Lindgomycin, an Unusual Antibiotic Polyketide from a Marine Fungus of the Lindgomycetaceae

An unusual polyketide with a new carbon skeleton, lindgomycin (1), and the recently described ascosetin (2) were extracted from mycelia and culture broth of different Lindgomycetaceae strains, which were isolated from a sponge of the Kiel Fjord in the Baltic Sea (Germany) and from the Antarctic. Their structures were established by spectroscopic means. In the new polyketide, two distinct domains, a bicyclic hydrocarbon and a tetramic acid, are connected by a bridging carbonyl. The tetramic acid substructure of compound 1 was proved to possess a unique 5-benzylpyrrolidine-2,4-dione unit. The combination of 5-benzylpyrrolidine-2,4-dione of compound 1 in its tetramic acid half and 3-methylbut-3-enoic acid pendant in its decalin half allow the assignment of a new carbon skeleton. The new compound 1 and ascosetin showed antibiotic activities with IC50 value of 5.1 (±0.2) µM and 3.2 (±0.4) μM, respectively, against methicillin-resistant Staphylococcus aureus.


Introduction
Marine-derived fungi living in a stressful habitat are of great interest as new, promising sources of biologically active products. Since marine organisms live in a biologically competitive environment with OPEN ACCESS unique conditions of pH, temperature, pressure, oxygen, light, nutrients, and salinity, the chemical diversity of the secondary metabolites from marine fungi is considerably high [1][2][3][4][5][6][7]. Various new genera were described from marine resources, however, marine isolates of known taxa reveal to be potent producers of novel chemistry as well. The fungal family Lindgomycetaceae (Pleosporales, Dothideomycetes) was introduced by Hirayama et al. [8] to accommodate four new species of the genus Lindgomyces. Another three Lindgomyces species were added by Raja et al. [9,10], and these authors also included Massariosphaeria typhicola, a closely related species, into the Lindgomycetaceae. Later on, the newly isolated species Lolia aquatica was added [11] and Clohesiomyces aquaticus was shown to belong to this family according to sequence data [12]. All members of the Lindgomycetaceae were isolated from submerged parts of decaying wood and plant material in a freshwater environment. Little is known on their metabolic capabilities and on their production of bioactive compounds. The only available data are from Raja et al. [10], who found a fatty acid, 6E,9E-octodecadienoic acid, and ergosterol peroxide as the major chemical compounds in Lindgomyces angustiascus.
The naturally occurring pyrrolidine-2,4-dione (tetramic acid) derivatives originating from a variety of marine and terrestrial species have attracted a great deal of interest due to their broad-spectrum biological activities and challenging structural complexity [13,14]. The majority of the compounds isolated to date exhibited antibiotic or antiviral activity. Tetramic acids possessing an octahydronaphthalene skeleton are rare in nature, showing activity against Gram-positive bacteria [13]. Recently, ascosetin was described as a new antibacterial tetramic acid derivative [15].
In this study, two marine strains of the family Lindgomycetaceae were shown to exhibit a diverse chemical profile, including tetramic acid derivatives having novel structures. The strains were isolated from a sponge of the Baltic Sea (Kiel Fjord) and from the Antarctic. The new compound lindgomycin (1) and ascosetin (2) showed antibiotic activities against human and plant pathogenic microorganisms.

Identification of Strains KF970 and LF327
Two fungal isolates, strains KF970 and LF327, which are members of the family Lindgomycetaceae, were isolated from different marine habitats. Both strains grew only as sterile mycelium on the media used, but did not produce conidia or ascomata on ten different media, e.g., on WSP30 ( Figure 1). Thus, morphological criteria for identification of the strain were lacking. The sequences of the 18S rRNA genes comprised 1546 nucleotides for KF970 and 1065 nucleotides for LF327, respectively. Both sequences exhibited a similarity of 99.7% to each other. The highest sequence similarity (100%) was observed for the 18S RNA gene sequence to "Phyllosticta" flevolandica AFTOL-ID 1786. Nevertheless, the deposited sequence of the P. flevolandica strain comprises only approximately a fourth of the length of 18S rRNA gene (441 nucleotides) and Crous et al. [16] have shown that "Phyllosticta" flevolandica does not affiliate with all members of the genus Phyllosticta, which belong to the Botryosphaeriales. Therefore, the strains most closely related were taken into consideration for the phylogenetic classification. Among these, Massariosphaeria typhicola CBS 609.86 (acc. no. KF314118) showed similarities of 99.9% to KF970 and 99.5% to LF327, respectively. The 18S rDNA sequences of strains KF970 and LF327 showed 99.2%-99.5% and 99.0%-99.2% similarity to the type of strains of the seven known Lindgomyces species [10]. Clohesyomyces aquaticus MFLUCC11-0092 (acc. no. JX276949) was 99.0% similar to KF970 and 99.78% similar to LF327. A comparison with the genus Lolia was not possible because no 18S rRNA gene sequence was deposited.
On the basis of the phylogenetic analyses, KF970 and LF327 were clearly assigned to the family Lindgomycetaceae (Pleosporales, Dothideomycetes).

