A study of biofilm formation in marine bacteria isolated from ballast tank fluids

Seawater-compensated fuel ballast systems maintain ship stability as fuel is spent but can introduce microorganisms that form biofilms, biodegrade fuel components, or enhance corrosion; all of which increase operating and maintenance costs. The aim of this study was to isolate planktonic bacteria from ballast tank fluids and taxonomically classify those that formed biofilm in culture. Twenty-two isolates were identified as belonging to seven genera based on 16S rRNA gene sequencing. Of the seven genera represented Alteromonas , Pseudoalteromonas , and Brevundimonas strains produced quantifiable biofilm in crystal violet assays. To test the hypothesis that the level of bulk nutrients would influence the extent of biofilm formation, isolates were grown in a conventional marine medium and marine medium supplemented with tryptone and yeast extract to represent standard and nutrient-replete media, respectively. While 80% of the Pseudoalteromonas strains produced 7 to 11-fold more biofilm in conventional medium vs. the supplemented medium, 64% of Alteromonas strains produced up to 50-fold less in the same medium. These results suggest that bulk nutrients influence the extent of biofilm formation in a taxa-specific fashion in these marine organisms. The sole Dasania marina isolate failed to display considerable biofilm growth in either media but was the only isolate to produce quorum sensing molecule(s), N-acyl homoserine lactone(s), in assays using an Agrobacterium tumefaciens reporter strain. Whether N-acyl homoserine lactones produced by D. marina could modulate biofilm formation in the other organisms isolated in this study would require further investigation.


Introduction
Biofilms are a form of fouling where microorganisms attach to a surface and produce a protective matrix that allows them to persist in hostile environments [1]. Several factors have been shown to influence the extent of biofilm formation including nutrient availability [2,3] and quorum sensing [4]. Biofilms present in marine systems can lead to increased maintenance costs or material damage [5]. For example, in maritime transport biofilm that accumulates on the hull of a ship promotes macrofouling by barnacles and mussels, and the increased drag increases fuel consumption by an estimated 30 to 50% [6]. Other examples of increases in biofouling-related maintenance costs include an annual increase of 2 to 4% due to fouling of ship ballast water [7] and 20 to 30% due to corrosion [8].
The Navy uses a compensated fuel ballast system in several classes of its ships, including cruisers and destroyers. Each system consists of a series of tanks connected in sequence and as fuel is spent seawater gets pumped in to compensate for the lost volume [9]. While this mechanism maintains ship stability during deployment, there can be unintended consequences to the ballast system infrastructure or fuel when seawater microorganisms are introduced. Several studies have demonstrated that some microorganisms can metabolize hydrocarbons present in fuels or enhance corrosion of the system [10][11][12].
The conditions for planktonic growth can be found within a ballast system [13] and include water, a permissible growth temperature and carbon source, and salt and nutrients which vary depending upon the route of the ship and type of water used for compensation [14]. In 2001, Drake et al. estimated that the ballast water from a ship can harbor up to 10 9 planktonic cells per L [15], and although the ballast water gets ejected during refueling operations, biofilm material remains attached to the walls of the system [16]. High-throughput DNA sequencing studies have shown that the composition and relative abundance of Abstract Seawater-compensated fuel ballast systems maintain ship stability as fuel is spent but can introduce microorganisms that form biofilms, biodegrade fuel components, or enhance corrosion; all of which increase operating and maintenance costs. The aim of this study was to isolate planktonic bacteria from ballast tank fluids and taxonomically classify those that formed biofilm in culture. Twenty-two isolates were identified as belonging to seven genera based on 16S rRNA gene sequencing. Of the seven genera represented Alteromonas, Pseudoalteromonas, and Brevundimonas strains produced quantifiable biofilm in crystal violet assays. To test the hypothesis that the level of bulk nutrients would influence the extent of biofilm formation, isolates were grown in a conventional marine medium and marine medium supplemented with tryptone and yeast extract to represent standard and nutrient-replete media, respectively. While 80% of the Pseudoalteromonas strains produced 7 to 11-fold more biofilm in conventional medium vs. the supplemented medium, 64% of Alteromonas strains produced up to 50-fold less in the same medium. These results suggest that bulk nutrients influence the extent of biofilm formation in a taxa-specific fashion in these marine organisms. The sole Dasania marina isolate failed to display considerable biofilm growth in either media but was the only isolate to produce quorum sensing molecule(s), N-acyl homoserine lactone(s), in assays using an Agrobacterium tumefaciens reporter strain. Whether N-acyl homoserine lactones produced by D. marina could modulate biofilm formation in the other organisms isolated in this study would require further investigation. dominant taxa in bulk seawater differ from that within biofilms that form on surfaces submerged in seawater [17,18] suggesting that certain taxa, not all, contribute to biofilm formation. The aim of this study was to isolate planktonic bacteria from ballast tank fluids, classify those that formed biofilm in culture, and determine if the level of bulk nutrients in growth media influenced the amount of biofilm produced.

