Bacterial strains and plasmids used in this study are listed in Table 1. In this study, as needed, we used two variants of V. cholerae N16961 strain, including N16961S (stands for smooth variant) and N16961R (stands for “rugose” variant). The method used to convert smooth variant to rugose variant has been described previously (Ali et al., 2002). As needed, bacterial strains were subcultured from glycerol stock stored at -80°C onto L-agar and the culture was incubated overnight at 37°C. For growth in L-broth, a single colony was transferred to 3 ml L-broth and the culture was grown overnight at 37°C with a shaking speed of 250 × g in an environmental shaker (New Brunswick Scientific, Edison, NJ, United States). Unless otherwise indicated, for culture in FSLW, 1 ml of overnight L-broth grown culture was centrifuged, washed 2× in 1 ml FSLW and finally the pellet was resuspended in 1 ml fresh FSLW. As needed, antibiotics were used at the following concentrations: ampicillin, 100 μg ml^-1 and polymyxin B, 50 U ml^-1.

For this study we used fresh water collected in a sterile 1,000 ml Nalgene bottle (Nalgene, Rochester, NY, United States) from a 30.2-acre natural lake (Wauburg Lake) in Gainesville, FL, United states as described previously (Jubair et al., 2012). For experiment, aliquot (300 ml) of water was filter sterilized using Nalgene 0.22 μm membrane filter unit. Chemical analysis of filter sterilized lake water was performed by Advanced Environmental Laboratories, Inc. (Gainesville, FL, United States). The analysis of lake water revealed following major components: total organic carbon (12.0 mg/l), total nitrogen (1.4 mg/l), ammonia (0.06 mg/l), nitrate plus nitrite (0.014 mg/l), sodium (7.3 mg/l), chloride (14.0 mg/l), calcium (7.0 mg/l), iron (0.03 mg/l), potassium (0.93 mg/l), total phosphate (0.26 mg/l), orthophosphate (0.002 mg/l), and magnesium (1.6 mg/l). Interestingly, low nutrient composition found in Wauburg Lake water was similar to what has been reported by us using the same lake water (Jubair et al., 2012), and in pond water obtained from Bangladesh and other fresh waters obtained from United States (Kierek and Watnick, 2003a; Nelson et al., 2008).

To determine if GASP-700D (Jubair et al., 2014) sustained a mutation in the flrA gene, we PCR amplified the gene from the GASP-700D genome with two convergent PCR primers [aa778 and aa780 (Supplementary Table 1)] using standard PCR conditions. The amplicon was purified using Qiagen PCR Purification Kit (Qiagen, Valencia, CA, United States) and sequenced with help from the Interdisciplinary Center for Biotechnology Research (ICBR) at University of Florida; sequence analysis revealed that GASP-700D sustained a 49-bp deletion mutation in flrA gene (we hereafter referred this strain as GSΔflrA).

For creating an identical 49-bp deletion mutation in flrA gene in the background of wild-type V. cholerae N16961S strain (Table 1), a one-step PCR cloning strategy was employed using primers, including aa778S and aa780S (Supplementary Table 1) carrying SalI and SacI site at the 5′-end of each primer, respectively. Genomic DNA of GSΔflrA was used as PCR template while primer pairs (aa778S and aa780S) flanking the 49-bp deletion in GSΔflrA were used to amplify the PCR product using standard PCR conditions. The purified PCR product was digested with SalI and SacI, the digested product was purified, and ligated into a similarly digested suicide vector, pCVD442 (Donnenberg and Kaper, 1991); the ligated product was transformed into E. coli SM10 λ pir using standard transformation procedures resulting in plasmid pAAS82. E. coli SM10 λ pir carrying pAAS82 was conjugated to V. cholerae N16961S; selection of transconjugants and counter selection for the flrA mutation resulting from homologous recombination were performed as described previously (Jubair et al., 2014). PCR and DNA sequencing methodologies were used to confirm the 49-bp mutation in flrA gene [AAS84 (N16961SΔflrA), Table 1]. An in-frame deletion in the ΔmshA gene was created in the backgrounds of N16961S and GSΔflrA by using primers and methodologies described previously (Thelin and Taylor, 1996). A two-step PCR cloning strategy, with a set of targeted primers, was used to create a null mutation in cheY-3 gene in the backgrounds of N16961S, N16961SΔflrA, and GSΔflrA strains. Mutation in vpsA gene (VC0917) encoding UDP-N-acetylglucosamine 2-epimerase (Ali et al., 2000b), a key component of vps-dependent biofilm in the background of V. cholerae N16961S, N16961R, and GSΔflrA strains was created as described previously (Jubair et al., 2014). To complement mutants, each wild-type gene of interest was PCR amplified using two convergent PCR primers targeting that gene; chromosomal DNA of N16961S was used as the template for PCR reactions. Following purification of the PCR product, the gene was cloned into pWSK29 vector (Wang and Kushner, 1999). For complementation of ΔmshA, we used plasmid vector pMMB67EH carrying wild-type mshA gene cloned into the vector (complementation plasmid is a kind gift from Andrew Camilli of Tufts University) (Table 1). The complementing plasmid was introduced into the V. cholerae mutant strain of interest using DNA electroporation (Ali et al., 2000a). All PCR primers used either in gene disruptions or in cloning wild-type genes are listed in Supplementary Table 1.