Metabolic Profiles of the Strains KF970 and LF327
Interestingly, both Lindgomycetaceae strains, which originated from different geographic regions, produced compounds 1 and 2. Cultivation experiments of LF327 were carried in two different media under static and non-static conditions. Mycelium and culture broth were not separated. Compounds 1 and 2 were produced in GYM4 medium. The comparison of static and non-static cultures showed that the production level in static cultures was much higher. The evaluation was done by the comparison of UV spectra and ESIMS data, obtained by analysing the extract by analytical HPLC-UV/MS.

Structural Elucidation
The MeOH extracts of the mycelia and the broth of strain KF970 were subjected to repeated column chromatography to purify the two polyketides 1 and 2 ( Figure 2).   (Table 1) in the upfield region of 1 were similar to those of the phomasetin and equisetin analogues [17]. These molecules contain two distinct domains, a bicyclic hydrocarbon and a tetramic acid, connected by a bridging carbonyl [13][14][15][17][18][19]. The 13 C NMR spectrum showed the presence of 18 signals for the hydrocarbon domain including a decahydronaphthalene basic skeleton and a sec-butyl moiety, with the remaining 11 resonances corresponding to a heterocyclic domain including a tetramic acid basic skeleton and a benzyl moiety. When compared to the NMR data of the polyketide phomasetin, isolated from the fungus Fusarium heterosporum, compound 1 showed similar chemical shifts of the main decalin skeleton. However, the structures of the tetramic acid and attached unit at the decahydronaphthalene ring were less similar to phomasetin.  In the 1 H NMR spectrum of 1, two methyl groups displayed two singlet signals, one of which was assigned at quaternary C-2 from the observation of HMBC cross peaks of Me-12/C-1 and Me-12/C-2. The long range correlations from the proton at δH 1.54 (s, Me-18) to the double bond carbon signals at δC 130.8 (s, C-4) and δC 129.5 (d, C-5) positioned the remaining methyl at C-4. The pendant attached at C-3 was obviously different from the known phomasetin. Instead of being a penta-diene, a 3-methylbut-3-enoic acid unit was attached at C-3. The carboxylic carbon at δC 169.6 (s) was attributed to C-15 from the observation of long range correlations from the proton signals at δH 5.63 (s, H-14) to carboxylic C-15. The methyl group in the pendant was assigned at C-13 from analysis of the HMBC cross peaks from Me-16 to olefinic C-13 and C-14. The substructure of the pedant was deduced to be a 3-methylbut-3-enoic acid unit, which was positioned at C-3 from the observation of HMBC correlation from the proton at The formula indicated that 11 resonances remained to be assigned to the tetramic acid half of molecule 1, corresponding to a tetramic acid basic skeleton and a benzyl moiety. However, the C-3′ signal was not detected due to rapid tautomerisations to multiple enol forms [18]. In the downfield region of the NMR spectra of 1, two sets of equivalent aromatic protons at δH 7.19, (d, J = 8.0 Hz, H-8′/12′) and 7.24 (t, J = 8.0 Hz, H-9′/11′) and an aromatic triplet at δH 7.21 (t, J = 8.0 Hz, H-10′) indicate that a single substituted benzene ring unit was added in the core structure of a tetramic acid moiety coupling with a decalin moiety. The structure of the five-member lactam ring at the tetramic acid unit was approved by the HMBC cross peaks of H-5′/C-2′. The five-member lactam ring proved to be linked at the quaternary C-2 via the oxygenated enolic C-1 from the observation of long range correlation from the proton signal at δH 1.38 (s, Me-12) to the enolic carbon signal at δC 201.4 (s, C-1). Owing to the formation of the intermolecular hydrogen bond between the hydroxyl at the olefin carbon and the carbonyl at the lactam ring, the C-1 was highly downshifted [14,17,19]. The benzylic methylene of H2-6′ showed long range two-dimensional 1 H-13 C correlations to the aromatic carbon C-8′/12′ and C-7′ of the phenyl ring and to the carbonyl C-4′ of the pyrrolone ring. This indicated that the benzene unit was linked with the five-member lactam unit via a methene bridge. This inference was confirmed by the observation of 1 H-1 H COSY cross peaks of H2-6′/H-5′. Thus, the planar structure of 1 was elucidated as shown in Figure 2.
The 1,3-diaxial NOESY cross peaks of H-6α/H-8α, H-8α/H-10α, H-11β/H-7β, and H-11β/H-9β revealed a 6,11-trans ring fusion between the cylcohexyl chair and cyclohexenyl boat rings of the bicyclic decalin with a β-oriented Me at C-8 ( Figure 4). The NOESY correlations from Me-12 to the axial H-6a and H-3 indicated a β-oriented 3-methylbut-3-enoic acid pendant at C-3 and a β-oriented tetramic acid pendant at C-2. The diagnostic NOESY correlation from the olefinic proton at δH 5.63 (s, H-14) to the proton at δH 3.46 s (s, H-3) revealed that the stereochemistry of the double bond between C-13 and C-14 was in an E configuration. Thus, compound 1 is found to be an unusual polyketide and is given the trivial name lindgomycin. The combination of 5-benzylpyrrolidine-2,4-dione of compound 1 in its tetramic acid half and the 3-methylbut-3-enoic acid pendant in its decalin half allowed the assignment of a new carbon skeleton. Compound 2 was identified as the recently described ascosetin by comparison of spectroscopic data reported by Ondeyka et al. [15].

