (TEF1-α ) exon genes. Detailed comparisons were made between the new taxa and their related fungi. Considering T. hypoxylon is a new fungicolous species and its special life style, we reasoned that it is valued to study the chemical profiles to get the linkage to the biological roles.

Results
Phylogenetic analyses. Alignment results show that the sequences ITS, RPB2 and TEF1-α of T. hypoxylon are less than 97% similar to other Trichoderma species. Based on phylogenetic analysis of single gene of ITS, RPB2 and TEF1-α , T. hypoxylon formed a clade with Trichoderma taxi and Trichoderma rubi. The position of this clade showed closed relationship with section Hypoceanum, Polysporum, Psychrophila and other species in Trichoderma (Fig. S1, and Fig. S2). The phylogenetic analysis of RPB2 showed that T. hypoxylon grouped with the Polysporum section (Fig. S3). Therefore, the combination matrix included 60 ingroup taxa of Trichoderma which are phylogenetically close to T. hypoxylon. This data matrix comprised 2839 characters. Bayesian inference (BI), maximum likelihood (ML) and maximum parsimony (MP) trees generated shared the same topology. In MP analyses, 1642 (57.8%) characters are constant, 216 (7.6%) characters are parsimony-uninformative, and 981 (34.6%) characters are parsimony informative. In MP analyses, 1642 characters were constant, 981 were parsimony-informative, and 216 variable characters were parsimony-uninformative. Maximum likelihood tree was Presented (Fig. 1).

Secondary metabolite analysis and characterization of compounds. HPLC analyses of extracts
of T. hypoxylon cultivated on PDA medium were carried out to assess the production of secondary metabolites. Four major peaks with high yields were obtained by HPLC chromatography (Fig. 5A). To characterize compounds from T. hypoxylon, 1 liter fermentation on PDA medium were performed. After the semi-preparative reversed-phase HPLC separation step, we isolated four known compounds trichodermamide A (1), aspergillazine A (2), aspergillazine C (3) and harzianum B (4) (Fig. 5B). The assignments of four compounds were based on the published data of proton NMR [24][25][26][27] . Notably, the yields of harzianum B were 50 mg out of 330 mg crude extracts.

Discussion
Traditionally, delimitation of Trichoderma species was mainly based on the morphology 1 , although, it could not well explain the taxonomic position of these species 6,9 . Phylogenetic analysis has resulted in the discovery of many new species and has been extensively used in fungal taxonomy 28 . Presently, more than 258 species of Trichoderma are accepted based on phylogenetic analysis 5,10 . However, many species of Trichoderma remain to be discovered and described 5,6 . To explore the taxonomic position of T. hypoxylon, phylogenetic tree containing all species of Trichoderma species was constructed and the putative position of T. hypoxylon was shown (Figs S1-S3). The single gene of ITS, RPB2 and TEF1-α could distinguish T. hypoxylon from other Trichoderma species, the results suggested that these three genes are effective in taxonomy of Trichoderma 5,6,9 . However, the single gene could not well delimit Longibrachiatum, Viride and some other sections in genus of Trichoderma. Multigene analysis was a popular and feasible approach to solve the problem 5,6,10 .
Based on analyses of the combined sequences of ITS, RPB2 and TEF1-α , 56 currently known species in Trichoderma clustered together (Fig. 1). Nine subclades, Brevicompactum, Deliquescens, Green spore group, Hypoceanum, Longibrachiatum, Polysporum, Psychrophilum, Viride and a new subclade including Trichoderma hypoxylon were recognized, which is basically congruent with the results by Jaklitsch and Voglmayr 6 . The new subclade contained T. hypoxylon, T. taxi and T. rubi together readily distinguished from other Trichoderma species 6 . Trichoderma hypoxylon clearly differed from T. taxi and T. rubi, resulting the sequences similarity of RPB2 and TEF1-α are less than 97%.
