Bioactive secondary metabolites produced by the emerging pathogen Diplodia olivarum

Copyright: © 2021 R. Di Lecce, M. Masi, B. T. Linaldeddu, G. Pescitelli, L. Maddau, A. Evidente. This is an open access, peer-reviewed article published by Firenze University Press (http:// www.fupress.com/pm) and distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Recently, Diplodia olivarum has emerged as an aggressive pathogen on different plant hosts in Italy. This fungus was first found on rotting olive drupes in southern Italy, and was described as a new species in 2008 (Lazzizera et al., 2008). It was later reported as a cause of canker on carob tree (Granata et al., 2011), lentisk  and wild olive (Manca et al., 2020). Symptoms caused by the pathogen in infected hosts include sunken cankers with characteristic wedgeshaped wood necroses on branches and stems. Foliar symptoms have also been observed especially on lentisk shoots ( Figure 1).
Given the expansion of severe dieback caused by D. olivarum in several natural ecosystems in Italy, and the limited information available about bioactive secondary metabolites produced by this emerging pathogen, the study described here was conducted to isolate, identify and evaluate phytotoxic, antifungal, antioomycetes and zootoxic activities of the main compounds produced by D. olivarum

Chemical characterization procedures
Optical rotations were measured in MeOH on a P-1010 digital polarimeter (Jasco, Tokyo, Japan), unless otherwise noted. IR spectra were recorded as a glass film deposits using a 5700 FT-IR spectrometer (Jasco), and UV spectra were measured in MeCN on a V-530 spectrophotometer (Easton). 1 H and 13 C NMR spectra were recorded, respectively, at 400 and 100 MHz in CDCl 3 , on a Bruker spectrometer (Billerica), using the same solvent as internal standard. The multiplicities were determined by DEPT spectrum (Berger and Braun, 2004). COSY, HSQC, HMBC and NOESY spectra were recorded using Bruker microprograms. HR ESIMS spectra were recorded on a 6120 Quadrupole LC/MS instrument (Agilent Technologies). Analytical (0.25 mm thickness) and preparative TLC (0.50 mm thickness) were performed on silica gel (Kieselgel 60, F 254 ,) and on reversed phase (Kieselgel 60 RP-18, F 254 , 0.20 mm tickness) plates (Merck). Resulting spots were visualized by exposure to UV radiation (253 nm), or by spraying first with 10% H 2 SO 4 in MeOH and then with 5% phosphomolybdic acid in EtOH, followed by heating at 110°C for 10 min. Column chromatography was performed using silica gel (Merck, Kieselgel 60, 0.063-0.200 mm).

Fungus strain
The D. olivarum strain used in this study was originally isolated from a cankered branch of lentisk collected in a natural area on Caprera Island (Italy). Representative genetic sequences from this strain were deposited in GenBank, with the accession numbers: ITS; KX833078), tef1-α; KX833079) and MAT1-2-1; MG015783 (Lopes et al., 2018). Pure cultures were maintained on potato dextrose agar (PDA) (Fluka, Sigma-Aldrich Chemic GmbH) and were stored at 4°C in the collection of the Dipartimento di Agraria, University of Sassari, Italy, as BL96.

Production, extraction and purification of secondary metabolites
Diplodia olivarum was grown on Czapek broth amended with 2% yeast extract or mineral salt medium (Pinkerton and Strobel, 1976), both at pH 5.7 in 1 L capacity Erlenmeyer flasks each containing 250 mL of medium. Each flask was seeded with 5 mL of a mycelium suspension and then incubated for 30 d at 25°C. Culture filtrates were obtained by filtering the cultures through filter paper in a vacuum system.
First, the conformers of (7S,15S)-1 and (7S,15R)-1 were investigated with the Monte Carlo algorithm using Merck molecular force field (MMFF). They were then screened by geometry optimizations at HF/3-21G level, single-point calculations at B3LYP/6-31G(d) level, and final geometry optimizations at the same level. Energies and populations were then estimated at the B97M-V/6-311+G(2df,2p) level. The procedure gave six energy minima for (7S,15S)-1 and ten minima for (7S,15R)-1 within the final energy threshold (10 kJ mol -1 at the B97M-V/6-31G(d) level). 13 C-NMR chemical shifts were then calculated with the GIAO method at the B3LYP/6-31G(d) level. An empirical correction was applied to each molecule depending on the number of bonds to the carbon and on the bond lengths (Hehre et al., 2019). 3 J coupling constants were determined as Boltzmann averages of all the DFT structures described above, either with Karplus equations or at B3LYP/pcJ-0 levels (Fermi contact term only).

