Diversity of cultivable fungal endophytes in Paullinia cupana (Mart.) Ducke and bioactivity of their secondary metabolites

Paullinia cupana is associated with a diverse community of pathogenic and endophytic microorganisms. We isolated and identified endophytic fungal communities from the roots and seeds of P. cupana genotypes susceptible and tolerant to anthracnose that grow in two sites of the Brazilian Amazonia forest. We assessed the antibacterial, antitumor and genotoxic activity in vitro of compounds isolated from the strains Trichoderma asperellum (1BDA) and Diaporthe phaseolorum (8S). In concert, we identified eight fungal species not previously reported as endophytes; some fungal species capable of inhibiting pathogen growth; and the production of antibiotics and compounds with bacteriostatic activity against Pseudomonas aeruginosa in both susceptible and multiresistant host strains. The plant genotype, geographic location and specially the organ influenced the composition of P. cupana endophytic fungal community. Together, our findings identify important functional roles of endophytic species found within the microbiome of P. cupana. This hypothesis requires experimental validation to propose management of this microbiome with the objective of promoting plant growth and protection.


Introduction
Paullinia cupana var. sorbilis (Mart.) Ducke is a Brazilian Amazonia plant species. Commonly known as guaraná, it is used as central nervous system stimulant due to its high concentration of caffeine [1]. The first P. cupana crops were grown by indigenous tribes, but the beverage Root associations, endophytic potential and functional traits P. cupana roots were washed, clarified with KOH (10%, v/v), and stained with trypan blue in lactoglycerol (0.05%, v/v) for visualization of fungal-root associations [73]. Fifteen root fragments from each individual were analyzed in the Nikon Eclipse E200 microscope coupled with a camera (Bel Photonics), in order to identify the presence of dark septate endophytic fungi (DSE) and endomycorrhizal structures, such as arbuscles and vesicles. The identity of the structures was confirmed by comparison with literature reports [74].
Endophytic fungi with dark mycelium and septate hyphae isolated from P. cupana were selected to examine their infection capacity and identify typical structures of dark septate endophytes. P. cupana seeds were not used due to the difficulty in promoting their germination [75], and alternatively, seeds of the Sorghum cultivar BRS 373 were used (Sorghum sp.). Sorghum sp seeds were disinfected [76] and incubated in minimum mineral medium (MM) [77] at room temperature (approximately 27˚C) under natural light. After 5 days, the rootlets of seedlings were inoculated with mycelium fragments. Non-inoculated seedlings were used as controls. The roots were cleared, stained, and analyzed after 15 days of interaction [73]. Stained roots were preserved in polyvinyl lactoglycerol and observed under light microscope [78].
The production of cellulase [79], esterase [80], amylase, phosphatase and protease [81] was analyzed in vitro in all the isolated endophytic fungal species, using essentially qualitative tests. The siderophore production was determined in cromoazurol S medium [82]. Discs of endophytic mycelium (⌀1 mm) were used as inoculum and the formation of halos was determined after 24 h incubation. The production of indoleacetic acid (IAA) was determined by a colorimetric method, using BD broth supplemented with tryptophan (0.73 μmol/ml) as the negative control [83]. The IAA concentration was determined using a standard curve prepared with commercial IAA.

Antibacterial activity
A lineage of each endophytic fungal species was grown in PDA medium for the production of ethyl acetate (EtOAc) extracts, as reported by Rosa [84]. The qualitative antibacterial activity of the endophytes' EtOAc extracts (20 mg/mL) was assessed as reported by Ichikawa [85], against the following strains: susceptible Escherichia coli (ATCC 25922), Pseudomonas aeruginosa (ATCC 9027) and Staphylococcus aureus (ATCC 6538), and multiresistant E. coli (1A), P. aeruginosa (2A) and S. aureus (3A). The extracts that suppressed bacterial growth the most strongly were selected for determination of the minimal inhibitory concentration (MIC)-the lowest sample concentration at which bacteria do not grow [86]-using serial dilutions ranging from 5,000 μg/mL to 39 μg/mL. Tetracycline (0.05 mg/mL) and LB broth were used as the positive and negative controls, respectively. The minimum concentration of death (MCD) was determined using Mueller-Hinton agar medium [87].