Swarming behavior of V. cholerae strains on motility agar (0.3% agar) was examined as described previously (Gardel and Mekalanos, 1996). Briefly, each strain of interest was grown overnight in L-broth at 37°C with a shaking speed of 250 × g. A sterile inoculating wire was dipped into the culture and the wire containing the culture was then inoculated onto motility agar. Zone of migration of each bacterial strain around the inoculating site was examined and the image was captured after incubation of the microorganism in motility agar at 37°C for 8 h.

Quantification of biofilm production by V. cholerae strains grown in L-broth was measured as described previously (Watnick and Kolter, 1999). For measurement of biofilm production in FSLW microcosm, a single colony of V. cholerae grown on L-agar was transferred into 3 ml L-broth and the culture was incubated overnight at 37°C with a shaking speed of 250 × g. One milliliter of culture was spun down and washed 2× with 1 ml of FSLW. A 24-well polystyrene plastic plate (Corning Incorporated, Corning, NY, United States) was used as the surface for bacterial attachment; 50 μl of culture was added to 450 μl of fresh FSLW (10-fold dilution) to obtain ca. 10⁸ cfu/ml in each well (six replicates). The culture was incubated overnight statically at ambient temperature. Following incubation, the culture was discarded from the wells and washed 2× with water to discard any residual bacterial cells. Crystal violet (CV) dye was added to each well and the plate was incubated at room temperature for 30 min; the CV dye was discarded and the well was washed (2×) with water. Finally, cell-associated CV dye was extracted with 600 μl of dimethyl sulfoxide (DSMO) (Sigma, St. Louis, MO, United States) and the quantitative biofilm production was determined as described previously (Jubair et al., 2014).

In order to determine the three-dimensional architectural structure and thickness of the biofilm, confocal microscopic analysis was performed as described previously (Jubair et al., 2014). V. cholerae strains, including N16961S, N16961SΔflrA, GSΔflrA, GSΔmshA, and GSΔmshA/pAA159 (Table 1) were included in the assay. Briefly, a single colony of each strain was inoculated into 3 ml L-broth and the culture was incubated and processed as described above. An inoculum (ca. 10⁸ cfu/ml) was transferred into a well containing 600 μl of fresh sterile FSLW in a 12-well polystyrene culture plate (Thermo Scientific Nunc, Pittsburgh, PA, United States). To provide a bacterial attachment platform, a 12 mm round glass cover slip (Fischer Scientific, Pittsburgh, PA, United States) was dipped into each culture well, and the culture was incubated statically overnight at room temperature. The following day, attached cells on cover slip were washed twice with Dulbecco’s PBS (DPBS) (HyClone Laboratories, Logan, UT, United States), and fixed in 10% neutral buffered formalin solution (Sigma-Aldrich, St. Louis, MO, United States) for 30 min. Cells were washed again with DPBS and stained with 500 μl/well of 1:1000 SYTO 9 dye (LIVE/DEAD BacLight Bacterial Viability Kit; Invitrogen, Grand Island, NY, United States) as described previously (Jubair et al., 2014). Following two washes with HyPure cell culture grade water (HyClone Laboratories, Logan, UT, United States), the glass cover slip was mounted on a 75 × 25 mm microscopic slide (Corning Inc., Corning, NY, United States). A Zeiss LSM 800 with Axio Observer Z1 inverted microscope (Carl Zeiss Iberia, S.L., Spain) was used to visualize the cover slip with an excitation and emission wavelengths of 484 and 500 nm, respectively. The attached microorganism was examined using Plan-Apochromat 63× objective lens. At least three biological replicates were examined during image analysis. All images and z-stacks were processed using Zen 2 (Blue edition) software and enhanced by adjusting the histogram of the image.

