Prevention of Biofilm Formation and Removal of Existing Biofilms by Extracellular DNases of Campylobacter jejuni

The fastidious nature of the foodborne bacterial pathogen Campylobacter jejuni contrasts with its ability to survive in the food chain. The formation of biofilms, or the integration into existing biofilms by C. jejuni, is thought to contribute to food chain survival. As extracellular DNA (eDNA) has previously been proposed to play a role in C. jejuni biofilms, we have investigated the role of extracellular DNases (eDNases) produced by C. jejuni in biofilm formation. A search of 2791 C. jejuni genomes highlighted that almost half of C. jejuni genomes contains at least one eDNase gene, but only a minority of isolates contains two or three of these eDNase genes, such as C. jejuni strain RM1221 which contains the cje0256, cje0566 and cje1441 eDNase genes. Strain RM1221 did not form biofilms, whereas the eDNase-negative strains NCTC 11168 and 81116 did. Incubation of pre-formed biofilms of NCTC 11168 with live C. jejuni RM1221 or with spent medium from a RM1221 culture resulted in removal of the biofilm. Inactivation of the cje1441 eDNase gene in strain RM1221 restored biofilm formation, and made the mutant unable to degrade biofilms of strain NCTC 11168. Finally, C. jejuni strain RM1221 was able to degrade genomic DNA from C. jejuni NCTC 11168, 81116 and RM1221, whereas strain NCTC 11168 and the RM1221 cje1441 mutant were unable to do so. This was mirrored by an absence of eDNA in overnight cultures of C. jejuni RM1221. This suggests that the activity of eDNases in C. jejuni affects biofilm formation and is not conducive to a biofilm lifestyle. These eDNases do however have a potential role in controlling biofilm formation by C. jejuni strains in food chain relevant environments.


Introduction
A biofilm is defined as a mono-species or multi-species population of bacterial cells, which is attached to a surface and surrounded by an extracellular polymeric substance (EPS) [1]. The matrix composition is highly variable, and is dependent on the microbial species populating the biofilm, but generally contains nucleic acids, proteins and polysaccharides [2]. The EPS is studentship, where academic and industrial partners collaborate and where the student has access to training opportunities at the industrial (CASE) partner. Author Roy P Betts contributed to the manuscript by critical analysis of data, suggestions for further research and by assistance with writing of the manuscript, and general supervisory assistance to the first author (who is a PhD student). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: Roy P. Betts, who is an employee of the organisation (Campden BRI) which provided a top-up funding to the PhD-studentship of first author Helen L. Brown. The authors want to confirm that this does not alter their adherence to all PLOS ONE policies on sharing data and materials.
Campylobacter growth for biofilm assay C. jejuni culture for biofilm formation was carried out as described previously [35]. Briefly, C. jejuni from Skirrow plates were used to inoculate Brucella broth then grown overnight as a shaking culture (37°C, microaerobic conditions). Following overnight growth, cell cultures were adjusted to an A 600 of 0.05 in Brucella medium. To allow biofilm formation, 1 ml of this solution was added to a sterile borosilicate glass test tube (Corning). Tubes were incubated at 37°C in either microaerobic or atmospheric air conditions for 48 hours before staining.
For biofilm formation on glass slides, 20 ml of C. jejuni culture of A 600 = 0.05 was added to a 50 ml tube (Corning) containing a sterile twin frost borosilicate glass microscope slide (VWR) and incubated statically at 37°C for 48 hours. Following incubation the slide was gently washed in sterile water and fixed by incubation in 4% formalin for 15 minutes before drying. Slides were stored at 4°C in the dark until use.
Biofilm degradation by strain RM1221 was performed by allowing biofilms to form for 24 h before adding a second (1 ml) volume of either fresh Brucella medium or diluted cell suspension. Biofilm cultures were then incubated for a further 24 h before viability assessment and crystal violet staining. Where spent media was used in the secondary incubation, instead of cell suspensions, the spent medium was prepared from overnight cultures of C. jejuni. The cells were subsequently pelleted by centrifugation, and the supernatant filter-sterilised using a 0.2 μm polyethersulfone filter, and frozen at -20°C until required.

