Correspondence: Tim Tolker-Nielsen, BioCentrum-DTU, Building 301, Technical University of Denmark, DK-2800 Lyngby, Denmark, UK. Tel.: +45 45 25 27 93; fax: +45 45 88 73 28; e-mail: email@example.com
Microbial biofilm formation often causes problems in medical and industrial settings, and knowledge about the factors that are involved in biofilm development and dispersion is useful for creating strategies to control the processes. In this report, we present evidence that proteins with GGDEF and EAL domains are involved in the regulation of biofilm formation and biofilm dispersion in Pseudomonas putida. Overexpression in P. putida of the Escherichia coli YedQ protein, which contains a GGDEF domain, resulted in increased biofilm formation. Overexpression in P. putida of the E. coli YhjH protein, which contains an EAL domain, strongly inhibited biofilm formation. Induction of YhjH expression in P. putida cells situated in established biofilms led to rapid dispersion of the biofilms. These results support the emerging theme that GGDEF-domain and EAL-domain proteins are involved in regulating the transition of bacteria between a roaming lifestyle and a sessile biofilm lifestyle.
We have previously described the dynamics of biofilm formation by Pseudomonas putida in a flow-chamber system (Tolker-Nielsen et al., 2000). In a subsequent study we characterized a dispersal process that occurs in P. putida biofilms in response to starvation (Gjermansen et al., 2005). In our previous study, we found that knockout of a GGDEF domain protein-encoding gene (PP0165) in P. putida resulted in a mutant that was unable to form biofilm, and that overproduction of a GGDEF domain protein in P. putida wild-type cells prevented starvation-induced biofilm dispersal (Gjermansen et al., 2005). These results might indicate that GGDEF domain activity in P. putida induces the formation of a biofilm-supporting matrix, and that EAL domain activity in P. putida induces biofilm dispersal. In the present report, we substantiate that elevated GGDEF domain activity in P. putida induces the synthesis of biofilm matrix material and the formation of biofilm, and we demonstrate that elevated EAL domain activity in P. putida prevents biofilm formation or leads to dispersal of established biofilm.
Materials and methods
Strains and media
Pseudomonas putida strains and plasmids used in this study are listed in Table 1. Pseudomonas putida strains were cultivated at 30°C, and E. coli strains were cultivated at 37°C in standard Luria–Bertani (LB) medium for overnight batch cultivation. AB minimal medium supplemented with citrate (1 mM) was used for cultivation of biofilms in flow chambers. Tetracyclin was added where appropriate for plasmid maintenance at a concentration of 10 μg mL−1. Colony morphology was assayed on LB plates (without NaCl) supplemented with 25 μg Congo Red mL−1 (Sigma).
E. coliyedQ (G178A, E179A) gene cloned in pRK404A TetR
E. coliyhjH (E136A) gene cloned in pBBR1-MCS3 TetR
Construction of plasmid pYhjH
The yhjH gene of E. coli MG1655 was amplified by PCR using the primers YhjH_up (5′-CTCTCAAGCTTATAGCGCCAGACC) and YhjH_down (5′-ATATAGCATGCTAAGGCAGGTTATCCA). The PCR product was cloned behind the artificial lac promoter PA1/04/03 replacing the gfp gene in the vector pJBA27 by restriction with HindIII and SphI. The expression cassette was excised from the resulting plasmid by restriction with NotI, and was inserted into the partially NotI digested broad host range vector pBBR1-MCS3. The resulting plasmid, pYhjH, was transferred to P. putida OUS82 Tn7-lacIq by conjugation.
Site-directed mutagenesis in pYedQ and pYhjH
Site directed mutagenesis was performed in the plasmids pYedQ and pYhjH according to the megaprimer strategy (Tyagi et al., 2004) to substitute critical residues in the GGDEF domain of YedQ and the EAL domain of YhjH. The residues chosen were previously shown to be essential for the activity of the proteins (Simm et al., 2004; Da Re & Ghigo, 2006). PCR was performed with pYedQ as a template using the primers YedQ.mut1.f (5′-GCGGGTCGGTGCTGCGGAGTTT) and YedQ.mut1.r (5′-GGTAACGCCAGGGTTTTCCCAG). The generated PCR product contained two point mutations corresponding to G178A and E179A amino acid changes in the active GGDEF site of the YedQ protein. The PCR product was used as a megaprimer for a second PCR round, together with the primer YedQ.mut2.f (5′-GAAACAGCTATGACCATGATTACGCCA). The resulting PCR product was digested with BamHI and HindIII and ligated into BamHI/HindIII-digested plasmid pRK404A. Clones were selected on 15 μg mL−1 tetracyclin and sequenced for fidelity and to verify that they contained the expected point mutations. One clone was selected and the plasmid was designated pYedQmut.
