Biosynthesis of Ditropolonyl Sulfide, an Antibacterial Compound Produced by Burkholderia cepacia Complex Strain R-12632

A library of 8304 transposon mutants was generated and the corresponding methanol extracts were screened for loss of inhibition of K. pneumoniae LMG 2095. Mutants showing a loss of activity (n = 658) or reduced activity (n = 139) against this indicator pathogen were retested using an overlay assay (Fig. 1). A total of 15 mutants showed complete loss of antibacterial activity, and an additional 22 mutants showed reduced activity (Table 1). Inhibition zone diameters for wild-type strain R-12632 and the 37 mutants with reduced or no antibacterial activity are presented in Table S2 in the supplemental material.

The complete genome sequence of the wild-type strain R-12632 was 8.11 Mbp in size and consisted of three large replicons (3.58 Mbp, 3.05 Mbp, and 1.35 Mbp) and one smaller plasmid (140 kb) with a total of 7,191 protein-coding genes. The R-12632 genome was analyzed using antiSMASH to identify known specialized metabolite gene clusters. A total of 20 biosynthetic gene clusters were predicted, of which 5 were located on the first replicon, 8 on the second replicon, 7 on the third replicon, and none on the plasmid (Table 2). Two NRPS clusters responsible for the production of the siderophores ornibactin and pyochelin were identified, as well as the biosynthetic genes for pyrrolnitrin production. The third replicon included one PKS gene cluster and two large hybrid NRPS-PKS gene clusters, none of which could be linked to the production of a known compound in the database. Finally, the genome of strain R-12632 contained six clusters encoding terpenes, three encoding bacteriocins, two encoding aryl polyenes, and a single cluster for each of the following functional classes: lasso peptide, ectoine, homoserine lactone, and phosphonate (Table 2).

In the 15 mutants that lost antibacterial activity, nine genes were disrupted by a transposon insertion (Table 1). Two mutants with an identical insertion site were identified for BCCR12632_00471, encoding an enoyl coenzyme A (enoyl-CoA) hydratase, and BCCR12632_02547, encoding a 5′-phosphosulfate (APS) reductase. Five mutants with distinct insertion sites were identified for BCCR12632_02955, encoding a glutathione synthetase. Through comparison with KEGG metabolic pathways, BCCR12632_02545, BCCR12632_02546, BCCR12632_02547, and BCCR12632_02549 were identified as part of the assimilatory sulfate reduction pathway, while BCCR12632_00250 and BCCR12632_00471 are involved in phenylacetic acid (PAA) catabolism (Table 1). BCCR12632_00396 is part of a type VI secretion system that is encoded by BCCR12632_03388 through BCCR12632_00404. While BCCR12632_00424 could not be linked to any known metabolic pathway, the gene immediately downstream (BCCR12632_00423) is predicted to encode a glutaredoxin.

In the 22 mutants with reduced antibacterial activity, 10 genes were disrupted (Table 1). Multiple mutants with unique insertion sites were identified for BCCR12632_00209 (n = 2), BCCR12632_00598 (n = 4), BCCR12632_02540 (n = 2), BCCR12632_02888 (n = 2), and BCCR12632_05920 (n = 2). BCCR12632_02540 encodes a cytosol aminopeptidase and is involved in glutathione metabolism. BCCR12632_00598, BCCR12632_02305, BCCR12632_02740, and BCCR12632_05920, encoding a glutamate dehydrogenase, an NADH:ubiquinone oxidoreductase subunit, a succinyl-CoA synthetase subunit, and a malate dehydrogenase, respectively, could be linked to energy metabolism (Table 1). For the five other genes, no link with known metabolic pathways was found.

Results from our earlier study (21) suggested that ditropolonyl sulfide was likely responsible for the observed antibacterial activity of strain R-12632. That study also showed that ditropolonyl sulfide could not be detected in crude agar extracts of strain R-12632 and that semipreparative fractionation combined with liquid chromatography–high-resolution mass spectrometry (LC-HRMS) was necessary to detect this compound. Therefore, wild-type strain R-12632 and three loss-of-activity mutants (R-75390, R-75501, and R-75718) were subjected to semipreparative fractionation to detect ditropolonyl sulfide production. Each selected mutant was impaired in a different metabolic pathway: glutathione metabolism (R-73590), PAA catabolism (R-75501), and assimilatory sulfate reduction (R-75718) (Table 1).

