Sesquiterpenes and Cyclodepsipeptides from Marine-Derived Fungus Trichoderma longibrachiatum and Their Antagonistic Activities against Soil-Borne Pathogens

Soil-borne pathogens, including phytopathogenic fungi and root-knot nematodes, could synergistically invade vegetable roots and result in serious economic losses. The genus of Trichoderma has been proven to be a promising reservoir of biocontrol agents in agriculture. In this study, the search for antagonistic metabolites from a marine-derived fungus, Trichoderma longibrachiatum, obtained two structural series of sesquiterpenes 1–6 and cyclodepsipeptides 7–9. Notably, the novel 1 was a rare norsesquiterpene characterized by an unprecedented tricyclic-6/5/5-[4.3.1.01,6]-decane skeleton. Their structures were elucidated by extensive spectroscopic analyses, while the absolute configuration of novel 1 was determined by the comparison of experimental and calculated ECD spectra. The novel 1 and known 2 and 3 showed significant antifungal activities against Colletotrichum lagrnarium with MIC values of 8, 16, and 16 μg/mL respectively, even better than those of the commonly used synthetic fungicide carbendazim with 32 μg/mL. They also exhibited antifungal potential against carbendazim-resistant Botrytis cinerea. Cyclodepsipeptides 7–9 showed moderate nematicidal activities against the southern root-knot nematode (Meloidogyne incognita). This study constitutes the first report on the antagonistic effects of metabolites from T. Longibrachiatum against soil-borne pathogens, also highlighting the integrated antagonistic potential of marine-derived T. Longibrachiatum as a biocontrol agent.


Introduction
Colletotrichum spp., Botrytis cinerea, and Fusarium oxysporum are three well-known and seriously damaging soil-borne phytopathogenic fungi worldwide, which cause anthracnose, gray mold, and wilt diseases of vegetables, respectively [1]. The southern root-knot nematode (Meloidogyne incognita) is also a typical soil-borne pathogen which invades vegetable roots [2,3]. More seriously, soil-borne fungi could synergistically interact with southern root-knot nematodes and therefore result in an even greater threat to vegetable cultivation [4,5]. Hence, the development of integrated control agents is always in demand.
Currently, the main strategy for controlling soil-borne pathogens is highly dependent on synthetic agrochemicals [6]. However, their ongoing overuse has resulted in a series of negative consequences, Mar. Drugs 2020, 18 such as residual toxicity, the developing resistance of targeted pathogens, and many other environmental issues [6,7]. Therefore, the search for environmentally friendly integrated control alternatives has attracted more and more attention. Notably, the genus of Trichoderma has been successfully applied to control soil-borne pathogens, such as Trichoderma harzianum and T. viride [8][9][10]. Recently, the species of T. longibrachiatum has also been suggested as a biocontrol agent against phytopathogens due to its potent antagonistic effect [11][12][13]. Research on its mechanism has mainly focused on its mycoparasitic and enzyme-producing abilities, with few reports on its antagonistic metabolites [12][13][14].
T. longibrachiatum is generally isolated from terrestrial soil and plants [11][12][13][14], being a rare marine resource [15]. Ji and co-workers isolated various Trichoderma species from marine alga which could produce structurally unique and biologically active metabolites [16][17][18][19], also indicating the research potential of marine-derived Trichoderma species. During our ongoing search for biocontrol agents in agriculture [20][21][22], a marine-derived fungus T. longibrachiatum attracted our attention due to its potent antagonistic ability to destroy soil-borne pathogens. Furthermore, the search for its antagonistic metabolites obtained two structural series of sesquiterpenes 1-6 and cyclodepsipeptides 7-9 ( Figure 1), especially including a new rare norsesquiterpene 1. The isolation, structural elucidation, and antagonistic evaluation of the isolated metabolites (1-9) are discussed herein.
Mar. Drugs 2019, 17, x FOR PEER REVIEW 2 of 9 even greater threat to vegetable cultivation [4,5]. Hence, the development of integrated control agents is always in demand. Currently, the main strategy for controlling soil-borne pathogens is highly dependent on synthetic agrochemicals [6]. However, their ongoing overuse has resulted in a series of negative consequences, such as residual toxicity, the developing resistance of targeted pathogens, and many other environmental issues [6,7]. Therefore, the search for environmentally friendly integrated control alternatives has attracted more and more attention. Notably, the genus of Trichoderma has been successfully applied to control soil-borne pathogens, such as Trichoderma harzianum and T. viride [8][9][10]. Recently, the species of T. longibrachiatum has also been suggested as a biocontrol agent against phytopathogens due to its potent antagonistic effect [11][12][13]. Research on its mechanism has mainly focused on its mycoparasitic and enzyme-producing abilities, with few reports on its antagonistic metabolites [12][13][14]. T. longibrachiatum is generally isolated from terrestrial soil and plants [11][12][13][14], being a rare marine resource [15]. Ji and co-workers isolated various Trichoderma species from marine alga which could produce structurally unique and biologically active metabolites [16][17][18][19], also indicating the research potential of marine-derived Trichoderma species.
During our ongoing search for biocontrol agents in agriculture [20][21][22], a marine-derived fungus T. longibrachiatum attracted our attention due to its potent antagonistic ability to destroy soil-borne pathogens. Furthermore, the search for its antagonistic metabolites obtained two structural series of sesquiterpenes 1-6 and cyclodepsipeptides 7-9 ( Figure 1), especially including a new rare norsesquiterpene 1. The isolation, structural elucidation, and antagonistic evaluation of the isolated metabolites (1-9) are discussed herein.