A chemotaxis assay was performed as described previously (Butler et al., 2006) with slight modifications. Briefly, a single colony of V. cholerae strain was grown in 3 ml L-broth overnight at 37°C with a shaking speed of 250 × g. One milliliter of culture was centrifuged; after decanting supernatant, the pellet was washed 2× with fresh FSLW and resuspended in 1 ml fresh FSLW. The culture was diluted (10-fold) in FSLW to obtain ca. 10⁸ cfu/ml and incubated overnight statically at room temperature. An aliquot of 150 μl of culture (three replicates) was transferred into 96-well plate. One microliter capillary tube (55 mm) (Drummond Scientific Company, Broomall, PA, United States) filled with either FSLW or FSLW supplemented with chitin (0.05%) or K₂HPO₄ (1.0 mM at final concentration) was inserted into each culture well. After 6 min of incubation at room temperature, the content of the capillary tube was expelled out into 99 μl of L-broth. The culture was spread onto L-agar plates and the plates were incubated overnight at 37°C; total bacterial number (colony forming units per milliliter) grown on L-agar was determined by standard plate count. The results were expressed as a comparative index (CI) for each strain. CI was calculated by the number of bacterial counts (colony forming units per milliliter) present in FSLW in the capillary tube supplemented with nutrient divided by the number of bacterial counts (colony forming units per milliliter) present in FSLW in the capillary tube not supplemented with nutrient.

One-way ANOVA along with Holm-Sidak’s multiple comparison test was performed using GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, CA, United States, http://www.graphpad.com. A p-value of <0.05 was considered as statistically significant.

After PCR amplification and sequencing of the flrA gene of GASP-700D, we found that the flrA gene sustained a 49-bp internal deletion mutation (we hereafter refer this strain as GSΔflrA) (Figure 1) causing a frame shift mutation. To determine if the 49-bp deletion in flrA gene in GSΔflrA inhibited motility, we created an identical 49-bp deletion mutation in the wild-type V. cholerae strain N16961S. Data presented in Figure 2 show that, in contrast to the wild-type V. cholerae N16961S strain (Figure 2A), the deletion mutation in the flrA gene in N16961S strain inhibited motility (Figure 2B). Furthermore, complementation of N16961SΔflrA with the wild-type flrA gene cloned and expressed from pWSK29 (pAA146) restored motility to the level seen with the wild-type strain (Figure 2C). Our data clearly indicate that the potential frame shift mutation (49-bp deletion) in flrA gene had no effect on downstream genes in motility. In contrast, inhibition of motility seen with GSΔflrA (Figure 2D) was only partially complemented by the same complementing vector (pAA146) (Figure 2E) indicating that GSΔflrA might have additional mutation(s) affecting motility. As expected, the N16961SΔflaA mutant was defective in motility (Figure 2F). In summary, our data confirmed that the deletion mutation of the flrA gene in GSΔflrA and N16961S (N16961SΔflrA) inhibited motility.