Wild-type strain R-12632 produced a characteristic green-brown pigment when grown on BSM-G (basal salts medium supplemented with glycerol) agar, which was not observed in any of the loss-of-activity mutants (Fig. S2). Mutant R-75718 was unable to grow on BSM-G agar, but supplementation of the growth medium with 0.5 mM l-cysteine allowed growth and restored characteristic green-brown pigment production and antibacterial activity similar to those of the wild-type (Table S2).

In wild-type strain R-12632, ditropolonyl sulfide presented as a sharp UV peak with characteristic absorbance at 400 nm and with a retention time between 19 and 20 min (Fig. 2). No characteristic absorbance at 400 nm was observed in any of the fractions of loss-of-activity mutants R-75501 and R-75390, and the ion with exact mass of 275.037 m/z corresponding to ditropolonyl sulfide was not detected (Fig. 2). In mutant R-75718 grown on BSM-G plus cysteine, ditropolonyl sulfide presented as a broad UV peak with characteristic absorbance at 400 nm and with a retention time between 17 and 20 min (Fig. 2).

To confirm its antibacterial activity, the MIC and minimum bactericidal concentration (MBC) of purified ditropolonyl sulfide were determined for 13 resistant bacterial pathogens. Ditropolonyl sulfide demonstrated antibacterial activity against K. pneumoniae LMG 2095 and against carbapenem-resistant strains or colistin-resistant strains of A. baumannii, C. freundii, and E. coli (MIC = 16 to 64 μg/ml) (Table 3). Antibacterial activity was also observed against vancomycin intermediate-resistant Staphylococcus aureus (VISA) Mu50 and vancomycin-resistant Enterococcus faecium (VRE) LMG 17188 (Table 3). Bactericidal activity was observed toward K. pneumoniae LMG 2095 (MBC = 32 μg/ml), A. baumannii R-67512 (MBC = 64 μg/ml), and all colistin-resistant strains tested (MBC = 64 μg/ml); all other pathogens had an MBC above 128 μg/ml (Table 3).

In the present study, we identified ditropolonyl sulfide as the main molecule responsible for the broad antibacterial activity in Bcc strain R-12632. Results obtained with transposon mutants lacking antibacterial activity and comparison with data from earlier studies indicated that the PAA catabolic pathway likely provides the tropolone backbone in ditropolonyl sulfide. Loss of activity observed in mutants defective in assimilatory sulfate reduction and glutathione biosynthesis suggested that cysteine and glutathione are potential sources of the sulfur atom linking the two tropolone moieties through a yet-uncharacterized mechanism. Its inhibition of multiple problematic Gram-negative bacterial pathogens warrants further investigation into the biosynthesis and the biological role of this unusual antimicrobial compound.

B. cepacia complex strain R-12632 (also known as strain MVP-C1-16) was isolated from maize rhizosphere in Italy in 1996 (22). This strain and all derived mutants were routinely grown on tryptone soya agar (TSA; Oxoid, USA) at 37°C and maintained at −80°C in 15% glycerol. Super optimal broth (SOB) medium was used for preparation of electrocompetent cells and for electrotransformation of R-12632 (43). For the preparation of selective plates, a 50-μg/ml kanamycin (Sigma-Aldrich, USA) stock was freshly prepared in distilled water and filter sterilized using 0.22-μm cellulose acetate filters (Whatman, Germany). Both TSA and a basal salts medium supplemented with glycerol (BSM-G) (44) were used as the media for antimicrobial production during the specialized metabolite analyses. BSM-G had a pH of 7 and contained the following per liter: 3.24 g K₂HPO₄, 1.13 g NaH₂PO₄·2H₂O, 2 g NH₄Cl, 0.2 g MgSO₄·7H₂O, 0.1 g nitrilotriacetic acid, 4.7 g glycerol, 3 mg MnSO₄·H₂O, 3 mg ZnSO₄·7H₂O, 1 mg CoSO₄·7H₂O, 7.7 mg FeSO₄·nH₂O (n = 1.5), and 15 g agar. After autoclaving and cooling to 55°C, the BSM-G medium was supplemented with a filter-sterilized l-cysteine hydrochloride solution to a final concentration of 0.5 mM when necessary.

