Anti-Mycoplasma Activity of Bacilotetrins C–E, Cyclic Lipodepsipeptides from the Marine-Derived Bacillus subtilis and Structure Revision of Bacilotetrins A and B

Mycoplasma hyorhinis most commonly causes polyserositis and arthritis in swine and is a common contaminant during the cell culture in the laboratory. In our continuing research for diverse bioactive compounds from Bacillus subtilis 109GGC020, we discovered uncommon cyclic lipotetrapeptides showing inhibitory activities against M. hyorhinis with similar structures to previously reported bacilotetrins A and B. Bacilotetrins C–E (1–3), new cyclic lipodepsipeptides, were isolated from the EtOAc extract obtained from the fermentation of marine-derived Bacillus subtilis isolated from a marine sponge sample collected from the Gageo reef, Republic of Korea. The structures of 1–3, consisting of three leucine residues, one glutamic acid, and a β-hydroxy fatty acid, were elucidated by detailed analysis of 1D, 2D NMR, and HR-ESIMS data. The absolute configurations of the amino acids and β-hydroxy fatty acid were established by advanced Marfey’s method and Mosher’s method, respectively. The localization of L- and D-amino acids within the compounds was determined by retention time comparison of each purchased dipeptide standard to the partial hydrolysate products using LC-MS. Compounds 1–3 exhibited anti-mycoplasma activity, with an MIC value of 31 μg/mL, twofold stronger than that of the positive control, BioMycoX®. Detailed analysis and comparison of the spectroscopic data between bacilotetrins A (4) and B (5) and 1–3 led us to revise the structures of 4 and 5.


