The Phosphodiesterase DipA (PA5017) Is Essential for Pseudomonas aeruginosa Biofilm Dispersion

Abstract

Although little is known regarding the mechanism of biofilm dispersion, it is becoming clear that this process coincides with alteration of cyclic di-GMP (c-di-GMP) levels. Here, we demonstrate that dispersion by Pseudomonas aeruginosa in response to sudden changes in nutrient concentrations resulted in increased phosphodiesterase activity and reduction of c-di-GMP levels compared to biofilm and planktonic cells. By screening mutants inactivated in genes encoding EAL domains for nutrient-induced dispersion, we identified in addition to the previously reported ΔrbdA mutant a second mutant, the ΔdipA strain (PA5017 [dispersion-induced phosphodiesterase A]), to be dispersion deficient in response to glutamate, nitric oxide, ammonium chloride, and mercury chloride. Using biochemical and in vivo studies, we show that DipA associates with the membrane and exhibits phosphodiesterase activity but no detectable diguanylate cyclase activity. Consistent with these data, a ΔdipA mutant exhibited reduced swarming motility, increased initial attachment, and polysaccharide production but only somewhat increased biofilm formation and c-di-GMP levels. DipA harbors an N-terminal GAF (cGMP-specific phosphodiesterases, adenylyl cyclases, and FhlA) domain and two EAL motifs within or near the C-terminal EAL domain. Mutational analyses of the two EAL motifs of DipA suggest that both are important for the observed phosphodiesterase activity and dispersion, while the GAF domain modulated DipA function both in vivo and in vitro without being required for phosphodiesterase activity. Dispersion was found to require protein synthesis and resulted in increased dipA expression and reduction of c-di-GMP levels. We propose a role of DipA in enabling dispersion in P. aeruginosa biofilms.

INTRODUCTION

In Pseudomonas aeruginosa, one of the principal pathogens associated with cystic fibrosis (CF) pulmonary infection, biofilm formation occurs in stages (44). The biofilm developmental cycle comes full circle when biofilms disperse (44, 45). Dispersion is a mechanism used by biofilm bacteria to successfully transit from the biofilm to the planktonic growth state and to spawn novel communities in new locales (44, 50). The transition is typically induced upon sensing of a myriad of environmental cues (45). Changes in oxygen or carbon substrate concentration, pH, iron, or other chemical parameters have been reported to induce dispersion of mature biofilms in various organisms (5, 14, 45, 53). In P. aeruginosa, biofilm dispersion can be triggered by the exogenous addition of nutrients and other components of media, such as ammonium chloride, as well as heavy metals (31, 45), and nitric oxide (5). These distinct dispersion-inducing conditions have in common that they require a step change in concentration to trigger dispersion (45).

Cyclic di-GMP (c-di-GMP) is emerging as an important intracellular signaling molecule, controlling the transition between a motile and a biofilm lifestyle. The current paradigm in the field is that high concentrations of this molecule correlate with a sessile lifestyle (e.g., biofilm formation), while its absence favors motility (e.g., twitching and swarming) and the free-swimming lifestyle (12). Overall, c-di-GMP plays an important role in biofilm formation by regulating the production of the exopolysaccharide (EPS) matrix, and by controlling motility (in particular swarming), autoaggregation, and adhesiveness (16, 21, 22, 23, 28, 31, 35, 36, 41, 42, 48, 52, 54). Levels of c-di-GMP are enzymatically modulated by diguanylate cyclases (DCG), proteins containing a GGDEF domain (36), and phosphodiesterases (PDE) containing either an EAL domain (51) or an HD-GYP domain (43).

Alterations in c-di-GMP levels have been shown to be associated with biofilm dispersal in a number of different bacteria. In Pseudomonas putida, two genes (PP0164 and PP0165) encoding a putative periplasmic protein and a putative transmembrane protein involved in c-di-GMP modulation, respectively, were found to be required for biofilm formation and starvation-induced dispersion (14). PP0164 mutants were unable to reduce their adhesiveness and disperse from biofilms in response to carbon starvation, while PP0165 mutant bacteria were unable to increase their adhesiveness and form biofilms (14). In Shewanella oneidensis, a rapid cellular detachment from the biofilm occurred upon activation of yhjH, encoding an enzyme having phosphodiesterase activity (52). In contrast, matrix attachment was shown to be dependent on mxdA, which encodes a diguanylate cyclase. The findings indicate that detachment is a result of reduced adhesiveness as well as controlled cessation/reduction of such activity due to reduced levels of c-di-GMP. In P. aeruginosa, only one protein, BdlA, has been identified so far to be essential for dispersion (31). BdlA harbors a Tar (methyl-accepting chemotaxis protein) domain and two PAS (Per-Arnt-Sint) domains involved in sensing. Inactivation of bdlA resulted in lack of dispersion in response to exogenous factors, such as nutrients or heavy metals (31). Recent evidence further suggested that the chemotaxis transducer BdlA is involved in the biofilm dispersal response induced by nitric oxide (6). While the protein lacks GGDEF or EAL domains, inactivation of bdlA resulted in increased levels of c-di-GMP in biofilms (31).

While the importance of c-di-GMP is apparent from the studies cited above, no pathway for dispersion has been identified. Moreover, it is also unclear which c-di-GMP-modulating enzymes are involved, how the activity of c-di-GMP-modulating enzymes is regulated during dispersion, and whether gene expression is required. By screening mutants inactivated in genes encoding potential phosphodiesterases or other c-di-GMP-modulating proteins for nutrient-induced dispersion, we identified two mutants (the previously reported ΔrbdA [3] and ΔdipA strains) that were dispersion deficient. DipA was further characterized to be a phosphodiesterase contributing to swarming motility, polysaccharide production, biofilm architecture, and nutrient-induced dispersion. Moreover, we demonstrate that induction of dispersion, requiring DipA, coincides with increased dipA expression and reduction of c-di-GMP levels.

MATERIALS AND METHODS

Bacterial strains, plasmids, media, and culture conditions.

All bacterial strains and plasmids used in this study are listed in Table S1 in the supplemental material. P. aeruginosa strains PA14 and PAO1 were used as parental strains. All planktonic strains were grown in Lennox broth (LB) (BD Biosciences) or Vogel-Bonner minimal medium (VBMM) (46) in shake flasks at 220 rpm in the absence or presence of 0.1 to 1.0% arabinose. Escherichia coli cultures were grown in LB in the absence or presence of 1 mM isopropyl β-d-1-thiogalactopyranoside (IPTG). Antibiotics were used at the following concentrations: 50 to 75 μg/ml gentamicin and 200 to 250 μg/ml carbenicillin for P. aeruginosa and 20 μg/ml gentamicin and 50 μg/ml ampicillin for E. coli.

