Introduction

Corals and related organisms are intimately co-evolved ‘holobionts’, complex symbiotic organisms formed by the invertebrate animal, the dinoflagellate, Symbiodinium spp. and their associated bacteria. The stability and effectiveness of this symbiosis determine the health of the whole reef ecosystem and its resistance to stresses and diseases. The differences in the microbial species composition of asymptomatic and diseased corals (Gil-Agudelo et al., 2006; Ritchie, 2006; Bourne et al., 2008; Koren and Rosenberg, 2008) imply a role for coral microbiota in coral health, although little is known about the mechanisms which govern the structuring and functions of coral-associated microbiota.

Coral-associated microbial communities are controlled by multiple factors. Nutrients and signals excreted by the coral host with coral mucus contribute to the structuring and functioning of the microbiota. Coral mucus is a polymer made in the specialized cells of the polyp from the photosynthate produced by their dinoflagellate symbionts and then excreted onto the coral surface (Brown and Bythell, 2005). Mucus has at least two roles in the associated microbial communities. Bacterial receptors can bind to mucus (Rosenberg and Falkovitz, 2004; Kvennefors et al., 2008), and this directly controls the composition of the associated microbiota. The mucus polymer itself and small molecular weight compounds within it serve as both nutrient sources for the microbes and signals that modulate behavior and gene expression in the associated microbial communities (Vacelet and Thomassin, 1991; Wild et al., 2004; Ritchie, 2006; Sharon and Rosenberg, 2008; Krediet et al., 2009a, 2009b). Although the fate of coral mucus in the reef environment and its role in coral-associated microbial communities are becoming more clear, significantly less is known about the microbe–microbe interactions within the coral surface mucus layer.

Analysis of other host–bacterial systems suggests that microbe–microbe interactions within coral-associated microbial communities may have important functions in coral health and the interactions of native microbiota with pathogens. Models and experimental evidence to support this hypothesis are beginning to emerge (Shnit-Orland and Kushmaro, 2009; Mao-Jones et al., 2010). For example, culturable microbes associated with the elkhorn coral Acropora palmata produced an array of anti-bacterials against a broad spectrum of pathogens, including the necrotizing coral pathogen Serratia marcescens (Ritchie, 2006). Thus, the in situ production of anti-microbials by coral-associated native microbiota may be one of the mechanisms by which interactions within the coral surface mucus layer are controlled. It is reasonable to hypothesize that in addition to the antibiosis, cooperative and antagonistic interactions take place within coral-associated microbial communities (Teplitski and Ritchie, 2009). Mechanisms that govern such interactions are not yet clearly understood.

One of the mechanisms known to be involved in microbe–microbe interactions is the intercellular exchange of small diffusible chemical molecules. The cell-to-cell signaling and the resulting changes in gene expression that link bacterial behaviors to the increases in their population densities within diffusion-limited environments are known as ‘quorum sensing’ (QS) (Fuqua et al., 2001). The best-characterized QS signals are N-acyl homoserine lactones (AHLs) (Eberhard et al., 1981; Fuqua et al., 2001). In many Gram-negative bacteria, including members of the Serratia genus, QS controls surface spreading, production of antibiotics and exoenzymes, attachment to surfaces and timing of virulence gene expression (Van Houdt et al., 2007). Production of QS signals has also been reported in cultures of coral-associated vibrios, although the role of QS in coral diseases caused by these microorganisms has not yet been established (Tait et al., 2010). Because QS has important roles in the interactions within microbial communities, various forms of QS manipulation have been documented (Givskov et al., 1996; Pasmore and Costerton, 2003; Skindersoe et al., 2008; Dobretsov et al., 2009). How these interactions carry out in natural habitats is far from being clear. The goal of this study was to test whether bacteria associated with marine invertebrates, and their endosymbiotic dinoflagellates, produce cell-to-cell signals capable of affecting behaviors in opportunistic pathogens by manipulating inversely regulated multicellular behaviors.

