Bacillaenes: Decomposition Trigger Point and Biofilm Enhancement in Bacillus.

Bacillaenes are a class of poly-unsaturated enamines produced by Bacillus strains that are notoriously unstable toward light, oxygen, and normal temperature. Herein, in an in-depth study of this highly unstable chemotype, the stability and biological function of bacillaenes were investigated. The structure change of the bacillaene scaffold was tracked by time-course 1H NMR data analysis coupled with the differential analysis of 2D-NMR spectra method, which was demonstrated to be a "domino" effect triggered by 4',5'-cis (2 and 3) configuration rearranged to trans (2a and 3a). These findings provide the possibility for stabilizing the bacillaene scaffold by chemical modification of its trigger points. In the biofilm assay, compounds 1 and 2 accelerated self-biofilm formation in Bacillus methylotrophicus B-9987 at low concentrations of 1.0 and 0.1 μg/mL. Interestingly, bacillaenes play dual roles as antibiotic and biofilm enhancers in a dose-dependent manner, both of which serve in the self-protection of Bacillus.

Introduction

Unstable secondary metabolites are a rarely explored area of natural product research. Bacillaenes are a class of poly-unsaturated enamines encoded by a trans-acyltransferase polyketide synthetase machinery in a wide range of Bacillus strains that are notoriously unstable toward light, oxygen, and normal temperature.[1−4] Since the first discovery in 1995,[1] their structures have not been identified for a decade because of chemical instability, until the arduous task was completed by Clardy and Walsh groups in 2007.[2] According to the literature studies, bacillaenes are considered as antibiotic weapons for Bacillus to resist other environmental microbes.[1,5−7] However, from the perspective of evolution, it does not seem to be enough because the bacillaene-producing Bacillus strains encode a variety of structurally stable antibiotics.[8−11]

In our recent study on bacillaenes, we identified a series of new glycosylated bacillaenes (gBAEs) (14′,15′-dihydro-bacillaene B, 12′,13′-trans-14′,15′-dihydro-bacillaene B, and bacillaene C) using a combined strategy of genome, 2D-NMR metabolome coupled with bioassay, and illustrated that glycosylation generally decreases the antibacterial activity.[12] Herein, as continuing studies, the stability and biological function of bacillaenes were investigated. The degradation trigger point of the bacillaene scaffold was caught and determined by time-course 1H NMR data analysis coupled with DANS (differential analysis of 2D-NMR spectra) method; moreover, the effect of bacillaenes on the biofilm formation of Bacillus strains was probed in a dose-dependent manner.

Results and Discussion

Because of the unstable bacillaene chemotype, the structural stability throughout the biological evaluation is one of the most concerning issues. Pure 1–3, which are the major bacillaene compounds, were prepared from the fermentation culture of the ΔbmmB mutant strain of Bacillus methylotrophicus B-9987. The molecular formulas of 1–3 were established as C34H48N2O6, C40H58N2O11, and C40H60N2O11, respectively, on the basis of the HR-ESIMS data ([M + H]+ at m/z 581.3595, 743.4163, and 745.4280). Compounds 1 and 3 were confirmed to be bacillaene A and 14′,15′-dihydro-bacillaene B, respectively, as reported in our earlier work,[12] and 2 was identified as bacillaene B[13] by 1D and 2D NMR data assignments (Figure S7 and Table S1). Compounds 1 and 2, being quantified by the standard curves, underwent time- (0–48 h) and dose-dependent (0.1–10 μg/mL) stability tests at normal bioassay conditions (aqueous solution system, 37 °C, in the dark) using UPLC-MS. As shown in Figure S8, bacillaene A (1) was rather unstable in aqueous solution. Bacillaene B (2), a 2″-O-glycosylated bacillaene A analogue, kept stable up to 10 μg/mL within 18 h, and in a longer period, its stability might be concentration-dependent. Decomposition of the non-gBAEs was more easily triggered compared to the glycosylated analogues. This result revealed that the glycosylation stabilizes the bacillaene scaffold, which supports our previous suggestion.[12]