Crystal violet staining
Cell suspensions were removed, and the tubes were washed with water and dried at 60°C for 30 min, followed by addition of 1 ml of 1% w/v crystal violet solution. Tubes were further incubated on a rocker at room temperature for 30 min. After incubation, the non-bound dye was removed from the tubes by thorough washing in water followed by drying at 37°C. Bound crystal violet was dissolved by adding 20% acetone/80% ethanol and incubating on a rocking platform for 15 min at room temperature. The resulting dissolved dye was measured at a wavelength of 590 nm using a Biomate 5 spectrophotometer (Thermo Scientific) [39].

Assessment of cell viability by culture
To determine the number of viable cells, the planktonic fraction was eight-fold serially diluted in PBS and 5 μl of each dilution spotted on Brucella agar plates. After two days of growth in microaerobic conditions, the dilution resulting in two or more colonies was recorded. Cell viability in biofilm assays was assessed upon initial addition of cultures into static culture and following static incubation, prior to crystal violet staining.
Creation of a C. jejuni strain expressing green fluorescent protein To generate a strain of C. jejuni that constitutively expressed GFP protein, strain NCTC 11168 was transformed with plasmid pCporAGFP + using standard protocols [40]. Plasmid pCpor-AGFP + contains the gfp gene from pWM1007 [41] under control of the C. jejuni porA promoter and a chloramphenicol resistance cassette, flanked by the 5' and 3' sequences of the cj0046 pseudogene [38]. Replacement of the cj0046 pseudogene with the GFP gene and chloramphenicol cassette was confirmed using the primers GFP fwdreadin, GFP revreadin, cat fwd readin and cat rev readin. Fluorescence was assessed by microscopy using a Zeiss 200M fluorescent and light microscope with Axiovision software.

DAPI staining of NCTC 11168 GFP biofilms
Biofilms previously grown on glass slides for 48 h were allowed to equilibrate to room temperature in dark, aerobic conditions, before staining with 4',6-Diamidino-2-Phenylindole Dihydrochloride (DAPI) using manufacturers guidelines (Invitrogen). Prior to addition of a coverslip, Slowfade Gold antifade reagent (Invitrogen) was added to the slide as recommended by the manufacturer. Slides were imaged using a Zeiss 200M fluorescent and light microscope with Axiovision software.

Visualisation of extracellular DNA from biofilms
Following static incubation to allow biofilm formation in microaerobic conditions, the supernatant was removed and the tubes were rinsed once with sterile PBS to remove loosely attached bacterial populations. Adhered cells were recovered from the surface of six borosilicate tubes and pooled: 1 ml of sterile PBS was added to the first tube and the adhered cells were gently resuspended using a sterile cotton wool swab. This suspension was removed and used to resuspend adhered cells from a second tube. This was repeated for all six tubes. The A 600 of the biofilm suspension was recorded, and the cells were diluted in sterile PBS to an A 600 of 0.3. A 20 μl aliquot of cells was mixed with 4 μl 6× gel loading buffer and loaded on a 0.9% agarose gel. A 1 kb ladder (NEB) was used for size comparison. Following 45 minutes electrophoresis in 0.5% TBE buffer at 100 V, the gel was stained with ethidium bromide, and DNA was visualised using a GelVue UV light and documented using a U:Genius gel documentation system (Syngene).
Creation of the C. jejuni RM1221 Δcje1441 mutant A C. jejuni RM1221 cje1441 mutant (hereafter referred to as Δ1441) was created by insertional inactivation of the cje1441 gene with a chloramphenicol resistance cassette. The cje1441 gene and flanking regions were PCR amplified using the primers 1441KO_RVPstI and 1441KO_F-DEcoRI and cloned into the pNEB193 plasmid (NEB). Subsequently the cje1441 was replaced with the cat cassette from pAV35 [40] by inverse PCR using primers 1441KO_2_fwd and 1441KO_2_rev. As strain RM1221 is non-transformable due to eDNase expression [16], in vitro methylation of the suicide plasmid was used to increase transformation efficiency [42]. Prior to electroporation, RM1221 cells were incubated on ice in 15% (v/v) glycerol, 272 mM sucrose, containing 10 mM EDTA for 1 hour. Following incubation the cells were washed with 15% glycerol, 272 mM sucrose to remove the EDTA and transformed using standard procedures [40].