Trichoderma rubi was found as a new saprobe on stems of Rubus ulmifolius, T. taxi was reported as new endophyte of Taxus mairei, whereas T. hypoxylon was an inhabitant on stroma of Hypoxylon anthochroum. In spite of their ecological niches, it showed closed phylogenetic relationship (Fig. 1, Figs S1-S3) and morphological similarity. However, they are a little different in conidiophores and conidia size, and T. hypoxylon did not produce pigment on PDA medium, whereas T. rubi produced brownish pigment and yellow crystals on PDA medium 6 .
Trichoderma species are a rich source of secondary metabolites 14,[15][16][17][18][19][20][21][22]29 , probably resulting their environments adaptation and lifestyles 15,16,30 . Comparative genome analysis revealed that fungicolous T. atroviride and T. virens are enriched in secondary metabolism-related genes compared with the biomass-degrading Trichoderma reesei 30 . Chemical analysis also showed that these fungicolous species could produce more peptaibols, peptides, polyketides, pyrones, siderophores, terpenoids/steroids than those non-fungicolous Trichoderma 15,30 . These compounds are ecologically and commercially important for their antimicrobial and anti-cancer properties, as well as their ability to induce systemic resistance in plants against microbial invasion 15,17,31 . As a new fungicolous species, the chemical diversity of T. hypoxylon aroused our attentions. Therefore, the secondary metabolites of this fungus were evaluated. Four major compounds were characterized (Fig. 5). Trichodermamide and aspergillazines are two kind of modified dipeptides 15,25 . Trichodermamide A has been found from a marine-derived T. virens 24 , it is also be obtained from marine-derived fungi Spicaria elegans 25 and Neosartorya pseudofischeri 32 and endophytic fungus Trichoderma spirale 33 . The yield of trichodermamide A produced by these fungi were 1.58 mg/L 24 , 0.73 mg/L 25 , 0.12 mg/L 32 and 8.98 mg/L 33 individually. In this study, we found that trichodermamide A was one of major compounds in the terrestrial fungus T. hypoxylon. It suggested that Trichoderma species are important resource for exploration of trichodermamide. Aspergillazines were firstly reported from a soil fungi Aspergillus unilateralis (MST-F867) 26 , it showed antibacterial and anti-cancer activity 15,19 . Aspergillazines A been found from two marine-derived fungi Spicaria elegans 26 and T. virides 25 , in which the productive rate of aspergillazines A was 3.17 mg/L 26 and 0.47 mg/L 25 . This compound has been found co-occurred with trichodermamide A in T. virides 25 . Trichothecenes are a well-studied class of sesquiterpene-based mycotoxins 34 . They are potent cytotoxins to eukaryotic cells which are mainly produced by fungal species of Fusarium, Myrothecium and Trichoderma in order Hypocreales 15,34 . Harzianum B is one of trichothecenes, which inhibit eukaryotic cell growth 27 and have the cytotoxic, antibiotic, and anthelmintic activities 35 . It is found that in a Hypocrea sp. (sexual morph of Trichoderma) strain F000527 yield of 18.13 mg/L 27 , while our results showed harzianum B was extremely high in T. hypoxylon with yield of 50 mg/L. The high amount of trichothecenes in T. hypoxylon, indicating that it has antagonistic potential against fungal hosts.

Materials and Methods
Isolates and specimens. Samples were collected on 5 May 2014 in Chiang Mai Province, Thailand.
Trichoderma strains were isolated from the host by single spore isolation as detailed in Chomnunti et al. 36  Morphological characterization. Methods and morphology were described basically following counterparts by Jaklitsch & Voglmayr 6 Colony radius and characteristics were determined on PDA as detailed in Manamgoda et al. 6 cornmeal dextrose agar (CMD; Difco cornmeal agar + 2% w/v dextrose) 10 and a defined low nutrient agar (SNA) 4 at 20, 25 and 30 °C in darkness for 7 d, then exposed to artificial light to stimulate conidia formation until 10 d. Microscopic observations and measurements were made from preparations mounted in 50% lactic acid. Photographs were taken with a Nikon DS-Fi2 CCD (Nikon, Japan) connected to a Nikon 80i microscope (Nikon, Japan) for anatomical structures. The statistics presented here are based on measurement of 30 mature conidia (± S.D.) and 30 phialides (± S.D.) at 100× magnification. To assess and describe their structure and morphology of conidiophores were taken from the edge of conidiogenous pustules or fascicles. Conidia were studied from cultures after 10 d of incubation.