K. pneumoniae LMG 2095 was used for activity testing of the mutant library and was cultivated in cation-adjusted Mueller-Hinton II broth (MHB; BD, USA) for the liquid inhibition assay or Iso-Sensitest broth (ISB; Oxoid, UK) for the overlay assay, both at 37°C. Strains for MIC and minimum bactericidal concentration (MBC) determinations were cultivated in MHB at 37°C. These included A. baumannii LMG 10520, A. baumannii R-67512 (carbapenem resistant), C. freundii R-67508 (carbapenem resistant), C. freundii 2200212 (low-level colistin resistant), E. coli R-67506 (carbapenem resistant), E. coli 1101443 (low-level colistin resistant), E. coli CD_R02067 (high-level colistin resistant), K. pneumoniae LMG 2095, K. pneumoniae AN0127/CP (high-level colistin resistant), K. pneumoniae SA0523 (low-level colistin resistant), Enterococcus faecium LMG 17188 (vancomycin resistant), and Staphylococcus aureus Mu50 (vancomycin-intermediate methicillin resistant).

(i) Preparation of electrocompetent cells. Electrocompetent cells of strain R-12632 were generated using a modified version of the protocol previously optimized for electrotransformation of B. cenocepacia J2315^T (45). Two well-isolated colonies from a 24-h culture on TSA were inoculated in 5 ml SOB and incubated at 37°C and 200 rpm for 18 h. This overnight culture was diluted to an optical density at 590 nm (OD₅₉₀) of 0.01 in SOB in an Erlenmeyer flask. The culture was incubated at 37°C and 200 rpm for 4 h to an OD₅₉₀ of approximately 0.2. Next, the culture was cooled on ice for 5 min and centrifuged at 4°C and 5,000 × g for 8 min. The resulting cell pellet was carefully washed twice using ice-cold 0.5 M sucrose, followed by centrifugation as described above. The cells were resuspended in 500 μl ice-cold 0.5 M sucrose with 10% glycerol, aliquoted, and frozen at −80°C.

The EZ-Tn5 <R6Kγori/KAN-2> Tnp Transposome kit (Lucigen, USA) was used to generate transposon mutants of strain R-12632. An aliquot of 55 μl electrocompetent cells was mixed with 1 μl EZ-Tn5 <R6Kγori/KAN-2> Tnp Transposome in a 0.1-cm-gap electroporation cuvette (Sigma-Aldrich, USA) and subjected to a pulse of 1,800 V using the preset “Ec1” protocol on the MicroPulser electroporator (Bio-Rad, USA). Nuclease-free water was used as a negative control. The electroporated cells were immediately resuspended in 950 μl SOB supplemented with 20 mM glucose and incubated while standing at 37°C for 4 h to allow the expression of the kanamycin resistance gene. Cells were diluted 1:1 in SOB broth, plated onto selective SOB agar supplemented with 50 μg/ml kanamycin, and incubated at 37°C for 24 h. Colonies were transferred with sterile toothpicks to 96-well microtiter plates (MTPs; Greiner Bio-One, Austria) containing SOB and were incubated at 37°C for another 24 h. The resulting liquid cultures were preserved at −80°C in 15% glycerol in 96-well MTPs using a Viaflo 96-well pipetting robot (Integra Biosciences, Switzerland).

