Design of Fungal Co-Cultivation Based on Comparative Metabolomics and Bioactivity for Discovery of Marine Fungal Agrochemicals

Microbial co-cultivation is employed for awakening silent biosynthetic gene clusters (BGCs) to enhance chemical diversity. However, the selection of appropriate partners for co-cultivation remains a challenge. Furthermore, competitive interactions involving the suppression of BGCs or upregulation of known, functional metabolite(s) during co-cultivation efforts is also common. Herein, we performed an alternative approach for targeted selection of the best co-cultivation pair. Eight marine sediment-derived fungi were classified as strong or weak, based on their anti-phytopathogenic potency. The fungi were co-cultured systematically and analyzed for their chemical profiles and anti-phytopathogenic activity. Based on enhanced bioactivity and a significantly different metabolite profile including the appearance of a co-culture specific cluster, the co-culture of Plenodomus influorescens (strong) and Pyrenochaeta nobilis (weak) was prioritized for chemical investigation. Large-scale co-cultivation resulted in isolation of five polyketide type compounds: two 12-membered macrolides, dendrodolide E (1) and its new analog dendrodolide N (2), as well as two rare azaphilones spiciferinone (3) and its new analog 8a-hydroxy-spiciferinone (4). A well-known bis-naphtho-γ-pyrone type mycotoxin, cephalochromin (5), whose production was specifically enhanced in the co-culture, was also isolated. Chemical structures of compounds 1–5 were elucidated by NMR, HRMS and [α]D20 analyses. Compound 5 showed the strongest anti-phytopathogenic activity against Xanthomonas campestris and Phytophthora infestans with IC50 values of 0.9 and 1.7 µg/mL, respectively.


Introduction
Plant diseases caused by bacterial and fungal pathogens pose severe losses to the global agroeconomy by directly reducing crop yield and quality [1]. The control of plant diseases is mainly achieved by spraying with conventional pesticides, although there is ample evidence of negative impacts on human health and the environment [2,3]. The development of resistance, as well as legal restrictions to conventional pesticides necessitate the search for natural, alternative plant protective agents [4]. Towards this aim, fungi and their secondary metabolites have a long-standing history. Their applicability for plant disease control involves direct introduction of the fungus into the soil as a biocontrol agent, as exemplified with Trichoderma sp. [5], and use of their secondary metabolites Table 1. Anti-phytopathogenic activity (%) of the fungal mono-and co-culture extracts at a test concentration of 100 µg/mL against Ps: P. syringae; Xc: X. campestris; Ea: E. amylovora; Rs: R. solanacearum; Mo: M. oryzae; Pi: P. infestans). The letters 'w' and 's' represent weak and strong activity, respectively. Strong activities are shown in bold letters. Reference compound for Xc, Ea, and Ps: 10 µM chloramphenicol; Rs: 10 µM tetracycline, Mo: 10 µM nystatin, Pi: 10 µM cycloheximide.  (Table S1). a Strains excluded from co-cultivation study; b confrontation zone.

Strain
The genus Penicillium is among the most fruitful and successfully studied fungal genera in terms of its ability to produce bioactive secondary metabolites [22]. With over 2200 documented isolated Mar. Drugs 2020, 18, 73 4 of 18 natural products reported in the Dictionary of Natural Products (DNP) [23] alone, new molecules are continually isolated from Penicillium [24]. Since the focus of our study was placed on less studied fungal genera, P. bialowiezense (strong) was excluded from further analyses. The yeast-like S. strictum (weak) was also excluded from further analysis due to having a very poor crude extract yield. The remaining four selected strains (two strong and two representative weak strains) were co-cultured on solid PDA medium at a small scale. This resulted in total six co-cultures, i.e. Helotiales sp. (strong)-P. influorescens (strong); P. nobilis (weak)-Lentithecium sp. (weak) and four combinations of one strong and one weak isolate: Helotiales sp. (strong)-P. nobilis (weak); Helotiales sp. (strong)-Lentithecium sp. (weak), P. influorescens (strong)-P. nobilis (weak) and P. influorescens (strong)-Lentithecium sp. (weak) (Figure 1, Table 1).

Selection of Co-Cultivation Pair for Large-Scale Co-Cultivation
Crude EtOAc extracts of the small-scale co-cultures ( Figure 1) were examined for their bioactivity against our in-house panel comprising six phytopathogens. Furthermore, tandem mass spectrometry (MS/MS) based comparative metabolomics including molecular networking (MN) was performed for all co-culture pairs against their respective single cultures ( Figures S2-S6).
