Modulation of Pseudomonas aeruginosa Biofilm Dispersal by a Cyclic-Di-GMP Phosphodiesterase with a Putative Hypoxia-Sensing Domain

Abstract

Pseudomonas aeruginosa encodes many enzymes that are potentially associated with the synthesis or degradation of the widely conserved second messenger cyclic-di-GMP (c-di-GMP). In this study, we show that mutation of rbdA, which encodes a fusion protein consisting of PAS-PAC-GGDEF-EAL multidomains, results in decreased biofilm dispersal. RbdA contains a highly conserved GGDEF domain and EAL domain, which are involved in the synthesis and degradation of c-di-GMP, respectively. However, in vivo and in vitro analyses show that the full-length RbdA protein only displays phosphodiesterase activity, causing c-di-GMP degradation. Further analysis reveals that the GGDEF domain of RbdA plays a role in activating the phosphodiesterase activity of the EAL domain in the presence of GTP. Moreover, we show that deletion of the PAS domain or substitution of the key residues implicated in sensing low-oxygen stress abrogates the functionality of RbdA. Subsequent study showed that RbdA is involved in positive regulation of bacterial motility and production of rhamnolipids, which are associated with biofilm dispersal, and in negative regulation of production of exopolysaccharides, which are required for biofilm formation. These data indicate that the c-di-GMP-degrading regulatory protein RbdA promotes biofilm dispersal through its two-pronged effects on biofilm development, i.e., downregulating biofilm formation and upregulating production of the factors associated with biofilm dispersal.

Pseudomonas aeruginosa is a ubiquitous Gram-negative bacterial pathogen, well known for its remarkable ability to live and survive in diverse environments. The pathogen produces a range of virulence factors, such as exotoxin A, exoenzyme S, pyocyanin, proteases, elastase, rhamnolipids, and lipopolysaccharides, and frequently causes acute and chronic infections in immunocompromised hosts. In addition, P. aeruginosa may switch from a planktonic growth mode to a surface-attached lifestyle, i.e., biofilms, in response to biotic or abiotic stresses (14). Biofilm bacterial cells are stuck to each other and embedded in a self-manufactured matrix of extracellular polymeric substance, enabling them to escape from human defense responses and withstand high-dose antibiotic treatments. P. aeruginosa has become a serious concern in intensive care units, largely due to its biofilm-related drug resistance and the potential of biofilm as a source of contamination (16, 41, 43, 46).

Biofilm formation by P. aeruginosa progresses through multiple developmental stages, beginning with attachment to a surface, followed by migration and division to form microcolonies, and then maturation involving expression of matrix polymers. The biofilm developmental life cycle comes full circle when the biofilm cells disperse (51). For the convenience of discussion, we define here that biofilm development covers two phases, i.e., formation and dispersal. Recent research has revealed a range of factors associated with biofilm dispersal, including matrix-degrading enzymes (5), activation of motility genes, nutrient level and microbial growth status (52), production of biosurfactants (4), activation of lytic bacteriophage (61), and changes in intracellular levels of cyclic di-GMP (c-di-GMP) (28, 30, 34).

Cyclic di-GMP is a ubiquitous second messenger identified in a growing number of bacterial species. It has been shown that intracellular levels of c-di-GMP influence a myriad of bacterial behaviors, with a common theme being that accumulation of c-di-GMP promotes sessile behaviors, i.e., biofilm formation (28, 56), while breakdown of c-di-GMP and a subsequent decrease in cellular levels of this signal favor motile behaviors, such as swarming motility and twitching motility (30, 33, 56). The cellular levels of c-di-GMP are controlled through the opposing activities of diguanylate cyclases, proteins containing a GGDEF domain (44, 56), and phosphodiesterases, which contain either an EAL domain (56) or an HD-GYP domain (48). A number of GGDEF domain proteins have been shown to synthesize c-di-GMP by using two molecules of GTP (44, 56), whereas EAL domain proteins or HD-GYP domain proteins hydrolyze c-di-GMP into GMP and pGpG (48, 56). The annotated genome of P. aeruginosa PAO1 encodes 17 proteins containing the GGDEF domain, 5 with an EAL domain, and 16 that carry both domains (34). A comprehensive survey study of the genes encoding diguanylate cyclases and phosphodiesterases showed that a subset of these c-di-GMP metabolic enzymes are associated with biofilm development (34). Among them, a few enzymes have been previously characterized at the molecular and biochemical levels (28, 30, 33, 38).

It was noticed that many of the enzymes implicated in c-di-GMP metabolism are fused to one or several types of signal-sensing domains or signal receiver domains at the N terminus, such as PAS, GAF, and BLUF (34). These findings suggest potential roles of these regulatory domains in the modulation of c-di-GMP metabolism in response to various environmental cues and signal molecules. In this study, by screening the transposon mutants of P. aeruginosa defective in biofilm dispersal, we identified the gene PA0861, designated rbdA for its role in regulation of biofilm disposal. rbdA encodes a regulatory protein consisting of PAS-PAC-GGDEF-EAL multidomains. Genetic and biochemical analyses were conducted to determine the role of RbdA in c-di-GMP metabolism and to investigate potential association of its signal-sensing domain PAS in the modulation of enzyme activity. In addition, we also determined the biological functions regulated by RbdA. Our data show that the conserved GGDEF domain of rbdA acts as an allosteric regulatory domain for the EAL-borne phosphodiesterase activity. We further present evidence that RbdA modulates P. aeruginosa biofilm dispersal through regulation of bacterial motility and production of rhamnolipids and exopolysaccharides (EPS).

