Biofilm Formation by the Fish Pathogen Flavobacterium columnare: Development and Parameters Affecting Surface Attachment

Abstract

Flavobacterium columnare is a bacterial fish pathogen that affects many freshwater species worldwide. The natural reservoir of this pathogen is unknown, but its resilience in closed aquaculture systems posits biofilm as the source of contagion for farmed fish. The objectives of this study were (i) to characterize the dynamics of biofilm formation and morphology under static and flow conditions and (ii) to evaluate the effects of temperature, pH, salinity, hardness, and carbohydrates on biofilm formation. Nineteen F. columnare strains, including representatives of all of the defined genetic groups (genomovars), were compared in this study. The structure of biofilm was characterized by light microscopy, confocal laser scanning microscopy, and scanning electron microscopy. F. columnare was able to attach to and colonize inert surfaces by producing biofilm. Surface colonization started within 6 h postinoculation, and microcolonies were observed within 24 h. Extracellular polysaccharide substances and water channels were observed in mature biofilms (24 to 48 h). A similar time course was observed when F. columnare formed biofilm in microfluidic chambers under flow conditions. The virulence potential of biofilm was confirmed by cutaneous inoculation of channel catfish fingerlings with mature biofilm. Several physicochemical parameters modulate attachment to surfaces, with the largest influence being exerted by hardness, salinity, and the presence of mannose. Maintenance of hardness and salinity values within certain ranges could prevent biofilm formation by F. columnare in aquaculture systems.

INTRODUCTION

Flavobacterium columnare is a Gram-negative bacterium, a member of the Cytophaga-Flavobacterium-Bacteroides group, and the causative agent of columnaris disease in freshwater fish. Columnaris disease affects many freshwater fish species worldwide but is particularly relevant for catfish, tilapia, and trout aquaculture, where it causes major economic losses (1–4). This bacterium is divided into three genetic groups or genomovars, with genomovar II being the group most virulent toward channel catfish and other fish species tested (5–7). Despite its economic impact, the ecological niche of F. columnare has not been clearly identified although most studies point to fish as the primary disease reservoir (8, 9). However, fish-to-fish contact is not required for disease transmission (10) and the long-term survival of F. columnare in lake water suggests that water can be the main reservoir of this pathogen (11). In a recent study, we have shown that although F. columnare can survive in water without nutrients for extended periods, its cells underwent drastic morphological changes that resulted in loss of fitness over time (12). Therefore, it is likely that this pathogen uses other niches in the aquatic environment besides the water column.

Biofilm is referred to as a community of microbes embedded in an organic matrix and attached to a physical surface. Adopting such a special lifestyle provides the bacteria with benefits such as protection from desiccation, enhanced antibiotic resistance, increased nutrient concentration, and protection against predators (13). Biofilm development requires several key steps, i.e., transport and attachment of planktonic bacteria to a surface, cell proliferation, formation of microcolonies, and dispersion of daughter cells into the water column (14). Many aquatic bacteria are capable of colonizing surfaces and forming biofilms that act as reservoirs of those bacterial populations (13, 15, 16). Aquaculture facilities are particularly prone to the development of biofilms and biofouling; thus, investigating what physicochemical factors favor the development of biofilm could reduce the presence of pathogens in the system (17). We have previously observed that F. columnare can be recovered from biofilms covering the glass edges of tanks after a columnaris outbreak (10), and we hypothesized that biofilm can act as a reservoir of this pathogen.

Despite the cosmopolitan distribution of columnaris disease in freshwater aquaculture facilities, very limited information is available regarding the ability of F. columnare to colonize surfaces and form biofilm (18). Previous studies have focused on in vivo attachment to host tissues and showed that both gill and skin tissues can be rapidly colonized after exposure to the pathogen (19, 20). However, significant differences were found between genomovars I and II in terms of both chemotaxis response and attachment to host tissues, with genomovar II showing a higher affinity for channel catfish tissues than genomovar I (20, 21).