Nuclear magnetic resonance (NMR) spectra were recorded using a Bruker Ascend 500 spectrometer operating at 500 MHz for 1 H nuclei and 125 MHz for 13 C nuclei. Chemical shifts were quoted in parts per million (ppm), referenced to the appropriate residual solvent peak. Two-dimensional spectroscopy (COSY, HSQC and HMBC) were used for structural determination. Gas chromatography coupled to mass spectrometry (GC-MS) were performed on a Shimadzu GC17A chromatograph coupled to a GCMSQP5050A detector. After structural determination, the purified samples were tested for antibacterial activity using the method reported in the previous section. MIC was determined with serial dilutions ranging from 30 μg/mL to 0.2 μg/mL.

Antitumor activity
The elucidated purified molecules (DEHP and 3HPA) were tested for antitumor activity. The Chinese hamster ovary cells (CHO) and Mus musculus skin melanoma B16F10 cell lines were kindly provided by Dr. Edgar Julian Paredes Gamero from Federal University of São Paulo (UNIFESP, São Paulo, SP, Brazil) and cultured in the Laboratory of Cell Culture and Molecular Biology at the Federal University of Grande Dourados (Dourados, MS, Brazil). CHO and B16F10 cells were cultured in DMEM + F10 and RPMI-1640 medium, respectively, containing 10% fetal bovine serum (FBS), 100 U/mL penicillin and 100 μg/mL streptomycin, in a humidified incubator at 37˚C and under 5% CO 2 .
The 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay provides a sensitive measurement of the cell metabolic status, in particular the mitochondrial function, which reflects early cellular redox changes [88]. CHO and B16F10 cells (6x10 3 and 1x10 3 cells/well, respectively) were grown in 96-well tissue culture plates and treated with the isolated compounds (3.12, 6.25, 12.5, 25, 50 and 100 μg/mL) for 24, 48 and 72 h. The cells were washed with 0.1 mL of 0.1 M phosphate buffered saline (PBS), pH 7.4, at 37˚C, and further incubated with 100 μL of MTT solution (1 mg/mL in culture medium) for 3 h, at 37˚C. The assay medium was removed and the dark-blue formazan crystals formed in intact cells were dissolved in DMSO. The absorbance was recorded at 630 nm in a microplate reader. Three independent experiments were performed in triplicate. Doxorubicin was used as positive control; it inhibits CHO and B16F10 cell growth by 50% (IC 50 ) at the concentration of 1 and 0.1 μg/mL, respectively. The results were expressed as the percentage of MTT reduction relative to the control group (untreated cells).
cause DNA damage that does not reflect the genotoxic effect. Thus, first we used the MTT assay to address whether compound 3A was cytotoxic towards V79 cells. Briefly, after a 24-h treatment with this sample (12.5, 25.0, 50.0 or 100.0 μg/mL) at 36.5˚C, the assay medium was removed and the cells were incubated with 100 μL of MTT (0.5 mg/mL) for 2 h at 36.5˚C. The assay medium was replaced by 100 μL of DMSO and, after five minutes, the samples were transferred to another plate to record their absorbance at 540 nm, using a microplate reader (Multiskan EX, Thermo Electron Corporation). Untreated cells and 10.0 μM doxorubicin (Sigma-Aldrich, St. Louis, MO, USA) represented the negative and positive controls, respectively. The results were expressed as the percentage of MTT reduction relative to the negative control group (100% survival).
