Enhanced Biofilm Formation and Increased Resistance to Antimicrobial Agents and Bacterial Invasion Are Caused by Synergistic Interactions in Multispecies Biofilms^†

Isolation of bacterial strains.

The bacterial strains were isolated from the surface of the marine alga Ulva australis as described by Rao et al. (41) and were subcultured at 25°C on complex Väätänen nine-salt solution (VNSS) marine medium (31) agar plates (containing 15 g agar/l) or in 5 to 10 ml of VNSS medium broths with shaking at 160 rpm.

Quantification of biofilm formation by use of CV.

In order to identify and select epiphytic isolates with poor biofilm formation and subsequently determine whether the biofilm formation ability differed when these isolates were grown individually compared to in multispecies consortia, quantification of cell adhesion and biofilm formation by single- and multispecies consortia was performed. The method used was a modified version of that described by O'Toole and Kolter, based on staining biofilms with crystal violet (CV) (36). CV binds to negatively charged molecules, including nucleic acids and acidic polysaccharides, and therefore serves as an overall measure of the whole biofilm. Strains were subcultured from frozen glycerol (30%) stocks onto VNSS agar plates for 48 h, and from these, colonies were transferred to 9 ml of VNSS broth and incubated for 48 h. Strains were then harvested by centrifugation (6,000 × g, 5 min) and resuspended in 4.5 ml VNSS. The cell densities of all suspensions were adjusted to an optical density at 600 nm (OD₆₀₀) of 0.15 by dilution in VNSS medium. The strains were inoculated into flat-bottomed 96-well Costar microtiter plates (no. 3595, “96 wells Cell Culture Cluster” polystyrene; Corning Inc.). The final volume added to each well was 200 μl. To some wells, 200 μl of VNSS medium was added to serve as negative controls and to obtain a background value, which was subtracted from values obtained from the wells containing cells. Plates were then sealed with Parafilm and incubated with shaking (100 rpm) at room temperature for 24 h.

To correlate biofilm formation with planktonic cell growth in each well, the planktonic cell fraction was transferred to new microtiter plates and the OD₆₀₀ was measured. The attached cells were rinsed three times with 200 μl phosphate-buffered saline (46) and stained with 200 μl of an aqueous CV solution (1%). After 20 min of staining, CV was removed and cells were rinsed once with 200 μl, once with 400 μl, and twice with 200 μl of PBS. To resuspend the CV, 200 μl of ethanol (96%) was added to each well, and the absorbance of CV at 600 nm was measured in a Wallac-Victor² 1420 Multilabel Counter (Perkin-Elmer, Boston, MA). When the CV absorbance increased to an OD₆₀₀ of above 1.5, the CV-ethanol was diluted 1:3 in 96% ethanol. Finally, the absorbance measurements obtained were related to the OD₆₀₀ of the planktonic cell fraction.

16S rRNA gene-based identification of bacterial strains.

On the basis of the results obtained in the screening of multispecies biofilm formation, four strains that interacted synergistically were chosen for species identification. Aliquots (1 ml) of overnight cultures of each of the isolates 2.04, 2.12, 2.3, and 2.34 were boiled for 10 min, chilled on ice, and centrifuged at 0°C and 14,000 g for 2 min. From the supernatants, 1 μl was used as the template in PCRs with primers designed to amplify the eubacterial 16S rRNA gene (27F and 1492R) (26). Sequencing reactions (DYEnamic ET dye terminator cycle sequencing kit [MegaBACE]) were performed using 0.1 μg purified PCR product and 10 pmol of primers (27F, 1100R, and 1492R) (26) in 10-μl reaction mixtures, and sequencing was performed using a MegaBACE 1000 sequencer (Molecular Dynamics, Sunnyvale, CA). Sequences obtained were compared to sequences available in the NCBI (National Center for Biotechnology Information) BLAST 2.0 database (Basic Local Alignment Search Tool) (2).

Resistance to antimicrobial agents in single- and multispecies biofilms.