Leaf puncture assays
Leaves of Phaseolus vulgaris L, Juglans regia L. and Quercus suber L. were used for this assay, and each compound was tested at 1.0 mg mL -1 . The assays were per-formed as previously reported , and each treatment was repeated three times. Leaves were observed daily and scored for symptoms after 5 d. The effects of the toxins on the leaves were observed up to 10 d . Lesions were estimated using APS Assess 2.0 software following the tutorials in the user's manual. Lesion size was expressed in mm 2 .

Antifungal assays
All compounds (1-6) were preliminarily tested on four different plant pathogens including the two fungi (Athelia rolfsii and D. corticola) and the two oomycetes (Phytophthora cambivora and P. lacustris). The sensitivity of all four species to these compounds was evaluated, depending on the species, on carrot agar (CA) or potato dextrose agar (PDA), as inhibition of the mycelium radial growth. The assays were performed as previously reported . Each metabolite was tested at 200 μg per plug. Methanol was used as negative controls. Metalaxyl-M (mefenoxam; p.a. 43.88%; Syngenta), a synthetic fungicide to which the oomycetes are sensitive, and PCNB (pentachloronitrobenzene) for ascomycetes and basidiomycetes, were used as positive controls. Each treatment consisted of three replicates, and the experiment was repeated two times

Artemia salina bioassays
All compounds were assayed on brine shrimp larvae (Artemia salina L.). The assay was performed in cell culture plates with 24 cells (Corning) as previously described . The metabolites were tested at 100 mg mL -1 . Tests were performed in quadruplicate. The proportions (%) of larval mortality was determined after 36 h incubation at 27°C in the dark.

