Pigmentosins from Gibellula sp. as antibiofilm agents and a new glycosylated asperfuran from Cordyceps javanica

In the course of our exploration of the Thai invertebrate-pathogenic fungi for biologically active metabolites, pigmentosin A (1) and a new bis(naphtho-α-pyrone) derivative, pigmentosin B (2), were isolated from the spider-associated fungus Gibellula sp. Furthermore, a new glycosylated asperfuran 3, together with one new (6) and two known (4 and 5) cyclodepsipeptides, was isolated from Cordyceps javanica. The pigmentosins 1 and 2 showed to be active against biofilm formation of Staphylococcus aureus DSM1104. The lack of toxicity toward the studied microorganism and cell lines of pigmentosin B (2), as well as the antimicrobial effect of pigmentosin A (1), made them good candidates for further development for use in combination therapy of infections involving biofilm-forming S. aureus. The structure elucidation and determination of the absolute configuration were accomplished using a combination of spectroscopy, including 1D and 2D NMR, HRMS, Mosher ester analysis, and comparison of calculated/experimental ECD spectra. A chemotaxonomic investigation of the secondary metabolite profiles using analytical HPLC coupled with diode array detection and mass spectrometry (HPLC–DAD–MS) revealed that the production of pigmentosin B (2) was apparently specific for Gibellula sp., while the glycoasperfuran 3 was specific for C. javanica.


Introduction
Nosocomial infections are often associated with the presence of S. aureus, generally transmitted either by direct contact with carriers or by medical procedures [1]. S. aureus is commonly considered as a cause of tissue-associated and medical devicerelated, in particular orthopedic implant-related infections, since implants coated with proteins facilitate bacterial attachment and biofilm development [1].
In general, bacteria are known to employ different strategies to cope with the presence of antibiotics, of which a biofilm, an aggregate of microorganisms held together within a self-produced matrix of extracellular polymeric substances, plays an important role as a main virulence determinant in staph infections [1,2]. Within a biofilm, bacteria become tolerant toward antibiotics and host immune responses greater than their planktonic (free-floating) cells, leading to an occurrence of reinfection once the antibiotic therapy is terminated [3][4][5].
In recent years, efforts to find new molecules that can selectively inhibit biofilms have steadily increased, based on the hypothesis that new agents can effectively disrupt biofilm formation and leave target microbes vulnerable to antibiotics [6]. A complementary approach of using a combination of an antibiotic with a biofilm inhibitor appears to be a promising solution to control biofilm-associated pathogens, as based on the evidence that traditional antibiotics were more effective when used in combination with biofilm inhibitors [7]. Since finding an effective strategy to control biofilm formation remains a challenge, the effort to search for an effective antibiofilm agent was herein made.
Invertebrate-pathogenic fungi, in particular the spider-pathogenic fungi, have recently proved to be a promising source of bioactive compounds [8][9][10]. Thus, during the current study, which is part of a project aiming to discover novel biofilm inhibitors from Thai fungi [11], a number of invertebrate-pathogenic fungi collected from various parts of Thailand were studied for production of bioactive secondary metabolites. Herein, we report on the isolation, structure elucidation, and biological activities of six compounds from Gibellula sp. and Cordyceps javanica. Furthermore, the species-specific patterns of secondary metabolite production were studied.