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The P. aeruginosa strains and other bacteria used in this study are listed in Table 1. Unless otherwise indicated, bacteria were routinely grown at 37°C in Luria-Bertani (LB) broth. Antibiotics were added when necessary at the following concentrations: carbenicillin, 300 μg/ml for P. aeruginosa and 200 μg/ml for Escherichia coli; gentamicin (Gm), 30 μg/ml for P. aeruginosa and 5 μg/ml for E. coli; ampicillin, 200 μg/ml for E. coli; tetracycline, 100 μg/ml for P. aeruginosa and 10 μg/ml for E. coli.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Relevant genotype or phenotype	Source or reference
P. aeruginosa strains
PAO1	Prototrophic laboratory strain	Holloway et al. (28a)
B1-93	Transposon inserted after nt 1020 of rbdA	This study
B4-55	Transposon inserted after nt 1980 of rbdA	This study
B4-56	Transposon inserted after nt 198 of rbdA	This study
B12-49	Transposon inserted after nt 16 of rbdA	This study
ΔrbdA mutant	rbdA deletion mutant of PAO1 with 1,446-nt internal coding region deleted	This study
ΔbifA mutant	bifA deletion mutant of PAO1 with 1,380-nt internal coding region deleted	This study
ΔpelD mutant	pelD deletion mutant of PAO1 with 1,002-nt internal coding region deleted	This study
ΔpslF mutant	pslF deletion mutant of PAO1 with 546-nt internal coding region deleted	This study
ΔrbdA ΔpelD mutant	ΔrbdA ΔpelD double-deletion mutant	This study
ΔrbdA ΔpslF mutant	ΔrbdA ΔpslF double-deletion mutant	This study
ΔpelD ΔpslF mutant	ΔpelD ΔpslF double-deletion mutant	This study
ΔrbdA ΔpelD ΔpslF mutant	ΔrbdA ΔpelD ΔpslF triple-deletion mutant	This study
PAO1::Tn7	Tn7 inserted into att Tn7 sites of PAO1	This study
PAO1::Tn7-rbdA	Tn7 fused with rbdA and its native promoter inserted into att Tn7 sites of PAO1	This study
ΔrbdA::Tn7	Tn7 inserted into att Tn7 sites of ΔrbdA mutant	This study
ΔrbdA::Tn7-rbdA	Tn7 fused with rbdA and its native promoter inserted into att Tn7 sites of ΔrbdA mutant	This study
E. coli strains
DH5α	spuE44 ΔlacU169(φ80lacZΔM15) hsdR17 λpir recA1 endA1 gyrA96 thi-1 relA1	Lab collection
S17-1	recA pro (RP4-2Tet::Mu Kan::Tn7)	Simon et al. (56a)
BL21(DE3)	F−ompT hsdS(rB− mB−) dcm+ Tetrgal (DE3) endA	Stratagene
Plasmids
pBT20	Mariner transposon mutagenesis vector; Gmr Ampr	Kulasekara et al. (35)
pME6010	pVS1-p15A shuttle vector	Heeb et al. (27)
pUCP19	E. coli-P. aeruginosa shuttle expression vector with Plac, Ampr Carr	ATCC
pUCP-rbdA	pUCP19 containing rbdA under control of Plac	This study
pUCP-GGDEF-5A	pUCP19 containing rbdA derivative with GGDEF motif replaced by alanine (G453A, G454A, D455A, E456A, F457A)	This study
pUCP-ΔPAS	pUCP19 containing rbdA derivative with the sequence encoding the PAS domain (aa 274 to 354) deleted	This study
pUCP-317A	pUCP19 containing rbdA derivative with I317A substitution	This study
pUCP-319A	pUCP19 containing rbdA derivative with G319A substitution	This study
pUCP-rbdA (339A)	pUCP19 containing rbdA derivative with L339A substitution	This study
pUCP-rbdA (317A-319A)	pUCP19 containing rbdA derivative with I317A and G319A substitutions	This study
pUCP-rbdA (317A, 319A, 339A)	pUCP19 containing rbdA derivative with I317A, G319A, and L339A substitutions	This study
pUCP-sadC	pUCP19 containing sadC under control of Plac	This study
pUCP-bifA	pUCP19 containing bifA under control of Plac	This study
pK18mobsacB	Broad-host-range gene replacement vector, sacB, Kmr	Lab collection
pK18GT	pK18mobsacB with the Gmr cassette inserted at NcoI site	This study
pK18GT-rbdAdel	pK18GT containing rbdA flanking regions for generation of rbdA in-frame deletion	This study
pK18GT-bifAdel	pK18GT containing bifA flanking regions for generation of bifA in-frame deletion	This study
pUC18-mini-Tn7T-Gm-lacZ	lacZ transcriptional fusion vector, Ampr Gmr	Choi et al. (10)
pUC18-mini-Tn7T-rbdA	Single-copy Tn7 insertion plasmid containing rbdA with its endogenous promoter	This study
pME2-lacZ	pME6010 carrying full-length lacZ	This study
PpelA′-lacZ	pME2-lacZ carrying pelA promoter in front of lacZ	This study
PrbdA′-lacZ	pME2-lacZ carrying rbdA promoter in front of lacZ	This study
pTNS2	R6K replicon, encodes the TnsABC+D-specific transposition pathway, Ampr	Choi et al. (10)
pGEX-6p-1	GST fusion protein expression vector, Ampr	Invitrogen
pGEX-rbdA	rbdA fused to GST gene in frame in pGEX-6p-1, Ampr	This study
pGEX-GGDEF-5A	pGEX6P1-rbdA with GGDEF motif replaced with alanine (G453A, G454A, D455A, E456A, F457A)	This study
pGEX-EAL-3A	pGEX6P1-rbdA with EAL motif replaced with alanine (E585A, L586A, L587A)	This study

Transposon mutagenesis.

Transposon mutants were generated with the Mariner transposon delivery plasmid pBT20 by using modified protocols from those reported by O'Toole and Kolter (40). Briefly, the plasmid was transferred from E. coli S17-1(λpir) into the recipient strain PAO1 by biparental mating at 37°C for 5 h. Transposon mutants were selected on minimal medium agar plates supplemented with Gm (for selection of transposon mutants) (64) and screened as described below.

Biofilm formation was first assayed by the ability of cells to adhere to the wells of 96-well microtiter plates made of polystyrene (Falcon 353072). Transposon mutants were inoculated into wells containing LB liquid medium (150 μl/well) using toothpicks and incubated at 37°C for 24 h. Bacterial suspensions were removed, and the plates were washed twice with water. Biofilm bacteria were stained by adding 200 μl of 0.1% crystal violet (CV) to each well. The plates were incubated at room temperature for approximately 15 min, rinsed thoroughly with water, and scored for the formation of biofilm (40). The mutants that showed less or more biofilm than the parental strain PAO1 were selected for further confirmation by using 14-ml polystyrene tubes. The selected biofilm mutants were sequenced as described below.

Arbitrary PCR was then used to identify the genes disrupted by transposon insertion as described previously (6). The DNA sequences flanking transposon insertion were analyzed with the NCBI BLAST server (http://www.ncbi.nlm.nih.gov/blast/) and the Pseudomonas Genome Database (62a).

Construction of mutant strains.