Assessment of swarming and autoagglutination
Motility of C. jejuni was assessed on 0.4% agar plates, as described previously [38]. Briefly, C. jejuni overnight culture (5 μl) was spotted onto Brucella medium supplemented with 0.4% agar and 0.05% TTC (2,3,5 triphenyltetrazolium chloride) before incubation at 37°C in microaerobic conditions for 48 h. The size of the halos were measured and compared to show relative motility between strains and mutants tested. Autoagglutination was measured as described previously [43] by monitoring the decrease in A 600 over a 24 h period following incubation in a cuvette at room temperature in aerobic conditions.

Degradation of extracellular DNA by C. jejuni RM1221
Degradation of exogenous DNA was investigated using two separate experimental approaches: assessment of a) eDNA degradation by C. jejuni RM1221 during growth and b) the ability of C. jejuni RM1221 to degrade purified DNA over a fixed time period. To assess eDNase activity in the supernatant of growing cultures, overnight cultures of C. jejuni were pelleted and an aliquot of the supernatant was removed for DNase activity assessment. DNase I (Fermentas) and RNase (QIAgen) treatments were carried out following manufacturers guidelines and incubated at 37°C in a water bath for one hour. Degradation of purified DNA by C. jejuni strains NCTC 11168, RM1221 and the RM1221 Δ1441 mutant was also assessed over a fixed time period. C. jejuni RM1221 cells were allowed to form a lawn on Skirrow plates. The cells were removed from the plate and suspended in 2 ml Brucella medium before pelleting and washing twice in sterile PBS. Following washing, the cell concentration was measured and the culture diluted to an A 600 of 0.5 in sterile PBS. To digest purified genomic DNA, 50 μl of cell suspension was added to approximately 2 μg of genomic C. jejuni NCTC 11168 DNA, and incubated at 37°C in a water bath for up to three hours. At 30 min intervals, an aliquot was taken, the cells pelleted and the supernatant removed and frozen at -20°C until analysis. For degradation of plasmid DNA, plasmid pET28a was purified from either E. coli strain Top10 (dam + dcm + ) or M147 (damdcm -) using a commercial miniprep kit (QIAGEN). To generate linear DNA, a 999 bp fragment was amplified from C. jejuni NCTC11168 genomic DNA with primers cj1388comp_Fwd (5'-GGAGAATTCATGT-CAAACTATCCAAAG-3') and pCASO51gDNARevScreen (5'-CCTACAGCTATAATGA-TAGGCAAGG-3') using HotStarTaq (QIAGEN). DNA concentration was determined using a Nanodrop 2000 spectrophotometer (Thermo Scientific). Assays contained 70 ng DNA substrate and 1 μl cell suspension in a total volume of 10 μl. EDTA was added at a final concentration of 50 mM. DNase (Fermentas) was added at a final concentration of 1 U and reactions were supplemented with 1x DNase buffer (2.5 mM MgCl 2 ). Prior to electrophoresis, samples were mixed with 6x gel loading buffer and loaded onto a 0.9% agarose gel. A 1 kb ladder (NEB) was used for size comparison. Following 45 minutes electrophoresis in 0.5% TBE buffer at 100 V, the gel was stained with ethidium bromide, and DNA was visualised using a GelVue UV light and documented using a U:Genius gel documentation system (Syngene).
DNase I treatment of C. jejuni RM1221 Δ1441 biofilms Biofilms were grown for 48 h in glass tubes. A volume of 4 μl DNase I, to give a final concentration of 4 U/ml (Fermentas) and 4 μl DNase I buffer (Fermentas) was added to test tubes at the start of the incubation. Following treatment, biofilms were either re-incubated for the remaining 48 h incubation, washed and crystal violet stained or washed and a new volume of cell culture or Brucella medium added to the tube.