(i) Preparation of agar extracts. Each step of the screening for loss of antibacterial activity was performed in 96-well MTPs. Frozen glycerol stocks of mutant cultures were thawed and 10-μl stock cell suspensions were inoculated into 190 μl SOB using the Viaflo 96-well pipetting robot and incubated at 37°C for 24 h. A 25-μl volume of this preculture was inoculated onto a 50-μl TSA agar plug using the Viaflo 96-well pipetting robot and incubated at 28°C for 4 days to allow for specialized metabolite production. Uninoculated SOB was used as a negative control (no growth and no antibacterial activity) and a liquid culture of strain R-12632 was used as a positive control for growth and strong antibacterial activity. Specialized metabolites were extracted from the 4-day cultures by adding 200 μl methanol per well. After 2 h of extraction, methanol extracts were transferred to new 96-well MTPs and left to air dry overnight. Dried extracts were resuspended in 40 μl sterile ultrapure water and immediately used in a liquid inhibition assay to determine the presence or absence of antibacterial activity.

Based on earlier results (21) (Table S1), K. pneumoniae LMG 2095 was chosen as an indicator organism to screen for loss of antibacterial activity in the transposon mutant library. The frozen glycerol stock of K. pneumoniae LMG 2095 was streaked on TSA for checking purity and incubated at 37°C for 24 h. Two well-separated colonies were used to inoculate 10 ml MHB which was incubated at 37°C overnight. This overnight culture was diluted in MHB to an OD₅₉₀ of 0.1 and then again diluted 1:100 in MHB to obtain a suspension of approximately 5 × 10⁵ cells/ml. In a 96-well MTP, a 90-μl volume of this cell suspension was mixed with 10 μl resuspended methanol extract and incubated at 37°C for 24 h. Noninoculated MHB served as a sterility control, an extract of the wild-type strain R-12632 was used as a positive control for antibacterial activity, and an extract of noninoculated growth medium was used as a negative control (no antibacterial activity).

Due to the inherent turbidity of the resuspended extracts, it was not possible to use turbidity alone to measure growth of the indicator pathogen K. pneumoniae LMG 2095. Therefore, a 10-μl volume of a freshly prepared filter-sterilized 0.125-mg/ml resazurin (Sigma-Aldrich, USA) stock was added to each well and plates were shaken at 200 rpm for 1 min and incubated at 37°C in the dark for 1 h. Plates were then examined visually and any color changes were noted. Resazurin is blue in its original oxidized state, but it is irreversibly converted to the pink compound resorufin through metabolic activity. Blue wells indicated an absence of pathogen growth and were interpreted as presence of antibacterial activity in the corresponding methanol extract. Pink or purple wells indicated pathogen growth and were interpreted as a reduction or absence of antibacterial activity in the corresponding extract.

Transposon mutants for which the methanol extract did not inhibit pathogen growth (i.e., resazurin was converted from blue to purple or pink) were tested for antibacterial activity using an overlay method as described earlier (21) to confirm and quantify the results. In short, transposon mutants were grown on TSA at 28°C for 48 h and were subcultured twice prior to performing the overlay assay. For each strain, a dense suspension (approximately 1 × 10⁸ CFU/ml) was prepared and 3 μl was inoculated onto the center of a TSA plate. Plates were incubated at 28°C for 4 days to allow the production of antimicrobials and were then exposed to chloroform vapors for 5 min. A diluted overnight culture of the indicator strain K. pneumoniae LMG 2095 was added to 100 ml soft agar (ISB plus 1% agar) to obtain approximately 1 × 10⁵ CFU/ml. Chloroform-exposed plates were overlaid with this inoculated soft agar and incubated at 37°C for 24 h. Antibacterial activity was scored by measuring the diameter of the inhibition zone, where no pathogen growth was observed. Loss of antibacterial activity was defined as the absence of an inhibition zone, whereas reduced activity was defined as an inhibition zone diameter that was at least 25% smaller than the inhibition zone produced by the wild-type strain R-12632.