Structural Elucidation
The molecular formula of compound 1 was obtained as C 14 H 22 O 6 by HRESIMS ( Figure S1 in materials, SI), implying four degrees of unsaturation. The one-dimensional NMR and HSQC data (Table 1 and Figure S5) exhibited one carbonyl carbon (δ C 214.7), one oxygenated (δ C 73.6) and one aliphatic (δ C 43.4) quaternary carbons, four CH (δ C 66.1, 48.0, 37.5, and 37.1), four CH 2 (δ C 35.9, 33.0, 20.6, and 20.3) and three CH 3 groups (δ C 28.0, 25.6, and 20.6). Therefore, in addition to one degree of unsaturation from the carbonyl group, three remaining ones could indicate the presence of a tricyclic system in 1. Firstly, the obvious COSY correlations between H-3a and H 2 -4 could confirm a structural fragment of CH 2 (3)-CH 2 (4) (Figure 2 and Figure S6). Further, due to the partially overlapped 1 H NMR signals of H-6 and H-3b, another CH(1)-CH(6) or CH 2 (3) could be deduced by the COSY cross-peak from H-1 to H-6 or 3b ( Figure S6). In order to clearly distinguish H-6 and 3b in the NMR spectra, the novel 1 was further determined using CD 3 OD with a higher sensitivity of 600 MHz. As shown in Figure 2 and Figure S11, the fragment could be unambiguously confirmed as CH(1)-CH(6) or CH(1)-CH 2 (3). However, the NMR signals of H-8, H 2 -10 and 9 were still seriously overlapped using CDCl 3 (Figures S2 and S6) or CD 3 OD (Figures S8 and S11), which were therefore difficult to clearly distinguish for the structure elucidation. from H-1 to H-6 or 3b ( Figure S6). In order to clearly distinguish H-6 and 3b in the NMR spectra, the novel 1 was further determined using CD3OD with a higher sensitivity of 600 MHz. As shown in Figures 2 and S11, the fragment could be unambiguously confirmed as CH(1)-CH(6) or CH(1)-CH2(3). However, the NMR signals of H-8, H2-10 and 9 were still seriously overlapped using CDCl3 ( Figures S2 and S6) or CD3OD ( Figures S8 and S11), which were therefore difficult to clearly distinguish for the structure elucidation. Secondly, the presence of a cyclohexanone residue (fragment 1) could be deduced by two groups of key HMBC correlations (Figures 2 and S7), one including HMBC signals from H-1/6 and H2-3/4 to CO-5 (Figures 2 and S7A), while another containing HMBC correlations from H-1, H2-4, H3-11 to C-2, from H3-11 to CH-1, from H-1 to CH2-3, and from H2-4 to CH-6 ( Figures 2 and S7B). The connection between the methylene residue (δC 35.9, CH2-10) and the quaternary carbon (C-2) could be confirmed by the HMBC cross-peaks from H-1 and H3-11 to CH2-10. Fragment 2 could also be deduced by the HMBC correlations shown in Figures 2 and S7B.
Thirdly, as shown in Figures 2 and S12, the obvious HMBC correlations from H-7 to CH-1 and CO-5 could indicate that the CH-7 group was connected to CH-6 of the cyclohexanone residue. The significant HMBC cross-peaks from H-7 to C-2, as well as from H-7 to CH-8 and C-12, could join fragments 1 and 2 to obtain fragment 3, through a CH(7)-CH2(10)-C2 bridge and a CH7-CH8 bond, respectively (Figures 2 and S12B).
As shown in Table 1 and Figure S8B, the large 5.6 Hz of H-6 was the vicinal coupling between H-6 (a bond) and H-1 (e bond), while the small 2.1 Hz of H-6 should be a long-range coupling between H-6 and H-4b, which usually appeared in bridged-ring or unsaturated compounds. The H-1 could be observed as a doublet peak with a coupling constant of 5.6 Hz, which was identical to that of H-6 ( Figure S8B). This "d" peak of H-1 indicated that there was no vicinal coupling between Secondly, the presence of a cyclohexanone residue (fragment 1) could be deduced by two groups of key HMBC correlations (Figure 2 and Figure S7), one including HMBC signals from H-1/6 and H 2 -3/4 to CO-5 ( Figure 2 and Figure S7A), while another containing HMBC correlations from H-1, H 2 -4, H 3 -11 to C-2, from H 3 -11 to CH-1, from H-1 to CH 2 -3, and from H 2 -4 to CH-6 ( Figure 2 and Figure S7B). The connection between the methylene residue (δ C 35.9, CH 2 -10) and the quaternary carbon (C-2) could be confirmed by the HMBC cross-peaks from H-1 and H 3 -11 to CH 2 -10. Fragment 2 could also be deduced by the HMBC correlations shown in Figure 2 and Figure S7B.
Thirdly, as shown in Figure 2 and Figure S12, the obvious HMBC correlations from H-7 to CH-1 and CO-5 could indicate that the CH-7 group was connected to CH-6 of the cyclohexanone residue. The significant HMBC cross-peaks from H-7 to C-2, as well as from H-7 to CH-8 and C-12, could join fragments 1 and 2 to obtain fragment 3, through a CH(7)-CH 2 (10)-C2 bridge and a CH7-CH8 bond, respectively (Figure 2 and Figure S12B).
As shown in Table 1 and Figure S8B, the large 5.6 Hz of H-6 was the vicinal coupling between H-6 (a bond) and H-1 (e bond), while the small 2.1 Hz of H-6 should be a long-range coupling between H-6 and H-4b, which usually appeared in bridged-ring or unsaturated compounds. The H-1 could be observed as a doublet peak with a coupling constant of 5.6 Hz, which was identical to that of H-6 ( Figure S8B). This "d" peak of H-1 indicated that there was no vicinal coupling between H-1 and H 2 -9, probably due to their mutual nearly perpendicular angles. The lack of some COSY correlations of H-7 and H 2 -9 ( Figures S6 and S11) should also be related to their perpendicular positions to nearby hydrogens [25].
The 13 C NMR shift of CH-1 was assigned as unusually large, 66.1, which should be related to the anisotropic deshielding effects of nearby C-C bonds, while the shift of CH-6 (δ C 37.1) to a higher field might be connected with the anisotropic shielding effect of the C=O bond, which also resulted in the higher-field-shifted CH 2 -4 (δ C 20.1) compared with CH 2 -3 (δ C 33.0). Similar results were usually found in bridgehead carbons of polycyclic terpenoids, such as CH-5 (δ C 61.5) in penicibilaene A [25], as well as CH-6 (δ C 66.0) and CH-15 (δ C 70.1) in conidiogenone G [26].
The relative configuration of compound 1 was deduced through the NOESY experiment ( Figure 2 and Figure S13). The consecutive NOE correlations from H 3 -11 to H-1, from H-1 to H-6, from H-6 to H-4b, H axial -9b and H 3 -14, as well as from H 3 -14 to H-7, suggested the co-face orientation of these protons. The absolute configuration of 1 was determined by the comparison of its experimental and calculated ECD (Electronic Circular Dichroism) spectra. As shown in Figure 3, the calculated ECD data of (1S, 2S, 6R, 7R, and 8S)-1 showed positive cotton effects (CEs) near 220 and 295 nm, as well as negative CE around 210 nm, the same as the experimental ones, while the calculated ECD spectra of (1R, 2R, 6S, 7S, and 8R)-1 exhibited opposite corresponding CEs.