Examination of the activities of single- and four-species biofilms of strains 2.04, 2.12, 2.3, and 2.34, with and without the addition of antimicrobial agents, was performed by use of the respiratory indicator 2,3,5-triphenyltetrazolium chloride (TTC). This compound is soluble and colorless in its original state but forms a red, insoluble salt when it is reduced by the oxidative enzyme complexes in the bacterial cell (16). Microtiter plates were inoculated and incubated as described above. After 24 h, the VNSS medium and planktonic cells were discarded, and the wells were rinsed once (to remove planktonic cells) with fresh VNSS. Then, 200 μl VNSS containing either hydrogen peroxide (1,700 μg/ml) or tetracycline (20 μg/ml) was added to the wells. To some wells, VNSS without inhibitory compounds was added. These wells served as measures of the activity of the biofilms without inhibition, and the activity measurements for biofilms being inhibited were related to these to obtain the percentage of activity in the exposed biofilms. Each treatment was performed in four replicates. Plates were then sealed with Parafilm and incubated with shaking (100 rpm). After 1 hour, 20 μl of TTC (0.1%, prepared in sterile water) was added to each well, and the plates were sealed with Parafilm, wrapped in foil, and incubated again. To optimize the time course of the assay, the presence of reduced TTC was quantified by measuring the absorbance at 490 nm (EL 340 microplate reader; Bio-Tek Instruments, Winooski, Vermont) at various time points after the biofilms were exposed to the antimicrobial agents.

Resistance to invasion by Pseudoalteromonas tunicata.

We examined the ability of preestablished single- and multispecies biofilms to resist invasion by the marine bacterium Pseudoalteromonas tunicata. This organism was selected for these studies because it aggressively colonizes and dominates biofilms on the surface of U. australis through the production of the potent antibacterial protein AlpP (41; unpublished data). Biofilms were grown in 20% VNSS broth in continuous-culture flow cells (channel dimensions, 1 by 4 by 40 mm) at room temperature as previously described (35). Biofilms were established by inoculating channels with 1 ml of overnight cultures (10⁶ cells ml⁻¹) of either strain 2.04, 2.12, 2.3, or 2.34 (single-species biofilms) or with 250 μl of each strain (four-species biofilm), resulting in a total volume of 1 ml. Cultures were incubated without flow for 1 h to allow cell attachment, followed by maturation of biofilms in the presence of flow (flow rate of 150 μl min⁻¹) for 2 h. The biofilms were inoculated with 10⁶ cells ml⁻¹ of an overnight culture of green fluorescent protein-labeled P. tunicata (41), and the flow was stopped for 1 h. After resumption of the flow, the biofilms were monitored at regular intervals for a period of 44 h. Biofilms were stained with Syto 59 (diluted to 3 μl ml⁻¹ in 20% VNSS) and visualized with a confocal laser scanning microscope (Olympus, Tokyo, Japan) using fluorescein isothiocyanate and tetramethyl rhodamine isocyanate optical filters. The flow cells were examined for red and green fluorescence, and the percent surface coverages of P. tunicata and of other cells were calculated using image analysis (ImageJ; NIH, Bethesda, Maryland). The flow cell experiments were repeated in three separate rounds with three independent flow cells running in parallel.

Effects of secreted compounds on biofilm synergy.

The supernatants of isolates 2.04, 2.12, 2.3, and 2.34 were examined in order to evaluate the role of the secreted compounds in the observed multispecies biofilm synergy and whether compounds mediating N-acyl homoserine lactone (AHL)-dependent quorum sensing were produced by any of the isolates. Four- and single-species biofilm attachment and formation experiments with isolates 2.04, 2.12, 2.3, and 2.34 were performed as described above. Four variants of the four-species biofilms were prepared, in which one isolate was replaced by its filtered supernatant and the remaining three isolates were inoculated as described above. After centrifugation of the cells, the spent supernatants were filtered through 0.2-μm-pore-size Minisart filters (Sartorius, Hannover, Germany). To ensure that no cells were present in the filtrates, 100 μl was spread onto VNSS agar plates, and 200 μl was inoculated in separate wells in the microtiter plate.