To understand the decomposition mode of the bacillaene scaffold, we used the time-course 1H NMR data collection method to capture the structural changes of the compounds. We prepared the pure sample (PS) and the decomposition-triggered sample (DTS) of compound 3 as described in the Experimental Section. Except for the light, the decomposition of 3 mostly occurred when the compound dose was over a threshold and exposed to oxygen, starting the initial conversion to a stereoisomer 3a (Figure A,B). Compound 3a was the first “shapeshift” of 3 once the degradation was triggered. According to the time-course 1H NMR data, PS-3 was stable up to 68 h at 274 K (Figure S9); however, compound 3 in DTS-3 was rapidly converted to 3a, which was determined by the signal decrease of H-2′ (δH 2.70, 2.45) and H-5′ (δH 6.55) in the 0 h 1H NMR spectrum of DTS-3 (Figure C). Moreover, the time-dependent (0–18 h) signal decrease of H-10 (δH 1.23) and H-3′ (δH 4.54) in the 1H NMR of DTS-3 suggested that 3a continuously changed even at 274 K (Figure C). To determine the structure of 3a, the Heteronuclear Single Quantum Coherence (HSQC), DQF-COSY, Heteronuclear Multiple Bond Coherence (HMBC), and NOESY data of DTS-3 were sequentially collected. In the differential analysis of the HSQC spectra of PS-3 and DTS-3, the protonated carbons of 2′, 3′, 5′–8′, and 17′ showed wide chemical shift differences between 3 and 3a, while other signals almost kept identical (Figure D). It strongly suggested that a structural change occurred in this part. However, because of the continuous changes of 3a during the 2D NMR data collection, the key HMBC correlations could not be fully observed. Considering that C-4′ and C-5′ are the “centers” of the NMR shift changes, 3a was proposed to be the 4′,5′-trans stereoisomer of 3, and it was supported by the 13C chemical shift prediction of 3/3a in ChemBioDraw 15.0 (Figure S12).

Figure 1

Determination of the decomposition trigger point of the bacillaene scaffold. (A) Structures of 3 and its stereoisomer 3a. (B) LC–MS TIC data of the DTS of 3 (DTS-3). (C) Time-course (0, 4, 10, and 18 h) 1H NMR spectra of DTS-3 and 1H NMR spectrum of PS-3 (in CD3OD, 600 MHz at 274 K). In DTS-3, majority of compound 3 had been converted to the stereoisomer 3a and other unknown compounds at the start (0 h) of NMR data collection, which was determined by significant signal decrease of δH 6.55 (H-5′) and δH 2.70, 2.45 (H2-2′). The continuous signal decrease of δH 1.23 (H-10) and δH 4.53 (H-3′) revealed that 3a continuously changed with the increasing time. (D) Differential HSQC analysis of PS-3 (red) and DTS-3 (blue). The carbons of 2′, 3′, 5′–8′, and 17′ showing wide chemical shift differences between 3 and 3a are labeled in red and blue, respectively. The bacillaene scaffold was determined to be “dominos” triggered by the 4′,5′-cis configuration rearranged to trans.

To further confirm the trigger point of structure changes in other bacillaenes, we prepared PS-2 and DTS-2 to conduct the time-course 1H NMR experiments. PS-2 was stable up to 84 h (Figure S13), whereas a large part of compound 2 in DTS-2 had been converted to 2a at the beginning of the NMR data collection (Figure S14). In order to obtain key HMBC correlations to assign the structure of 2a, the freshly prepared DTS-2 was first conducted to the HMBC data collection, and then the HSQC and the DQF-COSY data were sequentially obtained. According to the differential analysis of the HSQC spectra of PS-2 and DTS-2 (Figure S16), the peaks of 2′, 3′, 5′–8′, and 17′ of 2 underwent large chemical shifts when converting to 2a. These NMR shift changes were in accordance with those between 3 and 3a (Figure D). In the HMBC spectrum of DTS-2 (Figure S17), we observed the correlations from H-3′ (δH 4.54) to C-1′ (δC 170.7), C-2′ (δC 42.1), C-4′ (δC 139.2), C-5′ (δC 126.5), and C-17′ (δC 11.1), which undoubtedly determined the location of the structural changes. The experimental NMR data of 2a (Table ) were supported by the 13C chemical shift prediction of 4′,5′-trans stereoisomer of 2 calculated by ChemBioDraw 15.0 (Figure S18). Thus, we proposed that the bacillaene scaffold is “dominos” triggered by 4′,5′-cis configuration rearranged to trans. These findings provide the possibility for stabilizing the bacillaene scaffold by chemical modification of its trigger points.