The objectives of the present study were to evaluate the biofilm formation capability of F. columnare and to characterize its development from the early stages of cell attachment to the development of extracellular polysaccharide substances (EPS) in mature biofilms. In addition, the effects of different physicochemical parameters on attachment to and colonization of surfaces were evaluated. Strains representing all three genomovars were included in this study to cover the genetic diversity present in this species.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

Nineteen F. columnare stains isolated from different resources were used in this study (Table 1). All of the strains had been previously identified as F. columnare by standard methods (22–24), and they exhibited distinct genetic profiles according to previous genotyping analysis (24, 25). Representatives of all three genomovars were included in this study, although only two genomovar III strains could be obtained. Bacteria were routinely grown and maintained in modified Shieh (MS) broth (26) for 48 h at 28°C with gentle shaking. Stock suspensions of all isolates were stored in MS broth supplemented with 20% glycerol at −80°C.

Table 1.

F. columnare strains used in this study

Strain	Genomovar	Geographic origin	Source	Date
ARS-1	I	Alabama	Channel catfish	1996
ALM-05-122	I	Alabama	Blue catfish	2005
GA-02 −14	I	Georgia	Rainbow trout	2002
Grizzle	I	Alabama	Channel catfish	Unknown
ALM-05-114	I	Alabama	Threadfin shad	2005
ALG-03-063	I	Alabama	Channel catfish	2003
BioMed	I	Alabama	Channel catfish	1996
ALG-00-530	II	Alabama	Channel catfish	2000
BZ-1-02	II	Brazil	Nile tilapia	2002
ALM-05-202	II	Alabama	Blue catfish	2005
LT-ulcer	II	Alabama	Nile tilapia	2011
ALM-05-173	II	Alabama	Blue catfish	2005
BGFS-27	II	Alabama	Channel catfish	2005
ALG-00-527	II	Alabama	Channel catfish	2000
ALM-05-121	II	Alabama	Channel catfish	2005
CL-gill	II	Alabama	Nile tilapia	2011
BGFS-25	II	Alabama	Channel catfish	2005
AU-98-24	III	Alabama	Channel catfish	1998
Dickson-1	III	Unknown	Channel catfish	1995

Biofilm formation on glass slides.

To induce biofilm formation on glass surfaces, a microcosm was set up as follows. Each microcosm (500-ml glass beaker) contained 200 ml of MS broth. Four microscope glass slides (VWR Superfrost Plus; VWR, Radnor, PA) were suspended vertically in the broth with metal clamps attached to a glass rod. Approximately 3 cm of the glass slide was immersed in the broth. All components were assembled prior to sterilization by autoclaving. Each microcosm was inoculated with 200 μl of an 11-h-old culture of strains ALG-00-530 and ARS-1 to a final concentration of approximately 10⁵ cells ml⁻¹. The microcosm was kept at 28°C, the optimal temperature for F. columnare, to allow bacterial attachment and biofilm formation. Several microcosms, depending on the number of slides needed for the different microscopy methods used, were set up simultaneously. Slide samples were taken out at different time intervals, air dried, and observed with a Leica DM2500 microscope by differential interference contrast (Leica Microsystems). To visualize some of the characteristics of the biofilm, cells were treated with different chemical dyes, including calcofluor white (10 mg/ml; Sigma, St. Louis, MO), Congo red (0.1%, wt/vol; Sigma, St. Louis, MO), and dextran-rhodamine (Sigma, St. Louis, MO). Images were captured with a charge-coupled device camera and processed with Zeiss Smart software (version 3.0; Zeiss, Wetzlar, Germany).

SEM.

Samples were prepared for scanning electron microscopy (SEM) according to one of the following methods. For method 1, the colonized slides were carefully rinsed in distilled H₂O three times to eliminate planktonic cells, fixed in 10% neutral formalin solution for 10 min, and dehydrated in increasing concentrations of ethanol (30, 50, 80, and 100% for 10 min each). Samples were then placed into 1:1 ethanol-hexamethyldisilazane (HMDS) solution for 10 min and transferred to 100% HMDS for an additional 20 min. Samples were air dried under a chemical hood for another 10 min before sputter coating with a gold-palladium alloy in an EMS 550X (Electron Microscopy Science, Hatfield, PA). For method 2, the slides were rinsed as stated above and then fixed in 4% osmium tetroxide overnight under a chemical hood. Slides were air dried for 1 h and sputter coated for SEM observation with a Zeiss EVO 50 (Zeiss, Wetzlar, Germany). Planktonic cells were used as a control for morphological observations. Briefly, 5 μl of a 24-h-old culture was inoculated into 5 ml of MS broth, immediately fixed in 2.5% (vol/vol) glutaraldehyde, and filtered through an Isopore membrane (Millipore). Filters were then dehydrated in serial ethanol concentrations (50, 70, 90, and 100%) and subjected to critical-point drying before SEM as before.