We previously reported that the GASP-700D phenotype produced vps-independent biofilm in FSLW microcosms with a ∼1,000-fold down regulation of flrA gene (Jubair et al., 2014). However, in that study, we were unable to elucidate which genetic elements, if any, were involved in such biofilm formation. Previous investigations have demonstrated that V. cholerae employs its flagella and mannose sensitive hemagglutinin (MSHA) pilus, in concert for vps-dependent biofilm formation (Watnick et al., 1999; Watnick and Kolter, 1999). To test if flagella and MSHA pilus contribute similarly to the formation of vps-independent biofilm, we compared biofilm-forming ability between a wild-type V. cholerae N16961S strain and its flrA and mshA mutants both in nutrient-rich L-broth and in nutrient-deficient FSLW. As a control, we used a V. cholerae rugose strain (N16961R) that produces copious amounts of exopolysaccharide and robust biofilm both in L-broth and in FSLW (Jubair et al., 2014). As depicted in Figure 3A, except for the V. cholerae rugose strain, none of the tested strains, including wild-type N16961S and flrA mutants (N16961SΔflrA and GSΔflrA) were able to produce significant biofilm in L-broth. In contrast, when grown in FSLW (Figure 3B), both GSΔflrA and N16961SΔflrA produced significantly (p < 0.01) higher amount of biofilm relative to the wild-type N16961S strain. Furthermore, complementation of N16961SΔflrA with the wild-type flrA gene repressed biofilm formation (Figure 3B); however, complementation of GSΔflrA with wild-type flrA gene was unable to significantly repress biofilm formation (Figure 2B) reinforcing our hypothesis that GSΔflrA might have additional mutation(s) affecting motility and vps-independent biofilm formation. In summary, our results demonstrated that mutation in flrA gene in V. cholerae promoted increased biofilm formation specific to FSLW. Furthermore, we did not observe any significant changes in biofilm formation in FSLW among tested strains when biofilm was quantified after 48 h of incubation at room temperature (Supplementary Figure 1).

To determine if gene(s) required for vps-dependent biofilm formation enhances vps-independent biofilm formation both in L-broth and in FSLW, we compared biofilm-forming ability of vpsA mutants to that of corresponding wild-type strains both in L-broth and in FSLW. Data presented in Figure 3C exhibited that, as expected, in L-broth, vpsA mutants significantly inhibited vps-dependent biofilm formation compared to their respective wild-type variants; in contrast, vpsA mutants significantly increased biofilm production compared to their wild-type strains in FSLW (Figure 3D). In summary, consistent with our previous report (Jubair et al., 2014) we demonstrated that mutation in vpsA gene positively contributed to vps-independent biofilm in FSLW, but not in L-broth.

To determine the role for MSHA pilus, if any, for vps-independent biofilm production in FSLW, we created an in-frame deletion within the mshA gene both in GSΔflrA and in the N16961S background resulting in GSΔflrAΔmshA and N16961SΔmshA, respectively. Mutation in mshA gene in GSΔflrA (GSΔflrAΔmshA) had no effect on biofilm production in L-broth compared to its wild-type strain (Figure 3E), while N16961SΔmshA significantly reduced (p < 0.05) biofilm formation relative to its wild-type strain N16961S (Figure 3E). In contrast to L-broth, mutation in the mshA gene in GSΔflrA significantly (p < 0.01) repressed biofilm formation in FSLW (Figure 3F). Moreover, complementation of GSΔflrAΔmshA with the wild-type mshA gene cloned and expressed in trans from plasmid pAA159 completely restored biofilm formation as seen with GSΔflrA in FSLW (Figure 3F). Wild-type V. cholerae strain N16961S and its isogenic mshA mutant (N16961SΔmshA) exhibited no significant difference in biofilm formation in FSLW (Figure 3F). In summary, our results confirmed that, while FlrA negatively affects the expression of vps-independent biofilm production in FSLW, MSHA pilus positively contributes to vps-independent biofilm production elicited by GSΔflrA in FSLW. Unlike ΔflrA, a strain carrying mshA mutation in N16961S was unable to promote vps-independent biofilm formation in FSLW (Figure 3F).

To further examine the role for mshA gene in the expression of ΔflrA-driven biofilm in FSLW, we compared three-dimensional architectural structures and thickness of biofilm among wild-type N16961S, N16961SΔflrA, N16961SΔflrA/pAA146, GSΔflrA, GSΔflrAΔmshA, and GSΔflrAΔmshA/pAA159 using confocal microscopic analysis. As expected, N16961ΔflrA produced more biofilm (Figure 4B) compared to wild-type N16961S (Figure 4A). Complementation of N16961ΔflrA with wild-type flrA gene (N16961SΔflrA/pAA146) repressed biofilm formation (Figure 4C). In contrast to N16961ΔflrA, GSΔflrA produced much more biofilm when grown in FSLW (Figure 4D). GSΔflrAΔmshA completely inhibited biofilm formation (Figure 4E); complementation of GSΔflrAΔmshA strain with the wild-type mshA gene fully restored biofilm production at the level seen with GSΔflrA (Figure 4F). To rule out the possibility that the ΔmshA mutant may not have survived in FSLW, we enumerated microorganisms’ counts for each culture before making the slides used for confocal microscopic analysis using standard plate counts. Each culture yielded ca. 10⁸ cfu/ml suggesting that spotty occurrence of GSΔflrAΔmshA (Figure 4E) cells on slide resulted from the inability of the cells to form biofilm, and, not due to low cell counts present in the culture.