Bacterial strains and growth conditions
Bacterial isolates were obtained from the ballast tank fluids of four US Navy ships in 2014: USS Sampson (seawater in tank for 8 months), USS Higgins (seawater in tank for 4-6 months), USS Bunker Hill (seawater in tank 4 months), and USS Lake Champlain (seawater in tank 1 week). Pooled ballast tank water was plated onto Marine Agar 2216 (BD Difco) within 48 hours of receipt and plates were incubated at 26°C. Isolated colonies were re-streaked and grown for a minimum of three transfers to obtain pure cultures. Freezer stocks were made in Marine Broth 2216 (BD Difco) (MB), a common medium for culturing heterotrophic marine organisms, supplemented with 20% glycerol and stored at -80°C. Culturing of strains was carried out in MB and a nutrient-replete Marine Broth (MB/TY), which was MB supplemented with 5 g of tryptone and 4 g of yeast extract per L.

Taxonomic classification of ballast tank isolates
To identify bacterial isolates, genomic DNA was extracted using the PureLink Microbiome DNA Purification Kit (ThermoFisherScientific, US) following the manufacturer's instructions. The 16S rRNA gene was amplified by PCR using the forward primer (fd1) 5' AGAGTTTGATCCTGGCTCAG3' and reverse primer (rp2) 5' ACGGCTACCTTGTTACGACTT3' set [21] and DreamTaq PCR Master Mix (ThermoFisherScientific). The PCR reactions were sequenced at the sequencing facility at the University of Texas at Austin using the forward and reverse primers listed above and 5'CAGCAGCCGCGGTAA3' (519F) [22]. The Basic Local Alignment Search Tool (BLASTN) was used to align and identify each sequence. Sequences were submitted to GenBank under the accession numbers MK250492-MK250514. The sequence identifications were verified using the Ribosome Database Project Naive Bayesian rRNA Classifier Version 2.11 [23].

Assays for biofilm production and planktonic growth
Strains were grown overnight on Marine Agar and individual colonies were inoculated into 4 mL MB or MB/TY and incubated at 26°C with shaking (100 rpm) for 2 d. The starter cultures were observed for the presence of ring and floating biofilms at 2 d and 7 d, respectively and were used to inoculate 96-well plates for the crystal violet assay.
Crystal violet assays were performed as described [24]. Briefly, starter cultures were seeded into 96-well round-bottom plates at 1:100 dilution (OD 600 nm=1.0). The plates were incubated at 26°C for 2 d with shaking (100 rpm). The liquid culture was transferred to a 96-well flatbottom plate to measure planktonic growth at OD 600 using a Synergy HT microplate reader (Biotek Instruments, US), and the biofilm that remained attached to the wells of the plate was washed, stained with 0.1% crystal violet, and quantified by measuring the OD 550 as described by O'Toole [24].

AHL detection
A modified version of the procedure by Joelsson [20] was used to recover AHL from the supernatant of bacterial cultures. Briefly, 4 mL of log-phase cultures grown in MB were centrifuged at 13,500 x g for 3 min and the supernatant was subjected to extraction with 4 mL of ethyl acetate. The residue was resuspended in 400 μL (10X extract) acetonitrile and stored in a glass vial at -20 o C.
A variation of the method by Cha [25] was used to detect AHL on solid media. Briefly, the AHL-reporter strain A. tumefaciens KYC55 was grown on LB agar with antibiotics at 26 o C for 2 d, and isolated colonies were inoculated into 0.6 mL AT medium with antibiotics and grown for 24 h. The cells were seeded into 20 mL of warm melted AT agar (0.8%) supplemented with antibiotics and 40 µL of 20 mg/mL X-gal and poured to make AT agar plates. Five µL aliquots from each 10X extract were spotted onto AT agar plates, and plates were incubated at 26°C for 24 h. To quantify AHL in broth cultures, β-galactosidase assays were conducted as described by Joelsson [20]. The 10X extract from S. meliloti RM41 and 100 µM of the synthetic C8-HSL were included as positive controls.