Confocal laser scanning microscopy (CLSM).

Biofilm samples were prepared as stated above and rinsed with distilled water three times. Twenty microliters of a mixture of propidium iodide, SYTO 9 (Live/Dead cell viability assay kit; Invitrogen, Carlsbad, CA), and calcofluor white (10 mg/ml; Sigma, St. Louis, MO) was added on top of the biofilm, and the biofilm was incubated in the dark for 15 min. Slides were analyzed with a Nikon Eclipse confocal laser scanning inverted microscope (Nikon, Melville, NY), with a 60× oil immersion objective. To detect all three of the dyes used, propidium iodide, SYTO 9, and calcofluor white), excitation wavelengths of 528, 590, and 370 nm, respectively, were used. The three-dimensional architecture of the biofilm was assessed by z-stack images obtained in 0.9-μm increments. Images were acquired with a CoolSNAP HQ2 camera (Photometrics, Tucson, AZ) and analyzed with NIS-Elements AR software (version 3.0; Nikon, Melville, NY).

Biofilm formation under microfluidic conditions.

Biofilm formation under flow conditions was evaluated by using microfluidic chambers as previously described (27). Briefly, for microfluidic chamber fabrication, photolithography was used to etch a pattern of two parallel microchannels onto a silicon wafer, which was later replicated with polydimethylsiloxane (PDMS). The chamber was composed of a molded PDMS body attached to a cover glass and a supporting glass microscope slide. The microchannels were 80 μm wide, 50 μm deep, and 3.7 cm long (28). MS medium was injected by an automated syringe pump (Pico Plus; Harvard Apparatus, Holliston, MA) set at 0.25 μl/min. High-virulence strain ALG-00-530 (genomovar II) and low-virulence strain ARS-1 (genomovar I) were compared in parallel by introducing a cell suspension (optical density [OD] of 0.1) into each channel (6). The microfluidic chambers were mounted onto a Nikon Eclipse Ti inverted microscope (Nikon, Melville, NY) and observed at a magnification of ×40 with phase-contrast optics to monitor cell growth, attachment to glass surfaces, and biofilm formation for up to 4 days. Images were captured every 30 s by with a Nikon DS-Q1 digital camera (Nikon, Melville, NY) controlled by NIS-Elements software (Nikon, Melville, NY). The experiment was repeated two times independently.

Virulence study.

To determine if F. columnare cells retained virulence when embedded in biofilm, a virulence study was conducted with channel catfish fingerlings (for fish husbandry, see reference 5). To recreate the conditions under which channel catfish fingerlings may encounter F. columnare in the form of biofilm, the skin of the fish was slightly eroded with a scalpel (29). This abrasion was intended to mimic the cuts that fish suffer while rubbing themselves against the tank sides. Glass slides containing mature biofilm (about 72 h) of either strain BGFS-27 or ARS-1 were gently applied against the side of the fish (n = 20 per strain) for 5 s. Strains were chosen on the basis of their known virulence for channel catfish (BGFS-27 was preferred over ALG-00-530 because it achieves more consistent results under our challenge conditions) (6). Glass slides in pure MS broth served as a negative control (n = 20). Fish were monitored for signs of disease twice a day for 7 days.

Quantification of biofilm.

Biofilm formation by F. columnare was assessed as described previously (30), with some modifications to accommodate F. columnare growth. Briefly, a 28-h-old inoculum was diluted 100 times in the corresponding medium and 100 μl of each dilution was inoculated into the wells of 96-well microtiter polystyrene plates (NuncImmuno MaxiSorp; Nunc, Rochester, NY) in quadruplicate. The microtiter plates were incubated for 48 h at 28°C to allow bacterial attachment and biofilm formation, unless otherwise stated. Bacterial growth was then quantified by measuring the OD at 595 nm (OD₅₉₅) with a Synergy HT spectrophotometer (Bio-Tek Instruments Inc., Winooski, VT). Medium containing unattached cells was discarded, and the wells were washed four times with 300 μl of distilled water. After washing, the wells were stained with 150 μl of 1% (wt/vol) crystal violet. After 20 min, the crystal violet solution was removed and the wells were washed four times with 300 μl of distilled water. Finally, 200 μl of 96% (vol/vol) ethanol was added to dissolve the dye and after 10 min at room temperature, biofilm formation was quantified by measuring the OD₅₉₅.