The plasmids and primers used in this study are listed in Table S1 in the supplemental material. To generate the rbdA deletion mutant of P. aeruginosa, two PCR fragments flanking rbdA were amplified and fused by overlap extension PCR. The resulting fusion PCR fragments contained the truncated rbdA, in which the 1,447-nucleotide (nt) coding sequence (from bp 520 to 1967) of rbdA was deleted. After purification with the Quick PCR purification kit (Qiagen), the fusion fragments were digested with EcoRI and HindIII and cloned into the corresponding site of the vector pK18Gm. The resultant construct was introduced into E. coli S17-1 by electroporation and then P. aeruginosa strain PAO1 by biparental mating. The generated ΔrbdA deletion mutant was confirmed by PCR and DNA sequencing analysis. Single-deletion ΔbifA, ΔpelD, and ΔpslF mutants, double-deletion ΔrbdA ΔpelD, ΔrbdA ΔpslF, and ΔpelD ΔpslF mutants, and the triple-deletion ΔrbdA ΔpelD ΔpslF mutant were generated by the same procedure using corresponding primers (listed in Table S1 of the supplemental material).

Complementation analysis and construction of the reporter strain.

For mutant in trans complementation, the coding region was amplified and the PCR product was cloned downstream of the lac promoter in the shuttle vector pUCP19 digested by EcoRI and HindIII. The resultant construct was mobilized into E. coli and sequenced before introducing it into the corresponding mutants indicated below. For single-copy complementation, the rbdA coding region together with its native promoter were amplified, and the PCR products were digested by SpeI/HindIII and cloned into the vector pUCP18-mini-Tn7T-Gm-lacZ. The resultant construct pUCP18-mini-Tn7T-Gm-rbdA was introduced into P. aeruginosa by coelectroporation with the helper plasmid pTNS2 as described previously (10). To construct the PpelA′-lacZ reporter plasmid, a 501-bp fragment corresponding to bp −432 to +69 relative to the translational start site of the pelA gene was amplified from the genomic DNA of P. aeruginosa by PCR using the primer pair listed in Table S1 of the supplemental material. The fragment was digested with HindIII and EcoRI and cloned in the same sites of pME2-lacZ, which was derived from pME6010 (17). The resultant construct was used to transform P. aeruginosa by electroporation, and the transformants were selected on LB agar plates containing relevant antibiotics. The same method was used to construct PrbdA′-lacZ using the primer pair listed in Table S1.

Biofilm formation assay and quantification.

A biofilm formation assay was performed according to the method previously described, with minor modifications (40). Briefly, overnight bacterial cultures were diluted to an optical density at 600 nm (OD₆₀₀) of 0.002 with fresh LB broth or with LB plus KNO₃ for anaerobic analysis. The diluted cultures (2 ml) were transferred to a 14-ml polystyrene tube (17 by 100 mm; Falcon 352057) and incubated at 37°C with shaking at 250 rpm for the indicated periods of time. Bacterial cultures were carefully removed for measurement of the OD₆₀₀. The bacterial cells bound to the walls of the tubes (biofilms) were stained with 0.1% crystal violet (Sigma) for 15 min at room temperature, and the tubes were then rinsed several times with water. The tubes were air dried at room temperature and photographed. For quantification, biofilms were suspended in 4 ml of 75% ethanol, and absorbance at 570 nm was measured with a spectrophotometer. For anaerobic analysis, bacterial cells were inoculated and cultured as described above except that LB medium was supplemented with 1% KNO₃ (63) and 2 ml mineral oil (54) was added to the top of each bacterial culture. Each experiment was repeated at least three times with duplicates.

Motility assay.

Swimming and swarming motility assays were conducted, respectively, on tryptone-NaCl medium (10 g/liter tryptone, 5 g/liter NaCl) containing 0.3% (wt/vol) agarose and on tryptone-glucose medium (8 g/liter tryptone, 8 g/liter glucose) supplemented with 0.5% (wt/vol) Difco Bacto agar. The plates were inoculated with bacteria from an overnight culture grown on an LB agar plate at 37°C with a sterile toothpick. The swim plates were incubated at room temperature for 48 h, and the swarm plates were incubated at room temperature for 24 h and then at 37°C overnight. The experiments were repeated twice with six replicates for each strain.

Pellicle assay.

Standing cultures in 18- by 150-mm borosilicate glass tubes (Fisher) containing 5 ml of T-broth without salt (10 g/liter tryptone) were grown at 30°C for 24 h and at room temperature for 24 h. Pellicle formation was monitored by visual inspection on the air-liquid interface of the standing culture. Complete coverage of the surface of the culture by an opaque layer of cells and matrix material indicated pellicle formation. The culture liquid was removed, and the pellicle was washed with water and stained with 6 ml of 0.1% crystal violet for 15 min. This was followed by removal of unbound crystal violet and rinsing of the tubes with water (21). The tubes were air dried at room temperature and photographed. For quantification, the attached cells (pellicle) were suspended in 6 ml of 75% ethanol. Absorbance at 570 nm was measured with a spectrophotometer.

Congo red binding assay.

To observe colony morphology, overnight cultures were diluted to an OD₆₀₀ of 0.025 with T-broth without salt, and 5 μl of diluted culture was spotted on a Congo red plate (21). The plates were incubated at room temperature for 3 days. For the quantitative assay, bacterial cells from 2 ml of overnight culture were collected by centrifugation and resuspended in 1 ml T-broth without salt containing 40 μg/ml Congo red. The mixture was incubated at 37°C with shaking at 250 rpm for 180 min. The bacterial cells and bound Congo red were removed by centrifugation, and the amount of Congo red remaining in the supernatant was determined by measuring the absorbance of the supernatant at 490 nm, using T-broth supplemented with 40 μg/ml Congo red as a control. The relative EPS unit was defined as the amount (in μg) of Congo red bound to bacterial cells at an OD₆₀₀ of 1 (36).

Rhamnolipid assay.

For rhamnolipid analysis, the bacteria were inoculated in phosphate-limited proteose peptone-glucose-ammonium salts (PPGAS) medium (62). After incubation at 37°C for 48 h, rhamnolipids were determined as described previously (4). Briefly, 1 ml of culture supernatant was extracted twice with 2 ml of diethyl ether. The ether fractions were pooled, evaporated to dryness, and reconstituted in 0.1 ml H₂O. The samples were diluted 1/10 in aqueous solution containing 0.19% orcinol (Sigma) and 53% H₂SO₄. The samples were then placed in boiling water for 30 min and cooled at room temperature for 15 min, and absorbance at 421 nm was measured. The amount of rhamnolipids was calculated by using a standard curve generated with commercial rhamnolipids (JBR 599; Jeneil Biosurfactant Co., LCC).

Expression and purification of RbdA protein.

