Isolation of an antimicrobial compound produced by bacteria associated with reef-building corals

Bacterial isolation

Healthy colonies of the corals Pocillopora damicornis, Acropora millepora and Montipora aequituberculata (one colony per species) were collected in November 2011 from Davies Reef, Great Barrier Reef, Australia (latitude, 18°51′S; longitude, 147°41′E, Great Barrier Reef Marine Park Authority permit G12/35236.1) and maintained in aquaria for six days at the Australian Institute of Marine Science (Townsville, Queensland, Australia). Five replicate coral fragments (approximately 25 mm in length, containing 60–70 polyps) were collected from each colony and washed in sterile artificial seawater (ASW) to remove loosely attached microbes. Tissue slurries were produced by airbrushing (80 lb/in²) each coral fragment into 5 mL of ASW to remove coral tissues and associated microbes. These tissue slurries were homogenized to break down tissue clumps, and a dilution series was plated immediately on bacteriological agar (1%) in 1 L ASW supplemented with 0.3% casamino acids and 0.4% glucose (Hjelm et al., 2004). After two days of incubation at 28 °C, single colonies were transferred into Marine Broth (MB; Difco, BD, Franklin Lakes, NJ) and grown overnight. Liquid cultures were re-plated on minimal marine agar and the procedure was repeated until pure cultures were obtained.

Well diffusion assay with bacterial isolates

Fifty bacteria isolated from the coral tissue slurries of the three species (A. millepora = 16, P. damicornis = 17, M. aequituberculata = 17) were tested for growth-inhibitory activity against the known coral pathogens Vibrio coralliilyticus P1 (LMG23696) and V. owensii DY05 (LMG25443) in a well diffusion agar assay. In brief, the Vibrio strains were seeded into two different batches of minimal marine agar (after the agar temperature cooled to 40 °C). Following solidification, wells (diameter 5 mm) were cut into the agar and loaded with 20 μL of overnight cultures (10⁸ cells/mL) of the test isolates grown in MB (28 °C, 170 rpm). Plates were incubated at 28 °C and monitored every 24 h for a period of 72 h for inhibition zones. Phaeobacter strain 27-4 was used as a positive antagonistic control on each plate because of its broad spectrum inhibitory activity against Vibrio (Bruhn, Gram & Belas, 2007; Hjelm et al., 2004).

DNA extraction, gene sequencing genomic analyses

One strain, P12 isolated from Pocillopora damicornis, produced the strongest growth-inhibitory activity against the two target Vibrio strains. High molecular weight genomic DNA from P12 was extracted using a miniprep phenol-chloroform based extraction. Briefly, 5 mL of overnight liquid culture of P12 (10⁸ cells/mL) were spun in a micro-centrifuge (10,000 rcf) for 2 min. The pellet was then resuspended in 567 μL of TE buffer, 30 μL of 10% SDS and 3 μL of 20 mg/mL proteinase K. The tube was shaken thoroughly and incubated for 1 h at 37 °C. One hundred microliters of 5 M NaCl was subsequently added and the sample thoroughly mixed before adding 80 μL of CTAB/NaCl (10% CTAB in 0.7 M NaCl). The solution was incubated for 10 min at 65 °C, extracted with an equal volume of phenol/chloroform/isoamyl alcohol and centrifuged for 10 min (10,000 rcf). The supernatant was then extracted with an equal volume of chloroform/isoamyl alcohol and centrifuged again for 10 min. The aqueous phase was transferred to a new tube, DNA precipitated with equal volume of ice-cold isopropanol, washed with 70% ethanol and dried.

The near complete 16S rRNA gene of the strain was PCR amplified with bacterial specific primers 63F and 1387R, as outlined in Marchesi et al. (1998). Amplified PCR products were visualized by electrophoresis on 1% agarose gel stained with ethidium bromide. The amplified DNA was dried in a vacuum centrifuge (Savant DNA 120) and sequenced (Macrogen, Inc., Seoul, Korea). The 16S rRNA gene sequence of isolate P12 was used for phylogenetic comparisons and Maximum Likelihood trees were constructed using the ARB software.