Introduction
Marine micro-organisms are recognized principally as a significant resource producing new and bioactive compounds [1]. Marine Bacillus species produce structurally diverse secondary metabolites, including lipopeptides, polypeptides, macrolides, fatty acids, polyketides, carotenoids, and isocoumarins, which have various activities, such as antimicrobial, anticancer, and antialgal activities [2]. In particular, strong antimicrobial cyclic lipopeptides, including surfactins, iturins, and fengycins, from Bacillus subtilis have received great attention for potential biotechnological and pharmaceutical applications [3].
In our previous study, marine-derived Bacillus subtilis 109GGC020 has been reported to produce interesting secondary metabolites, including macrolactins (gageomacrolactins [4]), linear lipopeptides (gageotetrins A-C [5], gageopeptides A-D [6], and gageostatins A-C [7]), and cyclic lipopeptides (gageopeptins A and B [8], and bacilotetrins A and B [9]) with antibacterial and antifungal activities. Linear lipopeptides, including gageopeptides A-D and gageotetrin B, exhibited inhibitory effects on the wheat blast fungus Magnaporthe oryzae Triticum [10] and the results have revealed the potential of these compounds for agricultural antibiotics.
Mycoplasma is generally known as the smallest bacteria that can survive without oxygen and exists in various forms due to their lack of cell walls [11,12]. Mycoplasma species infect animals, plants, insects, and humans and are often found in research laboratories Mycoplasma is generally known as the smallest bacteria that can survive without oxygen and exists in various forms due to their lack of cell walls [11,12]. Mycoplasma species infect animals, plants, insects, and humans and are often found in research laboratories as contaminants in cell culture [11,13]. Among mycoplasmas, M. hyorhinis is a commensal bacterium of the upper respiratory tract of swine and it is a pathogenic mycoplasma species found in piglets [14]. In addition, it has been reported to cause polyserositis [15], arthritis [16], conjunctivitis [17], otitis [18], and cell culture contamination [13].
During the investigation for antimicrobial compounds against agricultural pathogens, we discovered new cyclic lipotetradepsipeptides (1-3) (Figure 1) from the strain 109GGC020 exhibiting inhibitory activities against Mycoplasma hyorhinis. The molecular formulae, 1 H, 13 C, and 2D NMR data of 1-3 were closely similar to previously reported bacilotetrins A (4) and B (5) [9]. By the detailed and careful analysis of NMR data, we also found that the planar structures of 4 and 5 were wrongly determined. Here, we described the isolation, structure determination, and anti-mycoplasma activity of 1-3, and structure revision of 4 and 5.
As the spectroscopic data, including NMR and MS, of 1-3 were very similar to previously reported bacilotetrins A (4) and B (5), we carefully compared and checked the NMR data of these compounds to discover that the planar structures of 4 and 5 were incorrectly determined. In the original paper for 4, the NMR signals for the carbonyl carbon (δC 173.0) and α-position (δC 51.6 and δH 4.58) of Leu-3 were misassigned to those of Glu (Table S4). These misassignments led to a wrong determination of the sequence of amino acids in 4. In addition, the methine signals (δC 30.6 and δH 1.50) for the anteiso-type β-OH acid in the original NMR data of 4 were not true signals. The methyl signals (δC 14.6 and δH 0.89) of the β-OH acid in 4 were in good agreement with those (δC 14.5 and δH 0.89) of   Bacilotetrins D (2) and E (3) were isolated as amorphous solids. Both the molecular formulae of 2 and 3 were determined to be C38H68N4O8 (unsaturation degree of 7) by HR-ESIMS. The NMR data for 2 and 3 are summarized in Table 1. The 1 H and 13 C NMR spectra of 2 and 3 were very similar to those of 1, except for the branched β-OH acids. The main difference in the 1 H and 13 C NMR spectra of 2 and 3 lies in the chemical shifts of the terminal methyls in the β-OH fatty chains. Bacilotetrin D (2) showed anteiso-methyl signals (δC 19.7/δH 0.85 and δC 11.8/δH 0.87), whereas 3 displayed iso-methyl signals (δC 23.1/δH 0.87 × 2). Compound 2 exhibited HMBC signals from H-13 (δH 1.29 and 1.09) of β-OH acid to C-11 (δC 30.7) and C-12 (δC 35.7) of β-OH acid and from H-14 (δH 0.87) and H-15 (δH 0.85) to C-13 (δC 37.8). These signals established that the branched-chain fatty acid in 2 is an anteiso type. Bacilotetrin E (3) also displayed HMBC signals from H-14 and H-15 (δH 0.87) of the β-OH acid to C-13 (δC 29.2) of β-OH acid and from H-12 (δH 1.16) to C-14 (δC 23.1) and C-15 (δC 23.1) of the β-OH acid, suggesting the presence of an iso-methyl branched fatty acid. In addition, anteiso-and iso-methyl signals of 2 and 3 were in good agreement with reference values (anteiso: δC 19.6/δH 0.86 and δC 11.8/δH 0.88; iso: δC 23.1/δH 0.88) [19]. Thus, the sequences of 1-3 were identified as cyclo-(R-β-OH acid-L-Glu-L-Leu-D-Leu-L-Leu).
As the spectroscopic data, including NMR and MS, of 1-3 were very similar to previously reported bacilotetrins A (4) and B (5), we carefully compared and checked the NMR data of these compounds to discover that the planar structures of 4 and 5 were incorrectly determined. In the original paper for 4, the NMR signals for the carbonyl carbon (δC 173.0) and α-position (δC 51.6 and δH 4.58) of Leu-3 were misassigned to those of Glu (Table S4). These misassignments led to a wrong determination of the sequence of amino acids in 4. In addition, the methine signals (δC 30.6 and δH 1.50) for the anteiso-type β-OH acid in the original NMR data of 4 were not true signals. The methyl signals (δC 14.6 and δH 0.89) of the β-OH acid in 4 were in good agreement with those (δC 14.5 and δH 0.89) of Bacilotetrins D (2) and E (3) were isolated as amorphous solids. Both the molecular formulae of 2 and 3 were determined to be C 38 H 68 N 4 O 8 (unsaturation degree of 7) by HR-ESIMS. The NMR data for 2 and 3 are summarized in Table 1. The 1 H and 13 C NMR spectra of 2 and 3 were very similar to those of 1, except for the branched β-OH acids. The main difference in the 1 H and 13 C NMR spectra of 2 and 3 lies in the chemical shifts of the terminal methyls in the β-OH fatty chains. Bacilotetrin D (2) showed anteiso-methyl signals (δ C 19.7/δ H 0.85 and δ C 11.8/δ H 0.87), whereas 3 displayed iso-methyl signals (δ C 23.1/δ H 0.87 × 2). Compound 2 exhibited HMBC signals from H-13 (δ H 1.29 and 1.09) of β-OH acid to C-11 (δ C 30.7) and C-12 (δ C 35.7) of β-OH acid and from H-14 (δ H 0.87) and H-15 (δ H 0.85) to C-13 (δ C 37.8). These signals established that the branched-chain fatty acid in 2 is an anteiso type. Bacilotetrin E (3) also displayed HMBC signals from H-14 and H-15 (δ H 0.87) of the β-OH acid to C-13 (δ C 29.2) of β-OH acid and from H-12 (δ H 1.16) to C-14 (δ C 23.1) and C-15 (δ C 23.1) of the β-OH acid, suggesting the presence of an iso-methyl branched fatty acid. In addition, anteiso-and iso-methyl signals of 2 and 3 were in good agreement with reference values (anteiso: δ C 19.6/δ H 0.86 and δ C 11.8/δ H 0.88; iso: δ C 23.1/δ H 0.88) [19]. Thus, the sequences of 1-3 were identified as cyclo-(R-β-OH acid-L-Glu-L-Leu-D-Leu-L-Leu).
As the spectroscopic data, including NMR and MS, of 1-3 were very similar to previously reported bacilotetrins A (4) and B (5), we carefully compared and checked the NMR data of these compounds to discover that the planar structures of 4 and 5 were incorrectly determined. In the original paper for 4, the NMR signals for the carbonyl carbon (δ C 173.0) and α-position (δ C 51.6 and δ H 4.58) of Leu-3 were misassigned to those of Glu (Table S4). These misassignments led to a wrong determination of the sequence of amino acids in 4. In addition, the methine signals (δ C 30.6 and δ H 1.50) for the anteiso-type β-OH acid in the original NMR data of 4 were not true signals. The methyl signals (δ C 14.6 and δ H 0.89) of the β-OH acid in 4 were in good agreement with those (δ C 14.5 and δ H 0.89) of 1 and the literature values for the linear-type β-OH acid [19]. Thus, we revise the planar structure of 4 to have a linear-type β-OH acid instead of the anteiso-type β-OH acid, and to be cyclo-(R-β-OH acid-L-Glu-L-Leu-L-Leu-L-Leu) instead of cyclo-(R-β-OH acid-L-Leu-L-Leu-L-Leu-L-Glu) ( Figure 5). Compound 5 was reported to have a 3-hydroxy-9,11-dimethyltridecanoic acid (HDTA, C 15 H 30 O 3 ) as a β-OH acid. However, by the detailed analysis of 2D NMR data, we found that the HDTA unit in 5 is a mixture of a 3-hydroxy-12methyltetradecanoic acid and a 3-hydroxy-13-methyltetradecanoic acid, which have the same molecular weight and formula to HDTA, as shown in Figure 5. This fact was also supported by the chemical shifts of the methyl signals (δ C 19.8/δ H 0.86, δ C 11.9/δ H 0.87, and δ C 23.1/δ H 0.87) of the β-OH acid in 5, which were well matched with the literature values for the anteiso-and iso-type β-OH acids [19]. Therefore, the planar structures of 4 and 5 should be revised to have the same planar core structure as baciloterins C-E (1-3). However 1 and the literature values for the linear-type β-OH acid [19]. Thus, we revise the planar structure of 4 to have a linear-type β-OH acid instead of the anteiso-type β-OH acid, and to be cyclo-(R-β-OH acid-L-Glu-L-Leu-L-Leu-L-Leu) instead of cyclo-(R-β-OH acid-L-Leu-L-Leu-L-Leu-L-Glu) ( Figure 5). Compound 5 was reported to have a 3-hydroxy-9,11dimethyltridecanoic acid (HDTA, C15H30O3) as a β-OH acid. However, by the detailed analysis of 2D NMR data, we found that the HDTA unit in 5 is a mixture of a 3-hydroxy-12-methyltetradecanoic acid and a 3-hydroxy-13-methyltetradecanoic acid, which have the same molecular weight and formula to HDTA, as shown in Figure 5. This fact was also supported by the chemical shifts of the methyl signals (δC 19.8/δH 0.86, δC 11.9/δH 0.87, and δC 23.1/δH 0.87) of the β-OH acid in 5, which were well matched with the literature values for the anteiso-and iso-type β-OH acids [19]. Therefore, the planar structures of 4 and 5 should be revised to have the same planar core structure as baciloterins C-E (  The structures of 1-3 have a similar structural composition to surfactins. Surfactins are cyclic lipopeptides consisting of seven amino acids (L-Glu-L-Leu-D-Leu-L-Val-L-Asp-D-Leu-L-Leu) and a β-OH fatty acid having 13 to 15 carbon atoms [22]. Likewise, 1-3 are also cyclic lipopeptides consisting of four amino acids (L-Glu-L-Leu-D-Leu-L-Leu) and a β-OH acid having 14 or 15 carbon atoms in a similar manner. These structural similarities suggest that 1-3 might be biosynthesized by a similar biosynthetic pathway, a non-ribosomal peptide synthetase (NRPS), to surfactins. The cyclic lipopeptide surfactins are synthesized by a complex of three surfactin synthetase subunits SrfA-A, SrfA-B, and SrfA-C [23]. These subunits consist of either three modules (SrfA-A and SrfA-B) or one module (SrfA-C) and each module contributes to the addition of one amino acid [24]. In the case of 1-3, it is predicted that one SrfA-B module is omitted and other modules are related to produce the structures ( Figure S37).