Identification of DNase-encoding genes in C. jejuni genome sequences
A total of 16 complete and 2781 draft genome sequences of C. jejuni were obtained from public collections such as pubMLST (http://pubmlst.org/campylobacter/) [44] (N = 2687), and the NCBI (http://www.ncbi.nlm.nih.gov/genome/browse/), and the Virginia Tech University PATRIC website (http://patricbrc.vbi.vt.edu/portal/portal/patric/Home) (N = 104) [45], and are listed with accession numbers and assembly status (S1 Table). Genomes were searched using MIST [46] and the BLAST+ (v2.28) suite with each individual gene of C. jejuni RM1221 CJIE1 (cje0213-cje0273), CJIE1 (cje0544-cje0601) and CJIE4 (cje1418-cje1474) [28]. Genes were considered to be present if matching 90% with the query sequence. Genomes were scored as positive for cje0256, cje0566 or cj1441 if a positive score for the respective gene was matched with 60% of the respective CJIE1, CJIE2 or CJIE4 genes being present. The MLST-clonal complex designation was determined for all genomes using MIST, with the definition file provided by the Campylobacter pubMLST website.

Statistical analysis
Statistical analysis was carried out using GraphPad Prism. At least three biological replicates (each with three technical replicates) were used to calculate median and interquartile range. Significance was measured using Mann-Whitney tests.

DNase-encoding genes are differentially distributed in C. jejuni strains
We investigated the distribution of eDNase genes in a collection of 2791 publicly available C. jejuni genome sequences, using the C. jejuni RM1221 CJIE1, CJIE2 and CJIE4 elements including the eDNase genes dns (cje0256), cje0566 and cje1441 as query sequences (Table 2). Of these genomes, 42% lacked any of the three eDNase genes. Orthologs of the dns gene were detected in 37% of genomes, whereas orthologs of cje0566 and cje1441 genes were detected in 22% and 14% of the genomes, respectively. Only 13% (353 of 2791) of the genomes contained more than a single DNase gene (Fig. 1), and only 25 genomes (0.9%) contained orthologs of all three DNase genes ( Table 2, S1 Table).
We also investigated whether the presence or absence of eDNase genes was associated with specific multi-locus sequence typing (MLST) clonal complexes [50].  (Table 2, S1 Table). Of the 25 genomes positive for all three eDNase genes, the majority (17/25) was of clonal complex ST-573. Some of the major MLST genotypes had none or relatively few isolates with DNase genes, such as ST-464, ST-283, ST-42 and ST-45. Most of these MLST-types are found within agricultural environments involved in food production [50], suggesting that genetic background and shared environments may play a role in transfer of the DNase gene-containing insertion elements.

C. jejuni strain RM1221 is unable to form a biofilm during static incubation
We selected C. jejuni strain RM1221 to further investigate the potential role of eDNase genes in biofilm formation, as it is one of the three isolates containing all three investigated eDNase genes. In previous studies investigating the role of chicken juice on biofilm formation by C. jejuni [39], we observed that chicken isolate RM1221 formed a poor biofilm in Brucella media alone. We confirmed this by comparing biofilm formation by C. jejuni strains NCTC 11168, 81116 and RM1221 using crystal violet staining, as there is a clear difference between the levels of biofilm formation of strains NCTC 11168 and 81116 versus strain RM1221, as the latter showed very little difference to the negative control (Brucella media only) (Fig. 2). Analysis by light microscopy showed that although RM1221 cells display initial attachment to the glass surface, this does not progress to the development of microcolonies (S1 Fig), unlike strains NCTC 11168 and 81116 [35,39]. Assessment of cell viability showed that there was no difference in viability between strains RM1221, NCTC 11168 and 81116 following static culture for up to 48 hours (S1 Fig). Strain RM1221 showed comparable levels of motility to NCTC 11168 in broth cultures, suggesting that the lack of biofilm formation was not due to reduced motility or absence of flagella [33,35]. The absence of biofilm formation by RM1221 was also not due to differences in growth in shaking cultures, nor chemotactic motility as measured by swarming, or autoagglutination, which were all comparable to strain NCTC 11168 (Fig. 2) and significantly higher than that of an aflaggelated mutant of NCTC 11168 (ΔflaAB).
C. jejuni RM1221 is able to degrade pre-existing biofilms of other C. jejuni strains We subsequently investigated whether the factors inhibiting biofilm formation by strain RM1221 are also able to affect biofilm formation of other C. jejuni strains. To test this, we grew   (Fig. 3). Biofilm levels were significantly reduced in tubes containing C. jejuni RM1221 in the secondary culture (Fig. 3), suggesting that not only is RM1221 a poor biofilm forming strain, but the presence of viable C. jejuni RM1221 can degrade a pre-existing Campylobacter biofilm.
To assess whether the negative effects of strain RM1221 on biofilm levels is due to the presence of the cells or an extracellular factor, cell-free media was prepared from C. jejuni RM1221 cultures grown under microaerobic conditions at 37°C overnight (see Materials and Methods). Cell-free supernatant from C. jejuni RM1221 culture was added to a 24 h biofilm culture of C. jejuni NCTC 11168, and this resulted in degradation of the biofilm to background levels (Fig. 3). As a control, cell-free supernatant from a C. jejuni NCTC 11168 culture did not affect biofilm formation. This suggests that the factor disrupting biofilm formation is soluble in spent media and is either actively secreted, results from cell lysis, or is a metabolic by-product.

Disruption of cje1441 restores biofilm formation and abolishes degradation of existing biofilms
One of the major differences between strains NCTC 11168, 81116 and RM1221 is the presence of the CJIE1-CJIE4 insertion elements (Fig. 1), of which each contain secreted proteins and secretion systems [15,16,51]. Since biofilms of C. jejuni strain 81-176 contain eDNA and are enhanced by the addition of exogenous DNA [33], we hypothesised that the ability of RM1221 to degrade biofilm is the result of a secreted DNase (cje0256 (dns), cje0566, or cje1441). Like strain 81-176, eDNA can be observed in a mature biofilm of strain NCTC 11168 (S2 Fig). We were able to inactivate the cje1441 gene in strain RM1221 by insertion of an antibiotic resistance cassette (see Materials and Methods). We were not able to inactive the dns or cje0566 genes despite repeated attempts, which confirms the proposed role of dns and cje0566 in preventing natural competence [15,16]. Likewise, we were unable to complement the cje1441 mutation, as constructs expressing the eDNase genes from a constitutive promoter invariably accrued spontaneous promoter mutations, suggesting that expression of C. jejuni eDNase genes in E. coli is toxic.
Inactivation of the cje1441 gene in RM1221 resulted in a significant increase in biofilm formation when compared to wild type RM1221, and produced similar levels of biofilm when compared to strain NCTC 11168 (Fig. 4). The level of biofilm eDNA of the Δ1441 mutant was comparable to stain NCTC 11168 (S2 Fig). Strain RM1221 did not form a biofilm and eDNA was not detected in these assays. Inactivation of cje1441 did not affect chemotactic motility, nor did it affect autoagglutination or growth (S3 Fig), suggesting that motility and flagellar expression were comparable to that of the parental wild-type strain RM1221. Biofilms formed by the Δ1441 mutant were sensitive to DNase I treatment (Fig. 4), supporting the observation that DNA is present in the ECM of the Δ1441 mutant and contributing to the biofilm structure. Furthermore, we observed that co-culture of strain NCTC 11168 and the Δ1441 mutant resulted in biofilm levels similar to those observed with NCTC 11168 alone, thus cje1441 contributes to the ability of RM1221 to degrade pre-formed C. jejuni biofilms (Fig. 4).

C. jejuni RM1221 is able to degrade exogenous DNA
To demonstrate the DNA-specific activity of the C. jejuni RM1221 eDNases, we mixed genomic DNA from C. jejuni strain NCTC 11168 with washed RM1221 cells, and incubated this mixture  at 37°C. No DNA degradation was observed when C, jejuni strain NCTC 11168 was mixed with genomic DNA (Fig. 5). However, addition of C. jejuni RM1221 resulted in time-dependent degradation of C. jejuni genomic DNA over a three hour time course (Fig. 5). DNA degradation was abolished in the C. jejuni Δ1441 mutant, suggesting that the CJE1441 eDNase makes an important contribution to the DNA degradation observed in the parental RM1221 strain (Fig. 5). We also tested whether RM1221 was able to degrade its own DNA, to exclude a role for DNA methylation. As with the NCTC 11168 genomic DNA, RM1221 genomic DNA was rapidly degraded, indicating that the DNase activity is non-specific (data not shown). Cell suspensions of RM1221 could also degrade linear PCR fragments and uncut plasmid DNA, both methylated and nonmethylated (S4 Fig). This DNase activity was inhibited in the presence of EDTA. This DNase activity was not detected in strain NCTC 11168 or the Δ1441 mutant.
Analysis of levels of extracellular DNA purified from overnight growth cultures of C. jejuni NCTC 11168, 81116 and RM1221 showed the presence of high molecular weight nucleic acids, running with the same mobility as genomic DNA, in the supernatants from strains NCTC 11168 and 81116 but not for RM1221 (Fig. 5). These fragments were sensitive to DNase I digestion but not RNase A treatment, suggesting that they are high molecular weight DNA molecules. As with previous experiments, we did not observe any significant differences in cell viability or growth, suggesting that variations in eDNA release between the C. jejuni strains is not caused by variations in viability (i.e. cell death and lysis) or rate of growth (data not shown).
Taken together these results suggest that C. jejuni NCTC 11168 releases DNA during growth, and that this DNA contributes to biofilm formation. RM1221 can degrade this DNA and thus disrupt both de novo biofilm formation and pre-formed biofilm. This activity is highly dependent on the CJE1441 eDNase, expressed from the cje1441 gene on the insertion element CJIE4. This DNase is likely an endonuclease that is dependent on metal ions for activity.

Discussion
Biofilms play an important role in the lifestyle of many bacteria, and cause both considerable problems in healthcare and the food industry. One problematic aspect of biofilm formation is its contribution to transmission and survival of bacterial and fungal pathogens. There is also now an increasing body of evidence that suggests biofilms may assist in C. jejuni food chain persistence [24,39,52] and recent work has shown that eDNA is important in the maturation of C. jejuni strain 81-176 biofilms [33]. A better understanding of the mechanisms involved in biofilm formation by C. jejuni could lead to development of applications targeting C. jejuni transmission in the food chain. In this study, we have shown that there are differences in biofilm formation between three C. jejuni reference isolates, and have shown that eDNase activity results in degradation of pre-formed C. jejuni biofilms, as well as prevention of de novo biofilm formation. This work highlights how naturally-occurring eDNase activity may be able to weaken or destroy natural biofilms, e.g. in food processing environments.
Most C. jejuni isolates are naturally competent, and readily take up DNA from the environment and in some cases, recombine this into their genome. One of the consequences is that C. jejuni shows a high level of genetic diversity, both at the sequence level and at the level of genetic content [29,53]. In this study we have used C. jejuni reference strain RM1221, which is not naturally transformable due to the expression of three eDNase genes from the CJIE1, CJIE2 and CJIE4 insertion elements [15,16]. Although the biological function of the eDNase activity in C. jejuni is yet to be elucidated, it is possible that it protects isolates with the insertion element against allelic exchange with insertion element-negative flanking sequences, as this incurs the risk of losing the insertion element, which offer some evolutionary advantage. Our investigation of a large (N = 2791) collection of C. jejuni genome sequences showed that the eDNase genes are differentially distributed in C. jejuni, and that very few genomes contain three copies of an eDNase gene. However, with 58% of the C. jejuni genomes included being positive for at least one eDNase gene, this suggests that there will be eDNase-expressing isolates present in many agricultural environments, and these may have a profound effect on biofilms produced or colonised.
We here also show that the expression of eDNase activity has another consequence, severely reducing biofilm formation by C. jejuni strain RM1221. Further support for a role of the eDNases in restricting biofilm formation was obtained by inactivation of the cje1441 eDNase gene in strain RM1221, which allowed RM1221 to form biofilms. The eDNase activity and lack of natural competence has so far precluded robust genetic manipulation of strain RM1221 (other than conjugation via tri-parental mating [41]), and our successful inactivation of cje1441 is to our knowledge the first genetic manipulation of the RM1221 chromosome. The eDNase genes pose technical problems for genetic manipulation and cloning, as their intracellular expression can lead to cytoplasmic DNase activity and cell death, thus hampering cloning and expression in E. coli. This has also been reported for eDNase proteins of other bacteria, such as the eDNase proteins from the predatory bacterium Bdellovibrio bacteriovorus, where expression was found to be lethal in E. coli [14]. Similarly, expression of active DNase I by E. coli could only be achieved by the use of the very tightly controlled expression plasmid pDOC55 [54]. Such plasmids are not available for C. jejuni, and hence genetic manipulation of the eDNase genes in C. jejuni is technically challenging.
The importance of eDNA in bacterial biofilms is now well recognised [4,33], and has attracted attention as a target for enzymatic or chemical treatment for disinfection purposes. DNase I is effective in interfering with the biofilms of the foodborne pathogens Listeria monocytogenes [55] and of E. coli [56], but also mixed species biofilms. Biofilms found in activated sludge flocs have eDNA from lysed cells forming close interactions with the viable cells within the biofilm [57], and affected microcolony formation within the biofilm. Similarly, mixed biofilms with Staphylococcus epidermidis and Candida albicans are also affected by DNase treatment [58], suggesting that DNase is able to modify both mixed species and mixed kingdom biofilms. Addition of exogenous DNase is effective in reducing biofilms of pathogenic bacteria, such Neisseria gonorrhoeae [59], Garderella vaginalis [6] and L. monocytogenes [60].
Many species which form biofilms are also able to produce and export extracellular DNase proteins, and eDNase proteins have multiple functions. The P. aeruginosa eDNase PA3909 is involved in DNA degradation, providing an additional nutrient source, and its expression is induced in phosphate limiting conditions [13], whereas in Shewanella oneidensis, expression of the nucleases ExeM and ExeS is strongly induced if DNA is the sole nutrient source, and deletion of the ExeM gene leads to a significantly reduced growth rate [61]. Finally, the eDNase genes of Staphylococcus aureus are involved in immune evasion, and their expression during host infection aids the escape of S. aureus from the DNA 'nets' which are secreted by neutrophils [8]. In the case of C. jejuni [15,16], and other bacteria such as Vibrio cholerae [62], the eDNase proteins restrict natural transformation.