The filtered supernatants were screened for presence of AHLs by the use of the Agrobacterium tumefaciens biosensor assay described by Zhu et al. (61). The AHL compound N-oxohexanoyl homoserine lactone at a final concentration of 10⁻⁷ M and sterile water were used as positive and negative controls, respectively.

Screening for plasmids.

Attempts were made to purify plasmids from strains 2.04, 2.12, 2.3, and 2.34 by use of small-scale preparations of plasmid DNA (46), the QIAprep spin miniprep kit (QIAGEN), the Plasmid Midikit (QIAGEN), and the Plasmid Mini AX (A&A Biotechnology, Gdynia, Poland). The strains were grown for 48 h in appropriate volumes of VNSS, and the methods were used as recommended and further optimized. Plasmid purifications were visualized by agarose gel electrophoresis.

Nucleotide sequence accession numbers.

The rRNA gene sequences of isolates 2.04, 2.12, 2.3, and 2.34 have been submitted to GenBank under accession numbers DQ328319 to DQ328322, respectively.

Biofilm formation by 17 epiphytic bacterial strains.

In order to determine the extent to which 17 marine epiphytic isolates formed biofilms, the strains were incubated as single species in microtiter plates (replicates of four wells) and biofilm formation was measured after 24 h (Fig. 1). Due to the high absorbance of CV in some wells (containing isolates 2.14, 2.19, 2.2, and 16), the CV was diluted (1:3) several times. Ten isolates with a biofilm/planktonic OD₆₀₀ value of less than 2 were categorized as poor biofilm formers. These strains were further examined for synergistic effects when coincubated with other isolates. A total of nine different combinations of three to six strains were grown in multispecies biofilm consortia and examined for synergy.

Synergistic interactions in a four-species biofilm.

In one combination of four isolates (2.04, 2.12, 2.3, and 2.34), synergistic interactions were observed (Fig. 2). Each of these four isolates was grown as single-species biofilms and in all possible combinations of, two-, three-, and four-species biofilm consortia in replicates of four. With three combinations as exceptions, the biofilm formation by two or more species was significantly greater than the biofilm produced by any of the single species. When the four isolates coexisted in the biofilm, the biofilm biomass increased by >167% compared to that of the single isolates after 24 h of biofilm formation. The colony morphologies of the four isolates were distinct and were used to verify the presence of all of the strains in approximately equal numbers after 24 h of coincubation in the wells (see the supplemental material). The experiment was repeated three times with similar outcome.

In two-species biofilms, each species interacted synergistically with at least one other organism. One isolate, strain 2.3, was able to cause synergy in all three dual-species combinations. In three-species biofilms, all combinations of strains, except that lacking strain 2.34, generated less biofilm biomass than that composed of the four isolates 2.04, 2.12, 2.3, and 2.34. While 2.34 appeared not to influence synergistic growth in four-species biofilms, we still included 2.34 in the four-species biofilm mix because it showed strong synergy with strain 2.3 in dual-species biofilms (Fig. 2).

Identification of the four biofilm-synergistic, epiphytic bacterial isolates.

The 16S rRNA genes of each of the four isolates 2.04, 2.12, 2.3, and 2.34 were sequenced (Table 1) and found to be identical (99 to 100%) to those of previously sequenced marine bacteria. Sequences of 1,321 to 1,389 base pairs were obtained, and Blast searches against sequences available in GenBank (http://www.ncbi.nlm.nih.gov/GenBank/) were performed. The closest match of isolate 2.04 was the gram-positive Microbacterium phyllosphaerae (3), belonging to the class Actinobacteria. The other three matched gram-negative bacteria, two of which belong to the Gammaproteobacteria: isolate 2.12, with closest homology to Shewanella japonica (22), and isolate 2.34, which most closely matched Acinetobacter lwoffii (56) The closest relative of isolate 2.3 proved to be a recently defined species, Dokdonia donghaensis (60) (also known as Dokdoa eastseensis), which belongs to the family Flexibacteriaceae in the Sphingobacteria class. The strains are referred to as these species below.