Physicochemical variables tested.

The effects of five variables (i.e., temperature, pH, salinity, hardness, and carbon source) on biofilm formation were investigated as follows. The effects of different temperatures were compared by incubating the 96-well microtiter plates at 21, 28, and 35°C. Five different pHs were assessed by adjusting (with 5 M NaOH or 1 M HCl) the MS broth pH to 5.8, 6.2, 6.6, 7.3, and 7.9. The effect of salinity was tested by adjusting the MS broth to 0, 5.0, 7.5, 10.5, and 14.0 ppt of NaCl. Similarly, MS broth hardness was adjusted to 12, 64, 120, 300, and 360 ppm with CaCl₂ · 2H₂O. Finally, MS broth was supplemented with the carbohydrates mannose, fucose, galactosamine hydrochloride, glucose, and N-acetyl-d-glucosamine at 50 mM (a previous study examined their role in the F. columnare chemotaxis response [31]). All of these modified culture broths were inoculated with F. columnare cells as described above.

Statistical analysis.

Data on biofilm formation on polystyrene plates were analyzed by one-way analysis of variance with SAS Software version 9.2 (SAS Institute, Cary, NC). The effects of salinity, hardness, pH, and carbohydrates were analyzed with Tukey's test, while the effect of temperature was analyzed with Fisher's least-significant-difference test. Significant difference was set at P < 0.05. Clustering analysis by the Pearson product moment correlation coefficient and the unweighted-pair group method using average linkages (UPGMA) were used to compare the similarity profiles of biofilm formation by the strains. Composite data sets, including the OD₅₉₅ values, of all of the parameters tested served as the input for pairwise comparisons and subsequent cluster analysis. Principal-component analysis (PCA) was used as a nonhierarchical clustering method to determine the contribution of each parameter to the overall variability observed (BioNumerics v. 6.6; Applied Maths, Austin, TX).

RESULTS

Biofilm formation on glass.

Studies using light microscopy confirmed that F. columnare can attach to and rapidly colonize microscope slides. Figure 1 shows colonization and biofilm formation by strain ALG-00-530 cells throughout the time course studied. Attachment to glass started within 6 h (Fig. 1A) postinoculation, although at this stage, cells were still gliding on the glass (see halos around the cells). Between 6 and 12 h postinoculation, individual cells were still visible on the slide but most of them were nonmotile; the first cell aggregates appeared, and cell division on the slide was noted. Between 18 h (Fig. 1C) and 24 h (Fig. 1D), the first microcolonies were observed. After 24 h, cell density was higher and microcolonies covered most of the slide (Fig. 1E). Finally, at 48 h postinoculation (Fig. 1F), the slide was completely covered with cells and what we speculated to be EPS on the basis of the drastic change in light refraction. Similar results were obtained with strain ARS-1 (data not shown).

Fig 1.

Colonization and biofilm development by F. columnare strain ALG-00-530 on glass slides. The bright-field light microscopy images in panels A to F display cell attachment and biofilm formation at 6, 12, 18, 24, 36, and 48 h postinoculation. Scale bars, 20 μm in panels A, C, D, E, and F and 10 μm in panel B.

Biofilm structure.

SEM provides higher resolution than light microscopy and was used to study the structure of biofilms. Figure 2 shows a closeup of the early stages (<24 h) of biofilm formation by strain ALG-00-530 on glass. Attached cells were approximately 20 μm in length, although they ranged in size from 5 to 50 μm (Fig. 2A). Conversely, the planktonic cells used as controls remained within the normal range for F. columnare (3 to 10 μm) at 24 h (data not shown). As microcolonies developed (Fig. 2B), cells formed complex three-dimensional structures (Fig. 2C). Blebbing-like vesicles were observed in some cells (Fig. 2D), as well as in the control planktonic cells. Cells in mature biofilms (>24 h) produced EPS that was clearly visible by SEM (Fig. 2D and 3). It is noteworthy that the different fixations methods used influenced the observed structure of the EPS (Fig. 3). The combination of formalin fixation with HMDS seemed to eliminate more EPS and extracellular residues and offered a better view of the individual cells (i.e., membrane blebbing) (Fig. 3A). However, samples fixed with osmium tetroxide maintained a large portion of the extracellular matrix and, as a result, presented a more complex biofilm structure with cells embedded in the organic secretions (Fig. 3B).