Inhibitory Activity of Isolated Compounds against Mycoplasma hyorhinis
The anti-mycoplasma activity of 1-3 was assessed by broth dilution assay (Table 2). Compounds 1-3 exhibited anti-mycoplasma activity, with an MIC value of 31 μg/mL. These results revealed that the type of branch of β-OH fatty acids does not affect their inhibitory activity against M. hyorhinis, and the cyclic lipodepsipeptide core plays a more important role. The structures of 1-3 have a similar structural composition to surfactins. Surfactins are cyclic lipopeptides consisting of seven amino acids (L-Glu-L-Leu-D-Leu-L-Val-L-Asp-D-Leu-L-Leu) and a β-OH fatty acid having 13 to 15 carbon atoms [22]. Likewise, 1-3 are also cyclic lipopeptides consisting of four amino acids (L-Glu-L-Leu-D-Leu-L-Leu) and a β-OH acid having 14 or 15 carbon atoms in a similar manner. These structural similarities suggest that 1-3 might be biosynthesized by a similar biosynthetic pathway, a non-ribosomal peptide synthetase (NRPS), to surfactins. The cyclic lipopeptide surfactins are synthesized by a complex of three surfactin synthetase subunits SrfA-A, SrfA-B, and SrfA-C [23]. These subunits consist of either three modules (SrfA-A and SrfA-B) or one module (SrfA-C) and each module contributes to the addition of one amino acid [24]. In the case of 1-3, it is predicted that one SrfA-B module is omitted and other modules are related to produce the structures ( Figure S37).