RESULTS AND DISCUSSION
The organic extract obtained from filtrates of D. olivarum culture grown on Czapek medium was purified to yield a new nor-diterpenoid cleistanthane (1; Figure 2), named here as olicleistanone, together with two known pimarane diterpenoids identified as sphaeropsidins A and C (respectively 2 and 3; Figure 2), and two known norpimarane diterpenoids identified as sphaeropsidin G and diplopimarane (respectively, 4 and 5; Figure 2). When the fungus was grown on mineral salt medium it produced (-)-mellein (6; Figure 2), sphaeropsidin A (2) and low amounts of sphaeropsidin G (4) and diplopimarane (5).
Olicleistanone (1) has a molecular formula of C 22 H 28 O 4 , as deduced from its HR ESIMS spectrum and consistent with nine hydrogen deficiencies. Preliminary investigation of its 1 H and 13 C NMR spectra (Table 1) showed that the compound is closely related to a tricyclic nor-diterpenoid, with aromatized and cyclohexadiene rings (C and B) joined to a dihydropyran ring (D) generated probably from a cleistanthane carbon skeleton (Devappa et al., 2011). The signal at δ 195.5 in the 13 C NMR spectrum also showed the presence of a conjugated ketone group (Breitmaier and Voelter, 1987). These results are in full agreement with the bands typical for carbonyl and aromatic groups observed in the IR spectrum (Najkanishi and Solomns 1977) and the absorption maxima observed in the UV spectrum (Pretsch et al., 2000).
The 1 H and COSY spectra (Berger and Braun, 2004) of olicleistanone (1) showed the presence of the typical signals of two ortho-coupled protons (H-11 and H-12) of a 1,2,3,4-tetrasubstituted C benzene ring, and the singlets of a methoxy group (CH 3 -22), a vinyl methyl (CH 3 -17) and two methyls (CH 3 -19 and CH 3 -18) bonded to a quaternary carbon. The two methyls represent the head of the geranylgeranyl biosynthetic precursor which generated the diterpenoid cleistanthane carbon skeleton. The same spectra showed the signal of an ethoxy group. A signal pattern due to pyran moiety (ring D) of the benzohydropyran system (C and D rings) appeared as an ABC system. The spectra also showed a signal typical of the three adjacent methylene groups (CH 2 -1, CH 2 -2 and CH 2 -3) of the A ring (Pretsch et al., 2000).
The correlations observed in the HSQC spectrum (Berger and Braun, 2004) allowed the chemical shifts to be assigned to the protonated carbons, as reported in Table 1 (Breitmaier and Voelter, 1987).
The long range couplings observed in the HMBC spectrum (Berger and Braun, 2004) (Table 1) allowed the quaternary carbons to be assigned. The signals at δ 34.0 correlated with H 2 -2, H 2 -3, H 3 -18 and H 3 -19 and were assigned to C-4, at 136.4 with H 2 -1, H 2 -3, H 3 -18 and H 3 -19 and assigned to C-5, at 195.5 with H 2 -1 and assigned to C-6, at 92.3 with H-15, H 2 -16 and H-20A and assigned to C-7, at 130.6 with H-12 and H 3 -17 and assigned to C-8, at 146.1 with H 2 -1, H 2 -2, H-11 and assigned to C-10, at 138.7 with H-11 and H 3 -17 and assigned to C-13, and at 130.2 with H-12, H-15 and H 2 -16 and assigned to C-14. The remaining signal at δ 132.9 was assigned to C-9 (Breitmaier and Voelter, 1987). The correlation between C-7 and H-20A allowed the ethoxy group to be located at C-7 and consequently the methoxy group at C-15. Ethoxy groups are relatively rare in natural products, but not unprecedented, including several ethoxy-containing ketals like 1 (Wang et al., 2006;Lim et al., 2013;Xiong et al., 2015;Shen et al., 2015;Zhang et al., 2016). We avoided the use of ethanol during the extraction or purification process, which could lead to 1 as an artifact (Maltese et al., 2009;Capon, 2020).
Thus, the chemical shifts were assigned to all the carbons and the corresponding protons, which are reported in Table 1, and olicleistanone (1) was formulated as 4-ethoxy-6a-methoxy-3, 8,8-trimethyl-4,5,8,9,10,11hexahydrodibenzo[de,g]chromen-7(6aH)-one. The structure assigned to 1 was supported by the other HMBC couplings reported in Table 1  Attempts to assign the relative configuration of 1 were made recording a NOESY spectrum. The measured NOESY correlations are reported in Table 2, but since there is no clear correlation between the protons of the methoxy and ethoxy groups, these data alone were not sufficient to assign the relative configuration of the two chiral centres (C-7 and C-15). To better interpret NMR data, a molecular modelling study was undertaken. First, two diastereomeric structures (7S,15S)-1 and (7S,15R)-1 were generated and their possible conformations were explored by means of a conformational search with molecular mechanics (Merck molecular force field, MMFF). Geometry optimizations were then run with the density functional method (DFT) at the B97M-V/6-311+G(2df,2p)//B3LYP/6-31G(d) level, using the computational protocol for the prediction of 13 C chemical shifts of flexible compounds, developed by Hehre et al. (2019). For the two diastereomers, six or ten conformers were found with detectable populations at room temperature. The various conformers differed in the conformation of the methoxy and ethoxy groups, but also in the puckering of ring A. A clear difference between the two diastereomers was the orientation of H-15, which was predominantly pseudo-equatorial in (7S,15S)-1 and pseudoaxial (7S,15R)-1. Thus, we presumed that the coupling constants between H-15 and H-16a/H-16b could be used to discriminate between the two isomers. Experimentally, H-15 appears as a doublet with splitting of 3.3 Hz, meaning that one J 15/16 was small (3.3 Hz) and the other was negligible. This agreed with a pseudo-equatorial orientation. 3 J 15/16 were then estimated with Karplus curve and spin-spin coupling calculations at B3LYP/pcJ-0 level. These results are shown in Table 3, and strongly support the assignment of 1 as (7S*,15S*)-1. 