The DNA fragment encoding RbdA was amplified with the primers listed in Table S1 of the supplemental material and subcloned into the expression vector pGEX-6p-1. The resultant glutathione S-transferase (GST)-RbdA fusion constructs were mobilized into E. coli and confirmed by DNA sequencing. For expression of the fusion protein, bacteria were grown at 37°C in LB medium supplemented with 200 μg/ml ampicillin in shaking flasks, and isopropyl-β-d-thiogalactopyranoside (0.1 mM) was added to the bacterial culture when the OD₆₀₀ reached about 0.6. After overnight culture at 18°C, the cell pellet was resuspended in phosphate-buffered saline (PBS) lysis buffer supplemented with Complete protease inhibitor cocktail (1 tablet/50 ml; Roche). The cells were then lysed by sonication, and the cell debris was removed by centrifugation at 18,000 rpm for 60 min. The supernatants were loaded onto an affinity column containing glutathione-Sepharose 4B beads for affinity binding, followed by washing with PBS lysis buffer. RbdA protein was separated from the bound GST protein by cleavage using PreScission protease (2 units per 100 μg of bound protein in PBS buffer) at 4°C overnight. The purity of recombinant RbdA was confirmed by SDS-PAGE analysis.

Enzyme reaction and HPLC analysis.

The enzyme assay and high-performance liquid chromatography (HPLC) analyses were performed following the methods described previously, with minor modifications (26). Briefly, 25 μg RbdA was added to reaction buffer containing 50 μM c-di-GMP, 60 mM Tris-HCl (pH 7.5), 10 mM MgCl₂, and 1 mM EDTA in a final volume of 500 μl. The reaction mixture was kept at room temperature for 10 min prior to incubation at 37°C. After the indicated periods of time, an aliquot of reaction mixture was taken and the reaction was terminated by boiling for 10 min. To test the effect of GTP, the molecule was added to the reaction mixture at a final concentration of 50 μM and the reaction was performed as described above.

Cyclic-di-GMP guanylate cyclase activity was determined following procedures described previously (44). Briefly, 20 μg RbdA or RbdA (EAL-AAA) and 50 μM GTP were added to reaction buffer (75 mM Tris-HCl [pH 7.8], 250 mM NaCl, 25 mM KCl, 10 mM MgCl₂). The reaction mixture was incubated at 37°C for up to 120 min. HPLC analysis was performed by injection of 10-μl aliquots of reaction supernatants to a 250- by 4.6-mm C₁₈ RP column (Waters) on a Waters LC chromatographic system. Separation was done by elution with an isocratic mobile phase (2% [vol/vol] acetonitrile in 50 mM NaH₂PO₄ [pH 5.2]) at a rate of 1 ml/min.

Site-directed mutagenesis.

The constructs pUCP19-rbdA and pGEX-6p-1-rbdA were used as templates for site-directed mutagenesis. PCR was carried out using Pfu Turbo DNA polymerase (Invitrogen) and relevant mutagenic PCR primers (listed in Table S1 of the supplemental material). The PCR products were treated with DpnI, to digest the methylated and hemimethylated DNA, prior to transformation of E. coli strain DH5α. Introduction of the desired mutation was confirmed by DNA sequencing.

β-Galactosidase assay.

For measurement of β-galactosidase activity, P. aeruginosa strain PAO1 containing the PpelA′-lacZ construct was grown overnight in LB with shaking at 37°C. The starter cells were then inoculated in LB at a ratio of 1:100 and grown at the same temperature to an OD₆₀₀ of 1.5. β-Galactosidase activity was determined using a standard protocol (50). Results are reported as Miller units (MU) of β-galactosidase activity per OD₆₀₀.

Western blot analysis.

For Western blot analysis, overnight cultures of wild-type P. aeruginosa and its derivatives were diluted 100-fold in LB broth and incubated at 37°C until the OD₆₀₀ reached about 1.5. Total soluble proteins were prepared by using CelLytic B cell lysis reagent (Sigma). For each sample, 5 μg total protein was loaded on an 8% SDS-PAGE gel for electrophoresis separation. Western blot analysis was performed following the standard protocol (50).

RESULTS

Mutation of PA0861 encoding a multidomain regulatory protein implicated in c-di-GMP metabolism resulted in a hyperbiofilm phenotype.

Transposon mutagenesis was conducted to identify the genes of P. aeruginosa implicated in biofilm development. Around 20,000 transposon insertion mutants were screened, resulting in the identification of 29 mutants with altered biofilm phenotypes. Although the majority of mutants arose due to insertions in the genes associated with polysaccharide production, such as pslA and pslC (20, 21), flagellar formation, such as flgJ and fleN (15, 16), and type IV fimbrial biogenesis, including pilW and pilY1 (1), four mutants were found to have transposon insertion at various positions of the gene PA0861. This gene encodes a hypothetical protein containing N-terminal PAS-PAC domains, a GGDEF (diguanylate cyclase) domain, and an EAL (phosphodiesterase) domain (Fig. 1A). A previous study showed that mutation of PA0861 resulted in increased biofilm formation (34), but the actual mechanism involved was unclear.

FIG. 1.

The biofilm phenotypes of P. aeruginosa PAO1 and its rbdA transposon insertion mutants. (A) Genetic organization and domain structures of RbdA. Vertical arrows indicate the relative transposon insertion sites in each mutant. The domain structure prediction was performed using the SMART program (http://smart.emblheidelberg.de/). Abbreviations: SP, signal peptide; TM, transmembrane domain; PAS, PAS domain; PAC, PAC domain; GGDEF, diguanylate cyclase domain; EAL, diguanylate phosphodiesterase domain. (B) Visualization of bacterial biofilm formation on polystyrene tubes by crystal violet staining. (C) Quantitative comparison of the biofilm formation by wild-type strain PAO1 and its rbdA mutants. The data shown are means of triplicates, and the standard deviations are shown by error bars.

When grown in LB medium for 16 h at 37°C, we found that these four mutants of PA0861 produced substantially more biofilm mass than their parental strain PAO1 (Fig. 1B). Quantitative analysis revealed a 3- to 4-fold difference between PAO1 and the mutants (Fig. 1C). Given its key role in the regulation of biofilm development, as described below, the PA0861 gene was designated rbdA. Genome organization analysis showed that its upstream gene, PA0860, encodes an ATP-binding/permease fusion unit of an ABC transporter, and its downstream gene, PA0862, encodes a hypothetical protein with about 32% amino acid identity to the pyrroloquinoline quinine synthesis protein C (gene locus HMPREF0014_03462) of Acinetobacter calcoaceticus. The above in silico analysis results do not seem to suggest a potential functional association between rbdA and its close neighbors.