Table 1

1H and 13C Chemical Shifts of Compounds 2a and 3a Obtained from the NMR Spectra of DTS-2 and DTS-3, Respectively (in CD3OD, 600 MHz)

	2a		3a
position	δC, type	δH (J in Hz)	δC, type	δH (J in Hz)
1	177.3, C		177.0, C
2	42.9, CH	3.15, m	42.9, CH	3.17, m
3	132.2, CH	5.72, m	130.9, CH	5.68, m
4	131.9, CH	6.18, m	a	a
5	a	a	a	a
6	126.6, CH	6.29, m	a	a
7	121.4, CH	5.74, m	121.0, CH	5.72, m
8	132.1, C		132.3, C
9	20.4, CH3	2.00, s	20.4, CH3	2.02, s
10	16.5, CH3	1.23, d, 7.2	16.9, CH3	1.23, d, 6.6
1′	170.7, C		a
2′	42.1, CH2	2.60, m	42.1, CH2	2.57
3′	74.0, CH	4.54, t (6.6)	73.9, CH	4.53, t (7.2)
4′	139.2, C		139.3, C
5′	126.5, CH	6.28, m	126.7, CH	6.27, m
6′	129.4, CH	6.64, m	a	a
7′	130.1, CH	6.28, m	a	a
8′	132.9, CH	6.33, m	131.9, CH	6.24, m
9′	135.4, C		135.4, C
10′	138.0, CH	6.39, m	137.5, CH	6.34, m
11′	123.2, CH	6.79, m	128.7, CH	6.34, m
12′	129.9, CH	6.10, m	131.8, CH	6.23, m
13′	127.9, CH	6.02, m	133.3, CH	5.78, m
14′	127.2, CH	6.78, m	29.9, CH2	2.21, m
15′	129.1, CH	5.75, m	28.5, CH2	1.67, m
16′	38.3, CH2	4.05, m	38.3, CH2	3.27, m
		3.91, m
17′	17.2, CH3	1.97, s	17.2, CH3	1.96, s
18′	11.3, CH3	1.94, s	11.1, CH3	1.90, s
1″	174.2, C		174.4, C
2″	77.1, CH	4.35, m	77.2, CH	4.30, dd (9.0, 3.6)
3″	42.3, CH2	1.65, m	42.2, CH2	1.62, m
		1.57, m		1.57, m
4″	23.8, CH	1.92, m	23.7, CH	1.92, m
5″	20.6, CH3	1.00, d (6.6)	20.5, CH3	1.00, d (6.6)
6″	22.5, CH3	0.95, d (6.6)	22.5, CH3	0.95, d (6.6)
1‴	101.9, CH	4.27, d (7.8)	102.1, CH	4.25, d (7.8)
2‴	73.5, CH	3.29, m	73.6, CH	3.29, m
3‴	76.4, CH	3.40, m	76.6, CH	3.38, m
4‴	69.7, CH	3.33, m	69.9, CH	3.34, m
5‴	76.5, CH	3.30, m	76.7, CH	3.28, m
6‴	61.0, CH2	3.88, m	61.1, CH2	3.88, m
		3.70, m		3.71, m

Signals not distinguishable.