The same strategy was used for site-directed mutagenesis in the EAL domain of pYhjH. The primers used were: YhjH.mut1.r: (5′-AGACGGATATGCGCCACCAGTTCG), Yhjh.mut1.f (5′-GAAGGGCGATCGGTGCGG) and YhjH.mut2.r (5′-GCATCAGCACCTTGTCGCCTTG). The resulting PCR product was ligated with the NcoI/PvuI-digested vector fragment of pYhjH to construct the plasmid designated pYhjHmut. The procedure resulted in an E136A substitution of the highly conserved residue of the EAL domain of pYhjH, which was verified by sequencing.
Construction of P. putida OUS82 Tn7-lacIq
The gene encoding the E. coli Lac repressor was inserted into the P. putida genome using a mini-Tn7 system (Koch et al., 2001). A four-parental mating between P. putida OUS82 (recipient), E. coli pRK600 (conjugation helper), E. coli pTn7- lacIq (donor) and E. coli pUX-BF13 (transposition helper) was carried out, and transconjugants were selected on Pseudomonas isolation agar (PIA) supplemented with 50 μg gentamycin mL−1. Site-specific insertion of the Tn7 transposon downstream of the glmS gene was verified by PCR.
Fluorescent tagging of P. putida OUS82 Tn7-lacIq
A gene encoding GFP was inserted into the P. putida genome using of a mini-Tn5 system (de Lorenzo et al., 1990). Tagging was performed by two-parental mating between E. coli S17-1 carrying pSM1695 and P. putida OUS82 Tn7-lacIq, and transconjugants were selected on LB plates supplemented with 10 μg tetracycline mL−1 and 50 μg kanamycin mL−1. Plasmid pSM1695 carries a mini-Tn5 transposon containing the gfpmut3b* allele under control of the E. coli ribosomal promoter PrrnBP1.
Cultivation of flow-chamber biofilms
Biofilms were cultivated at 22°C in flow chambers that were assembled and prepared as described previously (Wolfaardt et al., 1994). Flow chambers were inoculated with P. putida overnight cultures diluted 1000-fold in AB medium supplemented with citrate (1 mM) as described by Gjermansen et al. (2005). Where appropriate for plasmid maintenance, 10 μg tetracycline mL−1 was added to the growth medium. 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) was added for induction of the lac promoter in P. putida OUS82 Tn7-lacIq.
All microscopic observations and image acquisitions were performed on an LSM510 confocal laser scanning microscope (CLSM) (Carl Zeiss, Jena, Germany) equipped with detectors and filter sets for monitoring Gfp fluorescence. Images were obtained with a 63 × 1.4 objective or a 40 × 1.3 objective. Simulated three-dimensional images, shadow projections and sections were generated using the imaris software package (Bitplane AG, Zürich, Switzerland).
Crystal violet biofilm assay
Biofilm formation in static microtiter dishes was quantified by crystal violet staining as described by O'Toole & Kolter (1998). Briefly, 100 μL of LB supplemented with appropriate antibiotics was inoculated with O/N culture to an OD600 nm of 0.01 and incubated at 30°C. The wells were aspirated and washed with 120 μL 0.9% saline, followed by 15-min staining with 120 μL 0.1% crystal violet(Sigma) in 0.9% saline. After staining, the wells were washed twice with 150 μL 0.9% saline. The crystal violet was solubilized with 96% ethanol for 30 min before measuring the absorbance at 590 nm.
Pellicle formation assay
Pellicle formation was assayed by diluting overnight cultures 1000-fold in 50 mL ABT minimal medium supplemented with 0.2% glucose as a carbon source and supplemented with 10 μg tetracycline mL−1 for maintenance of the plasmids. The bottles were left under static conditions at room temperature for 4 days.
Swimming motility was tested in LB medium solidified with 0.3% agar and incubated at 22°C O/N. Induction of yhjH expression was achieved by supplementing the medium with 1 mM IPTG.