RbdA is implicated in regulation of P. aeruginosa biofilm dispersal.

To investigate how RbdA might influence biofilm development, we generated the rbdA in-frame ΔrbdA deletion mutant by using P. aeruginosa PAO1 as the parental strain. A time course analysis showed that during the first few hours of bacterial growth, both the wild type and the ΔrbdA mutant accumulated biofilm mass at a similar rate (Fig. 2A). The major difference appeared at 6 h after incubation. While the biofilm mass of wild-type strain PAO1 began to decline, the ΔrbdA deletion mutant maintained the increasing trend for biofilm accumulation over a period of 48 h (Fig. 2A). The rbdA mutant and the wild type grew at rates indistinguishable from each other under the same culture conditions used in this experiment (data not shown), suggesting that the enhanced biofilm mass accumulation in the ΔrbdA mutant was due to altered biofilm dispersal.

FIG. 2.

RbdA is associated with biofilm dispersal. (A) Time course analysis of bacterial biofilm mass. (B) In trans expression of rbdA from the multicopy vector pUCP19 in PAO1 and the ΔrbdA deletion mutant reduced biofilm accumulation. The empty vector pUCP19 was introduced into the wild type and ΔrbdA mutant, and these constructs were used as controls. (C) Expression of a single copy of rbdA in the ΔrbdA mutant restored the biofilm phenotype to the level of wild-type PAO1. The integration vector Tn7 was inserted in PAO1 and the ΔrbdA mutant, and these constructs were used as controls. The data shown are means of triplicates, and the standard deviations are shown by error bars.

For further verification, the wild-type rbdA gene was cloned under the control of the lac promoter in the plasmid vector pUCP19 for in trans complementation analysis. The resulting construct, pUCP-rbdA, was introduced separately into the wild type and ΔrbdA mutant for analysis of biofilm formation. The results showed that overexpression of rbdA in strain PAO1 resulted in an approximately 90% reduction in biofilm mass at 16 h after incubation, compared with the control strain PAO1 carrying an empty vector (Fig. 2B). It was noticed that in trans expression of rbdA in the ΔrbdA mutant and in wild-type PAO1 reduced biofilm accumulation to a similar residual level (Fig. 2B), despite the ΔrbdA mutant showing a much higher rate of biofilm accumulation than its parental strain PAO1. One plausible explanation is that rbdA was expressed in pUCP19, which is a high-copy-number vector.

We then tested the complementation effect of a single copy of rbdA inserted in the chromosome of the ΔrbdA mutant. The chromosomal integration vector pUCP18-mini-Tn7T and the expression construct pUCP18-mini-Tn7T-rbdA were introduced and integrated, respectively, into the chromosomal DNA of P. aeruginosa by insertion at a previously characterized chromosomal site in the intergenic region between PA5548 and glmS (10). The vector alone had no effect on biofilm formation by the wild type or ΔrbdA mutant (Fig. 2C). In contrast, expression of rbdA from the integrated vector on the chromosome in the ΔrbdA mutant fully restored its biofilm mass accumulation to the level of the wild-type strain (Fig. 2C). Expression of the same integrated rbdA on the chromosome in the wild-type strain, which has one original copy of rbdA, resulted in about a 70% decrease in the biofilm mass (Fig. 2C).

In silico analysis suggested that RbdA contains highly conserved diguanyl cyclase and phosphodiesterase motifs.

The rbdA gene encodes a protein of 818 amino acids (aa) with a predicted molecular mass of 91 kDa. Domain analysis showed that the protein contains two domains (i.e., GGDEF and EAL) associated with c-di-GMP metabolism. The GGDEF domain is the conserved feature of diguanyl cyclases, which catalyze the synthesis of c-di-GMP from two GTP molecules. On the other hand, the EAL domain is shared by many c-di-GMP phosphodiesterases that degrade c-di-GMP. To analyze the functionality of RbdA, we aligned the GGDEF domain of RbdA with the homologous domains of previously characterized diguanyl cyclases, including PleD of Caulobacter crescentus (44), YeaP of Escherichia coli (49), Dgc1 of Gluconacetobacter xylinus (58), HmsT of Yersinia pestis (55), and AdrA of Salmonella enterica serovar Typhimurium (22). The results showed that the predicted GGDEF domain of RbdA contains all the conserved signature amino acid residues, G, G, D, E, and F (the A site, also known as the GGDEF motif) (see Fig. S1A in the supplemental material), which are critical for diguanyl cyclase enzyme activity (8). In addition, the GGDEF domain of RbdA also contains the residues RXXD (I site), the inhibitory c-di-GMP-binding site situated 5 amino acids upstream of the GGDEF motif (12, 29). Similarly, we aligned the EAL domain of RbdA to its corresponding domains of YkuI of Bacillus subtilis (39), YhjH from Salmonella enterica (56), RocR of P. aeruginosa (47), and YahA and Dos of E. coli (53). The results showed that RbdA contains the identical residues E, L, and L as YhjH from Salmonella enterica, which is a functional phosphodiesterase (56), in the positions corresponding to the conserved EAL motif (see Fig. S1B in the supplemental material). Together, these findings suggest that RbdA may potentially have both diguanylate cyclase and phosphodiesterase activities.

Purified RbdA degrades c-di-GMP.

In order to determine the phosphodiesterase activity of RbdA, the recombinant RbdA was incubated with c-di-GMP for different lengths of time prior to HPLC analysis. The results showed that RbdA degraded c-di-GMP progressively, and after 4 h of incubation at 37°C, all the c-di-GMP molecules in the reaction mixture were consumed, generating pGpG and GMP (Fig. 3A to D). To test RbdA cyclase activity, we incubated RbdA with GTP alone. Surprisingly, RbdA also produced pGpG in the presence of GTP (Fig. 3E and F). After verification of this unexpected finding using newly purified RbdA protein, we decided to determine whether the GGDEF domain was responsible for pGpG production. For this purpose, we substituted the GGDEF and the EAL motif of RbdA with alanine (A) to generate the variants GGDEF-5A and EAL-3A, respectively. The variants were purified and reacted with GTP. As expected, HPLC analysis of the GGDEF-5A-GTP reaction mixture did not reveal any product peak (data not shown). However, a new product peak with a retention time of about 11.6 min was detected from the EAL-3A-GTP reaction mixture (see Fig. S4 in the supplemental material). This product is new and is neither pGpG (retention time, 9.6 min) nor c-di-GMP (retention time, 13.7 min). Mass analysis of this new peak showed that it was guanosine (see Fig. S5 in the supplemental material). Taken together, these results indicate that RbdA is a c-di-GMP phosphodiesterase. It can also produce pGpG by using GTP as a substrate, and this reaction requires both the GGDEF and EAL domains.