Genomic DNA of strain R-12632 and 37 selected mutants (Table 1) was prepared from a 48-h culture grown at 28°C on TSA. Cells were harvested using a 10-μl inoculation loop and suspended in 300 μl 4 M guanidine isothiocyanate solution (Fisher Scientific, USA). DNA extraction was performed using the Maxwell 16 automated nucleic acid purification system using the Maxwell 16 tissue DNA purification kit (Promega, USA) according to the manufacturer’s instructions. DNA was quantified using the QuantiFluor ONE double-stranded DNA (dsDNA) quantification system (Promega). A total of 3 μg genomic DNA of wild-type strain R-12632 was subjected to polyethylene glycol precipitation (final concentration: 5.5% polyethylene glycol 8000 [PEG 8000] and 0.8 M NaCl) to remove fragments shorter than 1 kb. Library preparation was performed using the ligation sequencing kit 1D (SQK-LSK109) according to the manufacturer’s instructions (Oxford Nanopore Technologies, UK). The library was sequenced for 3 h on a MinION R9.4 flow cell (Oxford Nanopore Technologies) to obtain approximately 700 MB of raw data. Reads were base called using Guppy 3.0.3 (Oxford Nanopore Technologies), retaining only reads with a qscore of >7, and quality reports were generated using NanoPlot (46). Short reads were generated for strain R-12632 and the 37 selected mutants by sequencing 150-bp paired-end libraries on an Illumina NovaSeq 6000 sequencer (Oxford Genomics Centre—Wellcome Centre for Human Genetics, Oxford, UK). Short reads were trimmed using fastp with –qualified_quality_phred 30 –length_required 50 –correction options, and adapters were automatically trimmed (47). Quality reports were generated using FastQC (48).

A hybrid assembly of Illumina and Nanopore long reads was performed using Unicycler (49). In the first step, Unicycler was used to build an initial assembly graph from Illumina short reads by using the de novo assembler SPAdes with a wide range of k-mer sizes. Long Nanopore reads were then aligned to the short-read assembly graph to bridge gaps between individual contigs and resolve repeats. The resulting assembly was subjected to several rounds of polishing by Pilon (50) using the high-quality short reads to reduce errors. Annotation was performed using Prokka v1.12 (51) with a genus-specific database based on reference genomes from the Burkholderia Genome Database v8.1 (52). Further functional annotation of protein-coding genes was performed using the KEGG Automatic Annotation Server (53), and KEGG pathways were visually reconstructed using KEGG Mapper (54). To confirm the correct annotation and functionality of genes in KEGG pathways, amino acid sequences were checked for the presence of conserved functional domains using the HMMER Web server (55). Gene orientation and operon organization were inspected visually using the Integrated Genome Viewer (56). antiSMASH 5.1.0 was used to mine genomes for biosynthetic gene clusters responsible for the production of specialized metabolites, including PKS, NRPS, siderophores, and bacteriocins (57).

The trimmed Illumina reads for the 37 selected mutants (Table 1) were assembled using shovill (58) with subsampling to 80× coverage depth and retaining only contigs longer than 500 bp. Ragout v2.2 (59) was used to scaffold the obtained contigs using the complete genome of strain R-12632 as the reference genome. Annotation was performed as described above. The insertion site of the transposon was determined via BLAST (60) and visually examined using Integrated Genome Viewer (56).

Crude extracts of the wild-type strain R-12632 and three loss-of-activity mutants grown on BSM-G (R-75390 and R-75501) or BSM-G supplemented with l-cysteine (R-75718) were prepared as described previously (21). Extracts were dissolved in 100% dimethyl sulfoxide (DMSO) and filtered (0.45-μm pore size) prior to semipreparative fractionation using a Gilson GX-281 322H2 instrument (Gilson Technologies, USA). Each extract was subjected to semipreparative reversed-phase high-performance liquid chromatography (HPLC) (Zorbax SB-C18 column, 9.4 by 250 mm, 5-μm particle size, 3.6 ml/min, UV detection at 210 and 400 nm) eluting with acetonitrile/water, in a linear gradient from 5% to 100% acetonitrile in 45 min and yielding 1.8 ml per fraction every 0.5 min to generate 80 central fractions. Based on the retention time and the obtained UV absorbance at 210 and 400 nm, four to six fractions per sample were subjected to LC-HRMS as described above to determine the presence of ditropolonyl sulfide. The extracted-ion chromatogram at m/z 275 was examined for the ion with exact mass of 275.037 ± 0.005 m/z, corresponding to ditropolonyl sulfide.