Fig 2.

Colonization and biofilm development by F. columnare strain ALG-00-530 on glass slides analyzed by SEM. (A) Cells attached at 12 h postinoculation (scale bar, 30 μm). (B and C) Details of cell aggregation at 48 h (scale bars, 20 μm). (D) Closeup of F. columnare cells in the early stages of EPS secretion (scale bar, 1 μm). The black arrows indicate bacterial surface blebbing (SB), and the white arrow indicates extracellular polysaccharide.

Fig 3.

SEM of F. columnare cells attached to glass slides and fixed with HMDS (A) and osmium tetroxide (B). Arrows indicate EPS and bacterial cells (BC). Scale bars, 2 μm.

The composition of the EPS was examined by staining the biofilm with calcofluor white (binds generic polysaccharides) and Congo red (specifically binds β-1,4-linked sugars). Positive staining was obtained with calcofluor white, while Congo red did not stain the biofilm, indicating that β-1,4-linked sugars were not predominant in the EPS. Water channels were visualized with the aid of dextran-rhodamine, which revealed a mountain-and-valley structure with the water channels in the valley regions (see Fig. S1 in the supplemental material).

Biofilm viability.

The combined use of the Live/Dead cell viability kit and CLSM allowed the differentiation of cells that were active from those that had their cell membranes compromised. In general, live cells tended to remain in the center of the microcolonies while “dead” cells (or cells with compromised membranes) appeared on the periphery of the cluster and were surrounded by EPS (Fig. 4). An examination of the three-dimensional structure by using the tiled view showed that the bottom layers of the biofilm (in contact with the glass) contained a large number of dead cells while in the sections toward the top of the biofilm, viable cells become predominant, with few or no dead cells in the outermost layers (see Fig. S2 in the supplemental material).

Fig 4.

Viability of F. columnare cells in biofilm determined with the Live/Dead cell viability kit and examination by CLSM. Biofilm was grown on glass slides for 48 h postinoculation. Live cells are stained green, dead cells are red, and EPS is stained blue with calcofluor white. Scale bar, 10 μm.

Biofilm formation under flow conditions.

Differences in the ability to attach to glass surfaces and form biofilm under flow conditions were observed between the two strains tested (Fig. 5). Strain ALG-00-530 attached to the glass surface rapidly, within 5 h, and formed cell aggregates resembling biofilm that filled all of the channels in a short time (15 to 17 h postinoculation) (see Movie S1 in the supplemental material). These aggregates resisted the shear force caused by liquid flow and were observed for 4 days (the whole duration of the experiment). On the other hand, strain ARS-1 showed weaker attachment to glass surfaces, and occasionally cell aggregates were observed but only at the side walls of the channels, where shear stress is reduced (Fig. 5; see Movie S2 in the supplemental material). ARS-1 cells were easily removed from the glass surface by liquid flow.

Fig 5.

Biofilm formation by F. columnare inside microfluidic chambers. F. columnare strains ALG-00-530 and ARS-1 were introduced into microfluidic chambers kept under constant flow of MS broth, and growth and biofilm formation were monitored microscopically. Time zero indicates that the images were acquired 24 h after the introduction of bacterial cells into the chamber. The following columns show images captured 1 and 3 h after the cells were introduced into the chamber. Note that strain ALG-00-530 formed a biofilm that occupied most of the channels, while ARS-1 formed cell aggregates only at the bottom of the channel under reduced shear stress. Scale bars, 20 μm.

Biofilm virulence.

Channel catfish fingerlings were challenged with biofilm produced by strains BGFS-27 and ARS-1. Biofilms were examined by light microscopy and SEM and found to have a morphology similar to that of the biofilms produced by ALG-00-530 (data not shown). Fish exposed to BGFS-27 biofilm exhibited columnaris signs in less than 12 h postchallenge, and 100% mortality was observed within 48 h postchallenge. Only a few fish exposed to ARS-1 biofilm contracted columnaris and died as result (3 out of 20 fish). All moribund and dead fish showed the typical signs of columnaris disease. F. columnare was recovered from all of the dead fish. Conversely, fish in the control tanks remained healthy and did not show any sign of disease. These results showed that F. columnare in biofilm maintained the ability to infect channel catfish and cause columnaris disease.