Macroscopically, all co-cultures involving the strong candidate Helotiales sp. showed distinct growth areas and a pronounced mycelia-free zone between the competing strains, described as distance inhibition ( Figure 1A-C). In contrast, a mutual type of interaction was observed between two weak candidates (P. nobilis -Lentithecium sp.) where fungal isolates displayed mycelial contact ( Figure 1D). The co-cultures of the strong partner P. influorescens showed different interaction types. When P. influorescens (strong) was co-cultured with Lentithecium sp. (weak), contacts between both mycelia were observed at certain spots on the plate ( Figure 1E), whereas the co-cultivation of P. influorescens with P.

Selection of Co-Cultivation Pair for Large-Scale Co-Cultivation
Crude EtOAc extracts of the small-scale co-cultures ( Figure 1) were examined for their bioactivity against our in-house panel comprising six phytopathogens. Furthermore, tandem mass spectrometry (MS/MS) based comparative metabolomics including molecular networking (MN) was performed for all co-culture pairs against their respective single cultures ( Figures S2-S6).
Macroscopically, all co-cultures involving the strong candidate Helotiales sp. showed distinct growth areas and a pronounced mycelia-free zone between the competing strains, described as distance inhibition ( Figure 1A-C). In contrast, a mutual type of interaction was observed between two weak candidates (P. nobilis -Lentithecium sp.) where fungal isolates displayed mycelial contact ( Figure 1D). The co-cultures of the strong partner P. influorescens showed different interaction types. When P. influorescens (strong) was co-cultured with Lentithecium sp. (weak), contacts between both mycelia were observed at certain spots on the plate ( Figure 1E), whereas the co-cultivation of P. influorescens with Mar. Drugs 2020, 18, 73 5 of 18 P. nobilis (weak) displayed a distance inhibition with a pronounced, dark-coloured zone of inhibition ( Figure 1F).
A detailed MN-based comparative metabolomics study showed that co-cultivation increased the size of several molecular families by inducing derivatives of compounds produced in the axenic mono-cultures ( Figures S2-S6). However, no distinct co-culture specific major peaks were observed in the UPLC-MS chromatograms of most co-cultures ( Figures S2-S6). As shown in Figure 2, only the co-culture of P. influorescens (strong) with P. nobilis (weak) coincided with the induction of a distinct major peak (5) that was also visualized in the MN (Figure 3). A molecular family cluster (MF) containing co-culture specific nodes was detected and putatively annotated as bis-naphtho-γ-pyrone ( Figure 3). Other putatively annotated clusters in this MN included terpenoids, cyclic macrolactones, azaphilones, and isocoumarins ( Figure 3). In total, 21 MFs (clusters of more than two nodes) were identified in the co-culture of P. influorescens (strong) with P. nobilis (weak).
Mar. Drugs 2020, 18, 73 5 of 19 nobilis (weak) displayed a distance inhibition with a pronounced, dark-coloured zone of inhibition ( Figure 1F). A detailed MN-based comparative metabolomics study showed that co-cultivation increased the size of several molecular families by inducing derivatives of compounds produced in the axenic monocultures ( Figures S2-S6). However, no distinct co-culture specific major peaks were observed in the UPLC-MS chromatograms of most co-cultures ( Figures S2-S6). As shown in Figure 2, only the coculture of P. influorescens (strong) with P. nobilis (weak) coincided with the induction of a distinct major peak (5) that was also visualized in the MN (Figure 3). A molecular family cluster (MF) containing coculture specific nodes was detected and putatively annotated as bis-naphtho-γ-pyrone ( Figure 3). Other putatively annotated clusters in this MN included terpenoids, cyclic macrolactones, azaphilones, and isocoumarins ( Figure 3). In total, 21 MFs (clusters of more than two nodes) were identified in the coculture of P. influorescens (strong) with P. nobilis (weak). Analysis of the UPLC-MS chromatograms of the