Results and Discussion
Structure elucidation Gibellula sp. was cultivated in liquid yeast, malt, and glucose (YMG) medium and extracted as described in the Experimental section. The extracts were purified by HPLC to give pigmentosin A (1) and pigmentosin B (2). Using a similar procedure, compounds 3-6 were obtained from the liquid culture of C. javanica ( Figure 1).
Compound 1 was obtained as pale green powder. Its molecular formula was determined as C 30  data. The presence of only 15 resonances in the 13 C NMR spectrum suggested a highly symmetric structure. The structure of 1 was then identified to be identical with pigmentosin A, a 3,4dihydro-α-naphthopyrone dimer with a 7,7′-dimethoxy pattern, by comparing its spectroscopic data with the published data for pigmentosin A [12]. Nevertheless, the chirality of the stereogenic centers C-3/C-3′ as well as the atropisomerism at the 6,6′ axis of pigmentosin A (1) were not elucidated previously. Therefore, electronic circular dichroism (ECD) measurements combined with time-dependent density functional theory (TDDFT) calculations of ECD data of compound 1 in MeOH at the B3LYP/6-311+G* level of theory were carried out. The CD spectrum of 1 showed strong Cotton effects: a first negative at 274 nm (Δε −196) and a positive second one at 252 nm (Δε +203), indicating the helicity of the 6-6′ axis as aR, according to the exciton chirality method [13]. Furthermore, the TDDFT-ECD calculations were performed on three isomers, namely (3R,3′R,6R)-1, (3S,3′S,6S)-1, and (3S,3′S,6R)-1. The calculated ECD spectrum of (3R,3′R,6R)-1 reproduced all tran-sitions of the experimental ECD spectrum. In contrast, the (3S,3′S,6S)-1 compound had a mirror image correlation with the experimental Cotton effects, which indicated that the main Cotton effects around 250 and 270 nm were due to atropisomerism (Figure 2), although the TDDFT-ECD curve of (3S,3′S,6R)-1 showed a further small positive Cotton effect around 228 nm. Nevertheless, due to the high similarity of both curves, we believe that the calculated ECD data could not distinguish between (3R,3′R,6R)-1 and (3S,3′S,6R)-1. Thus, the atropisomerism at the 6-6′ axis of pigmentosin A (1) was determined to be aR, while the absolute configuration of the stereogenic centers C-3/C-3′ remains unsolved.
The molecular formula of compound 3 was assigned as C 20  Hz, H-4′), suggesting the presence of a trisubstituted dihydrobenzofuran moiety in 3. COSY and HMBC correlations allowed the construction of a 1,3-pentadiene moiety (C-1′-C-5′), and the linkage to C-2 of the benzofuran ring was determined by a COSY correlation between H-1′ and H-2, as well as HMBC correlations from H-2′ to C-2 and H-1′ to C-3. Finally, the remaining signals were similar to the reported data for 4-O-methyl-β-ᴅ-glucopyranose in our previous reports [8,9]. The HMBC correlation from the anomeric proton H-1′′ to C-6 constructed the glycosidic bond. Thus, compound 3 was determined as a glycosylated derivative of the antifungal asperfuran [15], named glycoasperfuran. The absolute configuration of the sugar moiety was established by comparing the specific rotation of the aqueous layer of its acid hydrolysate ( ). This was in accordance with our previous reports on akanthopyrones [8]. Finally, the chiral center at C-2 was previously assigned for asperfuran to have R-configuration based on the CD spectrum, which showed a negative Cotton effect at 240 nm due to the chirality at C-2, while another asperfuran derivative ((S)-4,6-dimethyl-2-vinyldihydrobenzofuran) showed a positive Cotton effect at the same wavelength due to the S-configuration at C-2. Since the CD spectrum of glycoasperfuran (3) showed a negative Cotton effect at 240 nm, the absolute configuration at C-2 was assigned to be R. This was also confirmed by performing TDDFT calculations on the 2R/ᴅ-Glc-3 and 2S/ᴅ-Glc-3 isomers (Figure 2). The calculated ECD spectrum of 2S/ᴅ-Glc-3 showed a main positive Cotton effect at 242 nm, while 2R/ᴅ-Glc-3 had a negative Cotton effect at 237 nm, which was similar to the corresponding experimental ECD spectra. Thus, the absolute configuration of glycoasperfuran (3) was confirmed as 2R/ᴅ-Glc.
In addition, two known cyclotetradepsipeptides of the beauverolide family, namely beauverolides N (4) and I (5), and one new beauverolide, J b (6), were isolated from C. javanica BCC26304. Their structures were identified by comparing HRMS data as well as 1 H and 13 C chemical shifts to those reported by Kumza and co-workers [16] for 4 and by Mochizuki and co-workers [17] for 5. Beauverolide J b (6) showed the same molecular formula as beauverolide J a and very similar NMR data [18]. Nevertheless, comprehensive analysis of the 2D NMR data revealed that beauverolide J b (6) comprised a leucine moiety instead of isoleucine in beauverolide J a (see NMR data in the Experimental section and Figures S30-S34 in Supporting Information File 1).