Strain construction.

Complementation and overexpression of dipA (PA5017) was accomplished by placing the respective genes under the control of an arabinose-inducible promoter in the pMJT1 vector. C-terminal V5/6×His tagging of DipA and PA4843 was accomplished by subcloning into pET101D (Invitrogen). The tagged constructs were introduced into pJN105 and pMJT1. Site-directed mutagenesis of the indicated dipA sequences was accomplished by using the GeneArt site-directed mutagenesis kit (Invitrogen) according to the manufacturer's protocol. The identity of vector inserts was confirmed by sequencing. Plasmids were introduced into P. aeruginosa via conjugation or electroporation. The primers used for strain construction are listed in Table S2 in the supplemental material.

Biofilm formation.

Initial attachment was measured following 6 and 24 h of growth in LB medium using the microtiter dish assay system with crystal violet staining (34) and repeated four times with 12 replicates each. Biofilms were grown in a continuous flow tube reactor system (Masterflex 1-m-long, size 14 silicone tubing; Cole Parmer, Inc.) at 22°C for up to 6 days to obtain proteins and RNA. Biofilms were grown in flow cells to view the biofilm architecture by confocal laser scanning microscopy (CLSM) as previously described (1, 2, 37, 44, 49). Biofilms were grown at 22°C in 20-fold-diluted LB medium in the presence of 0.1% arabinose. Quantitative analysis of CLSM images was performed using COMSTAT (15).

Biofilm dispersion assays.

For biofilm dispersion assays, biofilms were cultivated in a once-through continuous flow tube reactor system composed of Masterflex size 13 silicone tubing (Cole Parmer, Inc.) at 22°C for 5 days. After 5 days of biofilm growth, biofilm dispersion was induced by the sudden addition of glutamate (18 mM), ammonium chloride (10 mM), and mercury chloride (2 mM) to the growth medium as previously described (31). In addition, 500 μM sodium nitroprusside (SNP) was used as a source of nitric oxide (5). Dispersion was indicated by an increase in turbidity at 600 nm in the effluent from the silicone tubing. Dispersion of flow-cell-grown biofilms was induced in a similar manner. Confocal images prior to and following induction of dispersion were analyzed using COMSTAT as previously described (31).

In vivo quantification of c-di-GMP from P. aeruginosa.

Cyclic di-GMP (c-di-GMP) was extracted in triplicate from wild-type and mutant strains using heat and ethanol precipitation (31) and quantitated essentially as previously described (39).

Motility assays.

Twitching motility was assessed using LB medium, and swarming motility was determined using M8 and nutrient agar (NA) (0.8% nutrient broth, 0.5% glucose) media, while swimming motility was assessed using M63 medium supplemented with Casamino Acids using 1%, 0.5%, and 0.3% agar, respectively, as previously described (24, 29, 33, 47, 49). Arabinose was added to a final concentration of 1% unless otherwise indicated.

Polysaccharides.

Psl polysaccharide was extracted from planktonic cells grown in VBMM essentially as described by Byrd et al. (9). Quantitation of Psl production was done by determining the anti-Psl dot blot spot volume (19) using the ImageMaster analysis software (GE Healthcare) (39). Production of polysaccharides by initially attached cells was assessed using a modified microtiter dish assay system following 6 h of growth in LB medium with 40 μg/ml Congo red.

Real-time qRT-PCR.

Quantitative reverse transcription-PCR (qRT-PCR) was used to determine the expression levels of dipA (PA5017), rbdA, bdlA, pelA, and pslA using 1 μg of total RNA isolated from wild-type P. aeruginosa cells grown planktonically, as a biofilm (prior to and postinduction of dispersion), and from dispersed cells following exposure to glutamate. Isolation of mRNA and cDNA synthesis were carried out as previously described (38, 39). qRT-PCR was performed using the Eppendorf Mastercycler ep realplex (Eppendorf AG, Hamburng, Germany) and the Kapa SYBR fast qPCR kit (Kapa Biosystems, Woburn, MA), with oligonucleotides listed in Table S2 in the supplemental material. mreB was used as a control. The stability of mreB levels was verified by 16S RNA abundance using primers HDA1 and HDA2 (27). Relative transcript quantitation was accomplished using the ep realplex software (Eppendorf AG) by first normalizing transcript abundance (based on the cycle threshold [C_T] value) to mreB followed by determining the transcript abundance ratios. Melting curve analyses were employed to verify specific single-product amplification.

Protein localization.

To determine cellular localization of DipA, cells expressing V5/His-tagged DipA were grown at 37°C and 220 rpm for 8 h in VBMM containing 1% arabinose. The cultures were centrifuged for 10 min at 6,000 × g, and the resulting pellet was resuspended in 500 μl TE buffer (10 mM Tris-HCl [pH 8.0], 1 mM EDTA, plus 0.3 μg/ml phenylmethylsulfonyl fluoride [PMSF]) and lysed by sonication. The samples were centrifuged for 2 min at 21,200 × g to pellet unbroken cells, and the resulting supernatant was spun again at 30,000 × g to remove any debris. The resulting supernatant was spun at 100,000 × g for 90 min at 4°C. The supernatant containing cytoplasmic proteins was retained, and the pellet was washed with cold TE buffer and centrifuged for another 90 min at 4°C at 100,000 × g. The final pellet containing membrane proteins was resuspended in TE buffer containing 1% CHAPS {3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate}. SDS loading buffer was mixed with the cytoplasmic and membrane samples, followed by heat denaturation at 100°C for 10 min. The samples were resolved on an 11% polyacrylamide gel and subsequently transferred onto polyvinylidene difluoride (PVDF) membrane using a TurboTransblot apparatus (Bio-Rad). Western blots were probed with anti-V5 antibodies and developed with LumiGlo detection reagents (Cell Signaling). Considering that catalase has been demonstrated to be located in the cytoplasm and periplasm but absent from the membrane (8, 20), proper fractionation was determined via the absence of catalase activity in the membrane fraction, but its presence in the cytoplasmic fraction was determined using 1 mM hydrogen peroxide as the substrate. Commercially available catalase (purified from Aspergillus) was used as a control.

Purification of His-tagged proteins.