Biological Activities
Both lindgomycin (1) and ascosetin (2) exhibited strong antibiotic activities with IC50 values in the range of 2 to 18 µM ( Table 2). The effects against the clinically relevant bacteria Staphylococcus epidermidis, Staphylococcus aureus, methicillin-resistant Staphylococcus aureus, and Propionibacterium acnes were two times less in comparison with chloramphenicol. The activity against the human pathogenic yeast Candida albicans was four times lower than it was observed for nystatin. The causative agents of black rot in crucifers (e.g., cabbage) and of leaf spot disease on crops (e.g., wheat), Xanthomonas campestris and Septoria tritici, were also inhibited by the new compounds. No inhibition of Gram-negative bacteria was observed.

Biotechnological Scale Up
In order to insure sustainable production of 1 and 2 in larger amounts, we have performed experiments to optimise and scale up the biotechnological production from Erlenmeyer flask cultures to controllable stirred tank reactors. As other fungi, strain KF970 showed variation in the metabolite production in various media and growth conditions. Among a number of culture media, the best production of 1 and 2 was found in GYM4 medium and in standing cultures as well as in shaken Erlenmeyer flasks at pH 7.2 and in the absence of added salt to the medium.
Interestingly, the culture medium and the age of the pre-cultures were highly important for good growth and production. GYM4 was the best medium for the pre-cultures, but if these were older than 21 days, no significant production could be obtained.
Based on the content of nitrogen in the candidate structures, a medium containing a high nitrogen/carbon ratio was introduced in order to further improve the production rate. Casamino acids medium proved to support very good growth in small brown pellets and improvement of the production (compared to GYM4) of 2-4 times. This was observed in standing cultures but more pronounced in shaken Erlenmeyer flasks.
For a sustainable large scale production, the transfer of the production into a fermenter system (stirred tank reactor) is essential. Therefore, the best conditions for the production of 1 and 2 by strain KF970 in Erlenmeyer flasks were used to establish the production in a 10 L fermenter. Low pH turned out to be best for production of secondary metabolites by strain KF970, and for all subsequent experiments in STR, the pH value was set to pH 5. The fungus showed comparable growth behavior in a 10 L fermenter system with controlled pH and air saturation (minimum value set to 30%), reaching the exponential phase of growth after 3 days of cultivation. Both the low pH and stable oxygen levels in the medium due to the stirring stimulate hyphal growth. Production of 1 and 2 occurred within the exponential growth phase with a maximum after 7 days of cultivation. This is a significant reduction of cultivation time for production as compared to a minimum of 14 days needed to reach maximum product levels of both compounds in Erlenmeyer flasks. After this period, growth of the fungus continued, but concentrations of 1 and 2 slowly decreased (with a half-life of 6 days). Considering the optimal time point for harvest, the production yield of 1 and 2 increased from 0.2 mg/L in Erlenmeyer flasks to 5.0 mg/L in a 10 L fermenter.