Materials and methods

Bacterial strains, media and growth conditions

Bacterial strains used in this study are listed in Table 1. Bacteria were isolated from Symbiodinium cultures via serial dilution in liquid f/2 media (Sigma, St Louis, MO, USA) followed by plating onto either Glycerol Artificial Sea Water medium (GASW; 356 mM NaCl, 40 mM MgSO4, 20 mM MgCl2·6H2O, 8 mM KCl, 60 μM K2HPO4, 33 μM Tris, and 7 μM FeSO4, with 0.05% peptone, 0.2% yeast extract and 2.0% glycerol, (Ritchie, 2006)) or Marine Broth (Difco-Becton, Dickinson and Company, Franklin Lakes, NJ, USA). Individual bacterial isolates were subcultured to purification on Marine Agar. Symbiodinium cultures were provided by T LaJeunesse (Penn State University) and maintained in f/2 medium (Sigma) at 25 °C under a12-h light:dark cycle. Coral-associated bacterial isolates were grown in either GASW or Marine Broth. All purified marine-derived isolates were aliquoted into 96-well plates in 25% glycerol seawater for frozen storage (−80 °C). Reporter strains were cultured in Luria-Bertani (LB) broth (Fisher Scientific, Pittsburgh, PA, USA) with appropriate antibiotics. As needed, ampicillin was added to the final concentration of 100 μg ml−1, tetracycline at 10 μg ml−1 and gentamicin at 30 μg ml−1.

Table 1 Bacterial strains used in the study
Full size table

The identities of marine bacterial isolates were confirmed by first PCR-amplifying fragments of their 16S rRNA genes with primers 8F (pA) 5′-AGAGTTTGATCCTGGCTCAG-3′ and 1489R 5′-TACCTTGTTACGACTTCA-3′ (Edwards et al., 1989; Bruneel et al., 2006) followed by cloning of the PCR fragments into pCR2.1-TOPO, and sequencing of the resulting constructs and BLAST searching against the NCBI (GenBank) database. Sequencing was carried out at the University of Florida Biotechnology Core Facility. Data are deposited in Genebank (Accession numbers JF263527; JF263528; JF263529; JF263530; JF263531; JF263532; JF346758; JF346759; JF346760; JF346761; JF346762; JF346763; JF3467640).

Field sample collection and handling

Coral mucus samples were collected from apparently asymptomatic Acropora palmata colonies at Looe Key Reef, Florida (24° 32.764′ N: 81° 24.304 W) using a needleless syringe in July 2009 and April 2010, as previously described (Ritchie, 2006). Mucus was pre-filtered through glass fiber filter followed by filtration through a 0.22-μm MCE filter, aliquots of mucus were stored frozen at −20 °C.

To collect hydrophobic compounds present within coral mucus surface layer in situ, reverse-phase C18 Si resin was wetted in ethanol, equilibrated in high-pressure liquid chromatography-grade water and then added into regenerated cellulose dialysis tubing (Fisher Scientific), pre-treated as per manufacturer's instructions. In parallel, cellulose dialysis tubing was filled with Diaion styrenic adsorbent resin HP20SS-1 (Sorbent Technologies, Atlanta, GA, USA), which was pre-treated as per manufacturer's instructions. Tubes with resin were clamped, weighed down on coral surfaces with small dive weights on cotton twines (Supplementary Figure S1). As a control, similarly constructed tubes (with C18 reverse-phase Si or HP20SS-1 resin) were deployed on the sand bottom, approximately 10–15 m away from nearest living coral (Supplementary Figure S1). Tubing with resin was deployed at Looe Key Reef, Florida for 3 days in May 2010. Upon completion of the experiment, samples were brought up to the lab, resin was scooped into flash chromatography columns and eluted sequentially with one volume of chloroform, isopropanol, 100% methanol, then 75% and 50% aqueous methanol solutions. Fractions were collected, rotary evaporated to dryness and stored at −20 °C.

Detection of quorum-sensing active compounds

The ability of tested bacteria to affect well-characterized QS reporters was tested using colony overlays and culture filtrate extracts with a suite of biosensors. For colony overlays, collections of bacteria isolated from marine invertebrates or from cultures of their dinoflagellate symbionts were replica plated into microtiter wells containing GASW medium solidified with 1% agar. After 2–3 days incubation at 30 °C, wells were overlaid with the suspension of Chromobacterium violaceum CV026 in soft LB agar with C4-HSL, as described before (McClean et al., 1997). Inhibition of violacein production in this reporter strain was scored visually. Interesting candidates were then re-tested by growing them on glass fiber disks placed onto GASW agar. After 3–5 days, when growth was observed on the surfaces of glass fiber disks, they were lifted from GASW agar and transferred onto the surface of LB agar, which was overlaid with a suspension of the C. violaceum CV026 reporter in soft LB agar with C4-HSL.