V5/6×His-tagged proteins were purified from supernatants following sonication of VBMM-grown planktonic cells and centrifugation at 21,200 × g. The supernatant was loaded onto nickel-nitrilotriacetic acid (Ni-NTA) affinity resin (Qiagen), washed with buffer, and eluted with an imidazole gradient according to the manufacturer's instructions for native protein purification. Since DipA was found to be membrane associated, the purification was carried out with the detergent Tween 20 (0.1% final concentration) present in all buffers. Protein preparations were examined for purity by SDS-PAGE, and fractions containing pure protein were pooled and desalted using VivaSpin centrifugal concentrator columns (10-kDa cutoff; Sartorius) (see Fig. S1 in the supplemental material). For phosphodiesterase and diguanylate cyclase assays, proteins were purified from P. aeruginosa cell extracts. PleD, purified from E. coli BL21, was used as a positive control in diguanylate cyclase assays.

Phosphodiesterase activity assay.

Phosphodiesterase activity of purified DipA was determined using the synthetic chromogenic substrate bis(p-nitrophenyl) phosphate (bis-pNPP) (Sigma-Aldrich) essentially as previously described (7, 22) in the absence or presence of 4 to 100 pmol of c-di-GMP, 100 μM cyclic AMP (cAMP), or 100 μM cGMP. Specific phosphodiesterase activity was determined using purified proteins and measuring the release of p-nitrophenol (pNP) at 405 nm. An extinction coefficient for p-nitrophenol of 1.78 · 10⁴/M · cm was used. Controls without extracts were included to account for any nonenzymatic bis-pNPP hydrolysis. Moreover, phosphodiesterase assays were performed as previously described using c-di-GMP as the substrate (10). Briefly, purified protein (100 μg) was added to reaction buffer consisting of 50 mM Tris (pH 8.0), 50 mM NaCl, and 50 mM MgCl₂. Reactions were initiated by addition of 1 μM c-di-GMP. Reaction mixtures were incubated at 25°C for 240 min, terminated by heating to 95°C, and subsequently analyzed by high-performance liquid chromatography (HPLC).

Diguanylate cyclase assays.

Diguanylate cyclase assays were performed essentially as previously described (30, 36) using purified PleD protein as a positive control. Briefly, the reaction mixtures contained 75 mM Tris-HCl at pH 7.8, 250 mM NaCl, 25 mM KCl, and 10 mM MgCl₂ in a 200-μl volume and were started by the addition of a mixture containing 50 μM GTP. A total of 100 μg of purified protein was used per assay. The reaction mixtures were incubated for 90 min at 25°C, terminated by heating to 95°C, and subsequently analyzed by HPLC.

Statistical analysis.

Student's t test was performed for pairwise comparisons of groups, and multivariant analyses were performed using a 1-way analysis of variance (ANOVA) followed by a post priori test using SigmaStat software. All experiments were performed at least in triplicate.

RESULTS

We previously demonstrated that the nutrient dispersion-deficient Pseudomonas aeruginosa ΔbdlA mutant biofilm had increased c-di-GMP levels (31), indicating that nutrient-induced biofilm dispersion may involve alteration of c-di-GMP levels. We therefore determined the levels of this intracellular signaling molecule in biofilms prior to and following dispersion and in suspended cells. Surface-attached P. aeruginosa biofilm cells were found to contain on average 84 pmol/mg c-di-GMP, which decreased to 30 pmol/mg in biofilm cells that remained attached following dispersion (Fig. 1A). c-di-GMP levels of dispersed cells were similar to those of planktonic cells and on average 10-fold lower than those of biofilms (Fig. 1A). To determine whether the decrease in c-di-GMP levels observed in dispersed cells compared to biofilm cells was due to increased degradation of c-di-GMP, phosphodiesterase (PDE) activity was determined using cell extracts obtained from biofilms prior to and following dispersion and from dispersed cells. Dispersed cells were found to have 3-fold-increased PDE activity compared to biofilms and biofilms following dispersion and 6-fold-increased PDE activity compared to planktonic cells (Fig. 1B).

Fig 1.

Modulation of c-di-GMP levels and phosphodiesterase activity over the course of biofilm development by P. aeruginosa wild-type and dipA mutant strains. c-di-GMP levels (A) and specific phosphodiesterase activity (B) were determined using cell extracts obtained from P. aeruginosa PA14 planktonic, dispersed, and biofilm cells and biofilms remaining attached following dispersion (“biofilms post dispersion”). c-di-GMP/mg refers to c-di-GMP levels per total cell pellet protein (in mg). *, significantly different from biofilms (P < 0.05), as determined by ANOVA and SigmaStat. Experiments were repeated at least 5 times. Similarly, specific phosphodiesterase activity (B) was determined using cell extracts obtained from ΔdipA mutants grown planktonically and as a biofilm prior to and following addition of glutamate. **, significantly different based on Student's t test (P < 0.01). Experiments were repeated at least 4 times. (C) Domain overview of DipA and RbdA. GAF, cGMP-specific phosphodiesterases, adenylyl cyclases, and FhlA domain; PAS, Per Arnt Sim sensory domain; EAL, phosphodiesterase domain; GGDEF, diguanylate cyclase domain; aa, amino acid.

Mutations in the phosphodiesterase-encoding genes PA5017 and PA0861 abolish the dispersion response by P. aeruginosa biofilms.

Since dispersion, the transition from the biofilm to the planktonic mode of growth, resulted in decreased levels of c-di-GMP and increased phosphodiesterase activity, we next asked whether dispersion requires the action of phosphodiesterases. We therefore screened transposon mutants described by Kulasekara et al. (23) to encode phosphodiesterases (PA0285, PA0575, PA0861 [rbdA], PA1181, PA2072, PA3311, PA4367 [bifA], PA5017 [dipA], and PA5442) in an in vitro assay for nutrient-induced biofilm dispersion using glutamate (31, 45). The screen is based on detecting increased turbidity and, thus, cells in the medium effluent of biofilm tube reactors upon challenging biofilms with sudden changes in the medium glutamate concentration (31, 45). In addition, two wsp mutants (ΔwspA and ΔwspR strains) and strains overproducing Pel and Psl polysaccharides were tested. The Wsp chemosensory system (WspAFR) has been shown to regulate biofilm formation through modulation of c-di-GMP levels. A bifA mutant has previously been shown to have a hyperbiofilm-forming phenotype, increased cellular pools of c-di-GMP relative to the wild type, and increased synthesis of the Pel polysaccharide (22, 28).

Exposure of biofilms to sudden changes in the medium glutamate concentration resulted in dispersion of wild-type P. aeruginosa and the majority of the mutants tested, including the bifA strain, both Wsp mutants, and the Pel- and Psl-overexpressing strains (see Fig. S2 in the supplemental material). Overall, we only identified two mutant strains, the PA5017 and ΔrbdA mutants, incapable of dispersion in response to a sudden increase in the medium glutamate concentration (Fig. 2A).

Fig 2.