P. influorescens and P. nobilis co-culture extracts showed unexpected pattern of distinctly suppressed peaks in the whole co-culture compared to monocultures. In the P. influorescens mono-culture ( Figure 2B, i), the majority of the peaks appear at tR 6.5-8.5 min, indicating the presence of mid-to non-polar compounds, whereas the P. nobilis mono-culture ( Figure 2B, ii) contained most of the peaks at tR 3-5 min, suggestive of rather polar compounds. In the whole co-culture (Figure 2A,B, iii) a newly induced peak at tR 7.37 min (compound 5) was observed, while almost all other peaks, except for the peak at tR 3.38 min, of the respective mono-cultures were significantly suppressed. The organic extract of the confrontation zone alone (Figure 2A,B, iv) contained three major peaks at tR 7.37 (5), 3.38 (1 and 2) and 3.70 (4). The latter (4) was present in the whole coculture extract ( Figure 2B, iii) in very low levels. The whole co-culture also contained compound 3, which is the constituent of P. nobilis, in minute amounts ( Figure 2B, iii and iv). Analysis of the UPLC-MS chromatograms of the P. influorescens and P. nobilis co-culture extracts showed unexpected pattern of distinctly suppressed peaks in the whole co-culture compared to mono-cultures. In the P. influorescens mono-culture ( Figure 2B, i), the majority of the peaks appear at t R 6.5-8.5 min, indicating the presence of mid-to non-polar compounds, whereas the P. nobilis mono-culture ( Figure 2B, ii) contained most of the peaks at t R 3-5 min, suggestive of rather polar compounds. In the whole co-culture (Figure 2A,B, iii) a newly induced peak at t R 7.37 min (compound 5) was observed, while almost all other peaks, except for the peak at t R 3.38 min, of the respective mono-cultures were significantly suppressed. The organic extract of the confrontation zone alone (Figure 2A,B, iv) contained three major peaks at t R 7.37 (5), 3.38 (1 and 2) and 3.70 (4). The latter (4) was present in the whole co-culture extract ( Figure 2B, iii) in very low levels. The whole co-culture also contained compound 3, which is the constituent of P. nobilis, in minute amounts ( Figure 2B, iii and iv). These observations were highly suggestive for a competitive interaction involving suppression of biosynthetic pathways of one strain, and upregulation of highly functional metabolite(s) mediating the competitive interaction by the other. The whole co-culture extract of P. influorescens-P. nobilis showed an enhanced activity (59%) against the phytopathogenic bacterium X. campestris (Table 1), which was further enhanced to 90% when the extract of the confrontation zone (iv, Figure 2B) was tested (Table 1). Hence, the whole co-culture P. influorescens-P. nobilis was prioritized for chemical work-up for targeted compound isolation.

MN-Guided Isolation and Structure Elucidation
As displayed in the UPLC chromatogram ( Figure 2), two distinct major peaks (compounds 1/2, tR 3.38 and 5, tR 7.37) were observed in the whole co-culture. Hence, these peaks were prioritized for isolation and chemical characterization. Large-scale cultivation of the co-culture of P. influorescens and P. nobilis on 200 PDA plates was carried out. The crude EtOAc extract was sequentially partitioned by a modified Kupchan extraction approach [25] to afford n-hexane, dichloromethane (DCM) and aqueous MeOH subextracts. The DCM subextract that exhibited the highest activity against the phytopathogen X. campestris (Table S2)   These observations were highly suggestive for a competitive interaction involving suppression of biosynthetic pathways of one strain, and upregulation of highly functional metabolite(s) mediating the competitive interaction by the other. The whole co-culture extract of P. influorescens-P. nobilis showed an enhanced activity (59%) against the phytopathogenic bacterium X. campestris (Table 1), which was further enhanced to 90% when the extract of the confrontation zone (iv, Figure 2B) was tested (Table 1). Hence, the whole co-culture P. influorescens-P. nobilis was prioritized for chemical work-up for targeted compound isolation.