Antagonistic Evaluation
The isolated metabolites of sesquiterpenes 1-6 and cyclodepsipeptides 7-9 were evaluated for their antagonistic potential (Table 2), including antifungal activities against three groups of representative soil-borne phytopathogenic fungi-Colletotrichum lagrnarium, Colletotrichum fragariae, carbendazim-resistant strains of Botrytis cinerea from grape (PTQ1) and strawberry (CMQ1), Fusarium oxysporum f. sp. cucumerinum, and Fusarium oxysporum f. sp. Lycopersici-as well as nematicidal effects against the southern root-knot nematode (M. incognita).  The novel norsesquiterpene 1 showed significant antifungal activities against two Colletotrichum species and two carbendazim-resistant strains of B. cinerea with MIC values ranging from 8 to 64 µg/mL. These results are better than those from the commonly used carbendazim, a benzimidazole fungicide Mar. Drugs 2020, 18, 165 6 of 10 which binds to the β-tubulin proteins and then inhibits cell division [36]. Therefore, although there are almost no reports concerning antifungal mechanisms of norsesquiterpenes, the novel 1 could show multiple-target potential.
The known trichothecene sesquiterpenes 2 and 3 exhibited broad-spectrum antifungal activities against all tested soil-borne phytopathogenic fungi, while the trichothecene congener 4 only showed far weaker effects compared to those of 2 and 3, suggesting that the 4-OH substituted in the cyclohexane ring of 4 might negatively modulate its antifungal activity. Trichothecenes are known as mycotoxins, which show antifungal, phytotoxic and cytotoxic activities [37][38][39]. Antifungal SAR research of trichothecene congeners showed that the 12-epoxide was essential to its activity [37], while the substituted groups in C-4 and C-8 could also modulate the effect [38], which was identical to the SAR of isolated 2-4. Trichoderma trichothecenes have been reported to be able to induce the expression of plant defense-related genes [39].