Inhibitory Activity of Isolated Compounds against Mycoplasma hyorhinis
The anti-mycoplasma activity of 1-3 was assessed by broth dilution assay ( Table 2). Compounds 1-3 exhibited anti-mycoplasma activity, with an MIC value of 31 µg/mL. These results revealed that the type of branch of β-OH fatty acids does not affect their inhibitory activity against M. hyorhinis, and the cyclic lipodepsipeptide core plays a more important role.

Micro-Organism and Fermentation
The bacterial strain Bacillus subtilis 109GGC020 (Genbank accession number JQ927413) was isolated from a marine sponge sample collected from the Gageo reef, Republic of Korea in 2010. The seed culture and production cultures were carried out in Bennett (BN)'s broth [9] (1% glucose, 0.2% tryptone, 0.1% yeast extract, 0.1% beef extract, 0.5% glycerol, 1.85% artificial sea salt, pH 7 before sterilization). The seed culture was performed in a 250 mL Erlenmeyer flask containing 100 mL BN broth at 28 • C, 120 rpm for 3 days. The seed culture was inoculated into a 100 L fermenter containing 70 L of the broth medium under the aseptic condition. The fermenter was operated at 28 • C, 55 rpm, and airflow rate of 20 L/min (LPM) for 7 days. The culture broth was separated by high-speed centrifugation (60,000 rpm) into cell mass and broth. The broth part was extracted with an equal volume of ethyl acetate (EtOAc, 70 L) twice.

Extraction and Isolation of Compounds 1-3
The EtOAc extract was concentrated in vacuo, and 28.4 g of a crude extract was obtained. A portion of the crude extract (9.7 g) was subjected to reversed-phase vacuum column chromatography (YMC Gel ODS-A, 12 nm, S 75 µm) with a stepwise gradient solvent system of 20, 40, 60, 80, and 100% MeOH in H 2 O. The 100% MeOH fraction, showing characteristic two exchangeable proton signals in 7-8 ppm of bacilotetrins, was selected for further purification.