Variables affecting biofilm formation.

All of the strains tested were able to attach to polystyrene plates to some degree, but we observed vast differences in attachment between strains. Because the primary objective was to assess the effects that individual parameters exert on biofilm formation at the species level, we averaged the results obtained with all of the strains and all of the replicates for each parameter tested. Figure 6 summarizes these results. To take into account how each parameter affected cell growth, the OD₅₉₅ was recorded prior to removal of the planktonic phase; thus, this value includes both planktonic and attached cells (total growth). It needs to be noted that OD was used as a proxy for measuring both cell growth and biofilm formation but the absorbance of crystal violet cannot be compared with that of unstained cells. Therefore, statistically significant mean cell growth and biofilm formation values were calculated separately.

Fig 6.

Adhesion (mean absorbance ± standard error) of F. columnare strains to microtiter plates. Light gray bars show cell growth, and dark gray bars represent attached cells at 48 h postinoculation. Panels A to E show the effects that temperature, pH, salinity, hardness, and the addition of carbohydrates had on cell attachment and proliferation. Different letters a to e indicate significant differences (P < 0.05) in cell growth. Different letters w to z indicate significant differences (P < 0.05) in cell attachment.

Figure 6A shows the effect of temperature on biofilm formation. Significantly more growth was observed at 28 and 35°C than at 21°C. No difference in biofilm formation was found between 21 and 28°C, but a significant decrease was observed at 35°C. The effect of pH on cell growth was more noticeable than that on biofilm formation. All of the pHs tested allowed the same amount of biofilm production. In contrast, salinity was able to modulate biofilm formation to a larger degree (Fig. 6C). Cells grew at a broad range of salinities, from 5 to 14 ppt, although significant differences were found. Cells growth was greatly inhibited by the absence of NaCl, and very little biofilm was detected under this condition. Biofilm formation was significantly promoted at 5 ppt but decreased at increasing NaCl concentrations. Hardness also played a significant role in biofilm proliferation (Fig. 6D). Bacteria incubated at 200 ppm or lower CaCl₂ · 2H₂O concentrations produced little biofilm, while a concentration of 360 ppm dramatically increased biofilm formation. Growth was similar at all of the concentrations tested. Finally, the addition of different carbohydrates to the culture medium had significant effects on both cell growth and biofilm formation (Fig. 6E). Culture medium supplementation with mannose significantly promoted biofilm formation compared to MS broth medium supplementation with other sugars, including fucose, galactosamine hydrochloride, glucose, and N-acetyl-d-glucosamine. Cell growth was reduced by galactosamine and glucose.

To determine which parameter(s) exerted a greater influence on the overall adhesion profile, we ran a PCA (see Fig. S3 in the supplemental material) in which all of the biofilm quantitative values of all of the parameters tested were used. Five components were responsible for 96% of the strain variability observed. Hardness at 360 ppm was the most influential component, explaining 67% of the diversity observed, followed by a temperature of 28°C (18%). Hardness at lower concentrations (64 to 200 ppm) along with mannose also increased biofilm formation. These six variables explained up to 98% of the variability found in all of the experiments.

Comparison of biofilm formation by the strains tested.

Overall, there were significant variations in individual strains when they were exposed to different medium conditions. To compare the biofilm formation patterns of all of the strains tested and to test our original hypothesis that different genomovars will exhibit different biofilm formation capabilities, we constructed a composite data set in BioNumerics with the information collected from the 23 parameters tested (three temperatures, five pH values, five salinities, five hardness concentrations, and five carbohydrates). The numerical value of each parameter was calculated as the arithmetic mean of four well replicates. The UPGMA tree derived from the clustering analysis is displayed in Fig. 7. Two groups could be inferred from the dendrogram, reflecting two main types of adhesion and biofilm formation patterns, with one strain (BGFS-25) clustering outside both groups. The adhesion profiles of strains grouped in cluster I showed higher overall values than those of strains in cluster II. The groups defined by this analysis do not correlate with the genomovar ascription, source, date of isolation, or virulence of the strains. Cluster I contained five and two genomovar II and I strains, respectively. Cluster II contained five genomovar I, four genomovar II, and two genomovar III strains. BGDS-25 is a genomovar II strain that showed an intermediate adhesion pattern.