We produced a draft genome assembly of P12. A paired-end library was prepared using the Illumina Truseq protocol (Illumina, San Diego, CA, USA), with an insert size of 169 bp and a read size of 150 bp. The library was sequenced on an Illumina MiSeq instrument at Monash University (Melbourne, Australia). The genome was assembled with the SPAdes assembler (v2.4.0) (Bankevich et al., 2012) and annotated with the Prokka software (v1.5.2) (Seemann, 2014). The presence of the genes involved in DMSP metabolism (dmdA, dddD, dddL, dddP, dddY, dddQ, dddW) and TDA production (tdaA-tdaH) was investigated by searching for homologs of the corresponding genes using reciprocal best blast hits.

DMSP metabolic capabilities of the isolate P12

Two different minimal media were used to examine the DMSP metabolic capabilities of P12: a modified marine ammonium salt medium (MAMS) (Raina et al., 2009) lacking a carbon source, and a modified basal salt medium lacking a sulfur source (Fuse et al., 2000) (25 g of NaCl, 0.7 g of KCl, 0.05 g of KH₂PO₄, 1 g of NH₄NO₃, 0.2 g of MgCl₂·H₂O, 0.02 g of CaCl₂·2H₂O, 0.005 g of FeEDTA, 1 g of Tris, 5 g of sodium succinate, 1.35 g of glucose in 1 L of distilled water). DMSP was added to both media (1 mm), acting either as the sole carbon or sulfur source. Five milliliters of each culture media were inoculated in triplicate with single P12 colonies and incubated at 28 °C for six days. Negative controls containing only the basal media and DMSP were used to account for possible chemical breakdown of DMSP. Bacterial growth was assessed via optical density measurement (NanoDrop, Thermo Fisher, Waltham, MA, USA). DMSP metabolism was assessed by ¹H Nuclear Magnetic Resonance spectroscopy (NMR). Methanol (CH₃OH; 40 mL) was added to each culture tube, the mixture shaken vigorously and sonicated for 10 min before being dried in vacuo using a rotary evaporator (Buchi, Flawil, Switzerland). The dried extracts were resuspended in a mixture of deuterium oxide (D₂O; 250 μL) and deuterated methanol (CD₃OD; 750 μL) (Cambridge Isotope Laboratories, Andover, MA, USA). A 750-mL aliquot of the particulate-free extract was transferred into a 5-mm Norell tube (Norell Inc., Landisville, NJ, USA) and analyzed immediately using quantitative NMR (Tapiolas et al., 2013).

Preparation of crude extracts for antagonist assays

An overnight culture of P12 (8 mL) was used to inoculate 4 × 250 mL of MB (total culture volume = 1 L). Bacterial cells were incubated for two days at 28 °C (120 rpm); the culture broth was then acidified to pH 2 with sulphuric acid before being exhaustively extracted with ethyl acetate (3 × 1.5 L). The extract was washed three times with MilliQ H₂O and dried in vacuo using a rotary evaporator (Buchi). The dried extract was then weighed and resuspended in CH₃OH (which was chosen for its ability to solubilize a wide range of compounds, its volatility and its innocuity in small volume towards both V. coralliilyticus and V. owensii) and tested in well-diffusion assays to confirm the extraction of the antimicrobial compound(s).

Purification and characterization of active compound

Purification of the crude extract was carried out using solid phase extraction on a reversed phase C₁₈ flash vacuum column (Septra C₁₈-E, Phenomenex, Torrance, CA, USA). Eleven fractions were eluted sequentially with 20, 40, 60, 80 and 90% CH₃OH in H₂O and 100% CH₃OH, followed by 20, 50 and 100% dichloromethane (CH₂Cl₂) in CH₃OH, 40% hexane in CH₂Cl₂ and finally 100% hexane. Each fraction was dried and resuspended in CH₃OH (1 mg mL⁻¹). Well diffusion assays were prepared as described above. On each plate, test wells were inoculated with 20 μL of each chromatographic fraction, or 20 μL of CH₃OH as a control, and Vibrio growth monitored. The most active faction (80% CH₃OH) presented an intense yellow color. Fine orange-red needles were crystallized from this active fraction to yield compound 1 (2.1 mg, 1.7% dry weight of organic extract).