Methanolysis of 1
Compound 1 (2.4 mg) was dissolved in 1.2 mL of 3M methanolic HCl and refluxed for 2 h. The completion of reaction was confirmed by LR-LCMS analysis. The mixture was concentrated under a N 2 gas stream and partitioned with Hex and water. The Hex layer was dried and 1.0 mg of a crude fatty acid ester 1a was obtained ( Figure S34).

Preparation of the (S)-and (R)-MTPA Esters (1b and 1c)
Crude fatty acid ester 1a was divided equally into two portions and dried under a N 2 gas stream. A few crystals of 4-dimethylaminopyridine (DMAP) and anhydrous pyridine (80 µL) were added to each vial and stirred at room temperature for 5 min. Then, 5 µL of R-(−) or S-(+)-α-methoxy-α-(trifluoromethyl)phenylacetyl chloride (MTPA-Cl) was added, respectively. The mixtures were stirred at room temperature for 16 h. The reaction mixtures were concentrated under a N 2 gas steam at 40 • C. Each mixture was dissolved in methylene chloride (MC) and washed with 1N HCl solution, saturated NaHCO 3 solution and brine. The MC layer was dried over anhydrous MgSO 4 and evaporated in vacuo. Each residue was purified by reversed-phase HPLC (YMC-Triart C 18

Measurement of Anti-Mycoplasma Activity
Anti-mycoplasma activity against Mycoplasma hyorhinis of 1-3 was evaluated by broth dilution assay. In brief, the test strain, Mycoplasma hyorhinis ATCC 17981, was cultured in PPLO broth medium [25] at 37 • C under a humidified atmosphere of 5% CO 2 . Stock solutions of 1-3 were dissolved in DMSO and diluted with PPLO broth medium to give serial twofold dilutions in the range of 500 to 1 µg/mL. The final DMSO concentration was maintained at 5% by adding DMSO to the PPLO broth. Culture broth (100 µL) containing approximately 2 × 10 4 CFU/mL of activated strain was added to each well of a 96-well plate. The plates were incubated for 7 days at 37 • C under a humidified atmosphere of 5% CO 2 . The color of the broth changes to yellow as bacteria grow. The minimum inhibitory concentration (MIC) values were determined as the lowest concentration of the test compound that inhibited bacterial growth. BioMycoX ® (CellSafe Co., Yongin, Republic of Korea) was used for a positive control.

Conclusions
Three new cyclic lipodepsipeptides, bacilotetrins C-E, consisting of four amino acids and a β-hydroxy fatty acid were isolated from the culture broth of Bacillus subtilis 109GGC020. Their spectroscopic data were very similar to those of previously reported for bacilotetrins A (4) and B (5). By the detailed and careful analysis of NMR data, we found that the planar structures of 4 and 5 must be reassigned and our comprehensive spectroscopic data analysis led to revision of their structures. In the revised structures of 4 and 5, the positions of Glu and the branch types of the β-hydroxy fatty acids are correctly determined.
The absolute configurations of the amino acids and β-hydroxy fatty acid in 1-3 were established by chemical derivatization, including Marfey's and Mosher's methods. The major difference between 1-3 and 4-5 lay in the fact that 1-3 consist of L-and D-amino acids, whereas 4-5 have only L-amino acids. In addition, as previously mentioned, 1-3 were expected to be synthesized through the similar biosynthetic pathway to surfactins, which are well known for their various biological activities, such as antifungal, antibacterial, anticancer, and anti-mycoplasma activities [22]. Compounds 1-3 also showed antimicrobial activity against M. hyorhinis, which is known to cause diseases, such as polyserositis, arthritis, conjunctivitis, and otitis in pigs, with an MIC value of 31 µg/mL, twofold stronger than that of the positive control, BioMycoX ® . The only difference between 1-3 was the type of branch in the β-OH acid. Therefore, on the basis of the result, it is supposed that the cyclic peptide core plays an important role in anti-mycoplasma activity, but the type of branch in the β-OH acid is not critical for activity. Further studies are needed to clarify the underlying mechanism of the activity for the development of antibiotics.  Figure S25 and Table S1: HPLC analysis of amino acids in 1 through the total hydrolysis and retention times of L-FDLA derivatives of hydrolysate of 1, Figure S26: HPLC chromatogram of partial hydrolysis of 1, Figure S27 and Table S2: HPLC analysis of L-FDLA derivatives of P1 and P3 hydrolysates and retention times of L-FDLA derivatives of hydrolysate of P1 and P3 hydrolysates, Figure S28: Structures of standard dipeptides, Figure S29 and Table S3: