Presence of Extracellular DNA during Biofilm Formation by Xanthomonas citri subsp. citri Strains with Different Host Range

Xanthomonas citri subsp. citri (Xcc) A strain causes citrus bacterial canker, a serious leaf, fruit and stem spotting disease of several Citrus species. X. alfalfae subsp. citrumelonis (Xac) is the cause of citrus bacterial spot, a minor disease of citrus nursery plants and X. campestris pv. campestris (Xc) is a systemic pathogen that causes black rot of cabbage. Xanthomonas spp. form biofilms in planta that facilitate the host infection process. Herein, the role of extracellular DNA (eDNA) was evaluated in the formation and stabilization of the biofilm matrix at different stages of biofilm development. Fluorescence and light microscopy, as well as DNAse treatments, were used to determine the presence of eDNA in biofilms and bacterial cultures. DNAse treatments of Xcc strains and Xac reduced biofilm formation at the initial stage of development, as well as disrupted preformed biofilm. By comparison, no significant effect of the DNAse was detected for biofilm formation by Xc. DNAse effects on biofilm formation or disruption varied among Xcc strains and Xanthomonas species which suggest different roles for eDNA. Variation in the structure of fibers containing eDNA in biofilms, bacterial cultures, and in twitching motility was also visualized by microscopy. The proposed roles for eDNA are as an adhesin in the early stages of biofilm formation, as an structural component of mature bacterial aggregates, and twitching motility structures.

eDNA is an important component of the extracellular matrix in biofilms for several Grampositive, Gram-negative bacteria [7][8][9][10][11][12] and archaea [13]. The presence of eDNA in the and 4) revealed similar extracellular structures to those observed with SYTO-9 and fluorescence microcopy. Microscopy observations of GFP tagged cells stained with propidium iodide Fig 1. Presence of eDNA at the early stages of biofilm formation in Xcc. Representative light (crystal violet, CV staining) and fluorescence (SYTO-9 staining) images of 72 h static cultures on LB or XVM2 media. Fibers were observed after both staining for strains Xcc 306 and Xcc 12879 A w at the early stages of biofilm formation. Fibers interconnected cells at different stages of aggregation, from one to several cells are shown. In XVM2 medium, fibers were thicker and more uniform after staining with CV and SYTO-9. eDNA in XVM2 medium appeared to cover the surface like a sheet in contrast to individual fibers produced in LB medium.
doi:10.1371/journal.pone.0156695.g001 (S1 Fig) indicated that fiber fluorescence is due to the presence of eDNA and not caused by bacterial cell aggregation. In all assays bacterial cells in a bacillary shape were distinguishable from the containing DNA fibers. These results confirm that most of the extracellular fibers have a high eDNA content.
Microscopic observations of biofilm development were in agreement with previously reported studies, since Xcc aggregation was higher in minimal than nutrient rich medium [31,42]. Less bacterial aggregation and biofilm formation for both wide and narrow host range strains of Xcc, Xcc 306 and Xcc 12879 A w , were observed in LB medium than in XVM2 (Figs 1 and 2). The difference in biofilm formation between culture media was more marked for Xcc 12879 A w than Xcc 306 [42]. eDNA appeared to be associated with long fibers of variable thickness and to an amorphous mass surrounding bacterial cells (Figs 1, 2, 3, 4 and S1 Fig). Fibers containing eDNA were observed in static cultures after 72 h (early stage of biofilm formation) as well as in mature biofilms (Figs 1 and 2). For strains Xcc 306 and Xcc 12879 A w on LB and XVM2 media, the fibers appeared to interconnect bacterial cells within the aggregates. However fibers in LB medium were longer, thinner and less connective than in XVM2 medium. The short and branched filaments in XVM2 appeared to provide higher bacterial connectivity and aggregate stability. Differences in the development of eDNA fibers were also observed between Xcc 306 and Xcc 12879 A w in exponential growth phase (Fig 3) and plate growth (S1 Fig) in LB medium. Similar results were observed in the twitching motility assay (Fig 4); fibers produced by strains Xcc 12879 A w and Iran 2 A Ã were longer than the fibers produced by Xcc 306 that correlated to the different phenotype of the twitching colonies demonstrated for narrow and wide host range strains.

Importance of eDNA in Xanthomonas biofilm formation
The role of eDNA in biofilm formation was estimated by measuring biofilm formation after DNAse I treatment at 0, 24, 48 or 72 hours post seeding (hps) by staining biofilms with crystal violet (CV) as described in Material and Methods. Transmission electron microscopy of Xcc cultures treated with DNAse showed reduction of extracellular fibers (Fig 5), confirming the DNAse activity and therefore the eDNA content. The bacterial population was not reduced by DNAse treatment (S2 Fig) and cell division in LB medium was shown under DNAse treatment ( Fig 5).
Biofilm formation was compared in XVM2 medium, XVM2 medium plus buffer, and XVM2 medium plus buffer and DNAse I at 20 or 40 Kunitz mL -1 . In most cases, reduction of biofilm formation was dependent on the timing of DNAse I treatment, but was not dependent on DNAse I concentration (Fig 6).
Reduction of biofilm was significant (p<0.05) for citrus xanthomonads and DNAse I application time. In contrast, DNAse I did not produce a statistically significant reduction of biofilm formation by Xc 1609 strain (p>0.05). Citrus Xanthomonas strains varied widely in response to DNAse I (Fig 6). In wide host range strains Xcc 306 and Xcc 62 biofilm formation was statistically reduced (p<0.05) when DNAse I was added at 0 hps but no reduction was observed at longer exposure times (p>0.05). For narrow host range strains, Xcc 12879 A w and Iran2 A Ã , Presence of eDNA in preformed biofilms of Xcc. Representative light (CV staining) and fluorescence (SYTO-9 staining) images of mature biofilms on LB or XVM2 media. Both Xcc 306 and Xcc 12879 A w strains were more aggregated in XVM2 than LB. A high level of aggregation for strain Xcc 12879 A w in XVM2 made it difficult to observe eDNA fibers in aggregates. In XVM2, strain Xcc 306 was less aggregated and CV and SYTO-9 staining revealed eDNA surrounding the cells. In LB both staining revealed long fibers interconnecting aggregates. and Xac F1 biofilm formation was reduced by DNAse I treatmentsat 0, 24 and 48 hps. Furthermore, biofilm produced by the Iran2 A Ã strain was reduced at 72 hps after DNAse I treatment. Representative fluorescence microscopy images of Xcc 306 and Xcc 12879 A w in LB broth at exponential growth phase stained with SYTO-9. A high level of eDNA fibers (marked with white arrows) were produced by both strains. Bacillary shape bacteria are distinguishable from the fibers that appeared to connect cells over a long distance. Xcc A w 12879 fibers were well developed while strain Xcc 306 fibers were more diffuse.  Twitching halos from Xcc 306, 12879 A w and Iran2 A* and light and fluorescence microscopy of twitching halos stained with CV or SYTO-9. Twitching motility was observed between the plate surface and the medium. CV and SYTO-9 staining of strains revealed similar twitching structures. Fibers observed after CV staining appeared to have a high eDNA content. Twitching fibres were longer for Xcc 12879 A w and Iran2 A* than for Xcc 306. Some fibers are marked with white arrows to highlight diference from bacillary shape bacteria associated.

Importance of eDNA in preformed biofilm
To evaluate the role of eDNA in mature biofilms, DNAse I treatments were performed for 1 h or overnight. Biofilm disruption was observed after both incubation periods but overnight treatment had greater effect (Fig 7). DNAse I treatment for 1 h did not affect strains Xcc 306 and Xc 1609, but significantly (p<0.05) ruptured biofilm for strains Xcc 62, 12789 A w , Iran 2 A Ã and Xac F1 (Fig 7A). Overnight DNAse treatment of Xcc strains affetcted up to 80-90% of the biofilm and no differences were observed between wide and narrow host range Xcc strains ( Fig 7B). Similar treatment did not significantly (p>0.05) disrupt biofilm of Xc 1609 compared to the buffer control. In summary, biofilm disruption by DNAse I was greater for Xcc strains than Xac F1 or Xc 1609.

Discussion
Biofilm formation is an important virulence factor in Xcc [3,31]. Furthermore several differences in biofilm structures and biofilm formation have been observed between wide and narrow host range strains of Xcc which was related to their ability to infect the host [42]. The variation in function of eDNA in biofilm formation between wide and narrow Xcc strains could also be associated with the ability to form biofilms and their structure. Extracellular DNA (eDNA) is an important component of the biofilm matrix in several bacteria, playing roles in adhesion, bacterial aggregate stability, nutrition, protection and gene exchange [9,10,[16][17][18][19]. In this study, eDNA was detected either at different stages of the biofilm or in cultures and during twitching motility in citrus Xanthomonas strains with different host range. The presence of eDNA in biofilms for citrus Xanthomonas was determined and compared to Xc.
eDNA in Xcc has been observed in fibers interconnecting bacterial cells, possibly associated with other extracellular components including proteins and polysaccharides. Amorphous eDNA was also observed which is possibly eDNA that had been recently secreted, but not processed into fibers. Fibers were observed during biofilm formation in LB and XVM2 media, however those fibers formed in XVM2 appeared to be thicker and more uniform than those in LB medium, which is consistent with previous reports of higher biofilm formation in XVM2 medium. In previous reports [31,42] biofilm formation varied between Xcc 306 and 12879 A w , especially in LB medium. In XVM2, a culture medium that mimics apoplastic conditions, several pathogenicity factors are expressed for Xcc strains [26,43] and among them, eDNA could be considered an important virulence component in the biofilm for plant infection.  Table 1). A, DNAse added at 0 h post seeding (hps); B, DNAse added at 24 hps; C, DNAse added at 48 hps; and D, DNAse added at 72 hps. The absorbance values were normalized to the control XVM2 plus buffer in order to compare the response for different strains. Error bars represent the standard deviation. Graphs are a representative assay of at least three assays with three replicates per assay. Statistical analyses were performed using STATGRAPHICS Plus, version 5.1 (Copyright Manugistics Inc.). doi:10.1371/journal.pone.0156695.g006 In P. aeruginosa, eDNA is a factor associated with twitching motility [44]. Twitching is enabled through the extension and retraction of the type IV pilus [45] which binds to DNA [46,47]. Herein, SYTO-9 staining of twitching assay plates revealed the presence of fibers containing eDNA during this type of movement recently been described for Xcc [32,48]. Moreover, twitching structures varied among narrow and wide host range strains of Xcc as also observed for biofilm formation on plant surfaces [42]. Twitching fibers of narrow host range strains were longer than those of the wide host range strains which had more interconnecting fibers, this may be related with the variable twitching halo shape between wide and narrow host range Xcc strains. Interestingly, Xcc cells were always connected by fibers containing eDNA. In P. aeruginosa, eDNA along with surfactants, were secreted onto the culture plate surface as the bacterium was undergoing motility [44]. In our study, Xcc appeared to track the eDNA in fibers possibly through the type IV pilus, since no surfactant activity was detected for Xcc (results not shown). Our hypothesis is that eDNA is associated with fibers on the plant surface in the initial stage of the colonization process, that involves motility and bacterial aggregation. Bacteria twitch across the surface by binding the type IV pilus with eDNA fibers deposited on the surface. Biofilms treatments with DNAse confirmed the role of eDNA in biofilm formation in citrus related Xanthomonas. In many cases, DNAse treatment reduced biofilm formation depending on the strain and the biofilm stage. DNAse did not cause mortality or reduce bacterial growth rate, but was limited to the effect on biofilm formation (S2 Fig). When DNAse treatment terminated, bacterial attachment to the surfaces and aggregation resumed thereafter. eDNA has been described as an adhesin in B. cereus [9], Staphylococcus epidermis, S. aureus, S. mutants [8] and P. aeruginosa [10]. Herein, similar eDNA properties were demonstrated for citrus strains of Xanthomonas. eDNA appeared to be important for the early stages of attachment by wide host range Xcc strains, whereas narrow host range strains required eDNA for biofilm development over an extended period, concurring with their higher requirements for biofilm formation. In planta, at 24 hpi, wide host range Xcc strains produced more complex structures in biofilms than the narrow host range Xcc strains and Xac strain which aggregated by producing long fibers [42]. The less interconnected fibers of the narrow host range Xcc strains and Xac strain may have rendered them more susceptible to DNAse disruption of aggregated cells and altered their biofilm formation. Other extracellular structures such as fimbria or flagella could influence biofilm formation and those structures may possibly be associated with eDNA since most of the fibers observed with CV stain were also stained with SYTO-9.
The role of eDNA in the stability of biofilms is important for bacteria including E. coli, B. subtilis [49], S. mutants [50], some strains of P. aeruginosa [11] and, as this study reports, strains of Xcc and Xac. Mature biofilms of citrus strains, but not the crucifer strain Xc 1609, were disrupted when exposed to DNAse. This observation suggests that the systemic pathogen Xc has a low dependence on eDNA for biofilm formation, surface attachment and stability of the biofilm or that eDNA is associated with other structures, making them inaccessible to DNAse I. Our hypothesis is that Xc moves systemically and aggregation would not be occurring after the initial stage of the plant colonization [51]. Hence, the role of eDNA in the different bacteria, depends on the structure and function of biofilm in plant colonization process in the Xanthomonas strain-host interaction.
Biofilm has been described as major factor for Xcc virulence and survival in citrus [3,31]. Several factors influence biofilm for wide host range Xcc strains. Mutants in the type IV pili produced biofilm but less structured than that produced by the wild type strain [32]. Non-fimbrilar adhesins have been described in the initial attachment of Xcc to abiotic surfaces [33]. Our findings provide a clue to a plausible role of eDNA acting in early stages of the biofilm as an adhesin in conjunction probably with other adhesins. Many studies describe the disruption of eDNA in animal bacterial pathogens, for examples, use of recombinant DNAse as a treatment against P. aeruginosa for cystic fibrosis patients [52] and extracellular DNAse (NucB) from B. licheniformis to inhibit and break bacterial biofilms [53]. Although biofilm formation and motility are well-described virulence factors for many plant pathogenic bacteria, occurrence and function of eDNA in plant pathogens in these processes is less well known. The determination of roles of eDNA in plant pathogenesis may lead to a better understanding of the plant-pathogen interaction and perhaps the development of tactics to disrupt eDNA to effect bacterial disease control.

Bacterial strains, culture media and growth conditions
Bacterial strains used in this study and their natural hosts are listed in Table 1. Two wide host range strains of Xcc A, and two narrow host range strains, A Ã and A w , were evaluated along with X. alfalfae subsp. citrumelonis and X. campestris pv. campestris. Bacterial strains were routinely grown in Luria Bertani (LB) broth (10 g tryptone, 5 g yeast extract and 5 g sodium chloride per litre) or on LB plates (1.5% bacteriological agar) at 27°C for 48 h.
In addition and to discriminate bacterial cells from eDNA contained in fibres, GFP tagged strains of Xcc 306 and Xcc 12879 A w were used [3,42]. GFP expressing strains were grown on LB broth or plates with the addition of kanamycin at 50 μl mL -1 (LB+K).

eDNA staining
To detect eDNA, samples were stained with SYTO-9 from the Live/Dead Bacterial viability kit (Molecular Probes Europe BV; Leiden, The Netherlands) and observed with fluorescence microscopy. SYTO-9 stains DNA and is also able to penetrate bacterial membranes [54,55]. Crystal Violet (CV) staining was also performed to visualize extracellular structures as previously described for flagella visualization of Xcc [56]. To determinate the presence of eDNA in biofilms, a Xcc 306 and Xcc 12879 A w colony was taken and seeded into 5 mL LB medium for an overnight period, a 60 μL aliquot was then seeded in 30 mL of LB medium an incubated for an overnight period. Finally bacteria in exponential growth phase (0.5-0.6 ODs) were washed twice in 10mM MgCl 2 and diluted to a concentration of 10 8 cfu mL -1 in 25 mL of XVM2 (20 mm NaCl, 10 mm (NH 4 ) 2 SO 4 , 5 mm MgSO 4 , 1 mm CaCl 2 , 0.16 mm KH 2 PO 4 , 0.32 mm K 2 HPO 4 , 0.01 mm FeSO 4 , 10 mm fructose, 10 mm sucrose, 0.03% casamino acid) [43,57] or LB media, and incubated without shaking at 27°C in a 50 mL borosilicate flask. In order to detect eDNA at the early stages of biofilm, bacteria growing and deposited on bottom of the flask were collected at 72 hours post seeding (hps) and stained either with SYTO-9 for 20 min in the dark or with CV. For mature biofilm eDNA detection, after 72 hps of static growth in 50 mL borosilicate flask, the liquid medium was decanted and the flasks were incubated an additional 72 h in dried condition in order to simulate natural situation. To stain mature biofilm, bacteria were collected by adding 50 μL of sterile distilled water (SDW) and scraping them from the bottom of the flask followed by the staining procedures described above.
To detect eDNA in exponential growth phase bacteria, an aliquot of 50 μL was collected from LB liquid cultures of Xcc 306 and Xcc 12879 A w and deposited onto a glass slide following the staining procedures described above.
To detect eDNA in plate growth Xcc 306 and Xcc 12879 A w carrying the GFP plasmid pUFZ75 [3,42] were collected from LB+K, stained with propidium iodide from the Live/Dead Bacterial viability kit (Molecular Probes Europe BV; Leiden, The Netherlands), deposited on a glass slide and observed with fluorescence microscopy.
To evaluate twitching motility, bacteria in exponential growth phase were washed twice with 10 mM MgCl 2 and suspended in 1.0 mL of MgCl 2 in a 1.5 mL centrifuge tube. Bacteria were then centrifuged at 10,000 g for 10 min and the supernatant was discarded. Bacteria were inoculated with a toothpick in PYM agar medium (Peptone 0.5%, Yeast extract 0.3%, Malt extract 0.3% and bacteriological agar 1%) supplemented with 2% glucose. Plates were incubated at 27°C for 7 d. In order to estimate the twitching production and observe the twitching halo the medium was removed and plates were washed and stained with 0.3% of CV. To observe extracellular structures and eDNA the twitching colony was stained with SYTO-9 or CV. After 20 min of staining, plates were washed with SDW and observed under the light and fluorescence microscope.

Determination of eDNA in biofilms
To confirm the presence of eDNA in biofilm, DNAse (Deoxyribonuclease I from bovine pancreas (2000 units mg -1 ); Sigma Aldrich Inc., St. Louis, MO, USA) was added at different concentrations to static cultures of Xanthomonas strains at different times.
In order to visualize the effect of DNAse in the extracellular matrix, bacterial broth cultures were treated with DNAse at a 40 Kunitz mL -1 concentration and observed by transmission electron microscopy (TEM) after staining with 1% uranyl acetate.
To evaluate biofilm formation, bacterial cultures in exponential growth phase, performed as described above, were washed twice with 10 mM MgCl 2 and suspended at 10 8 cfu mL -1 in XVM2 medium. XVM2 medium has been demonstrated to increase biofilm formation for Xcc strains when compared to other media such as LB [31,42]. Microtiter plate wells were filled with 200 μL of the bacterial culture and incubated without shaking for 72 h at 27°C. After removing the medium, bacteria were incubated for an additional 72 h in dried conditions, as described above for staining mature biofilms. Biofilm formation was measured after rinsing the plates with SDW and staining with 0.3% CV for 15 min. Excess stain was removed by rinsing the plates with SDW. Residual CV in each well was solubilized in 200μL of an acetone (20%), ethanol (80%) mixture and the absorption of the extract measured in a microtiter plate reader set at A570 nm wavelength. Absorption values for each strain and treatment were calculated as the mean of readings from three wells from at least three different assays. The means were compared by analysis of variance (ANOVA) and separated by Student-Newman-Keuls (SNK) multiple range test using Statgraphics Plus for Windows 4.1 (Statistical Graphics, Rockville, MD).
For the different treatments, DNAse I was suspended in 0.15 M of NaCl and then in DNAse I reaction buffer (10mM of MgCl 2 and 50% of glycerol suspended in 10mM of Tris-HCl at pH 7.5) as previously described [7]. DNAse I was added at concentrations of 20 or 40 Kunitz mL -1 at 0, 24, 48 and 72 hps to evaluate eDNA effect on biofilm development or on preformed biofilm. In order to determine if variation in biofilm was affected by the reaction buffer, a control treatment with the buffer without DNAse I was included.

Conclusions
Presence of eDNA in Xcc with different host range is described as an important component of filaments connecting bacteria during bacterial growth, biofilm formation and twitching motility. In addition, fibers comprising eDNA were thicker and more connective during biofilm formation by different Xcc strains in certain culture media.
The role of eDNA in biofilm formation was determined for citrus and crucifer pathogenic xanthomonads. Whereas DNAse treatments did not produce a significant effect on biofilm of Xc 1609 strain, for citrus xanthomonads, both Xcc and Xac, biofilms were reduced after DNAse treatments performed at early stages of biofilm formation as well as in mature aggregates. Furthermore, DNAse effects differed between narrow and wide host range strains of Xcc in early stages of biofilm; the narrow host range strains were affected over a longer period of time. Once the biofilm was established no differences were shown between the two Xcc strain types. Those results demonstrate the role of eDNA for biofilm formation and maintenance for citrus xanthomonads strains, at the early stage of biofilms and as structural component.
The presence and potential importance of eDNA in biofilm formation by citrus xanthomonads opens the door for future studies of new control methods for citrus bacterial canker disease based on the disruption of biofilms or interference in their formation.