The comet assay was performed using the method reported by Speit [91], with minor modifications. The V79 cells were treated with sample 3A (25.0, 50.0, or 100.0 μg/mL) for 3 h, at 36.5˚C, protected from light to prevent additional DNA damage. Cells treated with medium and 200 μM H 2 O 2 were used as the negative and positive controls, respectively. Aliquots of 20 μL of cell suspension were mixed with 100 μL of low-melting point agarose (0.5% in PBS) and layered onto a microscope slide previously thin-layered with 1.5% normal-melting point agarose. The slides were immersed in freshly prepared lysing solution (100 mM EDTA, 10 mM Tris, 2.5 M NaCl, 1% Triton X-100, and 10% DMSO pH 10 -all from Sigma-Aldrich, St. Louis, MO, USA) for at least 24 h, at 4˚C. Then, the slides were immersed in alkaline buffer (0.2 M EDTA, 10 M NaOH, pH >13, both from Sigma-Aldrich, St. Louis, MO, USA) for 25 min, at 4˚C, and submitted to electrophoresis at 0.92 V/cm and 300 mA, for further 25 min. Next, the slides were immersed in neutralization buffer (0.4 M Tris-HCl, pH 7.5, Sigma-Aldrich, St. Louis, MO, USA) for 15 min, and fixed in 100% ethanol for 5 min.
The slides were air-dried, stained with 0.3X Sybr Gold Nucleic Acid Gel Stain (Invitrogen, Carlsbad, CA), and analysed in an Ecliple Ci fluorescence microscope (Nikon, Japan) equipped with a 20X objective. Images of at least 100 nucleoids were obtained using the Lucia Comet Assay software (version 7.30, Laboratory Imaging, Czech Republic). The % DNA in the comet tail was considered as the parameter of genotoxicity. Three independent experiments were assayed in triplicate. Once the negative control of one experiment was excluded due to technical problems, we carried out an additional experiment without replicates. The data from each experiment are presented as the mean of median of 100 or 200 nucleoids.

Statistical analysis
The Past software [92] was used to determine diversity and equitability indexes and perform cluster analysis. The package vegan in the R software version 3.3.2 [93] was used to carry out the analyses of variance, similarity, permutation and correlation. Indicator species test was performed using package labdsv (http://ecology.msu.montana.edu/labdsv/R) within the R environment. Network was constructed and visualized using the software Gephi version 0.9. 1 [94]. Venn diagrams were generated with DrawVenn web tool (http://bioinformatics.psb.ugent.be/ webtools/Venn/). Data from the antitumor activity assay were analyzed by the Dunnett's multiple comparison test, using the GraphPad Prism software (GraphPad Software, Inc., San Diego, CA). Data from the comet assay (both cytotoxicity and genotoxicity) presented normal distribution and were statisticaly compared using one-way ANOVA and Bonferroni's posthoc test, at a significance level of 5%, with the aid of the GraphPad Prism software.

Diversity of endophytic fungi in P. cupana roots and seeds
We isolated and identified 256 strains and 34 species of endophytic fungi from P. cupana roots and seeds. The phylum Ascomycota, the class Sordariomycetes, and the order Diaporthales had the highest relative frequency ( Table 1). The aforementioned taxa often colonize tropical plant communities [28,95].
The genus Xylogone and the species Xylogone ganodermophthora had the highest total relative frequency in root endophytic communities ( Table 1). The genus Xylogone was described only once as endophyte of Taxus chinensis [96], and X. ganodermophthora was described as mycopathogen [97] and as an antifungal species against watermelon pathogens [98]. Here we report for the first time that the species X. ganodermophtora is an endophyte of P. cupana. Phomopsis asparagi, the species with the second highest total relative frequency ( Table 2), was reported as endophyte [99] and phytopathogen in asparagus crops [100].
Diaporthe phaseolorum and P. asparagi were the species with the highest relative frequency in seed endophytic communities (Table 2). Only the former was present in all the endophytic communities of P. cupana seeds ( Table 2). The genus Phomopsis and the species D. phaseolorum are predominantly endophytic in many tropical species [101][102][103][104].