DipA and RbdA are essential for the dispersion response of P. aeruginosa PA14. (A) ΔdipA and ΔrbdA biofilms do not disperse in response to addition of glutamate, ammonium chloride, nitric oxide, and mercury chloride compared to wild-type P. aeruginosa PA14. Biofilms were grown in tube reactors. (B) Confocal images of ΔbifA and ΔdipA biofilms prior to and 30 min following addition of glutamate. Biofilms were grown in flow cells for 5 days before induction of dispersion. Biofilms were stained with the LIVE/DEAD BacLight viability stain (Invitrogen Corp.). White size bars = 100 μm.

Inactivation of dipA and rbdA renders biofilms incapable of dispersing in response to glutamate, ammonium chloride, and nitric oxide.

RbdA (regulation of biofilm disposal A) harbors EAL, GGDEF, and PAS/PAC domains (Fig. 1C) (55) and was recently shown to only possess PDE activity, with inactivation of rbdA resulting in hyperbiofilm formation (3). While the authors suggested a role of this protein in dispersion, only indirect evidence was provided, as the effect of RbdA on biofilm formation, rather than on detachment in response to environmental cues, was assessed (3). PA5017, the second phosphodiesterase identified in our screen, was assigned the name DipA, for dispersion-induced phosphodiesterase A. The protein harbors GAF, PAS, EAL, and GGDEF domains (Fig. 1C) (55). Inactivation of dipA has been previously shown in P. aeruginosa PA68 to increase biofilm formation, probably due to impaired flagellum motility (24). To establish whether RbdA and DipA play a more general role in dispersion, dispersion assays were repeated using nitric oxide, mercury chloride, and ammonium chloride. While wild-type P. aeruginosa dispersed under all conditions tested, the two mutants did not disperse in response to glutamate, nitric oxide, mercury chloride, and ammonium chloride (Fig. 2A) with the dispersion-deficient phenotype of these two mutants being comparable to that of the ΔbdlA strain (6, 31).

To further confirm our findings of ΔdipA biofilms being dispersion deficient, the dispersion response by the ΔdipA strain to glutamate was visualized by confocal microscopy (CLSM) combined with COMSTAT analysis. A ΔbifA mutant, which was found to disperse in response to glutamate in our dispersion screen (see Fig. S2 in the supplemental material), was used as a control. While both ΔbifA and ΔdipA mutants formed biofilms on glass, only the ΔbifA mutant dispersed (Fig. 2B and Table 1), confirming our dispersion results obtained using the absorbance of medium effluent as an indicator.

Table 1.

Quantitative COMSAT analysis of biofilm architecture prior to and following induction of nutrient-induced dispersion^a

Condition and strain	Total biomass (μm3/μm2)	Thickness (μm)		Roughness coefficient
		Avg	Maximum
Prior to dispersion
PA14	34.5 ± 3.4	39.2 ± 7.8	81.3 ± 15.0	0.8 ± 0.1
ΔdipA mutant	45.1 ± 2.2	46.9 ± 2.3	95.0 ± 5.0	1.1 ± 0.1
ΔbifA mutant	51.9 ± 12.9	59.9 ± 12.2	142.1 ± 12.9	0.57 ± 0.12
Following induction of nutrient-induced dispersion
PA14	1.2 ± 0.4*	1.5 ± 0.6*	12.7 ± 3.3*	1.8 ± 0.1*
ΔdipA mutant	44.7 ± 0.9	46.9 ± 1.3	95.0 ± 5.0	1.1 ± 0.1
ΔbifA mutant	2.94 ± 0.05*	2.24 ± 0.08*	24.6 ± 8.6*	1.7 ± 0.37*

^aCOMSTAT analysis was carried out from biofilms grown in triplicate using at least 6 images per replicate. Dispersion was induced by addition of 18 mM glutamate to the growth medium. *, significantly different from biofilms prior to dispersion (P < 0.05), as determined by ANOVA and SigmaStat.

DipA contributes to the architecture and c-di-GMP levels of P. aeruginosa biofilms.

To initiate the characterization of DipA, we first generated a C-terminal V5/6×His-tagged DipA construct under the control of the PBAD promoter. The construct was capable of restoring the dispersion-deficient phenotype of the ΔdipA mutant to wild-type levels (Fig. 3C). We next analyzed the role of dipA in biofilms by CSLM and COMSTAT analysis. Compared to PA14 wild-type biofilms, ΔdipA biofilms were characterized by significantly increased biofilm biomass accumulation and increased average but similar maximum thicknesses (Fig. 3A and Table 2). However, the biofilm architectures of the PA14 and ΔdipA strains were comparable, with both biofilms being composed of large microcolonies (>100 μm). Complementation of the ΔdipA mutant by expression of dipA in trans under the control of an arabinose-inducible PBAD promoter (ΔdipA/pMJT-dipA) restored biofilm biomass accumulation to wild-type levels. A significant change in the biofilm architecture was observed, however, upon overexpression of dipA, as evidenced by PA14/pMJT-dipA biofilms being composed of only small microcolonies less than 50 μm in diameter and characterized by significantly reduced biofilm biomass accumulation and thickness compared to wild-type biofilms (Fig. 3A and Table 2). Similar results were obtained when dipA was overexpressed in P. aeruginosa PAO1 (Fig. 3A and Table 2).

Fig 3.

DipA plays a minor role in biofilm formation by P. aeruginosa. (A) Biofilms of strains PA14 and PAO1 inactivated in or (over)expressing dipA, grown for 144 h, were visualized by CLSM. P. aeruginosa biofilms in the absence/presence of the empty plasmid (pMJT1) were used as controls. Biofilms were stained with the LIVE/DEAD BacLight viability stain (Invitrogen Corp.). White size bars = 100 μm. (B) c-di-GMP levels of P. aeruginosa PA14 biofilms inactivated in or (over)expressing dipA. “c-di-GMP/mg” refers to c-di-GMP levels per total cell pellet protein (in mg) used. *, significantly different from PA14 biofilms (P < 0.05), as determined by ANOVA and SigmaStat. (C) The dispersion-deficient phenotype of ΔdipA biofilms is restored upon complementation with dipA in trans. Biofilms were grown for 5 days before dispersion was induced by a sudden increase in the medium glutamate concentration. Biofilm effluents were collected for a period of 18 min, and the absorbance was determined at 600 nm. Determination of c-di-GMP levels was repeated at least 6 times, while all other experiments were carried out in triplicate. Error bars indicate standard deviations.

Table 2.