Identification of ballast tank isolates
Twenty-two bacterial isolates were classified into three dominant phyla, with all but two belonging to Gammaproteobacteria. SDH6 was classified as Brevundimonas mediterranea belonging to Alphaproteobacteria and SD20 was classified as Bacillus adhaerens belonging to the Firmicutes (Table 1). SD20 was identified as the only gram-positive bacterium among the cohort. Of the 21 gramnegative strains, eight of the Gammaproteobacteria were classified as Alteromonas oceani S35 and three as A. tagae BCRC 17571. Five isolates were assigned to the genus Pseudoalteromonas, two were identified as Halomonas axialensis and the remaining Gammaproteobacteria were single organisms from the genera Dasania and Marinobacter. These data are consistent with a recent high throughput sequencing study showing that the most abundant Gammaproteobacterial OTUs from ballast waters from five separate ships belonged to Alteromonadales, which encompass the Alteromonas and Pseudoalteromonas genera [26].

Biofilm production in MB and MB/TY
All isolates were initially screened for biofilm production by observing cultures grown in glass tubes ( Figure 1). In doing so, two biofilm morphologies were observed: a ring and a floating disk, described by Mosharaf [27] as solid-air-liquid (SAL) and air-liquid (AL) biofilms, respectively. The ring morphology adhered to the wall of the glass tube and formed at the interface between the air and liquid culture (Figure 1, panel A). The floating disk morphology was attached to the surface of the culture and may or may not appear to adhere to the tube (Figure 1, panel B). All isolates except for the Marinobacter strain formed at least one of the two types of biofilm observed (Table S1). Alteromonas strains produced thick ring and floating biofilms with the exception of SD16, one of the three A. tagae isolates. It did not produce a visible ring and produced only a weakly visible floating biofilm. The five Pseudoalteromonas strains also produced both morphologies, but to a lesser extent than the Alteromonas strains. Finally, the single strains of Dasania, Marinobacter, and Bacillus produced very little or no detectable biofilms in glass tubes (Table S1).
Sixteen of the 22 isolates produced quantifiable biofilm in crystal violet assays in either MB or MB/TY, with OD 550 values greater than the media-only control (OD 550 =0.04) (Figure 2). This included all Alteromonas strains (Figure 2, panel A), four of the five Pseudoalteromonas strains (Figure 2, panel B), and the sole Brevundimonas strain (Figure 2, panel C). The remaining six strains produced very little to no biofilm when grown in either media and included strains belonging to the genera Halomonas, Dasania, Marinobacter, and Bacillus.
Although there were eight A. oceani strains and all produced quantifiable biofilm material, there was wide variation in their abilities to do so (Figure 2). SD8 produced significantly more biofilm material than any of the other strains when grown in MB (OD 550 =16.95±1.91). All but three isolates (SD4, SD17, SD18) produced more biofilm material when grown in MB vs MB/TY (Figure 2, panel A), with SD3 producing over 50-fold more. There was also variation in biofilm formation in the A. tagae isolates in that SD7 produced considerably more biofilm in MB than in MB/TY, and SD16, a weak biofilm producer, produced more biofilm in MB/TY. Interestingly, A. tagae (SD5), similar to SD7, produced more biofilm in MB (OD 550 > 4), but the biofilm material was not consistently stable to withstand the washes and resulted in some replicates being weakly stained in the same experiment. This accounted for the high standard deviation observed for SD5 grown in MB media. Furthermore, quantification of crystal violet in nutrientreplete medium was difficult since this strain produced a gel-like "pellicle" akin to the Alphaproteobacteria member Gluconacetobacter xylinus [28], which was ejected during the wash steps. Floating biofilms have been extensively studied in gram-positive bacteria such as Bacillus thuringiensis [29] or B. subtillus [28], where components of the matrix which allow it to float consist of proteins that form amyloid-like fibers and a hydrophobic layer. It is unknown if the gel-like matrix formed by SD5 in nutrient-replete marine medium consisted of similar proteins. Nevertheless, it was one of the only isolates to form this type of floating biofilm in microtiter plates.
In contrast to the Alteromonas strains which generally produced more biofilm in MB than in MB/TY, three of the five Pseudoalteromonas strains produced 7-to 11-fold more biofilm in MB/TY (Figure 2). This included the sole P. undina strain (SD31) and the two P. tetraodonis strains (SD1A and SD1B). Of the two P. shioyasakiensis strains (SD1 and SD10), only one produced quantifiable biofilm and did so at equivalent levels in both growth media. That the majority of the Alteromonas isolates produced more robust biofilms in MB, while isolates belonging to Pseudoalteromonas did so in MB/TY, supports the hypothesis that the level of bulk nutrients influence the extent of biofilm formation, and that it did so in a taxa-specific fashion. Finally, the sole B. mediterranea strain produced comparable levels of biofilm in both growth media (OD 550 =4.74±0.78 and 4.72±0.69) but the lowest overall planktonic growth (OD 600 ) of all strains in MB (Table  S2). The remaining five strains produced levels of biofilm that were comparable to background (OD 550 ≤ 0.04). For all isolates except SD5 (an Alteromonas strain), planktonic growth in the same cultures was higher in MB/TY than MB (Table S2).