Fig 7.

Clustering analysis of cell attachment using a composite data set with all of the parameters assayed. A similarity matrix was obtained by using the Pearson product moment correlation coefficient, and the dendrogram was generated by UPGMA. The scale bar shows the percent similarity between strains. The two main groups were delineated at 50% similarity.

DISCUSSION

In this study, we have characterized the dynamics of biofilm formation by the fish pathogen F. columnare under static and flow conditions. Previous studies of F. columnare adhesion mechanisms focused on host-pathogen interactions by using in vivo models (20) or proxies such as explants of living tissues (19) or polystyrene plates (32). Herein, we have shown that F. columnare can colonize a variety of man-made materials and have characterized the progression of biofilm development on different surfaces. Initial attachment to glass occurred between 4 and 6 h. During this first phase, the cells were observed gliding on the surface, as has been reported during the early stages of biofilm formation by other Gram-negative bacteria (33). An additional 6 h was required for the cells to lose motility and become immobile on the glass. This is longer that what has been observed in Pseudomonas aeruginosa (34) but probably explainable by the difference in growth rate between these two bacteria.

Interactions between aquatic bacteria and surfaces occurred at the boundary layer and typically under flow conditions since few aquatic systems are static (35). Using microfluidic chambers, we were able to mimic the conditions that F. columnare encounters in the aquatic environment. Formation of biofilm under water flow occurred on a time scale (5 h for early attachment and more than 12 for microcolony formation) similar to that of biofilm formation under static conditions. This contrasts with the short time (5 to 30 min) required for the cells to attach to host tissues while the host is actively swimming (8, 20). This time difference between in vivo and in vitro adhesion dynamics could be due to the positive chemotaxis toward host tissues that F. columnare displays (21). It is also plausible that cells needed more time to recognize and adapt to the artificial surface by inducing the expression of different cell membrane receptors required for adhesion. Interestingly, high-virulence strain ALG-00-530 was able to develop microcolonies that completely colonized the microfluidic chamber during the experiment while low-virulence strain ARS-1 attached to the chamber but remained confined to the sides. This contrasts with what was observed under static conditions since both strains developed microcolonies and mature biofilm in similar time frames (data not shown for ARS-1). In a previous study, we examined the adhesion capability of ARS-1 and observed that although it was able to attach to host tissues at high levels (equal to those of virulent strain BGFS-27), it failed to persist in the host over time (20). The mechanisms used by F. columnare to attach to a host or to man-made surfaces are still unknown because of the difficulty of creating mutants of this species (36). However, on the basis of our results, flow conditions both in vitro and in vivo exert different effects on biofilm formation, depending of the strain tested.

At our latitude, columnaris disease often occurs in late spring and early fall (8), which typically coincides with the optimal temperature for the pathogen (24 to 28°C). The temperature range in which F. columnare can grow actively is approximately 15 to 35°C. Because of the protective function of biofilm, it is possible that biofilm formation enhances the potential of these bacteria to survive in the aquatic environment. Kunttu et al. (37) found that adhesion was significantly enhanced when the temperature increased from 5 to 15 to 20°C but then decreased at 25°C. In our study, we also found that growth was influenced by temperature. At 21°C, cell growth and biofilm formation were lower than those observed at 28°C, probably because of a lower metabolic rate at this temperature. However, at 35°C, planktonic cell growth was not significantly different from that at 28°C but biofilm proliferation was significantly reduced, indicating that at higher temperatures cells prefer to remain in the water column. Our results agree with those of Kunttu et al. (37) in that an increase in temperature inhibited biofilm formation, but we observed the inhibitory effect at a higher temperature. This might be due to the sources of the strains. Their isolates were recovered from cold-water trout, while ours came from warm-water species.