NMR and FTMS analysis

Identification and structural elucidation of compound 1 was achieved using liquid chromatography–mass spectrometry (LC-MS), NMR, and Fourier Transform mass spectrometry (FTMS). Likewise these techniques were used to monitor for the presence of compound 1 in extracts and fractions. LC-MS analyses were performed on a Thermo Fisher Scientific Ultra High Performance Liquid Chromatography system connected to an LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific, San Jose, CA, USA). Samples were separated on a ACQUITY UPLC BEH RP-C₁₈ column (130 Å, 1.7 μM, 2.1 × 100 mm, solvents A = aqueous 0.1% formic acid and B = acetonitrile, gradient elution 80% A: 20% B for 0.5 min ramped up to 100% B over 10 min, then held for 4 min, 400 μL) and detected by positive mode electrospray ionisation using two different m/z ranges: 150–1,500 and 170–400. ¹H and ¹³C NMR spectra of compound 1 were acquired in a 5 mm 509-UP Norell NMR tube on a Bruker Avance 600 MHz NMR spectrometer (Bruker, Germany) with a TXI cryoprobe using standard Bruker pulse sequences. NMR spectra were referenced to residual ¹H and ¹³C resonances in deuterated chloroform (CDCl₃). High resolution mass spectra of compound 1 were measured with a Bruker BioApex 47e FTMS fitted with an Analytica of Branford ESI source; ions were detected in negative mode within a mass range m/z 200–1,000 via direct infusion at 120 μL h⁻¹.

Temperature-dependent activity

The antimicrobial activity of P12 grown at 32 °C (upper limit of coral thermal tolerance) was compared to that of the control incubated at 28 °C. The two cultures were grown overnight in MB at the two different temperatures, and their densities were determined by flow-cytometry (BD Accuri C6, Beckman Coulter, Brea, CA, USA). Cell numbers were normalized prior to inoculation into agar wells, and their activities against the two pathogens were compared using well-diffusion assays as described above. The same procedure was repeated with compound 1: two vials containing equal concentrations (2 μM of 1 in CH₃OH) were incubated overnight at 28 or 32 °C and their antimicrobial activities compared using the well diffusion assay.

Preparation of coral extracts

The coral species Montipora aequituberculata, M. turtlensis, Pocillopora damicornis, Acropora millepora, and Porites cylindrica (one colony each; 500 g of dry skeleton per species) were collected in July 2012 from Orpheus Island, Great Barrier Reef, Australia (latitude, 18°35′S; longitude, 146°20′E, Great Barrier Reef Marine Park Authority permit G12/35236.1). Coral tissues were airbrushed (80 lb/in²) into 0.2 μM filtered seawater (FSW) (total volume = 500 mL), acidified to pH 2 with sulphuric acid and the solution exhaustively extracted with equal volumes of ethyl acetate (3 × 750 mL). The combined organic layers were partitioned with MilliQ H₂O, dried and tested in well-diffusion assays, as previously described for the bacterial isolate extracts. The extracts of those coral species that exhibited antimicrobial activity were subsequently fractionated as described above for the crude extract from P12 and tested in well-diffusion assays. The active fractions were analyzed using ¹H NMR, FTMS and LC-MS.

Isolate P12: antimicrobial production, taxonomy and metabolic capabilities

A total of 50 coral-associated bacterial isolates were obtained from tissue slurry homogenates of the three coral species. Twelve of the 50 strains tested against the two pathogenic Vibrios (V. coralliilyticus and V. owensii) inhibited their growth in well diffusion assays. The bioactive isolate that exhibited the strongest in vitro activity against both pathogens, isolate P12, originated from Pocillopora damicornis and produced growth inhibition zones of 5 mm (±0.07 mm, n = 20) against V. owensii and 2 mm (±0.09 mm, n = 20) against V. coralliilyticus. The activity of P12 was temperature-dependent (Figs. 1A and 1B) and was significantly reduced when grown at 32 °C compared to 28 °C (Unpaired T-Test, n = 20, df = 38, t = 30.61, *p < 0.001 for V. owensii and n = 20, df = 38, t = 10.49, *p < 0.001 for V. coralliilyticus; Fig. 1C). Based on its bioactivity, the isolate P12 was selected for bioassay-guided fractionation.