COMSTAT analysis of P. aeruginosa wild-type and mutant biofilms inactivated in or differentially expressing dipA and truncated dipA constructs^a

Strain name or genotype	Total biomass (μm3/μm2)	Thickness (μm)
		Avg	Maximum
PA14 strains
PA14	30.10 ± 9.5	34.34 ± 11.5	104.83 ± 15.5
ΔdipA mutant	55.70 ± 11.6*	63.30 ± 16.2*	119.56 ± 13.8
ΔdipA/pMJT-dipA	39.76 ± 8.9	49.80 ± 8.9	103.26 ± 23.6
PA14/pMJT-dipA	12.00 ± 8.3*	18.64 ± 15.1*	33.14 ± 11.0*
PA14/pMJT-dipA-NoGAF	27.87 ± 18.9	34.50 ± 21.4	82.0 ± 15.12*
ΔdipA/pMJT-dipA-NoGAF	34.91 ± 12.8	46.34 ± 20.6	107.0 ± 23.7
PA14/pMJT-GAFonly	31.8 ± 4.4	40.48 ± 7.9	112.6 ± 38.1
ΔdipA/pMJT-GAFonly	47.2 ± 5.3*	54.4 ± 19.5	102.4 ± 11.6
PAO1 strains
PAO1/pMJT1	26.92 ± 5.6	37.18 ± 5.15	95.33 ± 19.5
PAO1/pMJT-dipA	16.0 ± 12.3*	18.64 ± 15.1*	53.14 ± 11.0*
PAO1/pMJT-dipA-NoGAF	24.45 ± 14.4	26.34 ± 15.5	74.91 ± 13.2

^aCOMSTAT analysis was carried out from biofilms grown in triplicate using at least 6 images per replicate. *, significantly different from the 144-h-old PA14 or PAO1 parental strain (P < 0.05), as determined by ANOVA and SigmaStat.

Overexpression of dipA also resulted in significantly reduced c-di-GMP levels, with PA14/pMJT-dipA biofilms containing no more than half of the c-di-GMP detected in wild-type biofilms (Fig. 3B). Despite the relatively low c-di-GMP levels, biofilms of this strain were capable of dispersing in response to glutamate addition (Fig. 3C). Compared to wild-type P. aeruginosa, ΔdipA biofilms were characterized by significantly increased c-di-GMP levels (on average, 110 pmol/mg), with complementation restoring c-di-GMP to wild-type levels (Fig. 3B).

To determine whether DipA contributes to the overall PDE activity of P. aeruginosa PA14, cell extracts obtained from the PA14 and ΔdipA strains grown planktonically and as biofilms prior to and post-glutamate addition were tested. Compared to the wild type under the same conditions, the overall PDE activity in untreated ΔdipA biofilms was reduced by 5%, suggesting that DipA contributes little to the overall PDE activity and, thus, the maintenance of c-di-GMP levels in intact biofilms. Interestingly, upon addition of glutamate (Fig. 1B, compare biofilms versus biofilms post-glutamate addition), the overall PDE activity in P. aeruginosa PA14 decreased by 33%, while that of ΔdipA biofilms remained unchanged. Considering that wild-type biofilms disperse, leaving behind a thin layer of cells that remain attached to the surface, while ΔdipA biofilms do not disperse, the finding indicates that cells may remain attached to the surface by modulating DipA activity. Overall, the highest PDE activity was detected in dispersed P. aeruginosa PA14 cells, which was 7-fold higher than that in planktonic cells. The finding strongly supports the notion that dispersion correlates with increased phosphodiesterase activity and reduced c-di-GMP levels (Fig. 1A). As ΔdipA biofilms do not disperse, PDE activity in dispersed cells was not determined. Under planktonic conditions, the ΔdipA mutant exhibited 80% of the PDE activity present in P. aeruginosa PA14 (Fig. 1B), indicating that DipA could possibly contribute up to 1/5th of the total PDE activity detectable in planktonic cells.

Dispersion results in increased dipA expression and requires protein synthesis.

Considering the growth mode-dependent contribution of DipA to the overall PDE activity, we next asked whether dipA was differentially expressed in biofilms and free-swimming cells. qRT-PCR analysis revealed dipA abundance to increase 3-fold in planktonic cells and 6.8-fold in dispersed cells compared to biofilm growth conditions (Fig. 4A). In contrast, bdlA expression was reduced in planktonic cells compared to biofilm cells and only increased 1.7-fold in dispersed cells (Fig. 4A). Similarly, rbdA expression only increased by 1.7-fold in dispersed cells (Fig. 4A).

Fig 4.

Dispersion by wild-type P. aeruginosa biofilms correlates with increased expression of dipA and requires protein synthesis. (A) Fold change in dipA, rbdA, and bdlA RNA levels in wild-type P. aeruginosa planktonic cells, the remaining biofilms following dispersion, and dispersed cells compared to biofilm growth conditions was assessed by qRT-PCR analysis. (B) P. aeruginosa biofilms following 4 days of growth under flowing conditions were treated for 30 min with tetracycline (arrow 1) prior to the induction of dispersion (arrow 2). Control biofilms were not treated. Dispersion was induced by a sudden increase in the medium glutamate concentration. The biofilm effluent was collected in 0.2-ml aliquots into microtiter plates, and the absorbance was read at 600 nm. All experiments were carried out in triplicate. Error bars indicate standard deviations. *, significantly different from biofilms (P < 0.05), as determined by ANOVA and SigmaStat.

This finding of dipA expression being induced upon dispersion or in dispersed cells prompted us to determine whether protein synthesis is required. P. aeruginosa PA14 biofilms were pretreated with tetracycline for 30 min prior to exposure to dispersion-inducing conditions. While untreated biofilms dispersed in response to increased glutamate concentration, tetracycline-treated biofilms did not disperse (Fig. 4B), suggesting that protein synthesis is required for biofilm dispersion to occur.

Inactivation of dipA impedes swarming motility, but enhances initial attachment and Psl polysaccharide production.