MN-Guided Isolation and Structure Elucidation
As displayed in the UPLC chromatogram ( Figure 2), two distinct major peaks (compounds 1/2, t R 3.38 and 5, t R 7.37) were observed in the whole co-culture. Hence, these peaks were prioritized for isolation and chemical characterization. Large-scale cultivation of the co-culture of P. influorescens and P. nobilis on 200 PDA plates was carried out. The crude EtOAc extract was sequentially partitioned by a modified Kupchan extraction approach [25] to afford n-hexane, dichloromethane (DCM) and aqueous MeOH subextracts. The DCM subextract that exhibited the highest activity against the phytopathogen X. campestris (Table S2)  azaphilone cluster, where two nodes representing compounds 3 and 4 were the main constituents. Therefore, we decided to isolate these compounds as well. and commercial databases. Predicted molecular formulae for all other nodes in this cluster returned no hits with a macrolactone scaffold, hence are putatively new. The node (b) m/z 249.1106 [M + Na] + represented the new compound 2. Node (c) m/z 253.0831 [M + Na] + had a mass difference of 2 amu from node (a), which is indicative of an unsaturation. We tentatively proposed a structure for (c) ( Figure 4A, c) based on the high spectral similarity score (>0.7) to (b) as implemented in the GNPS-MN platform (see MN in experimental section). Nodes (c) and (d) (m/z 367.1570 [M + Na] + ) were not purified because they were present only in very minute quantities. The other two nodes 254.094 and 252.079 are isotopic nodes for (c) and (a) respectively. In addition, we purified dendrodolide E (1) [26], which was not detected in the MN due to the low intensity of its fragment ions. H] + with a predicted molecular formula of C28H25O10, returned no hit and may represent a putatively new naphtho-γ-pyrone derivative. However, it could not be purified due to its insufficient quantity. The other two nodes with m/z 518.087 and m/z 520.101 are isotopic nodes for (b) and (c), respectively.
As mentioned above, another MF observed in the MN belonged to azaphilone class. A detailed MN analysis ( Figure 3) revealed a close relationship of two nodes within this cluster (compounds 3 and 4 with their isotopic nodes) that were also produced in the co-culture. Hence, we purified compounds As mentioned above, another MF observed in the MN belonged to azaphilone class. A detailed MN analysis (Figure 3) revealed a close relationship of two nodes within this cluster (compounds 3 and 4 with their isotopic nodes) that were also produced in the co-culture. Hence, we purified compounds 3 and 4, since the UPLC chromatogram ( Figure 2B) showed the specific presence of compound 3 in the whole co-culture (minor compound) and compound 4 in the confrontation zone (one of the major compounds). They were identified as the known polyketide spiciferinone (3) and its new derivative 8a-hydroxy-spiciferinone (4).
Compound 1 was isolated as a colorless film with the molecular formula C 12 H 16 O 4 deduced by HRESIMS. Its 1 H and 13 C NMR data ( Figures S7-S10) were identical to those of dendrodolide E, a 12-membered macrolide previously reported from the fungus Dendrodochium sp. associated with the sea cucumber Holothuria nobilis [26]. Compound 1 exhibited the same sign of specific rotation value 3 and 4, since the UPLC chromatogram ( Figure 2B) showed the specific presence of compound 3 in the whole co-culture (minor compound) and compound 4 in the confrontation zone (one of the major compounds). They were identified as the known polyketide spiciferinone (3) and its new derivative 8ahydroxy-spiciferinone (4). Compound 1 was isolated as a colorless film with the molecular formula C12H16O4 deduced by HRESIMS. Its 1 H and 13 C NMR data ( Figures S7-S10) were identical to those of dendrodolide E, a 12membered macrolide previously reported from the fungus Dendrodochium sp. associated with the sea cucumber Holothuria nobilis [26].   Table 2, Figures S11-S19). The 2D COSY spectrum of 2 revealed the presence of two clear spin systems ( Figure  6A). One spin system consisted of the secondary methyl group H3-12 (δH 1.22), the oxymethine proton H-11 (δH 4.97), and three methylene protons: H2-10 (δH 1.54), H2-9 (δH 1.94 and δH 1.64), and H2-8 (δH 2.3 and δH 2.5). The other spin network started from H2-2 (δH 2.72 and δH 2.57) that coupled with the oxymethine proton H-3 (δH 4.53). The latter proton had a scalar coupling with the olefinic proton H-4 (δH 5.83) which in turn coupled with the second olefinic methine H-5, (δH 5.73). Finally, H-5 was coupled with the methylene protons of H2-6 (δH 3.22 and δH 3.05) to finalize the spin network. The two spin systems were readily connected to each other with the aid of 1 H-13 C long-range correlations observed in the HMBC spectrum ( Figure 6A). The HMBC correlations between H2-8/C-7, H2-6/C-7, and H2-6/C-8 completed the upper end of the molecule (from C-8 to C-6). Diagnostic HMBC correlations from both H2-2 and H-11 to the ester carbonyl C-1 (δC 171.9) linked C-11 and C-2 thorough C-1, thereby establishing the full macrocyclic ring. The hydroxyl group was assigned to C-3 (δC 69.0), due to additional HMBC correlations observed between H-2/C-3, H-3/C-4, H-4/C-3 and H-5/C-3. The geometry of the double bond at C-4 was determined as E, due to the large coupling constant value (J4,5 = 15.4 Hz). Thus, 2 was assigned the planar structure as shown in Figure 6A, which is the C-8/C-9 saturated derivative of compound 1.  