Chemotaxonomic investigation
In order to investigate the distribution patterns of the secondary metabolite production among species of Cordycipitaceae, HPLC-UV-vis profiles of all fungal isolates were generated and compared to each other. This revealed that the individual species possessed unique secondary metabolite profiles. Pigmentosins A (1) and B (2) were detected in all Gibellula strains ( Figure 5), but the production rates of each compound varied among strains. Notably, Gibellula (class Sordariomycetes) and Hypotrachyna (class Lecanoromycetes), from which compound 1 was originally reported, are not phylogenetically close to each other [12,16,18,19], but nevertheless were found to produce the same compound. So far, pigmentosin A (1) was reported only from lichenized fungi [20,21], and thus this is the first report of this compound stemming from another group of fungi. Recently, we have reported on the two new β-carboline alkaloid derivatives gibellamines A and B from Gibellula gamsii. In the current study, our efforts focused on phylogenetic analysis in order to identify the producers of pigmentosins 1 and 2 as well as glycoasperfuran (3), and gibellamines producers were also included in the dataset. Both phylogenetic data and HPLC-based metabolite profiles supported discrimination between these two species, as they were phylogenetically distinct from each other and had individually unique chemotypes.
The comparison of HPLC-UV-vis profiles showed that in C. javanica, glycoasperfuran (3), and beauverolides I (5) [22] and J b (6) were present in all isolates, with the exceptions of the isolates BCC01857 and BCC29254, from which only glycoasperfuran (3) could be detected, while beauverolide N (4) [15] was produced in trace amounts and only seen in four out of eight isolates ( Figure 6). Beauveria and Cordyceps (Isaria) have been found to be phylogenetically close to each other [23], and they produce the same secondary metabolites according to the evidence from Kadlec and co-workers [24], Jegorov and co-workers [25], and Luangsa-ard and co-workers [26]. They unveiled the existences of beauverolides and beauvericin, originally described from Beauveria, in Isaria-producing Cordyceps species. Therefore, our results represent proof of finding the cyclotetradepsipeptides beauverolides as common metabolites in Cordyceps and Beauveria, and glycoasperfuran (3) as species-specific metabolite in C. javanica.
Recently, Helaly and co-workers [27] described the important role of chemotaxonomy in the modern taxonomy of fungi: 1) secondary metabolite profiles as high-informative data to support morphological and phylogenetic studies, 2) the success of using the polyphasic approach in species delimitation, and 3) the potency of chemotaxonomy in the discovery of numerous new secondary metabolites. In the future, chemotaxonomic studies should therefore be further expanded to other taxa in Gibellula, Cordyceps, and related genera, since this approach has been mostly restricted only to certain large ascomycete genera.

Bioactivities
Naphthopyrones are well-known to possess nonselective activities in biological systems and exhibit antimicrobial [28], cytotoxic [29], antimycobacterial [30], and antimalarial [31]   independently from their antimicrobial activity. Conclusively, even though the mode of action of the pigmentosins remains to be studied, they constitute promising candidates for combination therapy with existing or novel antibiotics.