General Experimental Procedures
Optical rotations were recorded on a Perkin Elmer 241 polarimeter. The IR spectra were run on a Perkin Elmer spectrometer with an attenuated total reflectance (ATR) unit. 1 H NMR (500 MHz) and 13 C NMR (125 MHz) spectra were measured at 25 °C on a Bruker AVANCE DMX 500 NMR spectrometer with tetramethylsilane (TMS) as internal standard. The signals of the residual solvent protons and the solvent carbons were used as internal references (δH 3.31 ppm and δC 49.0 ppm for methanol-d4). High-resolution mass spectra were acquired on a benchtop time-of-flight spectrometer (micrOTOF II, Bruker Daltonics, Germany) with positive electrospray ionization (ESI).

Isolation, Cultivation, and Storage of the Producer Strains KF970 and LF327
The fungus KF970 was isolated from samples of the expedition of RV Polarstern to the Arctic in 1991 by K. Schaumann [20]. The strain LF327 was isolated from the sponge Halichondria panicea, which was collected at the Kiel Fjord (Baltic Sea, Germany). Both strains were grown on WSP30 agar, a modified Wickerham-medium consisting of 1% glucose, 0.5% peptone, 0.3% yeast extract, 0.3% malt extract, 3% sodium chloride (pH = 6.8) [21]. The strains were stored in liquid nitrogen and in the Microbank System at −80 °C (MAST DIAGNOSTIKA, Reinfeld, Germany).

Comparison of the Metabolic Profiles the Strains KF970 and LF327
Cultivation of both strains was carried out in two L Erlenmeyer flasks containing 750 mL GYM4 (0.4% malt extract, 0.4% yeast extract, 0.4% glucose × H2O, 0.2% CaCO3, pH 7.2) for 21 days at 28 °C as static and shaken cultures in the dark. Circular agar pieces (WSP30, diameter of 1.8 cm) were used for inoculation. Extraction of mycelium and culture broth was performed using ethyl acetate added to the cultures in equal volume. The metabolite profiles of both strains were evaluated for compounds 1 and 2 using analytical HPLC-UV/MS.

Fermentation and Production of Extracts for the Purification of Compounds 1 and 2
Strain KF970 was inoculated onto agar-plates containing WSP30 medium. After incubation for 15 days at 22 °C the pre-culture was used for inoculation of two L Erlenmeyer flasks containing 750 mL SM medium (2% soy peptone, 2% mannitol, 1.5% NaCl, pH 7.0). The flasks were incubated for 32 days at 28 °C as static cultures in the dark. The mycelium was separated from the culture medium. The 22 L fermentation broth was extracted using ethyl acetate (12 L). After evaporation of the solvents the crude extract was re-dissolved in 5 mL methanol and stored at 4 °C until further use.
Semi-preparative HPLC was carried out using a HPLC-UV system (VWR Hitachi Elite LaChrom system, L-1230 pump, L-2450 diode array detector, L-2200 autosampler, Phenomenex Gemini-NX C18 P. acnes, S. tritici, and C. albicans. The lack of activity against Gram-negative bacteria has already been noticed for other tetramic acid derivatives by Lowery et al. [38] and maybe related to the difference in the cell wall structures. Apparently the outer membrane of Gram-negative bacteria is an efficient permeability barrier for these molecules. The WHO stated that 25,000 persons die each year due to infections with antibiotic resistant bacteria in 29 European countries [39]. Specifically, MRSA strains are causing infections with high mortality rates and a growing rate of resistance [40]. According to the data raised by the European Antimicrobial Resistance Surveillance Network (EARS-Net), the percentage of MRSA-positive isolates exceed 20% in 11 of 28 European countries that participate at the EARS-Net [41]. In addition to the highly demanded search for new natural products active against antibiotic-resistant bacteria, a number of other health issues also require new active compounds for improved treatment. These include new molecules against the human pathogens P. acnes or C. albicans. P. acnes is involved in the inflammatory skin disease acne vulgaris and causes opportunistic human infections [42]. Candida species are responsible for the infectious disease candidiasis, especially in immuno-compromised patients [43]. More than 90% of invasive infections are caused by C. albicans. Without a doubt, there still remains a strong demand for new antibiotics that might be more effective, less toxic, and show a lower risk of resistance. Lindgomycin (1) as well as ascosetin (2) could be promising candidates in this search.