Differential expression of dipA did not affect twitching motility (see Fig. S3B in the supplemental material). While inactivation of dipA in strain PA68 has been shown previously to result in reduced swimming motility (24), inactivation or overexpression of dipA in strain PA14 did not affect swimming motility (see Fig. S3A). However, consistent with previous findings (24), inactivation of dipA in strain PA14 significantly reduced swarming motility. The defect was comparable to that observed upon inactivation of bifA encoding a phosphodiesterase (Fig. 5A) and was observed using M8 and NA media (22). Complementation of ΔdipA by expression of dipA in trans (under the control of the PBAD promoter) correlated with increased swarming motility, with increasing arabinose concentrations (0.8 to 2%) resulting in increased swarming. Inactivation of dipA in strain PAO1 similarly impaired swarming motility (Fig. 5B). In this strain, complementation of ΔdipA fully restored swarming motility, while overexpression of dipA (PAO1/pMJT-dipA) increased swarming motility by more than 2-fold (Fig. 5B). Differential expression of dipA was furthermore found to alter attachment. Inactivation of dipA in strain PAO1 resulted in increased initial attachment following 6 h of incubation, while dipA overexpression impaired attachment by PAO1/pMJT-dipA (Fig. 5C). Differences in initial attachment at the 6-h time point were confirmed by bright-field microscopy (not shown). Similar results were obtained in PA14 strains inactivated in or overexpressing dipA (not shown). No difference in attachment, however, was noted following 24 h of incubation (see Fig. S3D). Growth curves in liquid media suggested that the increase in attachment was not a result of a general increase in growth rate (not shown). Increased attachment by the ΔdipA strain correlated with increased Congo red staining (see Fig. S3E), which is indicative of the increased presence of polysaccharides (25). To determine whether the ΔdipA mutant indeed produced more polysaccharide, Psl polysaccharides were extracted from PAO1 strains grown planktonically to the stationary phase and subsequently quantitated by immunoblot analysis using anti-Psl antibodies. Inactivation of dipA correlated with a 1.5-fold increase in Psl, while dipA overexpression reduced Psl levels 6-fold compared to those in the parental strain (Fig. 5D). However, the observed differences in Psl levels did not correlate with differences in the expression of pslA involved in Psl polysaccharide biosynthesis (Fig. 5E). Moreover, no difference in expression of pelA involved in Pel polysaccharide biosynthesis was observed in PAO1 and PA14 strains inactivated in or overexpressing dipA (Fig. 5E). The findings strongly suggested a role of DipA in polysaccharide regulation at the posttranscriptional level.

Fig 5.

Inactivation of dipA impairs swarming motility but enhances initial attachment and polysaccharide levels at the posttranscriptional level. (A) Swarming motility of P. aeruginosa PA14 strains inactivated or (over)expressing dipA following 48 h of growth on M8 agar. (B) Swarming motility of P. aeruginosa PAO1 strains inactivated or (over)expressing dipA, dipA-NoGAF, GAFonly, or dipA harboring an alanine substitution in either one of the EAL motifs. The swarming diameter was recorded following 24 h of growth on M8 agar. Insets above the graph are representative images showing swarming motility of selected strains. (C) Initial attachment of the P. aeruginosa PAO1, ΔdipA, and PAO1/pMJT-dipA strains following 6 h of growth using the microtiter plate dish assay and crystal violet staining. (D) Psl polysaccharide levels relative to the PAO1 parental strain as determined using anti-Psl dot blot analysis. (E) pelA and pslA are not differentially expressed in PAO1 and PA14 strains inactivated in or overexpressing dipA compared to the parental strains, as determined qRT-PCR analysis. *, significantly different from PAO1 (P < 0.05), as determined by ANOVA and SigmaStat.

DipA is a membrane-associated protein.

Predictions of subcellular localization gave conflicting results as PSORT indicated DipA to be located in the cytoplasm, while analysis by PSORTb (http://www.psort.org/psortb/) predicted DipA with high probability to be located in the cytoplasmic membrane. However, no transmembrane helices or extensive hydrophobic regions were detected by TMHMM and hydrophobicity plot analysis. To determine the subcellular localization of DipA, the protein was C-terminally V5/6×His tagged and subjected to subcellular fractionation and immunoblot analysis. The analysis revealed DipA, having an apparent molecular mass of 105 kDa (predicted relative molecular mass of 104.9 kDa, including the V5/His tag), to be located in both the cytoplasm and the membrane (Fig. 6A). Considering that catalase is absent from the membrane (8, 20) and that catalase activity was only detected in the cytoplasmic fraction (256 ± 8.5 U/mg) but not the membrane fractions, the observations suggest that DipA may be membrane associated. DipA was therefore purified as a protein-lipid-detergent complex using Tween 20, which resulted in DipA being partly soluble (Fig. 6B).

Fig 6.

dipA encodes a membrane-associated phosphodiesterase. (A) Immunoblot analysis of subcellular cell fractions (15 μg) demonstrating that V5/6×His-tagged DipA is located in the cytoplasm and associated with the membrane. Subcellular localization was achieved by ultracentrifugation. (B) V5/6×His-tagged DipA purified in the presence of the detergent Tween 20 is partly soluble, as indicated by a portion of the protein precipitating after ultracentrifugation and not remaining detectable in the supernatant. A total of 28 μg of purified DipA and 10 μg of the resulting supernatant fraction (Sup) and precipitate (pellet) were loaded onto the SDS gel. (C) DipA is a phosphodiesterase, as determined using purified V5/6×His-tagged DipA and bis(p-nitrophenyl) phosphate as a substrate. c-di-GMP concentrations ranging from 4 to 100 pmol were added to the enzyme assay. cAMP and cGMP were added at a final concentration of 100 μM. *, significantly different from DipA (P < 0.05). (D) Degradation of c-di-GMP by purified DipA and DipA-NoGAF (100 μg each) over a period of 240 min, as determined by HPLC analysis. cAMP was added at a final concentration of 1 μM. *, significantly different from DipA (P < 0.05). (E) Effect of overexpression of intact and truncated dipA and mutated dipA on aggregative behavior of PAO1/pJN-PA4843, as indicated by turbidity of liquid growth culture. *, significantly different from PAO1 harboring empty vector (P < 0.05). Aggregative behavior was assessed by turbidity. Following 3 h of growth under planktonic conditions at 37°C in LB, arabinose was added to a final concentration of 1%. Following 3.5 h of continued incubation, the bacterial suspensions were allowed to settle at room temperature for 10 min before the absorbance of the suspension was measured at 600 nm. Cultures not induced by arabinose were used as controls. (F) Effect of overexpression of intact and truncated dipA and mutated dipA on swarming motility of PAO1/pJN-PA4843 following 24 h of growth on M8 agar. *, significantly different from PAO1/pJN105 (P < 0.05). All experiments were carried out at least 5 times. Error bars indicate standard deviations. C, cytoplasmic fraction; M, membrane fraction; sup, supernatant after ultracentrifugation; pellet, pellet after ultracentrifugation; TCE, total cell extract; purified, DipA was purified using Ni-NTA affinity chromatography; No DipA, PAO1/pMJT1 cell extract not expressing any His-tagged DipA was purified using Ni-NTA affinity chromatography, and the respective eluates were used as controls; N.D., not detected. Significance was determined by ANOVA and SigmaStat.

dipA encodes a phosphodiesterase.