In addition to the Chromobacterium assay, the presence of compounds affecting AHL reporters was detected using direct assays with lux-based bacterial biosensors E. coli JM109 pSB401, pSB536 or pSB1075 (Winson et al., 1998). For preliminary screens, 1 ml samples of overnight bacterial cultures of each tested strain were pelleted at 16 000 g. Approximately 10 mg of charged Sample Prep C18 resin, 35–75U (Alltech, Deerfield, IL, USA) were added to the supernatants and gently mixed. Once resin settled, supernatants were aspirated and discarded. A dilution series of this C18 resin slurry was added to the wells of a black polystyrene 96-well plate (Corning Scientific, Corning, NY, USA) and luminescent biosensors were inoculated into the wells and incubated at 37 °C. Biossays were carried out as described previously (Alagely et al., 2011). Luminescence was measured with a multilabel plate reader (PerkinElmer Victor3, Waltham, MA, USA).

Bioassay-guided characterization of QS activities

A total of 13 bacterial isolates with the ability to promote or inhibit AHL reporters were selected for further characterization using thin-layer chromatography (TLC) aided by the AHL reporter Agrobacterium tumefaciens NT1 pZLR4. To prepare extracts, 5 ml aliquots of the overnight starter culture of each isolate were added to 0.5-l shake cultures of GASW and incubated with constant agitation at 200 r.p.m. at 30 °C (we note that AHL production or accumulation in Marine Broth was generally lower). After 48 h (when cultures reached OD600=0.9–1.5), two half volumes of acidified ethyl acetate were gently mixed with each culture, and organic phase was separated using a separatory funnel. The organic phases were frozen at −20 °C to remove residual water, and the solvent was evaporated at 37 °C in Rotavapor R-200 (Büchi Laboratorues, Flawil, Switzerland). The extracts were re-dissolved in a small volume of ethyl acetate, then spotted on C18-reversed phase TLC plates (Whatman, Piscataway, NJ, USA) and developed with 60:40 methanol:water. The TLC plates were dried and overlaid with A. tumefaciens NT1 pZLR4 in 0.6% M9 agar supplemented with X-Gal (40 mg l−1) as in Cha et al. (1998). The plates were incubated overnight at 30 °C, then air-dried and photographed.

To test whether the QS signals detectable with the Agrobacterium bioassay share structural homology with known AHL, the extracts were treated with the AHL-lactonase AiiA. E. coli DH5α pDSKaiiA and the control strain carrying just the pDSK vector, were grown in 1.5 l of LB with kanamycin (50 μg ml−1) and isopropyl-β-D-thiogalactoside (20 μg ml−1) at 37 °C to OD600=1.6 as previously described (Dong et al., 2000; Gao et al., 2007). The cells were centrifuged and reconstituted in 30-ml LB, then lysed using a French press and centrifuged to remove cellular debris. Assay conditions were optimized using synthetic 3-oxo-C6-HSL (Sigma). For the treatment of bacterial extracts, 50 μl of the enzymatic preparation were added to 30 μl of dried extracts and incubated at 30 °C for 3 h. After incubation, the contents were extracted twice with 200 μl of ethyl acetate and subjected to the bioassay-aided TLC as described above.

Swarming assays with S. marcescens

AB swarm agar 0.4% (wt/vol) was briefly solidified and then superficial indentations 1/3 along the diameter of the plate were made with a sterile bore (d=1 cm). Sterile glass filter disks (d=0.75 cm) with 30 μl of overnight cultures of bacterial isolates (in GASWA) were placed in the recess. A total of 10 μl of S. marcescens cultures (subcultured in LB with antibiotics for 2 h) were spotted 2 cm away from the glass filter disk containing another isolate. This method was chosen to promote diffusion of chemical compounds from the marine isolates without direct initial contact with the tested strain.

Negative controls consisted of sterile GASW on the glass filter disk. The plates were incubated at 30 °C for 24–60 h, monitored for the progression of swarming and documented when the control S. marcescens swarms covered 50–75% of the plate. Each assay was repeated at least three times.