Figure 6A). The HMBC correlations between H 2 -8/C-7, H 2 -6/C-7, and H 2 -6/C-8 completed the upper end of the molecule (from C-8 to C-6). Diagnostic HMBC correlations from both H 2 -2 and H-11 to the ester carbonyl C-1 (δ C 171.9) linked C-11 and C-2 thorough C-1, thereby establishing the full macrocyclic ring. The hydroxyl group was assigned to C-3 (δ C 69.0), due to additional HMBC correlations observed between H-2/C-3, H-3/C-4, H-4/C-3 and H-5/C-3. The geometry of the double bond at C-4 was determined as E, due to the large coupling constant value (J 4,5 = 15.4 Hz). Thus, 2 was assigned the planar structure as shown in Figure 6A, which is the C-8/C-9 saturated derivative of compound 1. The relative configuration of the stereogenic centers C-3 and C-11 was assigned on the basis of an NOESY experiment. A NOESY correlation between H-3 and H-5 suggested the proximity of both protons, and suggesting an α-orientation for H-3 [26]. Further NOESY correlations observed between H-8β/6β and H-8β/H-11 suggested H-11 to be β oriented. These data indicated that 2 is a new analogue of the dendrodolide class of compounds [26]. Based on the NOE data and the measured specific rotation value ([α] 20 D +146, c 0.0085, CHCl3), we suggest the trivial name (+)-(3S,11R)-dendrodolide N for compound 2. Compound 3 was isolated as a yellowish amorphous powder with a molecular formula of C14H16O3 assigned by HRESIMS. Further analysis of its 1D NMR and 2D NMR data (Table 3, Figures S20-S25), led to identification of 3 as spiciferinone, an azaphilone type phytotoxin that was reported from the terrestrial fungus Cochliobolus spicifer [27].  The relative configuration of the stereogenic centers C-3 and C-11 was assigned on the basis of an NOESY experiment. A NOESY correlation between H-3 and H-5 suggested the proximity of both protons, and suggesting an α-orientation for H-3 [26]. Further NOESY correlations observed between H-8β/6β and H-8β/H-11 suggested H-11 to be β oriented. These data indicated that 2 is a new analogue of the dendrodolide class of compounds [26]. Based on the NOE data and the measured specific rotation value (+146, c 0.0085, CHCl 3 ), we suggest the trivial name (+)-(3S,11R)-dendrodolide N for compound 2.
Compound 3 was isolated as a yellowish amorphous powder with a molecular formula of C 14 H 16 O 3 assigned by HRESIMS. Further analysis of its 1D NMR and 2D NMR data (Table 3, Figures S20-S25), led to identification of 3 as spiciferinone, an azaphilone type phytotoxin that was reported from the terrestrial fungus Cochliobolus spicifer [27] Figures S26-34). From the 13 C NMR spectrum, it was evident that 4 contained two carbonyl functions (C-6 δ C 201.9 and C-8 δ C 207.4), two double bonds (C-3 δ C 161.8, C-4 δ C 104.3, C-4a δ C 152.3 and C-5 δ C 116.0), and two oxygenated carbons (C-1 δ C 70.8, C-8a δ C 67.2) ( Table 3). This data was indicative of a bicyclic azaphilone structure, as found in compound 3. Comparison of the 1D and 2D NMR data of 4 suggested a close resemblance to compound 3. First major difference between 3 and 4 was the absence of the double bond at ∆ 1(8a) , hence the olefinic proton H-1 was replaced with an oxymethylene group (δ H 4.48, d, J = 12.4 Hz and 3.84, d, J = 12.4 Hz) in 4 (Table 3). Secondly, the molecular formula of 4 was 18 atomic mass units higher than 3, suggesting the presence of a hydroxyl substituent in 4. This assumption was further supported by a broad absorption band observed in the FT-IR spectrum of 4 at ν max 3220 cm −1 . The hydroxyl group was assigned to C-8a, based on its 13 C chemical shift (δ C 67.2), plus HMBC correlations detected between H-1/C-8a and H-5/C-8a, finalizing the planar structure of 4 ( Figure 6B). The specific rotation of 4 ([α] 20 D +75, c 0.0006, MeOH) was similar to that of 3. Given the similarity in specific rotation, structure and same biological source, it is biosynthetically reasonable to assume that 4 has the same (R) configuration at C-7. However, we were unable to assign the configuration of C-8a. We suggest the trivial name of 8a-hydroxy-spiciferinone for compound 4.