(i) Purification of ditropolonyl sulfide. For the extraction and purification of ditropolonyl sulfide, a dense suspension of wild-type strain R-12632 was prepared as described above. A total of 1.3 liters BSM-G agar, contained within 13 100-ml agar plates, was inoculated with this suspension and incubated at 28°C for 4 days. For each plate, the agar was ground, transferred to 500-ml glass flasks, and extracted by adding an equal volume of acetone (100 ml) and shaking at 220 rpm for 2 h. After centrifugation at 7,500 rpm for 10 min and filtration over paper in a Büchner funnel, the acetone extract was concentrated under a heated nitrogen stream to a final volume of 1.3 liters (100% water). The aqueous residue was loaded onto an SP207ss resin column (65 g, 100 by 22 mm) for reversed-phase flash fractionation with a 10% step gradient of acetone in water starting from 10% acetone to 100% acetone in 25 min at 10 ml/min. Aliquots of every flash fraction were analyzed and the presence of ditropolonyl sulfide was determined by LC-HRMS and the characteristic UV spectrum. Those fractions that presented the compound were dried, redissolved in 100% DMSO, and filtered (0.2-μm pore size) prior to HPLC fractionation using a Gilson GX-281 322H2 instrument (Gilson Technologies, USA). Enriched flash fractions were subjected to two preparative reversed-phase HPLC batches (Zorbax SB-C8 column, 21.2 by 250 mm, 7-μm particle size, 20 ml/min, UV detection at 210 and 280 nm) eluting with acetonitrile/water, in a linear gradient from 5 to 100% acetonitrile in 45 min, yielding 10 ml per fraction every 0.5 min to generate 80 central fractions. Enriched HPLC preparative fractions, as determined by LC-HRMS, were subjected to six semipreparative reversed-phase HPLC fractionation batches (Zorbax RX-C8 column, 9.4 by 250 mm, 5 μm particle size, 3.6 ml/min, UV detection at 210 and 280 nm) eluting with a linear gradient from 20 to 25% of acetonitrile in water for 35 min, yielding 10 ml per fraction every 0.5 min to generate 80 central fractions. Fractions were analyzed by LC-HRMS for the presence of ditropolonyl sulfide, which eluted in a sharp peak at 15 min with a broad long tail of 8 min. ¹H nuclear magnetic resonance (NMR) was used to confirm the presence of ditropolonyl sulfide (Fig. S1) (61). The purification yielded, in total, 2 mg of the pure compound (98% purity).

MICs and MBCs of ditropolonyl sulfide against A. baumannii LMG 10520, A. baumannii R-67512, C. freundii R-67508, C. freundii 2200212, E. coli R-67506, E. coli 1101443, E. coli CD_R02067, K. pneumoniae LMG 2095, K. pneumoniae AN0127/CP, K. pneumoniae SA0523, Pseudomonas aeruginosa PAO1, E. faecium LMG 17188, and S. aureus Mu50 were determined using flat-bottom 96-well MTPs (TPP, Switzerland) as described previously (62). Ditropolonyl sulfide was dissolved in ultrapure Milli-Q water (Millipore, USA) and was filter sterilized using 0.22-μm cellulose acetate filters (Whatman, Germany). Ditropolonyl sulfide concentrations tested ranged from 0.25 to 128 μg/ml. Blank wells with corresponding concentrations of ditropolonyl sulfide were used to correct for absorbance of the compound. The lowest concentration of antibiotic for which a similar optical density was observed in the inoculated and the blank wells was recorded as the MIC. A 100-μl aliquot from the inoculated wells for which an optical density similar to that of the blank wells was observed was used to determine the presence of surviving cells by plating on MH agar. The MBC was defined as the lowest concentration for which no CFU were recovered after 48 h of incubation at 37°C.

The data sets generated for this study can be found in the European Nucleotide Archive under accession number PRJEB40633. The whole-genome sequence of strain R-12632 was deposited in GenBank under the accession numbers FR989675 (replicon 1), FR989676 (replicon 2), FR989677 (replicon 3), and FR989678 (plasmid).