Only the genera Diaporthe, Fusarium, Glomerella and Phomopsis were identified in endophytic communities of both the roots and seeds of P. cupana ( Table 2). The genera Diaporthe, Pestalotiopsis and Phomopsis were previously identified in endophytic communities of phyllospheres [29]. Thus, we hypothesize that endophytic fungi that colonize P. cupana have organspecificity, as reported for other tropical and temperate host plants [105][106][107].
Regarding the ecological role of the isolated species (S1 Table), all the endophytic fungal communities isolated from P. cupana had at least one species that is reported as endophyte, phytopathogen, saprophyte, mycoparasite, entomopathogen or coprophile. Some fungal species have been reported as both endophytic and phytopathogenic. It is known that endophytic communities present latent phytopathogens [119] and several phytopathogens can behave as endophytes depending on the local biotic and abiotic conditions [120,121]. In addition, the genetic diversity of some endophytic fungi may enable them to act as phytopathogens or Several species reported as saprophytes, coprophiles and entomopathogens were isolated from P. cupana (S1 Table)    Analysis of the indicator value revealed that few species preferred a specific community. Regarding the geographic location, P. asparagi significantly preferred Maués (IndVal = 0.314, P = 0.01). Regarding the P. cupana organs, X. ganodermophthora significantly preferred root communities (IndVal = 0.475, P = 0.001), while D. phaseolorum significantly preferred seed communities (IndVal = 0.15, P = 0.001). H. fuscoatra was significantly associated with the root endophytic communities of the susceptible P. cupana genotype grown in Manaus (IndVal = 0.2, P = 0.012). No species was significantly associated with a specific genotype (susceptible or tolerant).
The P. cupana genotype influenced the endophytic fungal community diversity (Table 3). Considering the communities from seeds collected in both geographic locations (Manaus and Maués), those isolated from the susceptible genotype exhibited higher richness, Shannon index and evenness when compared with those isolated from the tolerant genotype. Regarding the root endophytic communities from Manaus, those isolated from the susceptible P. cupana genotype had higher Shannon index, evenness and colonization rate than those isolated from the tolerant genotype. The endophytic communities isolated from the roots and seeds of the susceptible P. cupana genotype have higher richness and Shannon index than those isolated from the same organs of the tolerant genotype. The richness of microbial communities in the tolerant genotype was greater than that of the susceptible genotype only in the root endophytic community of plants grown in Manaus. The root endophytic communities isolated from the susceptible genotype grown in Maués had higher richness, Shannon index and dominance but lower evenness and colonization rate than the root endophytic communities from the tolerant genotype grown in the same location.
The root endophytic communities of the tolerant and susceptible genotypes grown in Manaus shared 11 species (Fig 1A), while those grown in Maués shared 6 species ( Fig 1B). Communities isolated from the same organ shared a higher number of species in the susceptible ( Fig 1C) and tolerant (Fig 1D) genotypes. Only the species F. oxysporum, G. acutata and P. asparagi occurred in both organs, explaining why communities of different organs shared few species. The root endophytic communities of plants grown in Manaus had the largest number of unique species (Fig 1C and 1D).
As reported in tobacco, tomato and potato cultivars [39,40,145], we found that different endophytic communities colonized the susceptible and tolerant genotypes of P. cupana, and that the susceptible genotypes exhibited higher species diversity-the last finding may be related to their vulnerability to microbial infections, which colonize tissues faster [146,147]. Moreover, roots can produce exudates that promote colonization of certain groups of microorganisms [148,149]. Exudates produced by different genotypes can influence the microbial community in rhizosphere and consequently affect the composition and richness of endophytic communities [150]. Additional studies are required to determine whether susceptible and tolerant P. cupana genotypes differ with respect to the composition of exudates. The geographic location where P. cupana was grown also influenced the endophytic fungal community's diversity (Table 3). Compared with the same genotype grown in Maués, the endophytic fungal communities isolated: (i) from seeds of the tolerant genotype grown in Manaus had higher colonization rate, dominance and richness; (ii) from seeds of the susceptible genotype grown in Manaus had lower colonization rate, evenness, richness and Shannon index; (iii) from roots of the tolerant genotype grown in Manaus had higher dominance, richness and Shannon index; (iv) from roots of the susceptible genotype grown in Manaus had higher colonization rate, evenness, richness and Shannon index.