Characterization of P. putida in which a GGDEF domain protein is overexpressed
Expression of the E. coli YhcK protein from plasmid pYhcK has previously been shown to induce production of cellulose in Rhizobium leguminosarum and A. tumefaciens (Ausmees et al., 2001). This regulation was attributed to the GGDEF domain of the protein, and it was inferred that c-di-GMP was involved in the activation of cellulose synthesis (Ausmees et al., 2001). The diguanylate cyclase activity of the YhcK protein has subsequently been demonstrated, and the pYhcK plasmid has been used for overexpression of diguanylate cyclase activity in other bacteria (e.g. Simm et al., 2004; Gjermansen et al., 2005). However, in the original paper by Ausmees et al. (2001), the identity of the YhcK protein was not clear, and several different proteins have subsequently been identified as YhcK (e.g. Simm et al., 2005; Mendez-Ortiz et al., 2006). We have, by sequencing of the pYhcK plasmid, found that the GGDEF domain protein-encoding gene on the plasmid is identical to yedQ (b1956) in the E. coli MG1655 genome. In concurrence with the above-mentioned activity of the YhcK protein, the YedQ protein has recently been demonstrated to activate cellulose production in E. coli (Da Re & Ghigo, 2006). To avoid further confusion, we use the designations yedQ/YedQ/pYedQ instead of yhcK/YhcK/pYhcK for the gene/protein/plasmid in the present paper.
Because the YedQ protein has been shown to induce the synthesis of putative biofilm matrix material in various non-native hosts (e.g. Ausmees et al., 2001; Simm et al., 2004, 2005), we chose to use the pYedQ plasmid for our investigations of the effects of GGDEF-domain protein overproduction in P. putida. In order to investigate whether the observed phenotypes could be attributed to GGDEF-domain activity, we introduced point mutations in two critical positions in the GGDEF domain of the yedQ gene (as detailed in Materials and methods), and used the resulting pYedQmut plasmid as a negative control. When P. putida was grown on LB agar plates supplemented with the dye Congo Red, the pYedQ-containing strain formed wrinkled red colonies while the pYedQmut-containing strain and vector control strain formed smooth lightly colored colonies (Fig. 1a). When the YedQ-overproducing strain was grown in static liquid cultures, it formed thick robust pellicles unlike the pYedQmut-containing strain and the vector control strain (Fig. 1b). When P. putida harboring the pYedQ plasmid was grown in microtiter trays, overproduction of YedQ led to extensively increased biofilm formation compared with the pYedQmut-containing strain and the vector control strain in both rich and minimal media (Fig. 1c). In agreement with our previous observations (Gjermansen et al., 2005), in flow chambers perfused with citrate medium, P. putida harboring pYedQ formed dense biofilms with very large microcolonies, while the pYedQmut-containing strain and the vector control strain formed biofilms with loose protruding structures (Fig. 1d–f).
Characterization of P. putida in which an EAL domain protein is overexpressed
Based on sequence analysis of all EAL domain proteins of E. coli MG1655, we selected the gene yhjH to serve as our model for overexpression of phosphodiesterase activity in P. putida. This was done because the YhjH protein contains an EAL domain that complies with the consensus of known critical conserved residues of active EAL domains, and because the YhjH protein does not contain any other recognizable domains (Simm et al., 2004) that could interfere with our analysis. Recently, YhjH was shown to have phosphodiesterase activity towards c-di-GMP in Salmonella enterica and Shewanella oneidensis (Simm et al., 2004; Thormann et al., 2006). We fused the PCR-amplified yhjH gene to the synthetic lac promoter PA1/04/03 in the vector pJBA27, and the fusion was subsequently cloned in the broad host range vector pBBR1-MCS3, giving rise to the plasmid pYhjH, which was used for overproduction of YhjH in P. putida. As P. putida does not encode a lacI repressor, we anticipated that expression of the yhjH gene from pYhjH in P. putida would be constitutive. In order to be able to control expression of the yhjH gene in P. putida, we inserted the E. coli lacIq gene into the chromosome of the P. putida OUS82 strain using a mini-Tn7 system. In order to investigate whether the observed phenotypes could be attributed to EAL-domain activity, we introduced a point mutation in a critical position in the EAL domain of the yhjH gene (as detailed in Materials and methods), and used the resulting pYhjHmut plasmid as a negative control.