Conclusion
In the current study, the secondary metabolite profiling using HPLC-DAD-MS led to the isolation of three new compounds (2, 3, and 6), together with three known metabolites (1, 4, and 5), as well as species-specific patterns of secondary metabolite production in Gibellula sp. and C. javanica. Their chemical structures were elucidated based on the interpretation of their NMR and HRMS data. Pigmentosins A and B (1 and 2) were isolated from Gibellula sp., while glycoasperfuran (3) as well as beauverolides N, I, and J b (4-6) were obtained from C. javanica. The absolute configurations of the new compounds pigmentosin B (2, partially) and glycoasperfuran (3), as well as the atropisomerism in pigmentosin A (1), previously unassigned, were determined by a combination of Mosher ester analysis and comparison of the calculated and experimental ECD data. Since pigmentosins A and B (1 and 2) were able to significantly inhibit the biofilm formation of S. aureus, their bacteriostatic and bactericidal effects were further evaluated.
Remarkably, the inhibition toward S. aureus and target cell lines was not observed in pigmentosin B (2), but only in pigmentosin A (1). Nevertheless, pigmentosin A (1) displayed anti-S. aureus activity independently from its antibiofilm activity. These properties qualified them as promising candidates for alternative antibiofilm agents. We hope that our findings will also help to raise the general scientific interest in invertebratepathogenic fungi, and in particular in the taxonomy and secondary metabolism of the spider pathogens.

Experimental
General 1D and 2D NMR spectra were recorded on a Bruker Avance III 700 spectrometer with a 5 mm TXI cryoprobe ( 1 H NMR: 700 MHz, 13

Fungal material
The invertebrate parasitic fungal specimens were collected from Central and Northeastern Thailand. Their pure cultures were isolated and subsequently deposited at the BIOTEC Culture Collection (BCC), Pathum Thani, Thailand. Two nuclear DNA regions of all isolates including internal transcribed spacer regions of the ribosomal DNA (ITS) and translation elongation factor 1-alpha (EF1-α) were sequenced according to the protocols given by Kuephadungphan and co-workers [10] and Mongkolsamrit and co-workers [47]. The generated sequence data were submitted to GenBank. A list of fungal strains studied and species descriptions are provided in Supporting Information File 1.

Fermentation and extraction
Submerged fermentation was done as described by Chepkirui and co-workers [48], with minor modifications. Pure cultures were inoculated in YMG liquid medium by cutting seven mycelial plugs (1 cm × 1 cm) from an actively growing colony into a 500 mL Erlenmeyer flask containing 200 mL of the same medium, and incubated at 23 °C on a shaker at 140 rpm. The free glucose content of each fermented broth was tentatively monitored using Bayer Diastix Harnzuckerstreifen. After the glucose was depleted, the incubation was prolonged for half of the time each strain had taken for glucose consumption, and the cultures were then harvested. The fermented broths were separated from the mycelia by vacuum filtration, and were both subsequently extracted according to the procedure described by Phainuphong and co-workers [49]. The fungal mycelia were extracted with acetone, followed by EtOAc instead of MeOH and hexane. Thereafter, the secondary metabolite profiling was carried out on an Agilent 1260 UHPLC Infinity Systems.
Based on a comparison of HPLC profiles between each strain within species, the isolates BCC39707 and BCC26304, representing Gibellula sp. and C. javanica, respectively, were selected and fermented on a larger scale (4 L) using the procedure described above. After the fermentation was operated in 20 × 500 mL Erlenmeyer flasks containing 200 mL of YMG medium under shaking, the fungal cultures were harvested on day 22, and 5 for Gibellula sp. and C. javanica. The fermented filtrates from both strains were extracted with 4 L of EtOAc, giving dark brown oily residues, while their mycelia were extracted sequentially with acetone, followed by EtOAc to afford mycelial extracts as brown gum. They were all chemically profiled by HPLC-DAD-MS in order to optimize the chromatographic purification conditions.
Isolation and structure elucidation of compounds 1-6 The fractionation of the EtOAc extract of Gibellula sp. BCC39707 (dissolved in MeOH) was carried out on an Agilent 1100 series HPLC system (Agilent Technologies). The compounds were separated through a reversed-phase C 18 column (Kromasil, 250 mm × 20 mm, 7 µm, MZ Analysentechnik) using a mixture of deionized water (Milli-Q Millipore, solvent A) and acetonitrile (HPLC grade, solvent B) as eluent, applying a linear gradient of 10-90% solvent B for 50 min, continued to 100% solvent B for 10 min, followed by isocratic conditions of 100% solvent B for 5 min, with a flow rate of 20 mL/min. Thereby, UV detection was performed at 210, 280, and 354 nm.  Table 1.