The observed and estimated species richness (Chao index) in the endophytic fungal communities identified in Maués were similar to each other; in Manaus, the estimated richness was higher than that observed for most endophytic communities (Table 3). Therefore, the geographic location influences the endophytic fungal diversity in P. cupana. Manaus exhibited the highest species richness, more specifically in seeds of the tolerant genotype and roots of the susceptible genotype.
The level of particulate matter in the air of urban areas is high due to several anthropic activities, such as industrial and vehicular emissions, biomass burning and soil resuspension Functional traits and biological activity of endophytic fungi from Paullinia cupana [151,152]. Particulate matter suspended in the air carries a high diversity of fungi and bacteria [153-156] that can colonize the surface and integrate the epiphytic community of plants on which the particles deposit [157]. Several studies have reported epiphytic species as endophytes [28, 158,159], indicating that a variety of species can colonize the inner plant tissues. Considering that Maués is located in a rural area within the Amazonia forest relatively distant from urban centers, and that Manaus is a large urban center, we hypothesized that the P. cupana samples grown in Manaus received more particulate matter from the air and thereby a greater amount of inoculum, which may explain their greater richness of endophytic fungal species. Urban and rural areas may present distinct endophytic communities even among hosts belonging to the same species [160][161][162].
Other factors such as temperature, rainfall and soil composition can influence the diversity of endophytic species [163][164][165]. These parameters differed between Manaus and Maués, which may have impacted on the endophytic diversity of each community. According to Koppen climate classification, Manaus climate is ranked as Am (monsoon), with average annual temperature of 26.7˚C, average annual rainfall of 2.420 mm and soil characterized as yellow latosol. On the other hand, Maués climate is ranked as Af (humid equatorial), with average annual temperature of 25.5˚C, mean annual rainfall of 2.101 mm and soil characterized as eutrophic gleysoil [166,167].
The plant organ is an important determining factor of the endophytic community's composition in P. cupana, because the roots were colonized by communities with higher richness and diversity (S1 Fig). The isolation of seeds from the external environment for a long period [168] and the presence of antifungal compounds in seeds [9] may also lower endophytic species richness in this plant organ. P. cupana seeds are surrounded by a thick epicarp and partially surrounded by a thin membrane (the pith) [75], while the roots are in constant contact with the ground and can be infected by a great diversity of edaphic microorganisms.
Although the NMDS plot did not show separation between the endophytic fungal communities of distinct P. cupana genotypes and geographic locations, the communities of seeds and roots were clearly different and clustered as distinct groups (ANOSIM R 2 = 0.27, P = 0.001, stress = 0.064) (S2 Fig). The root endophytic communities had a narrower distribution because they shared greater similarity in composition and abundance, indicating that the plant organ influences the community's composition in P. cupana.
The Jaccard index of similarity (Fig 2) showed that (i) the communities from seeds and roots shared the lowest levels of similarity; (ii) the clonal type was crucial to distinguish the communities within each plant organ; and (iii) the location determined the highest rates of similarity between communities. The degree of similarity between endophytic communities may decrease as the geographical distance between same host species increases [28,34,35,165,169].