Kolter et al. reported in their recent review that many of the antibiotic secondary metabolites might have different functions (e.g. virulence, colonization, motility, stress response, and/or biofilm formation) in nature at their subinhibitory concentrations (SICs).[14] Interestingly, we observed that both of bacillaene A (1) and bacillaene B (2) accelerated self-biofilm formation in the B-9987 strain at low concentrations of 1.0 and 0.1 μg/mL (Figure ), which were far below than their antibiotic MICs (>10 μg/mL).[12,15] Then, to further evaluate the effects of bacillaenes on the self-biofilm formation in the B-9987 strain at higher concentrations, we used Fr. BAEs/Fr. gBAEs, which were more stable than pure compounds (Figure S19), to conduct the experiments. The Fr. BAEs enhanced the biofilm formation in B-9987 at the concentration of 0.01 μg/mL within 24 h of incubation, whereas at 100 μg/mL, it completely inhibited the B-9987 growth (Figure S20A). Then, the bacterial growth began to recover from 36 h of incubation (Figure S20A), at which timepoint Fr. BAEs had been extensively degraded (Figure S19). In the concentration range of 0.1–10 μg/mL, the Fr. BAEs demonstrated neither biofilm enhancement nor the growth inhibition in B-9987 (Figure S20A). Fr. gBAEs enhanced the biofilm formation in B-9987 at 0.01 μg/mL as well but did not show notable toxicity toward the cells up to 100 μg/mL (Figure S20B), indicating that the glycosylation has a detoxication effect on the host strain. Similar phenomena were observed when the bacillaene fractions were treated to another bacillaene-producing Bacillus sp. 5746 strain (Figure S21).

Figure 2

Microtiter plate assay of biofilm formation in the B. methylotrophicus B-9987 strain by addition of compounds 1 and 2 at concentrations of 0, 0.1, and 1 μg/mL, respectively. The B-9987 strain was grown in MSgg medium at 37 °C for 18 h.

Upon these results, we deduced that through long time evolution, the Bacillus strains might encode bacillaenes as important signal molecules contributing to self-biofilm formation at low SICs. Then, at inhibitory concentrations, bacillaenes resist outside pathogens or competitors. When the production of bacillaenes increased to poison the host strain, detoxication mechanisms of intracellular glycosylation together with extracellular photo/oxidative degradation might be inspired. More interestingly, bacillaenes are “pseudo poisons” for the host cells that can be recovered after compound degradation. Conclusively, the antibiotic and biofilm enhancement of bacillaenes are two sides of a coin, both of which serve in the self-protection of Bacillus.

Experimental Section

General Experimental Procedures

1D and 2D (DQF-COSY, HSQC, HMBC, and NOESY) NMR spectra were recorded on Bruker AVANCE III 600 spectrometers at 274 K. Chemical shifts were reported with reference to the respective solvent peaks and the residual solvent peaks (δH 3.31 and δC 49.0 for CD3OD). The NMR spectra were processed using the MestReNova 6.1.1 program. The Agilent series 1290 HPLC system equipped with Agilent 6430 triple quadrupole mass spectrometry was used for the quantitative LC–MS analysis. Analytical HPLC was performed on an Agilent 1260 Infinity apparatus with a diode array detector. Preparative HPLC was performed on a Hitachi Chromaster System.

Bacterial Strains and Culture Conditions

The wild-type and ΔbmmB mutant strains of B. methylotrophicus B-9987 (CGMCC no. 2095) (Figure S22) have been described previously.[12,16,17]Bacillus sp. B-5746 (Figure S22) was provided by Prof. Jianhua Ju (South China Sea Institute of Oceanology, Chinese Academy of Sciences). Bacillus strains were cultured in Luria-Bertani (LB) or Landy medium at 37 °C, 200 rpm. When appropriate, chloramphenicol (Chl; 5 mg/mL for Bacillus) and erythromycin (Erm; 5 mg/mL for Bacillus) were added to the medium.