According to its 16S rRNA gene sequence (NCBI accession number: KX198136), isolate P12 is an alphaproteobacterium belonging to the Rhodobacteraceae family and the Pseudovibrio genus. Its most closely related species is Pseudovibrio denitrificans (100% identity to the type strain; Fig. 2). Like other P. denitrificans strains (Enticknap et al., 2006), P12 colonies formed brown mucoid colonies when grown on Marine Agar. The brown color was absent when the strain was grown on minimal marine agar, with colonies appearing white. This strain was able use DMSP as either a sole carbon or sole sulfur source (Fig. 3). The complete utilization of DMSP from the liquid media after 2–3 days of incubation, as well as the presence of its metabolic byproduct dimethylsulfide (DMS), were confirmed by ¹H NMR. However acrylate, another possible byproduct of DMSP metabolism, was not observed.

Among the seven different DMSP degradation pathways currently identified (Moran et al., 2012), the full DMSP cleavage pathway (dddD, dddB, dddC, dddT, dddR; Table 1), involved in the conversion of DMSP into DMS without formation of acrylate (Todd et al., 2007) (Table 1), was identified in P12. We also identified possible orthologs for the demethylation pathway (dmdA, dmdB, dmdC and dmdD) used by marine bacteria to assimilate sulfur from DMSP, though these gene have low sequence identity to the genes originally identified in Ruegeria pomeroyi DSS-3 (Howard et al., 2006; Reisch et al., 2011) (Table 1). The presence of these two gene pathways corroborates the ¹H NMR measurements: the observed production of DMS without acrylate formation following DMSP metabolism (DddD pathway); and the ability to use DMSP as sole sulfur source (DmdA pathway) (Table 1).

Identification of antimicrobial compounds produced by P12

Well diffusion assays revealed that the crude extract from P12 retained the antimicrobial properties of the strain against both Vibrio species. Purification of the active fractions using reverse phase liquid chromatography yielded compound 1: optically inactive orange-red crystals; 2.1 mg (1.7% dry weight); IR (film) ν_max 3,420, 1,660, 1,280 cm⁻¹; UV (PDA, CH₃OH) λ_max 512 nm; ¹H NMR spectrum (600 MHz, CD₃Cl): δ 7.12, 7.44, 7.45 and 16.7; ¹³C NMR (150 MHz, CD₃Cl): δ 120.3, 132.0, 136.0, 138. 7, 149.5, 168.7, 171.7, and 183.5; HRESIMS m/z found 210.9534 (calculated for C₈H₃O₃S₂⁻ 210.9529, Δ 2 ppm). Combined spectroscopic techniques revealed that compound 1 was tropodithietic acid (TDA) (Brinkhoff et al., 2004; Penesyan et al., 2011) (Fig. 4A).

Orthologs for 11 genes involved in TDA biosynthesis (Geng et al., 2008) were present in the Pseudovibrio sp. P12 genome (Table 1). The biosynthesis of TDA correlated with production of the yellow-brown pigmentation in the culture medium and antimicrobial activity, similar to that previously reported (Brinkhoff et al., 2004; Bruhn et al., 2005; Porsby, 2010). Both coral pathogens were highly sensitive to TDA, with the pure compound still visually inhibiting their growth at 0.5 μg/mL (2.35 μM; Fig. 4B). In contrast to the decrease in antimicrobial activity exhibited by Pseudovibrio sp. P12 after incubation at 32 °C, TDA activity was not affected by exposure to this temperature (Unpaired T-Test, n = 20, df = 38, t = −0.94, p = 0.355 for V. owensii and n = 20, df = 38, t = 0.632, p = 0.531 for V. coralliilyticus; Figs. 1B and 1C).

Investigating the presence of TDA in coral samples

All of the extracts derived from the coral species investigated exhibited antimicrobial activity against the two pathogens, with the inhibition zones for P. cylindrica, M. aequituberculata, M. turtlensis and P. damicornis ranging from 3–5 mm in radius whilst the inhibition zones for A. millepora were much smaller (1 mm on average). ¹H NMR, LC-MS and FTMS analyses of the extracts and the active fractions of all coral species did not confirm the presence of TDA. The purified TDA could be detected by LC-MS in femtomolar concentrations when the coral fractions were artificially spiked, indicating that this lack of detection was not due to preferential ionization. Thus, TDA was either not present in the coral fractions tested or in concentrations below the LC-MS detection threshold.