Given the effect of dipA inactivation on swarming, attachment, and Psl polysaccharide production, our findings suggested that DipA is a phosphodiesterase. However, in addition to an EAL domain, DipA harbors a GGDEF domain (55) (Fig. 1C). This domain is lacking the typical GGDEF motif (ASNEF instead of GGDEF) (23). Consistent with the lack of the conserved GGDEF motif in the amino acid sequence of DipA, no cyclase activity for purified V5/6×His-tagged DipA was detected (not shown). In contrast, purified PleD used as positive control demonstrated cyclase activity under the conditions tested (not shown). PDE activity of V5/6×His-tagged DipA purified from P. aeruginosa was determined using the synthetic chromogenic substrate bis-pNPP. Under the conditions tested, DipA was found to have a specific PDE activity of 75 pmol/min · mg (Fig. 6C). Addition of cyclic di-GMP at low concentrations stimulated DipA PDE activity by up to 20%, while higher concentrations (100 pmol) resulted in inhibition, probably due to competition between the PDE substrates c-di-GMP and bis-pNPP in the colorimetric assay. Similar low activities were observed when c-di-GMP was used as a substrate (Fig. 6D; see Fig. S4 in the supplemental material). It is likely that the low activity was due to purified DipA being only partly soluble (Fig. 6B).

To further characterize the PDE activity of DipA, we tested the effect of dipA expression in a PAO1 strain overexpressing the cyclase-encoding gene PA4843. PAO1/pJN-PA4843 demonstrates aggregative behavior when grown in liquid culture, as indicated by clumping of cells and low medium turbidity (Fig. 6E), as well as impaired swarming motility (Fig. 6F). Consistent with a role of DipA as a phosphodiesterase, expression of dipA in this strain restored the cyclase-associated phenotypes to wild-type levels. Specifically, dipA expression increased swarming motility to wild-type levels and abrogated the hyperaggregative phenotype of PAO1/pJN-PA4843, as evidenced by unclumping of cells and increased turbidity of the growth medium (Fig. 6E and F). Growth curves in liquid media suggested that the difference in turbidity was not a result of a general increase in growth rate (data not shown).

Both EAL motifs are required for dispersion.

The DipA amino acid sequence harbors two EAL motifs. The first EAL motif (EAL1) is located at positions 631 to 633 upstream of the EAL domain, while the second EAL motif (EAL2) is located within the EAL domain at positions 675 to 677 (see Fig. S5 in the supplemental material). To determine their specific roles, both motifs were mutated by site-directed mutagenesis using alanine substitution (where EAL1 represents EAL631AAA and EAL2 represents EAL675AAA). Mutation of either of the conserved EAL motifs rendered DipA nonfunctional, as indicated by the finding that expression of the dipA variants from pMJT-dipA-EAL1 or pMJT-dipA-EAL2 did not restore swarming in a ΔdipA mutant (Fig. 5B). Moreover, expression of dipA-EAL2 in PAO1/pJN-PA4843 (PAO1/pJN-PA4843/pMJT-dipA-EAL2) failed to eliminate the autoaggregative behavior of PAO1/pJN-PA4843 cells in liquid and to restore swarming motility (Fig. 6E and F). Interestingly, similar results were obtained for the mutant construct PAO1/pJN-PA4843/pMJT-dipA-EAL1 (Fig. 6E and F), suggesting that both EAL motifs are essential for the phosphodiesterase activity of DipA. This was further supported by the finding that ΔdipA mutants complemented with pMJT-dipA-EAL1 and pMJT-dipA-EAL2 were dispersion deficient (Fig. 7A). It is of interest to note that the lack of activity in the swarming and dispersion assays was not due to improper folding (resulting in the formation of inclusion bodies), as both DipA-EAL1 and DipA-EAL2 proteins were found to be soluble (see Fig. S1 in the supplemental material).

Fig 7.

The EAL motifs are essential for dispersion, while the GAF domain is not essential for dispersion and biofilm formation. (A) Both EAL motifs are essential for dispersion. (B) Dispersion assays demonstrating that complementation of the ΔdipA mutant with dipA-NoGAF but not GAFonly restored the biofilm-deficient phenotype of the ΔdipA mutant to wild-type levels. (C) Biofilm architecture of biofilms expressing dipA-NoGAF. Biofilms were grown for 6 days in diluted LB medium, after which time, confocal images were acquired. Biofilms were stained with the LIVE/DEAD BacLight viability stain (Invitrogen Corp.). White size bars = 100 μm. All experiments were carried out at least in triplicate. Error bars indicate standard deviations.

The GAF domain modulates DipA activity.

DipA harbors a GAF domain (Fig. 1C), which is named after some of the proteins it is found in, namely, cGMP-specific phosphodiesterases, adenylyl cyclases, and FhlA. Cyclic GMP (cGMP)-regulated cyclic nucleotide phosphodiesterases have been shown not only to hydrolyze cGMP to 5′-GMP but also to be allosterically regulated by cGMP at noncatalytic sites (4, 17, 26). To determine whether cyclic nucleotides other than c-di-GMP (Fig. 6C) have an allosteric effect on DipA, PDE assays in the absence and presence of cAMP and cGMP were carried out. While cGMP had little effect, addition of cAMP increased DipA activity by 17% as determined using bis-NPP as a substrate (Fig. 6C). Moreover, increased degradation of c-di-GMP was observed in the presence of cAMP (Fig. 6D). Overall, the activity of DipA in the presence of cAMP was comparable to the in vitro activity of DipA lacking the GAF domain (DipA-NoGAF) (Fig. 6C and D). The PDE activity of purified DipA-NoGAF was not enhanced by the presence of cAMP or cGMP (data not shown).

Expression of dipA lacking the GAF domain in PAO1/pJN-PA4843 resulted in this strain demonstrating swarming motility at wild-type levels and increased turbidity of the growth medium indicative of cells not aggregating (Fig. 6E and F). The findings suggest that DipA lacking the GAF domain is an active phosphodiesterase, with the GAF domain playing a role in regulating DipA activity in vivo. In contrast, the GAF domain alone (“GAFonly”) was found to have no phosphodiesterase activity (Fig. 6E and F).

The GAF domain is not essential for dispersion and biofilm formation.

To determine whether DipA-NoGAF and the GAF domain alone can functionally replace intact DipA, the ΔdipA mutant was complemented with the dipA-NoGAF and GAFonly constructs and tested using dispersion assays. As shown in Fig. 7B, the GAF domain alone did not restore the dispersion phenotype of ΔdipA mutant biofilms. However, dipA-NoGAF was capable of restoring the dispersion-deficient phenotype of the ΔdipA mutant to wild-type levels. With respect to biofilm architecture, the DipA construct lacking the GAF domain partly restored the ΔdipA biofilm phenotype to wild-type levels (Fig. 7C and Table 2). Moreover, overexpression of dipA-NoGAF in PA14 resulted in reduced biofilm biomass accumulation and reduced thickness, but to a lesser extent than overexpression of dipA (Fig. 3A and 7C and Table 2). Similar results were obtained in PAO1 with overexpression of dipA-NoGAF resulting in biofilm biomass accumulation to a level between those observed for PAO1/pMJT-dipA and PAO1 biofilms (Fig. 3A and 7C and Table 2).