Swarming assays with S. marcescens PDL100 were more consistent if genetic analysis grade agarose (Fisher Scientific) was used instead of granulated molecular genetics bulk agar (Fisher Scientific). The former is typically used for agarose gel electrophoresis, and the latter is often used for laboratory media preparation. Swarm plates must be allowed to set at room temperature for 12 hrs before the assays (excessive dryness or wetness of the plates resulting from longer or shorter set times will confound the experiments). Further, subculturing the overnight culture for 2 hours with tetracycline and ampicillin before the assays resulted in a consistently observed swarming.

Biofilm assay of co-inoculation of marine isolates and S. marcescens PDL100 in coral mucus

A total of 50 μl of filter-sterilized coral mucus were added to the wells of a 96-well plate. After an hour, 40 μl were aspirated and the remaining mucus was left to dry in a laminar flow hood overnight as in Krediet et al., (2009a). The following day, overnight cultures of S. marcescens PDL100 and each of the marine isolates were diluted to OD600 of 0.4 and mixed 1:1 (S. marcescens PDL100: individual marine isolate) in sterile phosphate buffer solution and added to the wells to a total volume of 100 μl. After 48-h incubation, an aliquot of the supernatant was plated on Marine Agar (1.5% agar wt/vol). Each well was then washed three times with sterile phosphate buffer solution to remove un-attached bacteria; the remaining biofilms were dislodged using a sterile cotton swab, and then streaked to isolation on Marine Agar. The plates were incubated at 30 °C for 48 h and colonies were counted to determine the ratios between the coral pathogen and the marine isolates.

Co-inoculation experiments using marine isolates, S. marcescens PDL100 and a model polyp Aiptasia pallida

Clonal lines of Aiptasia pallida (from J. Pringle) were maintained in 10-gal saltwater aquaria at ambient temperature (22 °C) under blue actinic (460 nm) and super daylight white −6500 k fluorescent bulbs on a 12 h:12 h light:dark cycle. Polyps were fed weekly with brine shrimp. For the experiments, individual polyps were transferred into wells of six-well plates (Corning Scientific) with 10 ml of artificial seawater (Red Sea Coral Pro Salt, Eilat, Israel) sterilized by passing through a 0.2-μm filter. Polyps were acclimated in the wells for 2 days at room temperature on a shaker at 75 r.p.m. For the inoculations, overnight cultures of commensal strains were inoculated into filter-sterilized artificial seawater at 106 cfu ml−1; 10 ml of this suspension were added to the wells with A. pallida. After 24 h, the seawater with inocula was aspirated and replaced with filter-sterilized artificial seawater seeded with the overnight culture of S. marcescens PDL100 at 5 × 107 cfu ml−1. A. pallida sea anemones were photographed daily.

Results and discussion

Small molecular weight compounds present within coral mucus surface layer inhibit biofilm formation in S. marcescens

To survey biologically active substances that are present in situ on coral surfaces on the reef tract, we deployed ‘chemical traps’ consisting of cellulose dialysis tubing filled with reverse-phase Si or HP20SS-1 resin (Supplementary Figure S1). Hydrophobic substances were eluted with organic solvents, concentrated and tested for biological activity in the S. marcescens PDL100 biofilm assays. As shown in Figure 1, substances eluted with 75% methanol:25% water inhibited biofilm formation by the coral white pox pathogen S. marcescens PDL100. The corresponding fraction collected on the sand bottom did not have a statistically significant effect on the biofilm formation (Figure 1). Substances eluted from HP20SS with 100% methanol and 75% methanol similarly inhibited biofilm formation (data not shown). Other fractions did not inhibit biofilm formation. These observations indicate that the surface mucus layer of asymptomatic corals contains substances capable of inhibiting virulence-related behaviors in opportunistic pathogens.

Figure 1
figure 1

Inhibition of biofilm formation by Serratia marcescens PDL100 in the presence of compounds recovered from coral mucus surface layer. Left: Cellulose dialysis pouches containing reverse-phase C18 Si or HP20SS resin were deployed on surfaces of the boulder coral Montastraea faveolata and the compounds were eluted with a bed volume of chloroform, isopropanol, 100% methanol, 75% methanol and 50% aqueous methanol, evaporated, and reconstituted in 200 μl methanol. A total of 30 μl of serial 10-fold dilutions (left to right) were added to the wells of polystyrene microtiter plates where the pathogen was allowed to settle in colonization factor antigen media. The highest amount corresponds approximately to 37.5 cm2 of M. faveolata coral surface. Activities eluted from the reverse-phase C18 Si with 75% methanol:25% water exhibited noticeable inhibitory properties, shown here. Middle: As a control, signals present on the sand bottom 10–15 m from the nearest living coral were similarly tested. Right: Biofilm formation by S. marcescens PDL100 in the absence of extracts in colonization factor antigen media. Error bars are standard errors of four technical replications.