DNA extraction, PCR amplification and sequencing. Approximately 50 mg of fungal material from each culture was placed in 600 μ l of 2% CTAB buffer and ground with a plastic pestle. Genomic DNA was extracted using a modified CTAB extraction protocol 39 . Three primer pairs, ITS5 and ITS4 40 , fRPB2-5 f and fRPB2-7cr 23 , EF983F and EF2218R 41 were separately used to amplify fragments of ITS, RPB2 and TEF1-α . Each PCR was performed on a Votix thermal cycler (Bio-Rad, CA, USA) using easy tag (Tiangen, Beijing, China) in a final volume of 50 μ l containing 10 μ mol of each primer and 2 μ l of DNA (10 ng/L). Reactions were run with positive and negative controls to ensure accuracy and to detect contamination. Automated sequencing was performed by Sino Geno Max Co. (Beijing, China). The sequences used in this study are deposited in GenBank under the accession number provided in Table S1.
Phylogenetic analysis. The ITS and TEF1-α data sets used sequences of isolates CGMCC 3.17906, CGMCC 3.17907 and reference sequences were downloaded from GenBank (Table S1). The ITS, TEF1-α and RPB2 data sets were aligned by MAFFT ver.7.03 using the Q-INS-I strategy, individually 42 . The ambiguous areas Scientific RepoRts | 6:37369 | DOI: 10.1038/srep37369 of alignment were located and removed using Gblocks 0.91b 43 . Previous phylogenetic analysis of ITS, RPB2 and TEF1-α sequence data from 260 Trichoderma species was conducted seperately with N. berolinensis and N. eustromatica as outgroup. Then, single and combined genes analyses of ITS, TEF1-α and RPB2 sequence data of 60 phylogenetic closed Trichoderma species in Brevicompactum, Deliquescens, Hypoceanum, Longibrachiatum, Polysporum, Psychrophila and green spore sections were carried out. N. berolinensis and N. eustromatica were arranged as outgroup taxa.
Maximum parsimony (MP) analysis was conducted by PAUP 4.0b10 44 using a heuristic search with tree-bisection-reconnection branch swapping. All characters were treated as unordered and unweighted, gaps were treated as missing data, sequences were auto-increased and Maxtrees was 5,000. Topological confidence of resulted trees was tested by bootstrap proportion with 1,000 replicates, each with 100 replicates of random addition. Bootstrap proportion (BP) higher than 50% from maximum parsimony analysis from PAUP are given.
Maximum-likelihood (ML) analysis was performed in RAxML 45 implemented in raxml GUI v.1.3 46 . GTRGAMMAI was specified as the model. The analysis was run with a rapid bootstrap analysis using a random start with rapid bootstrap analysis with 1,000 replicates. Bootstrap proportion (BP) higher than 50% from maximum likelihood analysis from RAxML are given.
Bayesian Inference (BI) analysis was performed with MrBayes 3.1.2 47 using Markov chain Monte Carlo (MCMC) algorithm. Appropriate nucleotide substitution models was determined by MrModeltest 2.3 48 and the best fit model "GTR + I + G" was selected by Akaike Information Criterion for the investigated data set. Six chains (one cold and three heated) of 135,000 Markov chain Monte Carlo generations were run, sampling every 100 generation resulting in 1,350 total trees (in two simultaneous analyses). The initial 337 trees (25%) were discarded as burn-in phase of the analyses, and the remaining trees in each analysis were used to calculate posterior probabilities (PP) in the majority rule consensus tree 49 , posterior probabilities greater than 0.95 are given.
All trees were viewed in TreeView 1.6.6 50 and revised in Adobe Illustrator CS5.
Analytical methods and equipment overview. 1