AHL detection
Since quorum sensing is a variable that can influence biofilm formation and all of the biofilm-forming isolates in this study were gram-negative, they were subsequently screened on plates for the ability to produce AHL using the reporter strain A. tumefaciens KYC55. As shown in Figure 3, only D. marina (SD1D) produced AHL detectable by this reporter and did so in both growth media. Figure 3 panel C shows the controls used in this assay, demonstrating that the synthetic C-8 AHL and AHL produced by S. meliloti RM41 were detectable. When further experiments were performed to quantify β-galactosidase activity, the levels of activity of SD1D grown in either MB or MB/ TY were comparable to levels from an extract of the positive control strain, S. meliloti RM41, and the synthetic C8-HSL (Figure 4). Also shown in the figure are extracts from SDH6 (B. mediterranea) grown in MB and MB/TY. This strain produced comparable biofilm in both types of media, however AHL were not detectable. Extracts from the other isolates were negative for AHL production by the reporter strain (data not shown). This negative result for the Pseudoalteromonas strains was not expected as extracts from P. shioyasakiensis and P. tetraodonis strains isolated from sediments in the Zhejiang Province demonstrated a positive response in reporter assays using a different A. tumefaciens A136 reporter strain that responds to C6-to C14-HSL [30]. This range is within that of the reporter strain used in this study. That some of the isolates produced AHL that were not detected by the reporter cannot be ruled out but was beyond the scope of this study. Indeed, Huang et al. concluded that AHL bioassay results are not always consistent among different screening methods [31]. However, only the D. marina isolate (SD1D), which failed to produce quantifiable biofilm under   . AHL quantification using β-galactosidase assays. One hundred microliters of the reporter strain A. tumefaciens KYC55 were seeded into 2 mL of AT medium. Ten microliters of C8-HSL (positive control) or 10X extracts were added, the culture was grown to an OD 600 of 0.2 to 1, and β-galactosidase assays were conducted as described [20]. Data shown are the mean ± standard deviation (n = 2, two independent experiments). Standard deviations below 3 are not shown both nutrient conditions, tested positive for AHL using the reporter. Whether or not AHL released by SD1D could have an impact on biofilm formation when co-cultured with the other isolates is currently being investigated.

Conclusion
Approximately 70% (16 of 22) of isolates recovered from ballast tanks produced quantifiable biofilms in at least one growth media tested. Alteromonas isolates displayed enhanced planktonic growth in nutrient-replete medium, however biofilm formation was reduced under these same conditions for 70% (8 of 11) of them, suggesting that increased bulk nutrients limit biofilm formation for this group. Indeed, studies in other biofilm-producing strains, including Lysteria monocytogenes, showed a similar trend [32]. Alternatively, and unexpectedly, Pseudoalteromonads produced more biofilm when bulk nutrients were in excess.
A recent study demonstrated that Pseudoalteromonas strains were both cultivable on hydrocarbon-enriched media and abundant in oiled mesocosms [33], consistent with Pseudoalteromonas strains being abundant in hydrocarbon-laden environments, such as ballast tanks that hold fuel. Furthermore, Pseudoalteromonas, Alteromonas, Halomonas, and Marinobacter strains were identified as hydrocarbon- degrading bacteria enriched by the Deepwater Horizon oil spill in 2010 [34]. Whether or not these ballast fluid isolates can biodegrade fuel was beyond the scope of this study but is currently being investigated.
The reporter strain A. tumefaciens was selected for the screen since it detects a wide range of exogenous AHL [19] and those associated with biofouling include C4-to C12-HSL and 3-oxo-C6-to 3-oxo-C12-HSL [35]. AHL were not detected in culture extracts from the biofilmforming isolates in this study, suggesting that strains recovered from ballast tanks may not use AHL for biofilm formation, use AHL that is out of the range of the reporter strain, or possibly use a different quorum sensing system altogether if at all. Whether AHL produced by the D. marina isolate could modulate biofilm formation is being investigated. To the authors knowledge, very little is known about D. marina, outside of its identification and characterization in 2007 [36]. Thus, the possibility of this organism using its quorum sensing system has yet to be explored and is of interest for future studies.