In aquatic environments, many planktonic bacterial species use a combination of chemotaxis and flagellar motility to sense and acquire nutrients (Butler and Camilli, 2005). V. cholerae is no exception to this phenomenon. However, it is not known how non-motile bacterial phenotypes or phenotypes lacking productive motility (commonly found in biofilm communities) acquire nutrients from their aquatic reservoir. Specifically, we questioned whether non-motile V. cholerae, promoting ΔflrA-driven vps-independent biofilm in FSLW, can sense and move toward nutrients and, if so, does chemotaxis play any role in such mechanism? In our previous study (Jubair et al., 2012), we reported that nutrients, including chitin and phosphate (when added to FSLW microcosm) promoted the growth of highly starved (180 and 700 days surviving microorganisms) V. cholerae persisting in nutrient-poor FSLW microcosms. Based on that observation we hypothesized that chitin and phosphate serve as chemo-attractants in chemotaxis assays in FSLW. As illustrated in Figure 5, GSΔflrA and N1696S1ΔflrA mutants moved toward chitin (Figure 5A) and phosphate (Figure 5B) compared to their wild-type phenotype (N16961S). Strikingly, a V. cholerae rugose strain (N16961R), capable of producing robust vps-dependent biofilm (Ali et al., 2002), was unable to move toward nutrients in FSLW (Figures 5A,B). Complementation of GSΔflrA and N1696S1ΔflrA with a wild-type flrA gene inhibited movement to nutrients (Figures 5A,B). In contrast and, as expected, GSΔflrAΔmshA, that repressed vps-independent biofilm (described above), was unable to move toward nutrients (Figures 5A,B). Not surprisingly, unlike the N16961SΔflrA mutant, the N16961SΔmshA mutant, that was unable to induce vps-independent biofilm, failed to move to nutrients (Figures 5A,B).

We previously reported that vpsA mutants inhibiting vps-dependent biofilm promoted vps-independent biofilm in FSLW (Jubair et al., 2014). We questioned if ΔvpsA-driven biofilm can also move toward nutrients in FSLW. To test our idea we compared the ability of vpsA mutants to move toward nutrients relative to their corresponding wild-type strains. Data presented in Figures 5C,D indicated that ΔvpsA-driven biofilm akin to ΔflrA-driven biofilm moved to chitin and phosphate compared to their corresponding wild-type strains. Our data clearly indicate that in contrast to vps-dependent biofilm, vps-independent biofilm elicited by both ΔflrA- and ΔvpsA-driven biofilm can promote V. cholerae to move toward nutrients, including chitin and phosphate in FSLW.

Previous reports suggested that bacterial chemotaxis genes, including cheY-3 are involved in chemotaxis (Lee et al., 2001). To determine if cheY-3 plays any role to promote V. cholerae to move toward nutrients in FSLW, we created a null mutation in cheY-3 gene in the background of N16961S, GSΔflrA, and N16961SΔflrA. As depicted in Figures 5E,F, compared to wild-type strain N16961S, GSΔflrAΔcheY-3 and N16961SΔflrAΔcheY-3 significantly (p < 0.01) repressed their movement toward chitin and phosphate; complementation of mutation with the wild-type cheY-3 gene restored movement to chitin and phosphate (Figures 5E,F). Interestingly, we observed that there was no significant difference in biofilm production between N16961SΔflrA and N16961SΔflrAΔcheY-3 (Supplementary Figure 2). Our data indicate that CheY-3 of V. cholerae significantly (p < 0.01) contributed to movement toward nutrients, including chitin and phosphate in FSLW.