FIG. 3.

RbdA is a c-di-GMP phosphodiesterase. (A) The standard control mixture containing four nucleotides in a final concentration of 50 μm was prepared in the reaction buffer, and 10 μl was injected for HPLC analysis. (B to D) RbdA after reaction with c-di-GMP at room temperature for 10 min (B), at room temperature for 10 min followed by 37°C for 30 min (C), and at room temperature for 10 min followed by 37°C for 240 min (D). (E and F) RbdA after reaction with GTP at 37°C for 0 min (E) or 120 min (F). The peak areas (μV·s) of key molecules are provided under the corresponding HPLC peak for the convenience of comparison. (G) RbdA and the phosphodiesterase BifA, but not the diguanylate cyclase SadC, were functional homologues in regulation of biofilm development.

To verify the findings, we used a functional complementation approach by using in trans expression of a c-di-GMP phosphodiesterase and a diguanylate cyclase in the ΔrbdA mutant, respectively. For this purpose, we selected two characterized enzymes encoded, respectively, by sadC and bifA of P. aeruginosa. The gene sadC encodes an inner membrane diguanylate cyclase with a conserved GGDEF domain (38), and bifA encodes a dual EAL/GGDEF domain-containing protein but displays only phosphodiesterase activity (33). The results showed that expression of bifA in the ΔrbdA mutant restored the biofilm mass to a level similar to that of the mutant complemented with the wild-type rbdA, whereas expression of sadC in the ΔrbdA mutant further increased its biofilm biomass (Fig. 3G). Conversely, in trans expression of rbdA in the bifA in-frame deletion mutant of P. aeruginosa resulted in reduced biofilm formation (Fig. 3G). The results showed that RbdA is functionally similar to BifA and acts as a phosphodiesterase to hydrolyze c-di-GMP under both in vitro and in vivo conditions.

Addition of GTP enhances the phosphodiesterase activity of RbdA.

Given that RbdA is a functional phosphodiesterase, it was intriguing to determine whether the GGDEF domain of RbdA played a role in RbdA-dependent c-di-GMP signaling. We measured the phosphodiesterase activity of RbdA in the presence and absence of GTP under the same conditions described for Fig. 3. Like other proteins with GGDEF-EAL fusion domains (13, 30), the phosphodiesterase activity of RbdA was greatly enhanced when the reaction mixture was supplemented with 50 μM GTP. After 30 min of incubation at 37°C, almost all the c-di-GMP molecules were degraded by RbdA to generate pGpG in the presence of GTP (Fig. 4A). In contrast, only a small portion of c-di-GMP was digested by RbdA in the control mixture without GTP at 30 min postreaction (Fig. 4A).

FIG. 4.

GTP induction of RbdA phosphodiesterase activity requires the GGDEF domain. (A) The phosphodiesterase activity of RbdA and its variant GGDEF-5A in the absence and presence of GTP. The concentration of c-di-GMP after termination of the reaction was determined by measuring UV absorbance following HPLC separation. The experiment was performed twice with similar results, and the figure shows a representative set of data. (B) In vivo assay of the role of the GGDEF domain in modulation of biofilm development. The ΔrbdA deletion mutant was complemented by in trans expression of the wild-type rbdA or its variant, GGDEF-5A, from the vector pUCP19. The inset shows the Western blot data for RbdA and its variant, GGDEF-5A, expressed in the ΔrbdA mutant.

We then replaced the five signature residues, i.e., G, G, D, E, and F, of the GGDEF motif with alanine (A) to generate the variant GGDEF-5A, and we measured the phosphodiesterase activity of this mutated RbdA in the presence and absence of GTP. The results showed that the variant digested c-di-GMP at a similar rate, regardless of the presence or absence of GTP (Fig. 4A). To test the impact of the GGDEF motif substitution under in vivo conditions, we cloned the mutated rbdA gene under the control of the lac promoter to generate the construct pUCP-GGDEF-5A for complementation analysis. The results showed that the substitution of the GGDEF motif had no effect on protein expression level (Fig. 4B, inset) but compromised its ability to reduce the biofilm mass of the ΔrbdA mutant in comparison with that of the wild-type RbdA (Fig. 4B).

The PAS domain of RbdA is critical for regulation of biofilm dispersal.

Besides the GGDEF and EAL domains, RbdA also contains a PAS domain and a PAC domain. Several sensor proteins with PAS domains, including Dos of E. coli, FixL of Bradyrhizobium japonicum, and PdeA1 of Acetobacter xylinum, which is now called Gluconacetobacter xylinum, have been shown to play roles as oxygen sensors (9, 18, 24, 45). The sequence alignment of the PAS domains of these homologues showed that the most similar homologue of the PAS domain of RbdA is that of FixL, which shares about 25% identity and 62% similarity at the amino acid level. These two PAS domains also shared similar secondary structure folds (Aβ, Bβ, Cα, Dα, Eα, Fα, Gβ, Hβ, and Iβ), and the conserved residues I317, G319, and L339 (corresponding to I215, G217, and L236, respectively, in FixL) were involved in ligand binding (24, 31) (Fig. 5A).

FIG. 5.

The PAS domain is critical for RbdA activity. (A) The PAS domain of RbdA shares the conserved structural folds and the key residues of the PAS domain of FixL (NCBI accession no. CAA40143). The α-helix and β-sheet of FixL (dark shading) are based on previous reports (17, 23), and the secondary structure of RbdA was predicted using the program PredictProtein (http://cubic.bioc.columbia.edu/). The identical, highly similar, and similar residues are indicated by the symbols *, :, and ., respectively. The arrows indicate the key amino acid residues of FixL involved in heme and oxygen binding (30). (B) Quantification of biofilm formation of strain PAO1 and its derivatives under aerobic conditions. The ΔrbdA deletion mutant was complemented by in trans expression of the wild type rbdA or its variants. The inset shows the Western blot data for RbdA and its variants expressed in the ΔrbdA mutant performed with protein samples prepared from bacterial cultures when the OD₆₀₀ reached about 1.5.