Biofilm formation in microtiter trays of the OUS82 Tn7-lacIq pYhjH strain was significantly reduced when the yhjH gene was induced with IPTG in comparison to when it was uninduced, and in comparison to the OUS82 Tn7-lacIq pYhjHmut and OUS82 Tn7-lacIq pBBR1-MCS3 control strains under inducing or noninducing conditions (Fig. 2a). In order to confirm that the observed effect was not due to growth defects in the strain expressing the active YhjH protein, the growth rates of the OUS82 Tn7-lacIq pYhjH, OUS82 Tn7-lacIq pYhjHmut and OUS82 Tn7-lacIq pBBR1-MCS3 strains with and without IPTG induction were measured and found to be virtually identical (data not shown).
In order to investigate biofilm formation of the OUS82 Tn7-lacIq pYhjH, OUS82 Tn7-lacIq pYhjHmut and OUS82 Tn7-lacIq pBBR1-MCS3 strains in flow chambers using CLSM, we inserted a gfp gene into the strains using a mini-Tn5 transposon system. Because the Tn5 transposon inserts into random chromosomal sites, we performed all flow-chamber experiments with three different transposon-containing strains to rule out any effects of the location of the transposon insertion. We initially investigated the effects of induction of yhjH in the initial phase of biofilm formation. Flow chambers were inoculated with bacteria and irrigated with growth medium either supplemented with IPTG for induction of yhjH or without IPTG as controls. Induction of YhjH markedly decreased biofilm formation of the OUS82 Tn7-lacIq pYhjH strain as only a few surface-attached cells were present after 2 days in the flow chambers with IPTG induction compared to the flow chambers without IPTG induction (Fig. 2b). The control strains OUS82 Tn7-lacIq pYhjHmut (Fig. 2c) and OUS82 Tn7-lacIq pBBR1-MCS3 (data not shown) formed biofilms similar to the uninduced P. putida OUS82 Tn7-lacIq pYhjH strains in the flow-chambers with or without IPTG. Subsequently, we investigated the possible effects of induction of the EAL-domain protein in P. putida bacteria in established biofilms. Biofilms of P. putida OUS82 Tn7-lacIq pYhjH and the control strains OUS82 Tn7-lacIq pYhjHmut and OUS82 Tn7-lacIq pBBR1-MCS3 were grown in flow chambers for 3 days without IPTG, and on day 3 the medium was switched to an identical medium supplemented with IPTG. After the switch of medium, CLSM images of the biofilms were acquired at regular intervals. Three hours after the onset of IPTG induction most of the biomass had disappeared from biofilms formed by the OUS82 Tn7-lacIq pYhjH strain, whereas in biofilms formed by the OUS82 Tn7-lacIq pYhjHmut and OUS82 Tn7-lacIq pBBR1-MCS3 control strains, no dispersion was observed upon IPTG induction (Fig. 3). These observations suggest that YhjH overexpression decreases the adhesiveness of the bacteria in the initial attachment phase, and interferes with cell-to-cell connections in established biofilms, resulting in the dispersion of the biofilms.
Because overexpression of EAL domain proteins has also been associated with changes in cellular motility, we tested the swimming motility of the OUS82 Tn7-lacIq pYhjH strain under inducing and noninducing conditions, and consistently found an c. 30% increase in the diameter of swimming zones in a plate assay when YhjH expression was induced (Fig. 2d,e).
In the present report, we present evidence that overexpression of a GGDEF domain protein in P. putida induces the synthesis of biofilm matrix material and the formation of biofilm, and that overexpression of an EAL domain protein in P. putida prevents biofilm formation and induces biofilm dispersal. Control strains that produced proteins with point mutations in the GGDEF or EAL domains did not show these phenotypes.
Overexpression in P. putida of the GGDEF domain protein YedQ led to increased biofilm formation in both the static microtiter dish biofilm assay and in the hydrodynamic flow-chamber system. In addition, overexpression of YedQ in P. putida led to the formation of pellicles in static cultures and wrinkled red colonies on Congo Red agar plates. Evidence supporting a role of GGDEF domain proteins, including YedQ, in increasing the intracellular levels of c-di-GMP through their diguanylate cyclase activity is substantial (Chan et al., 2004; Paul et al., 2004; Simm et al., 2004, 2005; Hickman et al., 2005; Ryjenkov et al., 2005). Recent work in Salmonella enterica, V. cholera and P. aeruginosa showed that GGDEF domain protein-mediated increases in the intracellular levels of c-di-GMP led to increased production of specific surface proteins and exopolysaccharides, which resulted in increased cellular adhesiveness and biofilm formation (Simm et al., 2004; Tischler & Camilli, 2004; Hickman et al., 2005). Expression of specific surface proteins and/or exopolysaccharides has also been linked to changes in colony morphology resulting in wrinkly colonies that, in many cases, bound Congo Red (D'Argenio et al., 2002; Friedman & Kolter, 2004; Romling, 2005). In P. fluorescens, a close relative of P. putida, the production of an acetylated form of cellulose was shown to correlate with pellicle formation and wrinkly colony morphology, and the production of this exopolymer was shown to be regulated by the GGDEF domain protein WspR (Spiers et al., 2003). Later work has suggested that in addition to cellulose, an unidentified proteinacious component is also necessary for these phenotypes (Spiers & Rainey, 2005). On this basis, we suggest that the phenotypes of the P. putida YedQ-overproducing strain presented in the present paper are caused by increased production of Congo Red-binding EPS and/or surface proteins mediated by an increase in the intracellular c-di-GMP concentration. The targets affected by overproduction of YedQ in P. putida are presently unknown and subject to study in our laboratory.