Preparation of (S)-and (R)-MTPA esters of pigmentosin B (2)
Compound 2 (1.2 mg) was dissolved in deuterated pyridine (1 mL) and transferred into two clean 0.5 mL glass vials. (R)-MTPA-Cl (5 μL) was added into one vial to prepare the (S)-MTPA ester of 2, while (S)-MTPA-Cl (5 μL) was added into the other vial to prepare the (R)-MTPA ester. The reaction was performed at room temperature for 1 h. 1

Biological assays
To evaluate the biological effects of compounds 1-6, various assays were carried out. The antimicrobial activity of the isolated compounds against Bacillus subtilis DSM10, Escherichia coli DSM498, Candida tenuis MUCL29892, and Mucor plumbeus MUCL49355 was determined as described by Kuephadungphan and co-workers [8]. The nematicidal activity against Caenorhabditis elegans was investigated using a microtiter plate assay according to Helaly and co-workers [9], while the cytotoxicity was tested against murine fibroblast (L929) and human HeLa (KB3.1) cell lines according to Chepkirui and co-workers [50].
The isolated compounds were also tested for their ability to interfere in the biofilm formation of Staphylococcus aureus DSM1104 and Pseudomonas aeruginosa PA14 [51]. The biofilm inhibition assay was performed in 96-well microtiter plates using the microtiter dish biofilm formation assay described by O'Toole [52], with minor modifications, as outlined in our recent publications [9,48]. The antibiofilm activity is expressed as MIC values, which is defined as the lowest concentration of substance that prevents biofilm formation of a target microorganism by at least 50%.
Compounds that had shown inhibition of biofilm formation against either S. aureus or P. aeruginosa were further evaluated for their bacteriostatic and bactericidal activities, as described by Yuyama and co-workers [53], in 8 concentrations, ranging from 1.95-250 µg/mL. The assay was performed with each concentration tested in quadruplicates. The MIC was considered as the lowest concentration where the percentage of inhibition was higher than or equal to 90%. The minimum bactericidal concentration (MBC) of isolated compounds was also determined by transferring an aliquot of 2 µL from all concentrations tested onto nutrient agar (NA) plates, which were then incubated at 30 °C for 24 h. The MBC endpoint was defined as the lowest concentration of the compounds that killed microorganisms, where no visible growth of the microorganism tested was observed on the agar plates. The experimental procedures and the results are given in detail in Supporting Information File 1.

ECD theoretical calculations
TDDFT-ECD was used to perform theoretical ECD calculations. Conformational searches for the investigated compounds were first performed with a MMFF94S force field and an energy window of 10 kcal/mol using Omega2 software [54,55]. Each resulting conformer was then subjected to geometrical optimization and vibrational frequency calculation at the B3LYP/6-31+G* level of theory using the Gaussian 09 software [56]. Based on the optimized geometries, TDDFT calcula-tions were finally carried out at the B3LYP/6-311+G* level of theory, and the first 50 excitation states were considered. To consider the solvent effect, the polarizable continuum model (PCM) for methanol was applied. ECD spectra were obtained using SpecDis 1.71 [57,58] and averaged using Boltzmann factors evaluated at 293 K. In the calculated/experimental ECD comparison, wavelength shifts and intensity scaling were applied.

Supporting Information
Supporting Information File 1 LC-MS and NMR data of compounds 1-6, experimental procedures and detailed results for bioassays, as well as species identification of the pigmentosin and glycoasperfuran producers.