Anti-Phytopathogenic Activity of Isolated Compounds
All isolated compounds were tested in vitro against a panel of six phytopathogens, except for compound 1, which was obtained in very low amounts. Compound 5 that was specifically overexpressed in the co-culture exhibited the highest activity against X. campestris and P. infestans with IC 50 values of 0.9 µg/mL and 1.7 µg/mL, respectively ( Table 4). The new compound dendrodolide N (2) showed moderate activity against P. infestans (IC 50 13.9 µg/mL, Table 4), but was inactive against all other pathogens at the highest test concentration (100 µg/mL). Compounds 3 and 4 showed no activity against any of the plant pathogens within the range of concentrations tested. Table 4. Anti-phytopathogenic activity (IC 50 in µg/mL) of 2-5 against susceptible phytopathogens. Reference compound: chloramphenicol for X. campestris and P. syringae; cycloheximide for P. infestans.

Discussion
Co-cultivation is a cheap and straightforward approach, requiring regulatory mechanisms involving genetic and molecular information transfer to produce secondary metabolites [30]. Production of these metabolites is complex, dynamic, and unforeseeable, although it is the driving force for microbial survival in complex multispecies habitats. In microbial drug discovery, co-cultivation of microorganisms from the same environment represents an ecology-inspired approach to activate cryptic gene clusters resulting in biosynthesis of novel molecules, new analogs of known compounds and/or overexpression of previously known molecules [31]. However, co-cultivation may also lead to competitive behavior, thereby causing suppression of gene clusters in competing strains [32]. A recent study revealed that microbes respond to cues produced by competitors by suppressing antibiotic synthesis in competitors in order to reduce threats to themselves [33]. Co-cultivation experiments often follow different approaches including painstaking bipartite screening of a collection of microbes of the same or different genera, niches or environment [34]. Researchers also tend to co-cultivate well-studied species with other microbes, introducing a level of bias [19]. However, these approaches are rather time-consuming and untargeted for selection of the best co-cultivation pairs.
In this study, intra-and inter-categorial (strong and weak) solid-state co-cultivation allowed a direct visualization of the interactions on the plates. With the exception of P. influorescens (strong)-Lentithecium sp. (weak) co-culture, which showed slight mycelial contact ( Figure 1E), all co-cultures of strong candidates (Helotiales sp. and P. influorescens) displayed patterns of distance inhibition. This pattern is associated with the release of antimicrobial molecules into the medium to inhibit competitors [35,36]. In contrast, the co-culture of the weak candidates (P. nobilis and Lentithecium sp.) displayed mutual interaction [37] as fungal mycelia grew freely into each other.
A comparative metabolomics approach used in this study revealed the production of minor new metabolites (observed as blue nodes in MNs, Figures S2-S6) in all co-cultures, irrespective of the visual interaction type. This is in line with findings of previous studies reporting a non-significant difference in the number of induced metabolites in the confrontation areas of 600 co-cultures exhibiting different interaction types [36]. However, due to their low quantity, the induced metabolites were not detected in significant amounts in the UPLC chromatograms of the co-cultures. The co-culture of P. influorescens (strong)-P. nobilis (weak) was selected for large-scale cultivation and chemical analysis because of: (a) the significantly different metabolite profile compared to the axenic cultures; (b) the observation of a co-culture specific cluster in the MN (identified as the bis-naphtho-γ-pyrone cluster); and (c) the enhanced activity of co-culture against the phytopathogen X. campestris ( Table 1). The chemical profile of P. influorescens (strong)-P. nobilis (weak) co-culture showed a marked reduction in the chemical diversity compared to their respective mono-cultures. The biosynthesis of numerous compound clusters observed in the single fungal cultures (dereplicated as terpenes and isocoumarins) were suppressed, while the bis-naphtho-γ-pyrone type mycotoxin cephalochromin (5) was upregulated in P. influorescens. An earlier report involving the co-cultivation of mycoparasite Stachybotrys elegans and the soil-borne fungal pathogen Rhizoctonia solani revealed that the mycotoxins produced by the mycoparasite altered the metabolism of the fungal pathogen. The biosynthesis of many antimicrobial compounds in R. solani was downregulated, while trichothecene type mycotoxins were upregulated in S. elegans [32]. In this study, chemical signals produced by P. nobilis appeared to have activated PKS genes (non-reducing polyketide synthase [38]) responsible for cephalochromin biosynthesis in P. influorescens. Cephalochromin has previously been isolated from several fungi including Cosmospora vilior, Cephalosporium sp., Pseudoanguillospora sp., and Verticillium sp. [39]. It is however being reported/detected in the genus Plenodomus for the first time. Consistent with our bioactivity data (IC 50 value 1.7 µg/mL), cephalochromin has been shown to exert anti-phytopathogenic activity against the oomycete P. infestans [28]. Its activity against the phytopathogenic bacterium X. campestris (IC 50 value 0.9 µg/mL) is being reported for the first time herein.