Full size image

As most marine invertebrates are ‘holobionts’ (complex symbiotic organisms formed by the polyp, endosymbiotic dinoflagellates and the associated bacteria (Rosenberg et al., 2007; Rosenberg and Zilber-Rosenberg, 2008)), the bioactive compounds detected with this bioassay could have originated from any of the partners within the holobiont. To begin the characterization of these activities, we focused on testing the hypothesis that commensal bacteria associated with either the surface mucus layer or with the endosymbiotic dinoflagellates contribute to the synthesis of the activities capable of disrupting virulence-related behaviors in a model opportunistic pathogen Serratia marcescens PDL100.

Screen and identification of coral-associated bacteria capable of QS manipulation

A library containing over 300 culturable bacterial isolates from mucus of the elkhorn coral Acropora palmata (Ritchie, 2006), other marine invertebrates and their dinoflagellate Symbiodium spp. symbionts was first screened with the Chromobacterium violaceum CV026 reporter. Isolates capable of either stimulating or inhibiting QS-mediated pigment production in the reporter were then subject to direct assays using semi-synthetic luminescent reporters based on the LuxR, AhyR and LasR AHL receptors from Vibrio fischeri, Aeromonas hydrophila and Pseudomonas aeruginosa (respectively). Approximately 4% of the tested bacteria were capable of affecting at least one QS reporter. Culture filtrates of isolates 44B9, 47H1 and 52E5 (purified with reverse-phase C18 silica resin) stimulated all three luminescent reporters, while similarly prepared culture filtrates from isolates 44F6, 47A11, 47E6 and 52B8 inhibited the LasR reporter, and stimulated the LuxR and AhyR reporters (data not shown). The former observation is consistent with the possibility that the three strains produce AHLs with medium acyl chains that trigger positive responses in the three luminescent reporters (Winson et al., 1998). Stimulation of the LuxR and AhyR reporters with the concurrent inhibition of the LasR reporter is less common, and the reverse is usually true (long-chain AHLs that stimulate LasR typically inhibit AhyR (Swift et al., 1999)), although some non-AHL QS inhibitors can have diverse effects on bacterial AHL receptors (rev. Teplitski et al., 2011). Alternatively, this observation could be explained by the production of multiple signals, capable of either inhibiting or inducing different AHL receptors independently. To test these working hypotheses bioassay-guided thin-layer chromatography was carried out with 13 bacterial isolates.

The thirteen isolates representing different marine invertebrates or their dinoflagellate symbionts were selected for further characterization. Based on partial sequencing of the 16S rRNA gene, the isolates were identified as members of the genera Photobacterium, Marinobacter, Vibrio, Caryophanon and Planomicrobium; the identity of five α-proteobacteria (Table 1) was difficult to ascertain due to the low homology of their 16S rRNA genes to the sequences deposited and annotated in GenBank. The α- and γ-proteobacteria have been previously reported as over-represented clades in coral-associated microbial communities (Ritchie, 2006; Kooperman et al., 2007; Wegley et al., 2007); photobacteria and vibrios are commonly found in association with healthy or diseased corals (Ritchie, 2006; Rosenberg et al., 2007; Tait et al., 2010). However, Marinobacter spp. have not been reported in coral-associated microbial communities. Marinobacter spp. are most commonly characterized as members of hydrocarbon-degrading consortia. AHL-producing strains of Marinobacter have been recovered from ‘marine snow’ (Gram et al., 2002).

Bioassay-guided characterization of QS signals

To better characterize QS-active substances and to separate the inhibitory from the stimulatory QS activities produced by the selected bacteria, organic extracts of their culture filtrates were subjected to the bioassay-coupled reverse-phase C18 Si TLC. The biosensor, Agrobacterium tumefaciens NT1 pZLR4, produces β-galactidosidase in response to AHL with C6–C18 acyl side chains (Cha et al., 1998). Most of the tested strains produced one or two separable activities, which co-migrated with AHLs of medium (C6–C8) length (Figure 2). This lack of complexity in QS signal profiles of the tested bacteria is different from the reports, in which α- and γ-proteobacterial associates of sponges and marine dinoflagellates were shown to produce multiple AHLs, including signals with unusually long acyl side chains (Wagner-Dobler et al., 2005; Mohamed et al., 2008).