Preparation of Fr. BAEs and Fr. GBAEs

The ΔbmmB strain was cultivated in Landy medium at 37 °C (220 rpm) for 12 h. The combined culture broth (9 L) was extracted with EtOAc and concentrated by a vacuum evaporator. The EtOAc extract was partitioned between equal volumes of 90% MeOH and n-hexane to remove the nonpolar components. Then, the MeOH layer was concentrated and subjected to a stepped-gradient open column chromatography (ODS-A, 120 Å, S-30/50 mesh), eluting with 30, 35, 40, 45, and 50% ACN to yield five fractions. Then, each fraction was subjected to HPLC analysis with a linear gradient from 20 to 70% ACN/H2O in 60 min (YMC-Triart C18 column 150 × 4.6 mm, i.d. 5 μm; wavelength: 345 nm; flow rate: 1 mL/min). The gBAEs and nonglycosylated bacillaenes were separated completely by this fractionation protocol, which were concentrated in 40%-ACN and 45%-ACN fractions, respectively. The entire experimental process was strictly conducted in the dark.

Isolation of the Compounds

Isolation of the pure bacillaenes was conducted as described in our previous work.[9] Fr. BAEs were subjected to preparative HPLC (YMC-Pack ODS-A column 250 × 20 mm) eluting with 72% MeOH (flow rate: 4 mL/min) at 345 nm to afford bacillaene A (1), and Fr. gBAEs afforded bacillaene B (2) and 14′,15′-dihydro-bacillaene B (3). As the pure compounds 1–3 were very easily degraded during the vacuum evaporation process, each compound eluent was concentrated in batches (<2 mL/flask) to prepare PS. The DTS of compounds 2 and 3 were prepared by vacuum evaporation of each compound eluent for 20 mL/flask at one time. The entire experimental process was conducted in the dark.

Standard Curves for Compound Quantification

Different volumes (10, 30, 50, 70, and 90 μL for compound 1; 5, 10, 20, 40, and 80 μL for compounds 2 and 3) of compound eluents obtained by the above HPLC purification method were injected to analytical HPLC (YMC-Pack ODS-AQ column 250 × 20 mm) with a linear gradient from 30 to 70% B/A in 30 min (phase B: 100% MeOH; phase A: H2O; flow rate: 1 mL/min) at the wavelength of 345 nm. The peak area of each injected compound eluent in the HPLC profiles was measured. Then, a portion of each compound eluent was evaporated completely and weighed to get the accurate concentration. The standard curves of pure compounds were established by comparing peak areas with the concentrations of the compounds.

Time- and Dose-Dependent Stability Tests

Compounds 1 and 2 were prepared into 10, 1, and 0.1 μg/mL of aqueous solutions using the corresponding standard curve, respectively, and incubated at 37 °C for multi-timepoints (0, 18, 36, and 48 h) in the dark. The Agilent series 1290 HPLC system equipped with Agilent 6430 triple quadrupole mass spectrometry was used for the quantitative LC–MS analysis. The separations were carried out on a reversed-phase Thermos Hypersil GOLD C18 column (100 × 2.1 mm, 1.9 μm, 175 Å, Thermo Scientific Inc., USA) with a linear gradient from 65 to 100% B/A in 9 min for compound 2 and 15 min for compound 1 (phase B: 100% MeOH; phase A: H2O; wavelength 345 nm; flow rate: 0.2 mL/min). Fr. BAEs were prepared into 100, 10, and 1 μg/mL of aqueous solutions and incubated at 37 °C in the dark for multi-timepoints (0, 18, 24, 36, 48, and 60 h). HPLC analysis was performed on an Agilent 1260 HPLC system with a linear gradient from 35 to 100% B/A in 20 min (phase B: 100% MeOH; phase A: H2O; YMC-Pack ODS-A column 150 × 4.6 mm, i.d. 5 μm; wavelength: 345 nm; flow rate: 1 mL/min). The programs Thermo Xcalibur 2.2 and Agilent ChemStation for LC 3D systems Rev. B. 04. 03 were used for data analysis.

Biofilm Assay

A seed culture of the B-9987 or B-5746 strain was grown in LB medium to an optical density (OD600) of 1.0 at 600 nm. Then, the seed solution was diluted to 1:1000 with sterile deionized water, and 15 μL of the diluted culture and 1.5 μL of the different concentrations of the test compounds were added to 1.5 mL of MSgg medium contained in a well of a 24-well microtiter dish. The culture dish was incubated at 37 °C for the specified time.