DISCUSSION

Dispersion in response to environmental cues coincides with physical and physiological alterations, as evidenced by changes in the biofilm structure, differential gene expression, the requirement for protein synthesis, and the resulting decrease of c-di-GMP levels. Here, we demonstrate that the dispersion-specific modulation of c-di-GMP is independent of previously reported enzymes and regulatory systems known to modulate c-di-GMP, including BifA and the Wsp chemosensory system. Instead, our work resulted in the identification of the phosphodiesterase DipA essential for dispersion.

DipA has previously been identified to play a role in P. aeruginosa PA68 flagellum-driven motility and chemotaxis (24). Inactivation of the dipA gene led to a decreased chemotactic response, which correlated with a 2-fold decreased expression of genes carried by the cheYZAB operon from chemotaxis gene cluster 1. While still producing flagella, ΔdipA displayed reduced swimming and swarming motility compared to the parental strain, suggesting that DipA affected flagellar functions. Impaired flagellum-dependent motility was suggested to be the reason for the differences in biofilm architecture between the wild type and ΔdipA mutants. With citrate as the carbon source, the P. aeruginosa PA68 wild-type strain formed flat biofilms, whereas the ΔdipA mutant strain formed biofilms with irregularly shaped microcolonies. The difference in architecture was not observed with glucose as a carbon source (24). The authors did not link DipA to biofilm dispersion. While our findings support a role of DipA in swarming motility, no difference in swimming motility was observed. It is likely that the difference in swimming motility is due to strain differences or differences in the medium used for swimming motility assays. With respect to biofilm formation, our findings corroborate findings by Li et al. (24) in that the ΔdipA mutant displayed increased initial attachment and increased biofilm formation. However, the differences in biofilm architecture observed by us between mutant and wild-type biofilms grown on LB medium were less dramatic than those observed by Li et al. using a citrate-based growth medium (Fig. 3A) (24). Moreover, our findings suggest differences in swarming motility, attachment, and biofilm formation to be the result of the c-di-GMP-modulating activity of DipA.

While DipA contributed to the overall architecture of and c-di-GMP levels in biofilms, DipA functioned primarily in free-swimming cells. Our findings suggested DipA to be active upon exposure to dispersion-inducing conditions, with DipA PDE activity being elevated in free-swimming, dispersed cells. Considering that dipA expression resulted in decreased initial attachment and reduced Psl production, it is likely that DipA functions by rendering bacterial cells that are dispersing from the biofilm upon sensing dispersion-inducing cues less adhesive. In contrast, reduced DipA PDE activity and gene expression in biofilm cells that remained attached following induction of dispersion may assist in increasing Psl production and, thus, adhesiveness. While differences in polysaccharide production may contribute to the release of bacterial cells from the biofilm, elevated levels of Pel and Psl polysaccharide did not impair the dispersion response. Alternatively, it is likely that differences in attachment and Psl production contribute to the ΔdipA mutant transitioning at a different rate compared to the wild type from the reversible to the irreversible attachment stage. This is further supported by the finding that in strain PAO1, Psl has been demonstrated to contribute to attachment to glass and mucin-coated surfaces (11).

In many regards, DipA is an unusual phosphodiesterase. For one, among the 12 phosphodiesterase mutants tested for lack of dispersion, DipA was one of only two proteins harboring EAL domains that played a role in dispersion. This raises the question of whether there is functional redundancy with respect to dispersion or whether these proteins act in concert to enable dispersion. Considering that both proteins, similarly to the previously described BdlA (31), are essential for dispersion in response to a variety of dispersion-inducing conditions, it is likely that they do not represent functionally redundant activities but instead are part of a pathway or protein complex capable of modulating an intracellular c-di-GMP pool to enable dispersion. Future research will be required, however, to address this question. What makes DipA even more unusual is the finding of DipA harboring two EAL motifs, of which only one is located within the EAL domain and conserved in other phosphodiesterases, including BifA (see Fig. S5 in the supplemental material). While the function of EAL1 is unclear, our findings strongly suggested both EAL motifs to be essential for the DipA PDE activity in vivo as well as for biofilm dispersion.

Another interesting feature of DipA is its GAF domain. Considering that the GAF domain was essential for neither dispersion nor DipA phosphodiesterase activity but appeared instead to have some modulating effect on DipA activity, further research will be required to elucidate the role of this domain in DipA. We further demonstrate that the PDE activity of purified DipA is enhanced by the cyclic nucleotides c-di-GMP and cAMP. cAMP has not been previously linked to biofilm formation and dispersion by P. aeruginosa. Moreover, catabolite repression in this bacterium does not occur as a consequence of reduced intracellular cAMP levels, and repression is not alleviated by exogenously added cAMP (40). While the catabolite repression control (Crc) protein, which plays a role in the regulation of carbon metabolism, was determined to be necessary for biofilm formation, the inability to form a biofilm was due to a defect in type IV pilus-mediated twitching motility rather than cAMP modulation (32). cAMP levels have been reported, however, to affect biofilm formation in Vibrio cholerae, with exogenous addition of cAMP to the growth medium leading to a decrease in biofilm formation (18). Moreover, cAMP has been recently linked to c-di-GMP signaling in this bacterium via the cyclic cAMP receptor protein (cAMP-CRP) regulatory complex (13). cAMP-CRP negatively regulates transcription of VPS polysaccharide biosynthesis genes and genes encoding biofilm matrix proteins and is considered a negative regulator of biofilm formation in V. cholerae. Using mutational and phenotypic analysis, Fong and Yildiz (13) demonstrated that the diguanylate cyclase CdgA was largely responsible for the increased transcription of vps and biofilm matrix protein genes, as well as enhanced biofilm formation in a Δcrp mutant. Consistent with these findings, the PDE activity of DipA was elevated in the presence of cAMP.

In conclusion, our data suggest that the phosphodiesterase DipA contributes to the transition from the biofilm and the free-swimming mode of growth by modulating c-di-GMP and polysaccharide levels. While DipA played a role in the regulation of c-di-GMP levels under biofilm growth conditions, the phosphodiesterase contributed primarily to increased phosphodiesterase activity and reduced c-di-GMP levels detected in dispersed cells. To our knowledge, this is the first description of a dispersion-induced phosphodiesterase enabling dispersion in P. aeruginosa biofilms upon sensing environmental cues.