Figure 2
figure 2

Production of QS signals by bacteria isolated from marine holobionts. Bacterial cultures were extracted twice with half volumes of ethyl acetate acidified with glacial acetic acid (4 ml l−1), and concentrated by rotary evaporation. Samples were spotted on reverse-phase C18 Si TLC plates, developed with 60:40 methanol:water and overlaid with a suspension of the Agrobacterium tumefaciens NTL1 pZLR4 AHL reporter in soft agar with X-Gal. Synthetic AHL standards (all from Sigma) were similarly subjected to the bioassay-coupled thin-layer chromatography.

Full size image

Treatment with the AHL lactonase AiiA, significantly reduced appearance of the detectable TLC spots produced by the Marinobacter sp. 47G8 and α-proteobacteria 45A11 and 47H1 (Supplementary Figure S2). The sensitivity of these signals to the treatment with the AHL lactonase AiiA suggests that these compounds are likely to be AHL. Other signals produced by the marine bacteria and detectable with the Agrobacterium bioassay were resistant to the lactonase treatment (Supplementary Figure S2), suggesting that these compounds are not well-characterized AHLs.

As expected, the Gram-positive Caryophanon spp. 52E5 did not produce any compounds capable of stimulating the Agrobacterium reporter, although it produced yet un-identified substances that stimulated the LuxR- and AhyR-based reporters (pSB401 and pSB536, respectively) 600–200- and 115–138- fold over the background. We note that the production of QS-active substances in α-proteobacterium 46H6 was dependent on culture conditions: QS-active substances were consistently detected in cultures grown in GASW broth, but not in Marine Broth (data not shown).

Effect on swarming in Serratia marcescens

The ability of marine prokaryotes and eukaryotes to excrete compounds that stimulate or inhibit QS reporters is now well documented (Pasmore and Costerton, 2003; Wagner-Dobler et al., 2005; Skindersoe et al., 2008; Teasdale et al., 2009; Kwan et al., 2010); however, ecological roles of these compounds and of the bacteria that produce them are less understood. Therefore, our subsequent experiments focused on testing the behaviors of the isolates in the dual-species microbial consortia consisting of the coral pathogen S. marcescens PDL100 and the isolate of interest. We tested the hypothesis that the marine isolates capable of affecting QS reporters will also modulate those behaviors in the coral pathogen Serratia marcescens, which are known to be controlled by the AHL-mediated QS in Serratia spp. (Van Houdt et al., 2004).

In strains of S. marcescens, multicellular surface motility (‘swarming’) is facilitated by the QS-mediated production of a surfactant (Van Houdt et al., 2007). Wild-type S. marcescens MG1 swarms over semi-solid surfaces ((Lindum et al., 1998); Figure 3), while the disruption of the AHL synthase gene swrI significantly delays the appearance of the swarm ((Lindum et al., 1998); Figure 3). Despite the fact that 10 of the 13 tested strains produced QS signals detectable with the A. tumefaciens reporter, only 2 isolates (Vibrio sp. 52B8 and Marinobacter sp. 47E6) partially restored swarming in the swrI mutant of S. marcescens MG44 (Figure 3, bottom row). This suggests that either the QS activities detected with the Agrobacterium reporter are not AHLs, or that in addition to the QS signals the tested bacteria produce compounds that specifically disrupt swarming. Similarly to the well-characterized S. marcescens MG1, the coral white pox pathogen S. marcescens PDL100 was also capable of spreading over semi-solid surfaces, although the appearance of the swarm was distinct (Figure 3).

Figure 3
figure 3

Swarming of Serratia marcescens in the presence of marine bacteria. Swarming of the model opportunistic pathogen S. marcescens MG1 (top row), coral white pox pathogen S. marcescens PDL100 (middle row) and the SwrI AHL synthase mutant S. marcescens MG44 (bottom row) was tested. Controls (Serratia spotted next to a glass fiber disk without a marine bacterium) are in the far left column. Identities of the tested marine bacteria are listed above the figure.