Regarding the network analysis, the 34 nodes were connected by 107 edges, with average degree of 6.29 and clustering coefficient of 0.617 (S3 Fig). Ten species presented intermediate values of betweenness centrality, most of them were isolated from P. cupana roots and occured in both clonal types grown in both locations. X. ganodermophtora and P. asparagi were also identified in both clonal types from both locations and exhibited the highest betweenness centrality. X. ganodermophtora had the highest centrality degree, with 23 edges, binding to almost all species isolated from roots and acting as a local hub, while P. asparagi had the second highest centrality degree, with 20 edges mainly composed of species isolated from seeds. The network had two modules: the largest and the smallest ones were composed of endophytic fungal species isolated from P. cupana roots and seeds, respectively (S4 Fig). P. asparagi constituted an important node because it connected the seed and the root species modules. The formation of distinct modules for each organ, as well as the presence of different indicator species in each organ, stressed that different communities colonize the roots and seeds. Therefore, X. ganodermophtora and P. asparagi represented the central species of the endophytic fungal community of P. cupana and exerted the strongest influence on its structure, stability and dynamics.

In vitro endophytic ability
The presence of microsclerotia in the roots of all samples analyzed confirmed that P. cupana was associated with DSE (Fig 3). Associations with DSE fungi can become pathogenic, although most of them benefit the host by increasing nutrients absorption and promoting plant growth [170][171][172].
P. cupana roots contained typical structures of mycorrhizal fungi. Arbuscules with the arum type morphology were detected in the roots of the susceptible genotype grown in Manaus; coil hyphae and brown septate hyphae were exclusively detected in tolerant host plants; and vesicles were observed only in the clonal cultivar grown in Maués. The literature reports the association between P. cupana and mycorrhizal fungi [173,174]. Colonization and sporulation are seasonal events strongly influenced by increased rainfall [173], favored by increased soil acidity and manganese concentration, and inhibited by high iron content [174].
The strains C. gloeosporioides (6TSA), P. asparagi (22TSA) P. pinodella (28S), and Sydowiella fenestrans (104BDA) were selected for the evaluation of endophytic capacity because they presented dark mycelium and septate hyphae. P. asparagi is frequently reported as phytopathogenic against asparagus crops [100], causing discoloration of the stem or trunk, oval lesions and brown spots on the roots [175,176]. Although the endophyte P. asparagi (22TSA) produced microsclerotia (Fig 4A), it also caused the same pathogenic symptoms reported for asparagus, probably due to the long cultivation period used in this study. Sorghum roots colonized by P. pinodella (28S) contained microsclerotia (Fig 4B), but they did not show apparent symptoms of disease. Among the endophytic fungal species isolated from seeds, only P. pinodella (28S) exhibited dark mycelium and septate hyphae, the other fungal species colonized the seedling radicle without either developing dark septate structures or causing disease symptoms. There are no literature reports that the fungal species evaluated produce DSE structures; however, there are DSE species belonging to the genus Diaporthales [177], Phomopsis [178] and Pleosporales [179,180].
DSE are distributed in a wide variety of environments, such as sub-Antarctic regions, temperate and tropical forests, semi-arid regions, among others [181,182]. The currently available  Functional traits and biological activity of endophytic fungi from Paullinia cupana information on DSE mainly comes from studies carried out in arctic and temperate regions [183]. Several tropical plant families associate with DSE [181], but there are few reports on the association between DSE and species of the family Sapindaceae [184,185] and other species in the Amazonia region [186,187].
The EtOAc extracts of D. melonis (29TSA), D. phaseolorum (8S), D. terebinthifolii (24S), and T. harzianum (43BDA)-which suppressed the growth of the greatest number of bacterial strains in the qualitative test-were selected for determination of the minimal inhibitory concentration (Table 4). Analysis of the minimum concentration of death evidenced some microbial growth after treatment with all the extracts' concentrations tested, indicating that they exerted bacteriostatic activity and that their minimum concentration of death was higher than 5.000 μg/mL (Table 4). The antimicrobial activity of D. melonis [189], D. phaseolorum [103] and T. harzianum [202][203][204] has already been reported.