To determine the role of the PAS domain of RbdA, we generated a range of RbdA variants. In addition to the variant with the PAS coding sequences being deleted in frame (ΔPAS mutant, amino acids 274 to 354), we also generated three single-point variants (317A, 319A, and 339A), a double-point variant (317A319A), and a triple-point variant (317A319A339A), in which the three key residues (I317, G319, and L339) were replaced with alanine, respectively. In contrast to the wild-type RbdA, which reduced the biofilm mass by more than 10-fold when expressed in the ΔrbdA mutant, substitution of any one of the three key residues attenuated the ability of RbdA to suppress the mutant hyperbiofilm formation (Fig. 5B). In particular, deletion of the PAS domain or double or triple substitution completely abrogated the RbdA activity (Fig. 5B). Western blotting showed that substitution of the residues had no effect on protein expression (Fig. 5B, inset).

To further validate the role of the PAS domain in regulation of RbdA activity under different conditions, we compared the growth and biofilm phenotypes of P. aeruginosa wild-type strain PAO1 and its derivatives under aerobic and anaerobic conditions. The results showed that the wild type and its derivatives grew at a similar rate under the same conditions but proliferated more slowly under the anaerobic conditions than aerobic conditions during the first 20 h of growth (Fig. 6A). After 20 h, however, it was the aerobic strains that showed a slower growth yield than the anaerobic strains (Fig. 6A). In contrast to the growth phenotypes, the ΔrbdA deletion mutant produced substantially more biofilm than wild-type PAO1, especially under anaerobic conditions (Fig. 6B). Our results support the previous report that anaerobic conditions favor biofilm formation (63). The hyperbiofim phenotype of the ΔrbdA mutant was rescued by in trans expression of the wild-type rbdA. Particularly noteworthy is that the anaerobically grown ΔrbdA(rbdA) mutant showed hardly any biofilm formation, whereas the mutant complemented with the rbdA variant 317A319A failed to rescue the hyperbiofilm phenotype (Fig. 6B). Western blot analysis showed that under anaerobic conditions, rbdA and its variant were expressed at comparable levels (Fig. 6A, inset). Taken together, the above findings indicate that the PAS domain is essential for the functionality of RbdA and its activity is influenced by the oxygen level in bacterial culture.

FIG. 6.

Bacterial growth and biofilm development of strain PAO1 and its derivatives under aerobic and anaerobic conditions. (A) Time course analysis of bacterial growth. For Western blot analysis (inset), total protein samples were prepared from anaerobically grown bacterial cultures when the OD₆₀₀ reached about 1.5. (B) Time course analysis of biofilm formation.

RbdA positively regulates swimming, swarming motility, and rhamnolipid production.

Bacterial motility is known to affect biofilm formation in P. aeruginosa. Swarming motility decreases biofilm formation (7), whereas swimming motility increases initial bacterial attachment to surfaces during biofilm development (40). To examine the relationship between enhanced biofilm formation and the motility of the rbdA mutant, we compared its swimming, swarming, and twitching motilities with its wild-type strain PAO1. The results showed that deletion of rbdA reduced the swarm zone by about 60% (Fig. 7A) and decreased the swimming motility by 30% (Fig. 7B). We did not observe a significant difference in twitching motility between the ΔrbdA mutant and PAO1. Given that swarming motility is positively influenced by production of the biosurfactant rhamnolipids in P. aeruginosa (32), we then determined the biosurfactant level in the bacterial culture supernatants. The data showed that the ΔrbdA mutant had an approximate 30% decrease in rhamnolipid production compared with the wild-type PAO1 (Fig. 7C).

FIG. 7.

RbdA regulates bacterial motility, EPS production, and pellicle formation. (A) Swarming motility; (B) swimming motility; (C) rhamnolipid production; (D) EPS production; (E) pellicle formation in static culture (top right) and after crystal violet staining (bottom right). Three plates were used for each swarming and swimming test, and data shown are the averages of three independent experiments and the standard deviations.

RbdA negatively regulates EPS production and transcriptional expression of the pel operon.

Two lines of evidence indicate that RbdA negatively regulates EPS production. The plate assay using Congo red adsorption, which is positively correlated with the presence of EPS in a number of bacterial species, including P. aeruginosa (21), showed that the rbdA mutant developed a deeper red color than its parental strain PAO1 (Fig. 7D). Quantitative analysis found that the rbdA mutant bound about 70% more Congo red than PAO1 (Fig. 7D). Pellicle formation is another indicator of EPS production in P. aeruginosa (21). The results showed that deletion of rbdA caused substantially more pellicle formation than the wild-type PAO1 (Fig. 7E).

P. aeruginosa PAO1 has two operons, pel and psl, that have been reported to contribute to EPS production (21). To determine whether these two operons are associated with the EPS overproduction phenotype of the rbdA mutant, we generated the pelD and pslF single- and double-deletion mutants in PAO1 and ΔrbdA backgrounds. The Congo red and pellicle formation assays showed that deletion of pelD or pslF alone did not cause an obvious change in Congo red binding compared with wild-type PAO1. However, deletion of the two genes separately in the background of the ΔrbdA mutant, especially pelD, reduced the Congo red-binding ability to the level of PAO1 (Fig. 8A). Similarly, the double-deletion ΔrbdA ΔpelD mutant showed substantially decreased pellicle formation (Fig. 8B).

FIG. 8.

RbdA modulates biofilm formation through its influence on the genes encoding EPS biosynthesis. (A) Congo red-binding analysis of EPS production by PAO1 and its derivatives. (B) The effects of pel and psl operon deletions on pellicle formation. (C) pelA′-lacZ fusion gene expression assay. Different bacterial strains containing the PpelA-lacZ construct were grown in LB broth at 37°C to an OD₆₀₀ of 1.5, and the cells were then collected and assayed for β-galactosidase activity. (D) Quantitative comparison of biofilm formation by wild-type strain PAO1 and corresponding mutants.

We then tested the effect of RbdA on the expression level of the pel operon by generation of a reporter construct, PpelA-lacZ, in which the lacZ gene was placed under the control of the pelA promoter. The assay results showed that expression of pelA in the rbdA mutant was higher than in parental strain PAO1 and the corresponding complemented strains (Fig. 8C). This increased expression of the pel operon might contribute to the hyperbiofilm phenotype of the ΔrbdA mutant. For verification, we compared biofilm formation in strain PAO1 and the relevant single- and double-deletion mutants. The results showed that deletion of pelD alone in the background of PAO1 did not cause much difference in biofilm formation, but deletion of pelD in the genetic background of the rbdA mutant reduced biofilm accumulation close to the level of PAO1 (Fig. 8D).