Dendrodolides E (1) and N (2) were observed in the monoculture of P. influorescens and also produced in the co-culture. They belong to the 12-membered macrolide family, commonly produced in fungi with antimicrobial and cytotoxic activities [40][41][42]. Their biosynthesis involves a highly reducing polyketide synthase from a hexaketide starter unit with a cyclization of their linear chain elongation [26,43]. The dendrodolides have been reported as a subfamily of the 12-membered ring macrolides from the fungus Dendrodochium sp. isolated from a sea cucumber Holothuria nobilis Selanka [26]. The current study adds another new member to this subfamily as dendrodolide N, with moderate inhibitory activity (IC 50 value 13.9 µg/mL) against the phytopathogenic oomycete P. infestans. It is the first report of such macrolides from the fungal genus Plenodomus.
Spiciferinone (3) and its new hydroxylated derivative (4) belong to the rare class of azaphilone type polyketides with a meta-quinone ring. They were produced in the monoculture of P. nobilis and detected in comparatively smaller amounts in the co-culture. Spiciferinone (3) was previously characterized as a phytotoxin produced by a terrestrial fungus Cochliobolus spicifer [27]. It is the first observation of such azaphilones in the fungal genus Pyrenochaeta. In our bioactivity tests against bacterial and fungal phytopathogens, spiciferinone (3), as well as its new derivative (4), remained inactive. Azaphilones display a wide range of activities including antiviral, anti-inflammatory, anticancer and antioxidant among others [44], thereby warranting further bioactivity investigations.
In summary, we employed a new, alternative co-cultivation approach, which exploits the anti-phytopathogenic potency of axenic cultures as proxies to categorize and subsequently select co-culture partners by comparative MN and bioactivity assessments. Our results describe an example of biosynthetic gene suppression in one strain and significant induction of a known mycotoxin in another strain simultaneously. Our future efforts will include identifying the chemical signals responsible for overexpression of polyketides, such as cephalochromin and characterizing the individual metabolites of the mono-cultures. Genome analysis will also be included to understand the influence of co-cultivation on the expression of functional metabolites in these two fungi.

General Procedures
Specific rotations of compounds 1-5 were measured on a monochromatic light source in MeOH or CHCl 3 at 20 • C on a Jasco P-2000 polarimeter (Jasco, Pfungstadt, Germany). IR spectra were recorded on a PerkinElmer Spectrum Two FT-IR spectrometer (PerkinElmer, Boston, MA, USA). HRMS was recorded on micrOTOF II-High-performance TOF-MS system (Bruker ® , Billerica, MA, USA) equipped with an electrospray ionization source. The NMR spectra were acquired on a BRUKER AV 600 spectrometer equipped with a Z-gradient triple resonance cryo-probehead (Bruker ® , Billerica, MA, USA). All spectra were run in solvents as specified in the text with referencing to residual 1 H and 13 C signals in the deuterated solvent. HPLC separations were performed on an Acquity UPLC HSS T3 column (High Strength Silica C18, 1.8 mm, 100 2.1 mm, Waters, Milford, MA, USA). HRMS/MS data were recorded on an Acquity UPLC I-Class System coupled to a Xevo G2-XS QToF Mass Spectrometer (Waters ® , Milford, MA, USA) in the positive mode at a mass range of m/z 50-1600 Da. Semi-preparative HPLC separations were performed using a VWR Hitachi Chromaster system (VWR International, Allison Park, PA, USA) consisting of a 5310 column oven, a 5260 autosampler, a 5110 pump, and a 5430 diode array detector connected in parallel with a VWR Evaporative Light Scattering Detector (ELSD 90, VWR International, Allison Park, PA, USA). Separation was achieved using a C18 semi-preparative column (Onyx, 10 mm × 100 mm, Phenomenex, Torrance, CA, USA) or on a Synergi 4u Polar-RP 80A column (250 × 4.6 mm).