13 C-NMR calculations were then run at the B3LYP/6-31G(d) level. The estimated root-mean-square (rms) error between experimental and calculated 13 C chemical shifts was acceptable (2.4-2.5)  but similar for both isomers, confirming the (7S*,15S*)-1 assignment but without further supporting it. Nevertheless, we consider that the argument based on J-couplings is accurate enough to assign the relative configuration. For the absolute configuration, the ECD spectrum of a solution of 1 in acetonitrile (1 mM, 0.01 cm cell) was measured. The ECD spectrum was not distinguishable from the baseline over the whole range (185-400 nm, data not shown), despite the optimal absorption (0.3 to 0.8 for the absorption peaks). It therefore must be concluded that the isolated sample of 1 was a racemate. Racemic natural products are rare, and are thought to result from nonenzymatic reactions (Zask and Ellestad, 2018). The chirality centre at C-7 of 1 is a tertiary benzylic carbon in α position to carbonyl group and it is therefore easily subject to racemization. However, racemization of this centre does not occur in a post-synthetic step, otherwise two diastereomers would be obtained. On the other hand, the isolated (7S,15S)-1 isomer was more stable than its (7S,15R) diastereomer by about 2 kcal mol -1 at the present level of calculation, suggesting that if the chiral centre at C-15 was biosynthesized in a later step than C-7, its configuration would be dictated by that at C-7.
All metabolites (1 to 6) isolated in this study were screened for phytotoxic, antifungal, antioomycete and zootoxic activities.
Except for compound 2, phytotoxicity was not detected for any of the metabolites (at 1 mg mL -1 ) when applied to leaves of Phaseolus vulgaris, Quercus suber, or Juglans regia. Sphaeropsidin A (2) caused necrotic lesions on leaves of all the plant species tested, with mean lesion sizes of 75.6 mm 2 on P. vulgaris, 163.3 mm 2 on J. regia and 15.1 mm 2 on Q. suber.
In the assays of antifungal activity, sphaeropsidin A (2) inhibited mycelium growth of all the plant pathogens tested (100% inhibition rate). Diplopimarane (4) completely inhibited growth of Athelia rolfsii and partially inhibited growth of D. corticola, P. cambivora and P. lacustris, inhibition from 56% to 75%. No colony growth inhibition was observed for the other four metabolites at the concentration used.
Cleistanthane-type diterpenoids are produced by different fungi and plants, but few examples of cleistanthane nor-diterpenoids are reported. Among them there are aspergiloids A, B, F and G isolated from the fermentation broth extract of Aspergillus sp. YXf3, an endophytic fungus from Ginkgo biloba. However, no biologi-cal activities have been reported for these compounds (Guo et al., 2012;Yan et al., 2013).
Melleins are 3,4-dihydroisocoumarins, which are produced by many fungi of various genera as well as plants, insects and bacteria. These compounds have several phytotoxic, zootoxic and antifungal effects (Reveglia et al., 2020). (-)-Mellein was toxic on grapevine leaves and grapevine calli (Djoukeng et al., 2009;Ramirez-Suero et al., 2014, and was detected in symptomatic and asymptomatic grapevine wood samples and green shoots (Djoukeng et al., 2009) from plants with Botryosphaeria dieback and leaf stripe. The role of this compound in pathogenesis was investigated by examining the extent to which it caused expression of defenserelated genes in grapevine calli (Ramirez-Suero et al., 2014). Recently, (-)-mellein was also identified as a metabolite of Lasiodiplodia euphorbiaceicola during screening of phytotoxic metabolites isolated from Lasiodiplodia spp. infecting grapevine in Brazil (Cimmino et al., 2017b), and from Sardiniella urbana, a pathogen from declining European hackberry trees in Italy .
It is well known that fungal phytotoxins are closely related to host plant interactions. These compounds play key roles inducing virulence and pathogenicity of fungi. The host/pathogen interaction is the first process of the complex mechanism of infection. Fungal pathogens produce enzymes to degrade host wood cell walls, and the fungal toxins penetrate vessels and metabolize dead host tissues. The toxins then translocate to branches and leaves distant from the infection points, inducing chlorosis and necrosis (Durbin et al., 1989;Ballio, 1991;Möbius and Hertweck 2009;Keller, 2019). Thus, the phytotoxicity of D. olivarum was probably due to production of sphaeropsidin A (2), as outlined in the present study.
Phytotoxins have also been shown to possess herbicidal, antimicrobial and insecticidal activities (Spara-pano et al. 2004;Evidente et al. 2011;Cimmino et al. 2015;Barilli et al., 2017;Aznar et al. 2019). Drug-based phytotoxins can also be used in medicine against some important human diseases, such as cancer, malaria, dengue and yellow fevers, and against fungal and bacterial infections (Bajsa et al., 2007;Evidente et al., 2014Masi et al., 2018;Roscetto et al., 2020). Some of these toxins could also be produced in industrial large scale, and be formulated for applications in agriculture and medicine. Among the toxins previously isolated, and also from D. olivarum, sphaeropsidin A (2) is a phytotoxin with strong potential for drug development (Ingles et al., 2017;Masi et al., 2018;Roscetto et al., 2020).
In conclusion, this study was the first to investigate secondary metabolites produced by D. olivarum, an emerging pathogen of forest trees in the Mediterranean region. The results confirm that Botryosphaeriaceae are sources of bioactive secondary metabolites, some of which have potential for applications in biotechnology sectors.
Among the metabolites produced in vitro by D. olivarum, sphaeropsidin A and diplopimarane inhibited vegetative growth of four plant pathogens belonging to different phyla. Additionally, the strong activity of the newly identified metabolite, olicleistanone (1), against A. salina deserves detailed investigation, because several applications of A. salina in toxicology and ecotoxicology continue to be widely used.