Fungal Material
The fungal strain T. longibrachiatum was isolated from the root of Suaeda glauca, a highly halophile plant collected from the intertidal zone of Jiaozhou Bay, Qingdao, China in October 2015. The fungus was identified on the basis of morphological characteristics and molecular analyses of ITS (internal transcribed spacer)-5.8S Rdna region sequence [20]. The strain was preserved in the Natural Products Laboratory, College of Chemistry and Pharmacy, Qingdao Agricultural University.

Fermentation, Extraction and Isolation
Fresh mycelia of the fungus were statically fermented at 28 • C for 30 days on liquid Potato Dextrose Broth (PDB) media. The liquid culture was conducted in 50 × 1 L conical flasks containing 300 mL of PDB medium (2% glucose and 20% potato juice in natural seawater).

Antagonistic Evaluation
In order to evaluate the antagonistic potential of T. longibrachiatum metabolites against soil-borne pathogens, its antifungal activities were tested against three groups of representative soil-borne phytopathogens, including Colletotrichum spp. (C. fragariae and C. lagenarium), carbendazim-resistant strains of B. cinerea from grape (PTQ1) and strawberry (CMQ1), and Fusarium oxysporum, using a broth microdilution method in 96-well plates [21,40].
Another synergetic soil-borne pathogen, the southern root-knot nematode (M. incognita), was also selected for nematicidal bioassay in 24-well plates. Briefly, J2s of M. incognita were collected to prepare the nematode suspension based on the protocol reported previously [21]. The isolated metabolites (1-9) were dissolved and diluted in DMSO to obtain sample solvents with a series of different concentrations. The sample solvents (5 µL) were added to each well containing the nematode suspension (495 µL) with about 60 J2s, while the same amount of DMSO (5 µL) was added for the negative control. The plates were maintained at 25 • C for 48 h and then observed using a stereomicroscope to evaluate the nematode mortalities. Nematodes were defined to be dead if their bodies became straight and did not react to mechanical touches. The experiment was repeated three times under the same conditions.

Conclusions
The investigation of antagonistic metabolites from marine-derived fungus T. longibrachiatum obtained two structural series of sesquiterpenes 1-6 and cyclodepsipeptides 7-9. Notably, the novel 1 was a rare norsesquiterpene possessing an unprecedented tricyclic-6/5/5-[4.3.1.0 1,6 ]decane skeleton. Its absolute configuration was determined by the comparison of experimental and calculated ECD spectra. The novel 1 and known 2 and 3 showed significant antifungal activities against two Colletotrichum species and two carbendazim-resistant strains of B. cinerea with MIC values ranging from 8 to 64 µg/mL, even better than those of the commonly used fungicide carbendazim. Cyclodepsipeptides 7-9 showed moderate nematicidal activities against the southern root-knot nematode (M. incognita). The antifungal activities of 1-4 and nematicidal effects of 5-9 were reported for the first time and further revealed the synergistically antagonistic potential of marine-derived T. longibrachiatum against soil-borne pathogens.