Fungal Materials
All eight fungi were isolated from sediment samples obtained from the Windebyer Noor, Schleswig-Holstein, Germany in November 2015 [21]. Identification of the fungal strains by Sanger sequencing of the internal transcribed spacer (ITS) regions or 18S rRNA gene was reported previously [21]. Isolates used in this study had GenBank accession numbers MH791233 (ITS sequence), MH791253 (ITS sequence), MH791254 (ITS sequence), MH791292 (ITS sequence), MH791174 (18S rRNA gene sequence), MH791258 (ITS sequence), MH791244 (ITS sequence), and MH791275 (ITS sequence). The fungal strains were identified at species, genus or order level based on their closest relative according to BLAST [45]. To allow an assignment to the highest possible taxonomic rank, two new BLAST searches (access of NCBI nucleotide database: 31.07.19, Table S1) were conducted for all eight strains, one searching for highly similar sequences and the second searching for sequences from type material only. The combined information from both searches allowed four cases of assignment to the species level, to the genus level in one case, and to order level only in three cases. Fungi were cryopreserved in liquid nitrogen using Microbank TM (PRO-LAB Diagnostics, Richmond Hill, Canada). Each fungal culture was maintained on potato dextrose agar (PDA-potato starch 4 g, glucose monohydrate 4 g, agar 15 g, and dH2O ad 1 L) at 22 • C. phase B: 99.9% acetonitrile (MeCN, ULC/MS grade, Biosolve BV, Dieuze, France) / 0.1% formic acid in a linear gradient: 99% A for 11.5 min, 0% A for 1 min and 99% A until minute 15.
MS and MS/MS fragmentation spectra were acquired in a data dependent analysis (DDA) mode with an electrospray ionization source in the positive mode using the following parameters: A mass range of m/z 50-1600 Da, capillary voltage of 0.8 KV, cone and desolvation gas flow of 50 and 1200 L/h, respectively, source temperature at 150 • C and desolvation temperature at 550 • C with sampling cone and source offset at 40 and 80, respectively. Collision energy (CE) was ramped: Low CE from 6-60 eV to high CE of 9-80 eV. As controls, solvent (MeOH) and non-inoculated medium were injected under the same conditions.

Molecular Networking
Data obtained from the UPLC-MS/MS system were converted a to mzXML format using Proteo Wizard msconvert (version 3.0.10051, Vanderbilt University, Nashville, TN, USA) [46], uploaded to Global Natural Products Social Molecular Networking (GNPS) platform, and analyzed using the molecular networking workflow (http://gnps.ucsd.edu). The data were filtered by removing all MS/MS peaks within +/− 17 Da of the precursor m/z. MS/MS spectra were window filtered by choosing only the top six peaks in the +/− 50 Da window throughout the spectrum. The data were clustered with MS-Cluster with a parent mass tolerance and MS/MS fragment ion tolerance 0.02 Da. Further, consensus spectra that contained less than two spectra were discarded. A network was then created where edges were filtered to have a cosine score above 0.7 and more than six matched peaks. Further edges between two nodes were kept in the network only if each of the nodes appeared in each other's respective top 10 most similar nodes. The spectra in the network were then searched against GNPS spectral libraries. The library spectra were filtered in the same manner as the input data. All matches kept between network spectra and library spectra were required to have a score above 0.7 and at least 6 matched peaks [47]. The MS/MS spectra were then searched against an in-silico MS/MS database (ISDB) using the fragmentation tool CFM-ID [48]. The molecular networking data were visualized in Cytoscape 3.5.1. program using 'directed 'style [49]. All compounds (nodes) originating from PDA media and solvent control (MeOH) were deleted from the original network enabling visualization of metabolites coming from mono-and co-cultures.

Isolation of Compounds 1-5
Pre-cultures of P. influorescens and P. nobilis were grown for 14 days at 22 • C on a PDA medium. Mycelial plugs (1 cm 2 ) of both pre-cultures were transferred onto a PDA plate at 5 cm apart. For large scale cultivation, 200 PDA plates were inoculated and incubated at 22 • C for 21 days. The same extraction protocol described in Section 4.3 was used to yield a pooled EtOAc extract (1.2 g). The crude