The use of alkaline pH has been shown to have an inhibitory effect on biofilm formation by Staphylococcus sp. (38). In our experiment, the pH of the medium had little effect on planktonic cell growth and biofilm formation, which agrees with previous studies that reported a broad pH tolerance of F. columnare (2). In contrast, salinity exerted a significant influence on the development of biofilm. Altinok and Grizzle (39) have previously studied the effect of salinity on bacterial growth and on in vitro attachment to polystyrene plates. They found that bacterial growth at 1.0 or 3.0 ppt salinity was significantly higher than in control medium (0.3 ppt), but in vitro adhesion of bacteria was reduced with increasing salinity. Our findings are in agreement with theirs, as we found that absence of NaCl significantly inhibited bacterial growth and affected both planktonic multiplication and biofilm proliferation. However, planktonic cells were able to maintain similar growth patterns under different NaCl concentrations while biofilm formation was significantly inhibited by the increased salinity. Several studies have reported that the growth of F. columnare was completely inhibited by 10 ppt NaCl (40, 41), but we observed bacterial growth at salinities as high as 14 ppt NaCl, which might be explained by the use of different culture conditions.

F. columnare can survive for long periods in distilled water (11, 12), but calcium has a positive effect on both its survival (42) and its growth (43). A calcium carbonate hardness of 70 ppm is considered optimum for F. columnare growth, while levels lower than 50 ppm have been shown to decrease its survival (44). Our study showed that planktonic cells had optimum growth at 64 ppm, whereas biofilm formation was greatly promoted at a higher hardness level (360 ppm). Calcium has been shown to increase surface attachment and biofilm production by many other bacteria (45–47), although in a few bacterial animal pathogens it has been shown to have an inhibitory effect (48, 49). It is suggested that divalent cations may be important for the maintenance of biofilm structure by acting as bridging agents within the three-dimensional EPS matrix (13). In addition, calcium modulates EPS production in different bacteria (50, 51).

Decostere et al. (19) studied the adhesion of F. columnare to gill tissue and found that carbohydrates were the main components of the host receptors. In their study, the addition of d-glucose and d-galactose significantly reduced bacterial attachment to host tissues. Similarly, Klesius et al. (31) found that d-glucose, d-galactosamine, and d-mannose inhibited the chemotactic response of F. columnare to catfish mucus. They concluded that carbohydrate-binding receptors on the bacterial capsule were involved in the chemotactic response. We found that none of the carbohydrates tested reduced biofilm formation. This can be explained if the bacterial ligands that attached the cells to the polystyrene wells lack carbohydrate-binding domains. A positive effect was observed when d-mannose was added to the medium, which significantly increased biofilm formation. Mannose is known to be a receptor for different bacterial adhesins; thus, the addition of mannose can block attachment to surfaces but the opposite effect was observed in F. columnare (52). Why d-mannose enhanced biofilm formation warrants further investigation.

All of the strains used in this study were divided into two groups on the basis of their adhesion profiles when all of the parameters tested were taken into account. The effects that each parameter had on the two groups were similar, but overall, the strains in group I exhibited greater adhesion to polystyrene plates than those in group II. No correlation was observed between the adhesion of groups I and II and the genomovar, virulence, or the date or source of isolation. Previous studies have shown that adhesion to host tissues by F. columnare was correlated with virulence (6, 53) while attachment to polystyrene was not (32, 37). We have further confirmed this by using a broader diversity of isolates of F. columnare.

The results obtained in this study describe for the first time the formation of biofilm by the fish pathogen F. columnare under both static and flow conditions. Mature biofilms were characterized by presenting complex three-dimensional structures with water channels and abundant EPS similar to those of other Gram-negative bacteria (33, 54). SEM observations were affected by the sample preparation methods used, which should be taken into account in future studies. An important outcome of this study is that F. columnare retained its virulence in biofilms. Therefore, best-management practices in aquaculture facilities should avoid the water quality parameters identified in this study as biofilm promoters. High temperature (>28°C) and high salinity (>5 ppt NaCl) significantly inhibited biofilm formation and could be used as prophylactic measures. In contrast, high hardness (360 ppm) had a striking positive effect on biofilm formation; therefore, the use of lower water hardness is recommended to prevent columnaris disease. All of these parameters, temperature, salinity, and water hardness, are difficult to control in commercial catfish ponds but can be maintained in hatcheries, in which columnaris disease can cause more than 90% mortality in catfish fingerlings (4). Although further studies are needed to assess if F. columnare persists as biofilm in aquaculture settings, our study provides the baseline to determine the effects that temperature, salinity, and hardness will have in a commercial operation.