Many bacteria which produce eDNase enzymes are still able to form biofilms and appear to utilise the enzymes in order to modify their biofilm structures. Two well-studied examples of eDNase-positive bacterial species that can form biofilms are V. cholerea [11] and S. aureus [63]. Since the eDNase genes of RM1221 are classified as non-specific DNA/RNA endonucleases, they are not expected to have stringent specificity regarding the source, methylation or sequence of the DNA targeted for digestion, and this was confirmed by absence of eDNA in RM1221 cultures (Fig. 5). We hypothesize that rapid degradation of eDNA restricts its deposition on surfaces, and as such inhibits the initial stages of attachment of C. jejuni to these surfaces [39].
It is important to note that not all bacterial species show reduction of biofilm formation following treatment with DNases. When the opportunistic pathogen Burkholderia cenocepacia was exposed to DNase it produced significantly denser biofilms [64], while Helicobacter pylori biofilms remain unaffected following treatment with DNase I [65]. Finally, the presence of eDNA on a surface inhibits biofilm formation by Salmonella enterica serovars Typhimurium and Typhi [66]. These examples show that DNase treatment may not be effective in the case of all single species biofilms. However many naturally occurring biofilms, such as are found in processing plants, are comprised of multiple species and so DNase treatment should still be considered an effective mechanism of at least partially degrading biofilms and allowing better penetration of antimicrobials.
Treatment of biofilm-based bacterial infections with DNases has increased in recent years. Impregnation of the biomaterial polymethylmethacrylate with DNase I lead to reduced adherence of P. aeruginosa and S. aureus, without a detrimental effect on adhesion and proliferation of human cells [67]. Human recombinant DNase dornase alpha (brand name Pulmozyme) is frequently used in the treatment of cystic fibrosis [68], and it also degrades DNA within biofilms isolated from children with recurrent acute otitis media [69]. DNase I treatment has also been shown to reduce established Bordetella bronchiseptica and B. pertussis biofilms from the mouse respiratory tract [70]. In vitro treatment of biofilms of non-typeable Haemophilus influenzae with DNase I also allowed increased bacterial killing by β defensins [71], this suggests that even in biofilms where DNase I treatment does not have a direct biofilm reducing effect it can still be a useful addition to a treatment regimen.
Treatment with DNase enzymes is becoming a common intervention in treatment of some biofilm infections and chronic conditions such as cystic fibrosis, but DNase production is costly. This is not considered problematic within the medical industry, but the high cost of production severely limits its potential for use in the food chain. Within the food industry, the use of naturally produced bacterial eDNases could be a suitable alternative to DNase I use. Bacteria such as Aeromonas sp. produce several secreted DNase enzymes [72] and in species such as Streptococcus agalactiae, some of these eDNase proteins are heat stable [73]. Many of the DNase-positive bacteria have low complexity growth requirements and do not have the ethical or legal issues, which may preclude or limit the use of DNase obtained from animals, or recombinant products from genetically modified organisms. The cell-free extracts of C. jejuni RM1221 retain their DNase activity, and are able to degrade C. jejuni biofilms even after a ten minute heat treatment (data not shown). This suggests that the eDNase enzymes of RM1221 are relatively heat stable and could potentially be a source of easily obtainable DNase proteins for use during food chain cleaning, although such an application requires further consideration and investigation to ensure that any supernatant derived products is safe for use, particularly from pathogenic bacteria.
In conclusion, eDNase activity inhibits biofilm formation by C. jejuni RM1221, and this eDNase activity can be utilised to degrade biofilms formed by other C. jejuni strains, using either live RM1221 cells or cell-free supernatant. Since DNase treatment has been proved to be so effective against both bacterial and fungal biofilms, extraction of eDNase enzymes from C. jejuni strains such as RM1221 could in future provide a cost effective alternative source of DNase enzymes, and assist in developing applications improving food safety by prevention of biofilm-assisted transmission of foodborne pathogens such as C. jejuni.  Table 1) counter stained with DAPI. A diffuse blue dye can be seen around the GFP-expressing cells suggesting that there is a large quantity of eDNA present within the mature biofilm. (B) Three biological replicates showing ethidium bromide-stained DNA isolated from biofilm samples from strains NCTC 11168, RM1221, and Δ1441 after agarose gel electrophoresis. (TIF) S3 Fig. C. jejuni RM1221 Δ1441 and its parent strain show no significant difference in swarming, autoagglutination or growth. C. jejuni strains NCTC 11168 (white), its non-motile ΔflaAB mutant (dark grey), RM1221 (black bars) and the Δ1441 mutant (light grey) were compared for their ability to swarm (A) and autoagglutinate (B). In both tests no statistical difference was observed between Δ1441 and the wild-type. Panel C shows growth over a 24 hour period for Δ1441 (light grey triangles), RM1221 wild-type (black circles) and NCTC 11168 (white squares). Bars represent the median, error bars show range and significance was measured using Mann-Whitney tests. (TIF)

Supporting Information
S4 Fig. C. jejuni RM1221 cell suspension has EDTA-dependent endonuclease (DNase) activity. Plasmid DNA (70 ng) was incubated with cell suspensions for 60 minutes at 37°C prior to agarose gel electrophoresis. Plasmid DNA is almost entirely degraded in reactions containing RM1221 cell suspension, but not NCTC 11168 or Δ1441. (TIF) S1 Table. Presence/absence analysis for homologs of the cje0256, cje0566 and cje1441 eDNase genes in C. jejuni. (XLSX)