We did not identify esterase-, phosphatase-, and siderophore-producing endophytic fungal species in P. cupana under the conditions assessed. Thirteen endophytic fungal species did not display any enzyme activity; however, enzyme release by some of them were already reported, such as the synthesis of (i) amylase and cellulase by C. gloeosporioides and P. macrospinosa [205,223]; (ii) amylase and protease by M. camptospora [214,224]; and (iii) amylase, cellulase and protease by T. asperellum [225][226][227]. The species D. terebinthifolii, M. elegans, N. rigidiuscula, N. mackinnonii, P. arecacearum, P. macrospinosa, P. pinodella, P. lagerstroemiae and S. fenestrans-which did not present enzymatic activity in our work-, are not reported in the literature with any enzymatic activity.
Only the species F. oxysporum, M. terrestris and T. harzianum, which were isolated from tolerant and susceptible genotypes of P. cupana, exhibited amylolytic, cellulolytic and proteolytic activity. Although protease-, cellulase-, and amylase-producing endophytic fungal species Diaporthe melonis -29TSA 312.5 39.06 > 5.000 5.000  Functional traits and biological activity of endophytic fungi from Paullinia cupana colonized both P. cupana genotypes, the production of enzymes and antifungals in the roots may not be sufficient to prevent the development of anthracnose-this disease affects the aerial organs, mostly leaflets, petioles and young stems [18,30,228]. We identified 17 IAA-producing endophytic fungal species in P. cupana (Table 4), among which F. oxysporum, F. solani, T. harzianum and T. asperellum are known IAA producers in vitro [229][230][231]. Although the other endophytic fungal species analyzed did not produce detectable levels of IAA, several species of Glomerella, Mycena and Pestalotiopsis were previously reported to synthesise IAA in vitro and promote plant growth [232][233][234]. Endophytes can release IAA to promote plant growth in several crops such as rice, sugar cane [235] and coffee [236], and to favor germination [237]. Together, our findings demonstrate that endophytic fungi isolated from tolerant and susceptible genotypes of P. cupana also synthesize IAA in vitro, which could benefit the host.

Purification and structural elucidation of special metabolites
The CH 3 OH-H 2 O (3:7) fraction from D. phaseolorum (8S) afforded a colorless solid, named sample 3A. The 1 H NMR spectrum showed two signals at 3.07 ppm (2H, t, J = 6.02 Hz) and 4.68 ppm (2H, t, J = 6.05 Hz); the 13 C NMR spectrum showed three signals at 30.74, 69.29 and 174.51 ppm; and the 2D NMR spectrum showed two triplets coupled with each other (COSY). HMQC correlations demonstrated that protons with chemical shifts at 4.68 and 3.07 ppm were attached to the carbons that gave the signals at 69.29 and 30.74 ppm, respectively, while the HMBC correlation demonstrated that both proton signals gave long-range correlations to each other's carbons as well as to the carbon at 174.51 ppm. The aforementioned spectral data are similar to those reported in the literature for 3-HPA [103,238] (Table 5). Therefore, sample 3A corresponds to 3-HPA (3-hydroxypropionic acid).
The CH 3 OH extract fraction obtained from T. asperellum (1BDA) provided a solid lightyellow sample named 17A after a HPLC purification step. Analyses of 13 C and 1 H NMR spectra (  67, respectively. Mass spectrum showed a molecular ion peak at m/z = 254. Thus, sample 17A corresponds to 1-hydroxy-8-methoxyanthraquinone, corroborating literature reports [240].
DEHP is used as plasticizer in the plastics industry and is considered as a persistent environmental pollutant due to its extensive usage [241]. DEHP was isolated from endophytic microorganisms [242,243], but not from those belonging to the genus Diaporthe. Hence, this is the first report on the production of DEHP by species of the genus Diaporthe. As DEHP is present in laboratory equipment and accessories [244,245] and it is considered a common contaminant of analytical equipment [246,247], several authors have questioned whether DEHP is a natural product or an analytical contaminant [248][249][250]. We believe that DEHP is a fungal product because it was exclusively detected in this sample.