Full size image

When inoculated in the vicinity of the wild-type Serratia strains, all Marinobacter spp. isolates (46E2, 47A11, 47E6 and 47G8) and α-proteobacterial isolate 46H6 inhibited their surface spreading. The α-proteobacterial isolates 44F6 and 44B9, Photobacterium sp. 34E11, Caryophanon sp. 52E5 and Planomicrobium sp. 34D8 inhibited swarming in S. marcescens MG1, but not in the coral pathogen S. marcescens PDL100 (Figure 3). The α-proteobacterium 47H1 stimulated swarming in S. marcescens MG1; however, this phenotype is not likely to be due to the production of AHLs or a surfactant by the strain as evidenced by the lack of swarming restoration in the swrI mutant S. marcescens MG44 (Figure 3, bottom row). Co-culture of the coral pathogen S. marcescens PDL100 with Planomicrobium sp. 34D8 and α-proteobacterial isolate 44F6 stimulated surface spreading in the coral pathogen (Figure 3, middle row). In a co-culture of S. marcescens PDL100 with Vibrio sp. 52B5, a swarm, likely composed of the two organisms, was observed. These results suggest that in addition to producing compounds that trigger responses in QS reporters, tested strains secrete substances that inhibit swarming in S. marcescens, either by interfering with the regulation of the flagellar regulon or by disrupting the synthesis of the surfactant. Both of these functions are required for surface swarming (Lindum et al., 1998; Verstraeten et al., 2008).

Biofilm formation on coral mucus

In addition to controlling surface spreading, cell-to-cell signaling also has important roles in the formation of sessile microbial communities (‘biofilms’) and interactions within them. Both swarming and biofilm formation are the behaviors required for surface-associated growth; however, they represent disparate strategies in surface colonization and are inversely regulated in γ-proteobacteria (rev. Verstraeten et al., 2008). If the invertebrate-associated bacteria secrete a compound that affects global regulatory systems required for surface-associated growth, then—we hypothesized—in the absence of other interactions, there should be a measurable effect on biofilm formation by S. marcescens in the presence of the tested bacterial isolates.

The effects on the biofilm formation by S. marcescens PDL100 were tested using dual-species microbial consortia containing bacteria capable of inhibiting swarming. Experiments were carried out in polystyrene microtiter plates coated with mucus of A. palmata, as in (Krediet et al., 2009a). Consistent with our hypothesis, in dual-species consortia, α-proteobacterium 44B9 stimulated swarming (Figure 3) and also inhibited biofilm formation by the coral pathogen S. marcescens PDL100, even though growth of the pathogen in the suspension cultures was not affected (Figure 4). Similar trends (stimulation of swarming and reduced biofilm formation) were observed for α-proteobacterium 47H1, Photobacterium sp. 34E11 and Planomicrobium sp. 34D8, although statistically significant differences were not observed on all batches of mucus collected from spatially separated A. palmata colonies on the same reef (Figure 4). Of the isolates capable of promoting swarming and inhibiting biofilm formation, α-proteobacterium 44B9 both reduced biofilm formation by the pathogen and also dominated the biofilms. Also consistently with the working hypothesis, Marinobacter sp. 47A11 inhibited swarming and modestly stimulated biofilm formation by S. marcescens PDL100 on some batches of coral mucus (Figure 4). Collectively, these results indicate that the compound(s) secreted by these organisms may target a global regulatory switch involved in the regulation of swarming and biofilm formation, without affecting growth of another organism.

Figure 4
figure 4

Biofilm formation by the white pox pathogen S. marcescens PDL100 on coral mucus in the presence of marine bacteria. Dual-species consortia consisting of the coral pathogen and a marine isolate were set up in microtiter plates, surfaces of which were coated with mucus of Acropora palmata. S. marcescens PDL100 was inoculated into each well, marine bacteria present within each dual-species consortium are listed at the bottom of the figure. The relative numbers of S. marcescens PDL100 and marine isolates were enumerated by dilution plating. S. marcescens PDL100 was distinguished from the other marine bacteria based on colony morphology: S. marcescens PDL100 forms small cream-color circular convex colonies with an entire edge on Marine Agar (1.5% agar). The black bars indicate the percentage of the coral pathogen and the gray bar is the percentage of the marine isolate in each sample. In each data set, the left bar indicates un-attached bacteria, and the right bar is the biofilm. Inhibition of biofilm formation by the coral pathogen was scored as the decrease of the relative percentage of S. marcescens PDL100 within biofilms, compared with the suspension. Data from the dual-species biofilms with Vibrio spp 52B8 are not shown because mixed swarms containing the two bacteria formed on agar plates, thus making enumeration of bacteria nearly impossible. Each panel represents data from three independent experiments using mucus harvested from three different colonies of A. palmata in April of 2010. Error bars represent standard errors of four biological replications (independent dual-species consortia).