DISCUSSION

Biofilm dispersal is the last stage of biofilm development, in which a subpopulation of biofilm cells detach and swim away from mature biofilms and revert to a planktonic lifestyle (57). The event can be triggered by various environmental cues. One of them is the oxygen depletion and the by-products of anaerobic metabolism, such as nitric oxide (NO) (2, 3). Production of rhamnolipids by P. aeruginosa during the stationary phase can also lead to biofilm dispersal (17, 42). The signaling and regulatory mechanisms underlying these events are poorly understood, but recently the two-component regulatory system BqsS/BqsR was shown to be involved in the modulation of biofilm dispersal (19). In our study, we showed that RbdA, a fusion protein containing the PAS-PAC-GGDEF-EAL multidomain, plays a key role in the regulation of biofilm dispersal by functioning as a phosphodiesterase that degrades c-di-GMP. We also presented evidence that the functionality of RbdA is tightly modulated by its PAS domain.

Our data show that RbdA is a c-di-GMP phosphodiesterase under both the in vitro and in vivo conditions used in this study. Our findings are further consolidated by sequence comparison with the c-di-GMP phosphodiesterase YkuI of B. subtilis, which has been characterized using crystal structure analysis (39). Sequence alignment of RbdA with YkuI shows that, besides the signature motif EXL, where X can be either A, L, or V (39, 47, 56), RbdA also shares other important residues, including the substrate-binding residues Gln565 and Arg585, the putative general base Glu762, the active site Lys726, and four residues which coordinate the divalent cation (39) (see Fig. S1B in the supplemental material).

Although RbdA contains a highly conserved GGDEF domain, the purified RbdA did not show detectable diguanylate cyclase activity. Instead, RbdA reacted with GTP to produce pGpG (see Fig. S3 in the supplemental material). We considered the possibility that the GGDEF domain of RbdA reacts with GTP to form c-di-GMP, which is then digested by the EAL domain to produce pGpG. However, two lines of evidence did not support this hypothesis. First, the time course analysis did not reveal the presence of the putative intermediate c-di-GMP in the reaction mixture (see Fig. S3). Second, the RbdA variant containing a functional GGDEF domain and an inactivated EAL domain did not produce c-di-GMP (see Fig. S4 in the supplemental material). Although it has been established in this study that both the GGDEF and EAL domains are required for production of pGpG, using GTP as substrate, the detailed catalytic mechanism and biological significance remain to be further investigated. The case of RbdA is somewhat similar to the GGDEF-EAL fusion protein CC3396 from Caulobacter crescentus, which contains an altered active site motif GEDEF and has no detectable diguanylate cyclase activity in vitro but does possess phosphodiesterase activity (12). The altered GGDEF motif is capable of binding GTP, which increases the phosphodiesterase activity of the protein by about 40-fold. A mutant version of CC3396 in which the GEDEF motif has been mutated to GQNEF showed only a 3-fold increase of phosphodiesterase activity in response to GTP (12). We found that the phosphodiesterase activity of RbdA was also stimulated by GTP, and the GGDEF motif was essential for the GTP-inducible enzyme activity. The findings suggest that the GGDEF motif of RbdA plays an analogous role to the GEDEF sequence of CC3396.

Bacteria commonly rely on PAS domains to monitor changes in oxygen, redox potential, and light (59). Among these environmental cues, hypoxia is a generic stress that bacterial pathogens may encounter in the process of infection (25). One of the well-characterized PAS domain proteins is FixL from the nitrogen-fixing rhizobia. In FixL, the N-terminal PAS domain controls the activity of the C-terminal histidine kinase. When the heme of the PAS domain is not bound with the ligand O₂, the kinase is active and phosphorylates its cognate response regulator, FixJ, which subsequently promotes the expression of a range of genes required for nitrogen fixation (23). Given the structural and sequence similarities between the corresponding PAS domains of RbdA and FixL (Fig. 5A), we speculated that the PAS domain of RbdA might play a similar role as its counterpart FixL in perceiving hypoxia stress. By analogy, it is likely that at the early stage of bacterial growth, the c-di-GMP degradation activity of RbdA is suppressed by the O₂-bound PAS domain, and the suppression is relieved along with bacterial proliferation due to oxygen depletion. This hypothesis is consistent with several lines of evidence presented in this study. First, the effect of mutant RbdA on biofilm phenotype became obvious only when bacterial cells entered the late exponential growth phase. Second, deletion of the PAS domain or replacement of the residues associated with ligand binding abrogated the biological activity of RbdA. Third, the wild-type rbdA-complemented ΔrbdA(rbdA) strain showed hardly any biofilm formation under anaerobic conditions, whereas the same strain growing under aerobic conditions still produced basal levels of biofilm. This is in contrast to the ΔrbdA mutant expressing the RbdA variant 317A319A, which formed similar hyperbiofilms as the parental ΔrbdA strain under either aerobic or anaerobic growth conditions. However, direct evidence for how RbdA interacts with the putative ligand O₂ and induces subsequent changes in enzyme activity is needed to establish the role of its PAS domain as an oxygen sensor. At this stage, the possibility of another factor(s) being involved in modulation of RbdA activity cannot be excluded. Further biochemical and crystal structure analyses will be critical to understanding the subtle mechanisms of regulation of this multidomain protein.

Several molecules, including EPS (37, 65), flagella (40), type IV pili (40), and Cup fimbria (60), are known to be involved in biofilm formation, whereas rhamnolipids and alginate lyase are associated with biofilm dispersal in P. aeruginosa (4, 5, 17). Among these molecules, our results showed that RbdA upregulates swarming and swimming motility and rhamnolipid production and downregulates EPS production. Cumulatively, the results from this study showed that upon perceiving environmental stress through its PAS domain, the activated RbdA degrades c-di-GMP and promotes biofilm dispersal through a two-pronged effect on biofilm development, i.e., downregulating biofilm formation and upregulating the production of the factors associated with biofilm dispersal. The molecular and regulatory mechanisms of several c-di-GMP metabolic enzymes of P. aeruginosa have been reported over the last few years (11, 28, 30, 38). Among them, FimX and BifA have been shown to function as phosphodiesterases (28, 30). It is interesting that these phosphodiesterases regulate different sets of genes and may play different roles in P. aeruginosa physiology. Similar to the mutant of RbdA, null mutation of BifA in P. aeruginosa results in a hyperbiofilm phenotype (33). BifA was also shown to regulate biofilm formation by modulating the expression of the pel operon (33). Unlike BifA, RbdA contains a PAS domain, and our results showed that the phosphodiesterase activity of RbdA is tightly controlled by its PAS domain. Further characterization of the localization of these regulatory enzymes, the corresponding c-di-GMP receptors, the mechanisms of domain interaction, and the downstream regulators will add to the current understanding of the diverse roles these c-di-GMP metabolic enzymes play in bacterial genetics and physiology.