Overexpression in P. putida of the EAL domain protein YhjH resulted in decreased initial adhesion and decreased biofilm formation in both the microtiter dish biofilm assay and the flow-chamber system. In addition, induction of YhjH in P. putida resulted in the dispersal of established biofilms within a relatively short time span. Evidence supporting a role of EAL domain proteins, including YhjH, in decreasing the intracellular level of c-di-GMP through their phosphodiesterase activity is substantial (Simm et al., 2004; Tischler & Camilli, 2004; Bobrov et al., 2005; Christen et al., 2005; Schmidt et al., 2005; Tamayo et al., 2005; Thormann et al., 2006). EAL domain protein-mediated decreases in the intracellular levels of c-di-GMP have recently been correlated with downregulation of cellular adhesiveness and biofilm formation in P. aeruginosa, Y. pestis and Salmonella enterica (Simm et al., 2004; Bobrov et al., 2005; Hickman et al., 2005). The present study, and a very recent similar study in Shewanella oneidensis (Thormann et al., 2006), suggest that in addition to downregulation of adhesiveness, a decrease in c-di-GMP levels may act as a signal for the initiation of biofilm dispersion.
The dispersion process observed in P. putida biofilms in response to increasing cellular YhjH levels could occur as a simple consequence of down-regulation of exopolysaccharide production (e.g. if biofilm-forming cells are anchored to exopolysaccharide via the polysaccharide synthase complex), but could also be caused by the activation of specific dispersion factors. In a previous study, we characterized a dispersion process occurring in P. putida biofilms in response to carbon starvation, and showed that it could be prevented by overexpression of YedQ in the P. putida cells (Gjermansen et al., 2005). In the present study it was demonstrated that overexpression of YhjH in P. putida is sufficient to induce the biofilm dispersion process without the need for external environmental stimuli such as carbon starvation. We have previously shown that carbon starvation-induced dispersion of P. putida biofilms is associated with extensive cellular motility (Gjermansen et al., 2005), and in the present study we presented evidence that overexpression of YhjH in P. putida leads to upregulation of cellular motility. The evidence indicating that overexpression of EAL or GGDEF domain proteins can result in overriding or uncoupling of the naturally occurring dispersion process strongly suggests that c-di-GMP is a central signal in this process.
The existence of multiple GGDEF and EAL domain proteins (36 and 21, respectively, among these 17 containing both domains) in P. putida (Galperin, 2005) suggests the potential for integration of multiple signals to modulate the intracellular c-di-GMP concentration. It has been suggested that the activities of GGDEF and EAL domain proteins might be tightly controlled in the cellular space and that crosstalk between the different systems is avoided by compartmentalization of the signals (Paul et al., 2004). Overexpression of the GGDEF domain protein YedQ or the EAL domain protein YhjH presumably results in very high levels of diguanylate cyclase or phosphodiesterase activity, and hence highly increased or decreased intracellular c-di-GMP levels. This may lead to overriding of the naturally occurring regulation of c-di-GMP–dependent proteins. However, under the conditions used in the present study the increased or decreased intracellular c-di-GMP levels evidently resulted in upregulation or downregulation of cellular adhesiveness as the dominant phenotype.
The emerging theme that GGDEF/EAL domain proteins and c-di-GMP are key regulatory elements involved in the transitions of bacteria between the planktonic and biofilm mode of growth suggests that modulation of these elements might represent novel targets for control of biofilm development.
This work was supported by grants from the Lundbeck Foundation and from the Danish Research Agency (2052-03-0013). We thank Rikke Tange and Anna-Kathrine Fevre for construction of the pYhjH plasmid.