Resistance to antimicrobial agents in single- and four-species biofilms.

Hydrogen peroxide and tetracycline are common antimicrobial agents used to inhibit bacterial growth. Hydrogen peroxide causes oxidative stress in the bacterial cell, and the broad-spectrum antibiotic compound tetracycline inhibits protein synthesis. In order to assess the fitnesses of the single- and four-species biofilms composed of M. phyllosphaerae, S. japonica, D. donghaensis, and A. lwoffii, the activities of these biofilms when exposed to either hydrogen peroxide or tetracycline were examined. Biofilms were established in microtiter plate wells for 24 h and then exposed to hydrogen peroxide or tetracycline. The activities of these biofilms were determined by addition of the respiratory indicator TTC, and the activities obtained were related to those of the corresponding untreated biofilms. The data shown in Fig. 3 represent the TTC absorbance after 16 h of hydrogen peroxide or tetracycline exposure. A marked difference was observed in the relative activities (biofilms exposed to the antimicrobial agents versus nonexposed control biofilms) of the single- and four-species biofilms; the four-species biofilm was significantly more active, as deduced by the TTC measurements, in the presence of the inhibitory compounds than any of the single-species biofilms.

Resistance to invasion by Pseudoalteromonas tunicata.

The marine epiphytic bacterium P. tunicata produces a range of biocidal compounds, including a broad-spectrum, antibacterial, high-molecular-weight protein, AlpP, that is effective against gram-negative and -positive isolates from various environments (9-11, 23). AlpP is expressed in P. tunicata biofilms and is known to play a role in the competitive dominance of the organism during mixed-species biofilm formation (41). Based on this as well as the recent finding that P. tunicata aggressively replaces Cytophaga funicola and an Alteromonas sp. in mixed-species marine biofilms (41), P. tunicata was included to determine resistance of established biofilms against invasion by another bacterial species. The four-species biofilm consortium was exposed to bacterial invasion by addition of a green fluorescent protein-expressing P. tunicata to preestablished single- and four-species biofilms. At various time points over 44 h, the degree of P. tunicata invasion was determined from confocal laser scanning microscope visualization and image analysis of invaded and noninvaded biofilms. Corresponding invaded and noninvaded biofilms were compared, and the results are presented as percent invasion in Fig. 4. At every time point, the P. tunicata invasion was significantly lower in the four-species biofilm, indicating that this biofilm resisted the invasion to a greater extent than did the biofilms composed of single species. Examples of confocal micrographs obtained during the invasion of mixed-species biofilms with P. tunicata are shown in the supplemental material.

Effects of secreted compounds on biofilm synergy.

In order to assess whether extracellular compounds produced by the epiphytic bacteria would induce biofilm synergistic effects, the supernatants from each of the four strains, M. phyllosphaerae, S. japonica, D. donghaensis, and A. lwoffii, were used in add-back experiments during the course of biofilm formation. The results are presented in Table 2. Significant reductions in biofilm formation, approximately 50%, were observed when S. japonica and D. donghaensis were replaced by their filtered supernatants. Only minor reductions were observed when M. phyllosphaerae and A. lwoffii supernatants were added instead of cells. Enhanced biofilm formation by supernatant addition was not detected.

None of the cell-free supernatants induced the AHL biosensor above the background level, indicating that these strains did not produce AHLs.

Add-back experiments with purified AlpP from P. tunicata were not performed because AlpP is a large, 190-kDa protein which exhibits limited diffusion through agar media (and likely also within cell-dense and polysaccharide matrix-encased biofilms). Our data also suggest that AlpP activity is principally cell associated and mediated by cell-cell contact with competing organisms (unpublished data). Exposure of biofilms to free exogenous AlpP is therefore unlikely to occur within the environment.