3-HPA is commercially available and used in a range of industrial applications [251]. This compound can be synthesized through different metabolic pathways, from different precursors, and by both prokaryotes and eukaryotes [252], including endophytes [103,238,253]. There are few literature reports on 1-hydroxy-8-methoxyanthraquinone. It was produced through chemical synthesis [254] and isolated for the first time as a natural product from the Table 5. NMR data of isolated compounds from P. cupana endophytic fungi. phytopathogenic fungus Leptographium wageneri [240]. To the best of our knowledge, this is the first report on the isolation of 1-hydroxy-8-methoxyanthraquinone from an endophytic fungus and from a microorganism belonging to the genus Trichoderma.

Antibacterial, antitumor and genotoxic activity of the isolated compounds
Here we addressed whether the three special metabolites isolated from endophytic fungal species that colonize P. cupana display antibacterial, antitumor and genotoxic effecs. First, we used the minimal inhibitory concentration test to examine their antibacterial activity. 3-HPA and DEHP, at all the concentrations tested, exerted the strongest antimicrobial activity against P. aeruginosa susceptible and multiresistant strains ( Table 6). The growth of bacterial colonies in culture medium indicated that 3-HPA and DEHP were bacteriostatic. The antibacterial and nematicidal activity of 3-HPA was previously reported [103,238], but the mechanisms underlying its bactericidal action remain unclear [252]. DEHP exerts antibiotic activity against resistant and susceptible pathogenic bacterial strains [255][256][257][258][259]. In contrast, 1-hydroxy-8-methoxyanthraquinone did not suppress the growth of the bacterial strains tested and there are no literature reports on its antimicrobial activity. Second, we assessed the antitumor effect of the three special metabolites against CHO and B16F10 cells in vitro. 3-HPA reduced the viability of both tumor cell lines in a concentrationdependent manner (S5 Fig). A 72-h treatment with 3-HPA at 100 μg/ml diminished CHO and B16F10 cell viability to 80% and 40%, respectively. 3-HPA is strongly cytotoxic towards prokaryotic cells [252], but there are no reports on its antitumor effect. Our data suggest that 3-HPA is toxic to both tumor and non-tumor cell lines and exerts stronger cytotoxic effect on the former.
There As we did not find information about its genotoxicity in the scientific literature, we recommend performing additional mutagenic tests to examine whether it is safe. 3-HPA was cytotoxic towards CHO cells but not towards V79 cells, corroborating literature reports that these cell lines have different sensibility to cytotoxic compounds, and that CHO cells can be several folds more sensible than V79 cells to toxic molecules [266]. Table 6. Minimal inhibitory concentration (μg/mL) of the special metabolites isolated from endophytic fungi of P. cupana.

Conclusion
We isolated and identified thirty-four endophytic fungal species in P. cupana: eight in seeds, twenty-three in roots and three in both organs. Eight species were not previously reported as endophytic, including X. ganodermophthora, which was the most abundant species in roots. These findings demonstrate the potential of the Amazonia forest as an environment with high diversity of endophytic fungi. The plant geographic location, clonal type, and organ are factors that influence the structure of the endophytic fungal community of P. cupana. The plant organ is the most important factor that causes differentiation in the community's composition. We confirmed that mycorrhizae and DSE colonize susceptible and tolerant genotypes of P. cupana. P. asparagi (22TSA) and P. pinodella (28S) develop microsclerotia on roots under in vitro conditions. The endophytic fungal community of P. cupana harbors species bearing some plant growth-promoting traits, including the synthesis of enzymes, phytohormones and biologically active molecules. We isolated and identified 3-hydroxypropionic acid and di-(2-ethylhexyl) phthalate from Diaporthe phaseolorum (8S) and 1-hydroxy-8-methoxyanthraquinone from Trichoderma asperellum (1BDA). This study opens possibilities for further investigations, because some species have never been explored for the biological control of crop pests like insects, weeds or parasites, and for the production of chemically active constituents. Hence, they represent potential sources of new and valuable organic molecules.
Supporting information S1