Full size image

The phenotypes of seven isolates did not fit the working hypothesis. The α-proteobacterium 44F6 stimulated both swarming and biofilm formation. Caryophanon sp. 52E5 promoted swarming of S. marcescens PDL100 but had no appreciable effect on biofilm formation on mucus of A. palmata (Figure 4). Marinobacter spp. 47E6, 46E2, 47G8, α-proteobacteria 45A11 and 46H6 reduced both swarming and biofilm formation (Figure 4). The latter phenotype would be consistent with the production of multiple compounds, each capable of reducing swarming and inhibiting biofilm formation, likely by affecting independent regulatory or functional cascades.

Antagonistic isolates reduce appearance of serratiosis in Aiptasia pallida

To test whether the isolates capable of affecting biofilm formation and swarming in the model opportunistic pathogen S. marcescens PDL100, individual polyps were inoculated under laboratory conditions with the pathogen with or without the antagonistic marine isolates. As shown in Figure 5 (column B), S. marcescens PDL100 can completely degrade the polyp within 3–5 days. Pre-inoculation of the polyps with either a cocktail of isolates (containing α-proteobacteria 45A11 and 46H6 and Marinobacter spp. 47E6, 46E2 and 47G8) or a monoculture of the α-proteobacterium 44B9 reduced the appearance of the disease symptoms in the polyps infected with the white pox pathogen (columns C and D). These results indicate that the native microbiota associated with the invertebrates or their endosymbiotic dinoflagellates are capable of producing activities that reduce susceptibility of marine invertebrates to opportunistic pathogens.

Figure 5
figure 5

Inhibition of serattiosis in Aiptasia pallida by antagonistic holobiont-associated bacteria. Clonal lines of Aiptasia pallida were maintained in aquaria. For the experiments, individual polyps were acclimated in six-well plates with 10 ml of sterile artificial seawater for 2 days. Antagonistic marine strains (columns C and E were added as a cocktail containing Marinobacter spp. 47E6, 47G8 and 46E2 as well as α-proteobacteria 46H6 and 45A11); polyps shown in columns D and F were treated with α-proteobacterium 44B9 at 106 cfu ml−1. Wells in column A contain polyps that were not exposed to S. marcescens or marine isolates; polyps in columns B, C and D were infected with S. marcescens PDL100 at 5 × 107 cfu ml−1 on day 1 of the experiment. Signs of the disease progression (darkened polyp and retracted tentacles) were documented daily for a week. Brightness and color balance of the images were adjusted in Adobe Photoshop CS4 using default settings. Experiments were performed with three biological and six technical replications.

Full size image

Conclusions

Interactions within the microbiota associated with marine holobionts are undoubtedly complex. Results presented in this manuscript suggest that bacteria, which inhabit the holobiont, can modulate behaviors in the model multi-species opportunistic pathogen S. marcescens. In dual-species microbial consortia, swarming of the coral pathogen S. marcescens PDL100 was generally inhibited by the bacteria recovered from the associations with the symbiotic dinoflagellate Symbiodium spp. (Figure 3). The ability of the antagonistic bacteria to inhibit the progression of a disease caused by S. marcescens in a model marine invertebrate holobiont A. pallida (Figure 5), is consistent with the fact that swarming is often co-regulated with virulence in γ-proteobacterial pathogens (Verstraeten et al., 2008). Unlike virulence genes that are generally co-regulated with genes involved in swarming, biofilm formation is inversely regulated (Verstraeten et al., 2008). The ability of some strains to inhibit biofilm formation and promote swarming is consistent with the possibility that the native bacteria produce compound(s) that disrupt global regulatory cascades involved in switching from surface motility (swarming) to biofilm formation. Many of the tested native bacteria inhibited both swarming and biofilm formation by S. marcescens PDL100 (but not growth of the pathogen), indicative of the ability to produce inhibitory activities capable of disrupting multiple behaviors in the pathogen. Even though bacteria were selected for this study based on their ability to affect well-characterized QS reporters, no strong correlation between their QS activities and the effect on swarming and/or biofilm formation by S. marcescens PDL100 was observed.